Identifying impacts of interventions aimed at promoting walking and cycling; directions for increasing non-motorized transportation in U.S. cities

Material Information

Identifying impacts of interventions aimed at promoting walking and cycling; directions for increasing non-motorized transportation in U.S. cities
Piatkowski, Daniel Philip ( author )
Place of Publication:
Denver, CO
University of Colorado Denver
Publication Date:
Physical Description:
1 electronic file (216 pages). : ;


Subjects / Keywords:
Cycling -- Government policy -- United States ( lcsh )
Walking -- Government policy -- United States ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


This dissertation aims to contribute to the field of transportation planning by answering the: what is the impact of interventions aimed at promoting bicycling and walking? Employing a multi-methods approach, this research identifies population-level impacts of interventions put in place during the Non-Motorized Transportation Pilot Program (NTPP), from 2006-2010. The NTPP was a federally sponsored program, unique in United States history for its scale and goals. Four pilot cities were awarded funds for promoting bicycling and walking, with the stated goal: to demonstrate the extent to which bicycling and walking can carry a significant part of the transportation load, and represent a major portion of the transportation solution. The impacts of the NTPP have far-reaching implications for research and practice. This dissertation is the most extensive evaluation of the NTPP to date and contributes to the field in the following ways: first, identifying new metrics for evaluating population-level impacts of bicycle and pedestrian interventions. Second, operationalizing measures of different types of interventions using a novel application of survey data and structural equation models. Finally, uses a multiple-methods approach to weigh intervention efficacy against ease of implementation, offering directions for practice.
Thesis (Ph.D.)--University of Colorado Denver. Design and planning
Includes bibliographic references.
System Details:
System requirements: Adobe Reader.
General Note:
College of Architecture and Planning
Statement of Responsibility:
by Daniel Philip Piatkowski.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
892523468 ( OCLC )


This item is only available as the following downloads:

Full Text


IDENTIFYING IMPACTS OF INTERVENTIONS AIMED AT PROMOTING WALKING AND CYCLING; DIRECTIONS FOR INCREASING NON MOTORIZED TRANSPORTATION IN U.S. CITIES by DANIEL PHILIP PIATKOWSKI B.A., English, Arizona State University 2001 Master's of Urban and Environmental Planning (MUEP) Arizona State University 2006 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirement s for the degree of Doctor of Philosophy Design and Planning 2013


ii This thesis for the Doctor of Philosophy degree by Daniel Philip Piatkowski has been approved for the Design and Planning Program by Jeremy Nemeth Chair Kevin J. Krizek, Advisor Thomas Clark Debbi Main Wesley Marshall Date: May 13 th 2013


iii Piatkowski Daniel Philip (Ph.D., Design and Planning ) Identifying Impacts of Interventions Aimed at Promoting Walking and Cycling; Directions for Increasing Non motorized Transportation in U S Cities Thesis directed by Professor Kevin J. Krizek ABSTRACT Bicycle and pedestrian transportation has been on the national agenda since the passage of the Intermodal Surface Transportation Equity Act of 1991, stemming from concerns about the harmful effects of automobile reliance. Since its passage, planners and policy makers have struggled to spur significant mode shifts from driving to bicycling and walking. This dissertation aims to contribute to t he field of transportation planning by answering the question: what is the impact of interventions aimed at promoting bicycling and walking? Employing a multiple methods approach, this research identifies population level impacts of interventions put in place during the Non motorized Transportation Pilot Program (NTPP), from 2006 2010. The NTPP was a federally sponsored program, unique in United States history for its scale and goals. Four pilot cities were awarded funds for promoting bicycling and walkin g, with the stated goal: "to demonstrate the extent to which bicycling and walking can carry a significant part of the transportation load, and represent a major portion of the transportation solution."


iv The impacts of the NTPP have far reaching implicati ons for research and practice. This dissertation is the most extensive evaluation of the NTPP to date and contributes to the field in the following ways: first, identifying new metrics for evaluating population level impacts of bicycle and pedestrian inter ventions. Second, operationalizing measures of different types of interventions using a novel application of survey data and structural equation models. Finally, uses a multiple methods approach to weigh intervention efficacy against ease of implementation offering directions for practice. The form and content of this abstract are approved. I recommend its publication. Approved: Kevin J. Krizek


v DEDICATION I wish to dedicate this work to my family and friends f or their inspiration, motivation, and support along the way Most importantly, to Melissa.


vi ACKNOWLEDGMENTS This work would not have been possible without the support of the inte rdisciplinary graduate program i n sustainable urban infrastructure at the University of Colorado Denver, which funded much of my studies through a National Science Foundation Integrative Graduate Research Education Traineeship. I would like to thank all those affiliated with the Non motoriz ed Transportation Pilot Program particularly the University of Minnesota Center for Transportation Studies The Rails to Trails Conservancy the US Department of Transportation's Volpe National Transportation Systems Center and the communities of Columbia, Missouri; Marin County, Cali fornia; Minneapol is, Minnesota; Sheboygan, Wiscons in; and Spokane, Washington. I would also like to thank Thomas G š tschi Loren Cobb, Sarah Schmiege and Billy Fields for their assistance with the myriad obstacles I faced in both quantitative and qualitativ e research and analysis I owe a great deal of gratitude to my advisor, Kevin Krizek for his outstanding mentorship throughout my doctoral studies I would also like to thank Wesley Marshall Debbi Main Jeremy Nemeth, and Tom Clark for their invaluable insights, advice, and service on my committee Finally, I would like to thank the Active Communities Transportation Research Group for fostering a supportive and rigorous academic environment for transportation research at the U niversity of C olorado Denve r.


vii TABLE OF CONTENTS CHAPTER I DISSERTATION OVERVIEW ................................ ................................ ...................... 1 Introduction ................................ ................................ ................................ ................................ 1 Questions and Approaches ................................ ................................ ................................ .......... 2 Data ................................ ................................ ................................ ................................ .............. 3 Objectives, Methods, and Organization ................................ ................................ ...................... 4 II UNDERSTANDING CONTEXT AND IDENTIFYING IMPACTS OF THE NON MOTORIZED TRANSPORTATION PILOT PROGRAM: A QUALITATIVE APPROACH ................................ ................................ ................................ ....................... 8 Introduction ................................ ................................ ................................ ................................ 8 Methods ................................ ................................ ................................ ................................ ..... 10 Study Respondents ................................ ................................ ................................ ................ 11 Data Collection and Analysis ................................ ................................ ................................ 11 Background, Context, and Goals ................................ ................................ ............................... 1 4 Overview of the NTPP Cities ................................ ................................ ................................ 14 Intervention Accounting ................................ ................................ ................................ ........ 16 The Moderating Effect of the NTPP ................................ ................................ .......................... 28 Behavior Change ................................ ................................ ................................ ................... 30 Attitude Change ................................ ................................ ................................ .................... 31 Conclusion ................................ ................................ ................................ ................................ 35 III ATTITUDES, NEIGHBORHOOD PERCEPTIONS, AND NON MOTORIZED TRANSPORTATION: EXAMINING CHANGES IN ATTITUDES AND PERCEPTIONS OF BICYCLING AND WALKING IN FIVE US CITIES ................... 39


viii Introduction ................................ ................................ ................................ ............................... 39 Literature Review ................................ ................................ ................................ ...................... 40 Changing Attitudes to Change Behaviors ................................ ................................ ............. 42 Study Background ................................ ................................ ................................ ..................... 44 Non motorized Transportation Interventions ................................ ................................ ........ 47 Lag Effects Between Interventions and Outcomes ................................ ............................... 48 Quantitative Analysis ................................ ................................ ................................ ................ 50 Data ................................ ................................ ................................ ................................ ............ 51 The NTPP Community Survey ................................ ................................ ............................. 52 Sample Population Characteristics ................................ ................................ ........................ 53 Methods and Findings ................................ ................................ ................................ ............... 55 Difference in Means and Regression ................................ ................................ .................... 56 Dimension Reduction ................................ ................................ ................................ ............ 59 Results ................................ ................................ ................................ ................................ ....... 63 Statistical Model 1 ................................ ................................ ................................ ................. 63 Statistical Model 2 ................................ ................................ ................................ ................. 66 Discussion ................................ ................................ ................................ ................................ .. 68 Conclusions and Future Directions ................................ ................................ ............................ 69 IV CARROTS, STICKS, AND NON MOTORIZED TRANSPORTATION: TESTING THE MEDIATING EFFECT OF DISCOURAGING DRIVING ON PROMOTING WALKING AND BICYCLING ................................ ................................ ....................... 72 Introduction ................................ ................................ ................................ ............................... 72 Literature Review ................................ ................................ ................................ ...................... 75 Non motorized Transportation Interventions Carrots ................................ ......................... 76 Interventions Aimed at Dis incentivizing Driving Sticks ................................ ................... 77


ix Mode Choice and Decision Theory ................................ ................................ ....................... 79 Intervention Theories ................................ ................................ ................................ ............ 82 Data ................................ ................................ ................................ ................................ ............ 85 Methods ................................ ................................ ................................ ................................ ..... 86 Factor Analysis ................................ ................................ ................................ ...................... 87 Mediation Analysis Overview ................................ ................................ ............................... 91 Testing Mediation with Structural Equation Models ................................ ............................ 97 Results ................................ ................................ ................................ ................................ ....... 99 Latent Factor Mediation ................................ ................................ ................................ ...... 103 Indirect Effects ................................ ................................ ................................ .................... 105 Discussion ................................ ................................ ................................ ................................ 106 Interpretation ................................ ................................ ................................ ....................... 107 Socio demographics ................................ ................................ ................................ ............ 108 City of Residence ................................ ................................ ................................ ................ 109 Regular NMT Use ................................ ................................ ................................ ............... 110 Conclusion ................................ ................................ ................................ ............................... 111 V CARROTS VERSUS STICKS: ASSESSING INTERVENTION EFFECTIVENESS AND IMPLEMENTATION CHALENGES FOR NON MOTORIZED TRANSPORTATION ................................ ................................ ................................ ..... 114 Introduction ................................ ................................ ................................ ............................. 114 Literature Review ................................ ................................ ................................ .................... 115 Carrot Interventions ................................ ................................ ................................ ............ 116 Stick Interventions ................................ ................................ ................................ .............. 119 Combined Approaches: Carrots and Sticks ................................ ................................ ......... 120 Relevant Theory ................................ ................................ ................................ ...................... 125


x Study 1: Quantitative Evidence for "Carrots or Sticks" vs. "Carrots and Stick s" to Encourage Non motorized Transportation ................................ ................................ ................................ 130 Background and Data ................................ ................................ ................................ .............. 130 Methods ................................ ................................ ................................ ................................ ... 131 Results ................................ ................................ ................................ ................................ ..... 132 Study 2: Assessing Implementation Challenges for Various Types of Interventions Aimed at Promoting Non motorized Transportation ................................ ................................ .............. 134 Background and Data ................................ ................................ ................................ .............. 134 Findings ................................ ................................ ................................ ................................ ... 135 Focus on Carrots ................................ ................................ ................................ ................. 135 "Providing Options" ................................ ................................ ................................ ............ 136 Carrots Informing Social Norms ................................ ................................ ......................... 138 Limited Examples of Sticks ................................ ................................ ................................ 144 Combining Carrots and Sticks ................................ ................................ ............................ 145 Discussion and Conclusion ................................ ................................ ................................ ...... 150 VI CONCLUSIONS AND CONTRIBUTIONS ................................ ........................... 156 Introduction ................................ ................................ ................................ ............................. 156 Expanding Definitions of Interventions Impacts ................................ ................................ ..... 158 Carrots versus Sticks in the United States ................................ ................................ ............... 159 Future Research ................................ ................................ ................................ ....................... 161 Behavioral Outcomes and Changes Over Time ................................ ................................ .. 162 Lag Effects ................................ ................................ ................................ .......................... 163 REFERENCES ................................ ................................ ................................ ............... 164 APPENDIX


xi A CHAPTER 2 SUPPLEMENTAL DOCUMENTATION ................................ .......... 184 B CHAPTER 3 SUPPLEMENTAL DOCUMENTATION ................................ .......... 191 C CHAPTER 4 SUPPLEMENTAL DOCUMENTATION ................................ .......... 195


xii LIST OF TABLES TABLE 1 NTPP Community Summary Statistics. ................................ ................................ ........ 15 2 Columbia, MO: Key NTPP Interventions. ................................ ................................ .... 19 3 Marin County, CA: Key NTPP Interventions. ................................ .............................. 21 4 Minneapolis, MN: Key NTPP Interventions. ................................ ............................... 23 5 Sheboygan County, WI: Key NTPP Interventions. ................................ ...................... 25 6 Descriptive Statistics by Phase of Data Collection. ................................ ...................... 54 7 Reference Trip: Attitude/Perceptions ................................ ................................ .......... 55 8 Neighborhood Perception Factor. ................................ ................................ ................. 60 9 Cycling Barriers Fac tor. ................................ ................................ .............................. 61 10 Attitude Variable Descriptives. ................................ ................................ ................... 62 11 Model 1: Neighborho od Perception Factor Regression. ................................ ............. 63 12 Model 2: Cycling Barrier Factor Regression. ................................ ............................. 66 13 Attitude/Perception Variables. ................................ ................................ .................... 88 14 Exploratory Factor Analysis: Results and Model Fit Indi ces. ................................ .... 89 15 Exploratory and Confirmatory Factor Analysis Results. ................................ ............ 90 1 6 Phase 2 Model Results Mediation SEM. ................................ ............................... 100 17 Gas Prices in NTPP Cities (1 Gallon in US Dollars). ................................ ............... 147 18 Intervention Accounting: Columbia, MO (Source: Krizek, et al., 2007). ................ 186 19 Intervention Accounting: Marin County, CA (Source: Krizek, et al., 2007). ........... 187 20 Intervention Accounting: Minneapolis, MN (Source: Krizek, et al., 2007). ............ 188 21 Intervention Accounting: Sheboygan County, WI (Source: Krizek, et al., 2007). ... 189 22 Intervention Accounting: Spokane, WA (Source: Krizek, et al., 2007). .................. 190 23 Coding Scheme for Household Survey Variables. ................................ ................... 191


xiii 24 Independent Samples T tests of Attitude Variables by NTPP City. ......................... 192 25 Backward Stepwise Linear Regression Attitudinal Var iable Models. ................... 194 26 Sobel's Test of Mediation: Significant Direct and Indirect Effects. ......................... 19 5 27 P hase 1 Model Results SEM Medi ation Models. ................................ .................. 196 28 Phase 2 Model Results Direct Path SEM ................................ .............................. 197


xiv LIST OF FIGURES FIGURE 1 The Moderating Effect of the NTPP. ................................ ................................ ............ 29 2 Conceptual Model. ................................ ................................ ................................ ........ 41 3 Model of Positive Attitudes. ................................ ................................ ......................... 45 4 Model of Negative Attitudes. ................................ ................................ ........................ 46 4 Phase x Income in Model 1. ................................ ................................ .......................... 65 5 Phase x Race in Model 2. ................................ ................................ .............................. 68 7 Theory of Routine Mode Choice Decisions (source: Schneider, 2013) ...................... 82 8 Mediation Conceptual Model. ................................ ................................ ...................... 92 9 Single Mediator Path Diagram (Source: Baron and Kenny, 1986). ............................. 92 10 Mediation Conceptual Mode l. ................................ ................................ .................... 94 11 Mediation SEM Diagram. ................................ ................................ ........................... 98 12 Mediation Effects City of R esidence. ................................ ................................ .... 102 13 Mediation Effects Socio demographics. ................................ ................................ 102 14 Mediation Effects Regular NMT Use. ................................ ................................ ... 103 15 The ABC Model (Source: Stern, 2000). ................................ ................................ ... 129 16 Mediation Structural Equation Model (Source, Chapter 4). ................................ ..... 131 17 Ideal Bicyclist Behavior in BLIP (Photo Credit: Ted Curtis). ................................ .. 140 18 Bicycle Advisory Lane (Ph oto Credit: Simon Blensky). ................................ .......... 143 19 Summary of Multiple Methods Findings. ................................ ................................ 151 20 Transtheoretical Model (Prochaska & Diclemente, 1983). ................................ ...... 152 21 Overview of Research Completed. ................................ ................................ ........... 157


xv GLOSSARY The dissertation uses a variety of words and phrases to refer to specific issues regarding transportation, land use, and intervention evaluation. It is helpful to provide initial definitions to these terms at the outset. Interventions refer to an y project, program, policy, or infrastructure meant to influence travel behavior. Non motorized Transportation (NMT) refers specifically to walking and bicycling for transportation purposes this includes trips such as shopping, commuting, socializing, et c. any travel in which some part of the purpose is to get from one place to another (as opposed to purely recreational walking and cycling). Travel behavior is a used to describe various trip types and purposes together. Taken together, NMT interventions are interventions specifically meant to impact transportation walking and cycling. The primary data sources for this dissertation come from evaluation efforts of the Non motorized Transportation Pilot Program (NTPP) The NTPP includes four treatment cit ies and one control city. All references to the NTPP cities specifically means all five cities, otherwise treatment or control cities are specified. The context within which interventions are put in place refers to both the social and physical environment. One goal of this research is to better understand how context influences the impact of interventions; that is, how do factors external to the individual and the intervention impact behavior. Automobile accessibility and mobility are discussed in contras t to NMT accessibility and mobility. High levels of automobile accessibility are associated with


xvi areas that include numerous destinations and activities easily reached by car, whereas high levels of NMT accessibility are associated with numerous destinatio ns easily accessed by NMT. Frequently these areas are mutually exclusive. Mobility refers to the relative ease of movement through a transportation system by mode. High auto mobility may refer to high speed limits and low congestion, whereas high NMT mobil ity frequently refers to the availability of rapid, direct routes via NMT specific infrastructure. Relative accessibility and mobility by mode auto, walk, or bicycle in the context of this dissertation, can in turn influence mode share the share of tota l travel completed in a city or region by various modes.


1 CHAPTER I DISSERTATION OVERVIEW Introduction There is a great deal of interest in the United States in encouraging Non motorized Transportation (NMT) Non motorized modes are seen as one solution to a variety of problems that plague US cities, which are disproportionately reliant on the automobile. Indeed, increased NMT may help to reduce congestion and greenhouse gas emissions, and may increase public health through increased physical activity. However, to realize these goals, NMT mode share must significantly increase in US cities. How can US cities best use limited strategies and resources to realize the greatest transportation impacts? In most of the US, planner s and advocates focus on modest infrastructure enhancements or social programs interventions aimed at encouraging walking and bicycling. Infrastructure such as shared use paths and bicycle lanes can create a more attractive, and possibly safer experien ce for NMT users. Social programs, such as bike/walk to work events may improve social norms and acceptance of non motorized commuting. The literature confirms that modest interventions typically have modest impacts on travel behavior. Alternately, interve ntions could serve to discourage driving, but such actions face severe opposition from a public accustomed to auto supportive policies and infrastructure. In this context of few options and limited funds, what can planners and policy makers do to reduce au to reliance?


2 To address this question, the dissertation examines a federally funded pilot program, unique in US history for it's scale and goals. In 2006, four US cities each received $25 million to invest in interventions aimed at promoting NMT. Once the primary barrier to intervention implementation funding was removed, each of these cities embarked on significant infrastructure programs and social policies to reduce driving and increase walking and bicycling. The dissertation closely examines the NT PP, focusing on the types of interventions put in place during the NTPP. The research goals are to identify the impacts of a one time influx of funding for NMT in different types of cities, and draw conclusions about how planners and policy makers can bes t utilize limited resources to reach city scale transportation goals. Questions and Approaches The over arching research question this dissertation addresses is: "what are the impacts of interventions aimed at promoting walking and bicycling?" In the pro cess of shedding light on this primary question, the assumptions regarding the purposes and impacts of interventions are also questioned and evaluated. The question is addressed in four parts, using multiple data sources and methods of inquiry: 1. Understandi ng context and identifying city scale impacts of NMT interventions. 2. Measuring impacts of NMT interventions in the general population. 3. Testing the interrelationship of different types of interventions, with lessons for intervention effectiveness. 4. Weighing i ntervention effectiveness against ease of implementation.


3 Data The dissertation employs primary and secondary data collected to evaluate the Non motorized Transportation Pilot Program (NTPP). The NTPP was a federal program funded under the Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA LU) in 2005. Four pilot cities were awarded funds for promoting bicycling and walking, with the stated goal: "to demonstrate the extent to which the bicycling an d walking can carry a significant part of the transportation load, and represent a major portion of the transportation solution." As the first of its kind in the US, the impacts of the NTPP have far reaching implications for both research and practice. A community survey of five NTPP cities comprises the secondary quantitative data source. The community surveys were conducted as an evaluation effort of the NTPP. Utilizing a quasi experimental research design and probability based sampling frame, with cont rol, surveys were administered in each of the four pilot communities (Columbia, Missouri; Marin County, California; Minneapolis, Minnesota; Sheboygan County, Wisconsin; as well as in Spokane, Washington, the control group) in 2006 (Phase 1), at the start o f the program, and then again in 2010 (Phase 2). Primary data was collected through key informant interviews in the NTPP communities. Each of the NTPP cities allocated oversight of funds to committees of local stakeholders and applicable government agenci es, typically overseen by a few government employees in public works and/or planning departments. The treatment city


4 also participated in coordinated meetings overseen by national transportation and advocacy organizations. Interviews were conducted in each city with members of this group of advocates, planners, and engineers with expert knowledge of the NTPP, and interventions put in place during the course of the NTPP. Objectives, Methods, and Organization This research is the most extensive evaluation of the NTPP to date and has the combined goals of providing a novel evaluation of the impacts of the NTPP itself, as well as using lessons and findings from the NTPP to inform future NMT intervention evaluatio n and implementation efforts. The dissertation is organized into six chapters. Chapter two provides background, context, and directions for the rest of the dissertation. Chapters three, four, and five each represent semi independent research efforts, addre ssing a distinct research question. An overview of chapters two through five is presented below: Chapter 2: What are the impacts of the NTPP: this is an exploratory analysis of impacts of the NTPP on the general population, the community context, and loc al institutions. This chapter uses qualitative data and methods to provide background, and context of the NTPP in each city. Additional attention was paid to exploring possible impacts of the NTPP that either were not addressed in community surveys, or co uld not be addressed using quantitative methods. This chapter illustrates the wide array of goals and challenges faced in each community, as well as possible reasons that behavior change was not


5 measurable in population surveys. Nonetheless, the NTPP had a significant impact on the quality and quantity of NMT infrastructure in the treatment communities, and may have implications for institutional capacity in both the treatment and control communities. Such capacity building may ease the implementation proce ss for future NMT interventions. Chapter 3: Can population level impacts of the Non motorized Transportation Pilot Program be quantified: are there measurable impacts of the NTPP on the general population of the NTPP cities? By closely examining the NTPP community surveys, administered in 2006 and 2010, this chapter aims to identify quantifiable changes in the survey population that are correlated with the NTPP. Linear and logistic regression confirms behavior change cannot be identified at the population level; however, combining principal components analysis with linear regression yields a significant change in attitudes and perceptions toward NMT from 2006 to 2010. Attitude changes may be indicative of preliminary impacts of the NTPP that precede behavi or changes. Chapter 4: Is there a relationship between carrots that encourage NMT and sticks that discourage driving: do carrots and sticks interrelate and lead to more effective interventions? There are a numerous assertions that sticks which discourage driving may be more effective at impacting NMT mode share than carrots that encourage NMT, and a great deal of anecdotal evidence from practitioners


6 that combining carrots and sticks may be most effective. Chapter four first identifies latent variables tha t represent carrots and sticks using exploratory and confirmatory factor analysis. Then, structural equation modeling provides empirical support for the most effective types of interventions. Chapter 5: What are the relative challenges of implementing dif ferent types of interventions, and what are the relative impacts of these interventions: intervention effectiveness depends on the ability to implement said interventions. The final research effort presented in the dissertation uses a multiple methods ap proach to two related but distinct research questions. A quantitative evaluation of the NTPP using structural equation models to assess the impact of carrots, sticks, or a combination of carrots and sticks on mode choice perceptions (Research question 1). A qualitative approach, using in depth, key informant interviews in the NTPP cities is presented to understand the challenges of implementing carrot or stick interventions (Research question 2). The combined findings shed light on what types of interventio ns are most effective at influencing NMT behavior, and how challenging such interventions are to implement. The first two chapters described above are, generally speaking, aimed at understanding the impacts of the NTPP specifically. In contrast, the latte r two chapters use the NTPP and accompanying data sources to draw conclusions regarding NMT intervention research and practice. The final concluding chapter of the dissertation highlights key


7 findings from the dissertation, describing the significance of t he findings in terms of contributions to research and practice, and offers directions for future research.


8 CHAPTER II UNDERSTANDING CONTEXT AND IDENTIFYING IMPACTS OF THE NON MOTORIZED TRANSPORTATION PILOT PROGRAM: A QUALITATIVE APPROACH Introduction Promoting bicycling and walking has been on the national agenda in the United States for over 20 years Bicycling and walking non motorized transportation (NMT) are seen as key alternatives to driving (Pucher & Komanoff, 1999) Reducing driving by encouraging NMT may in turn reduce congestion, mitigate greenhouse gas emissions, improve safety, and increa se physical activity (Krizek & Handy, 2009) For NMT to offer a partial solution to any of these vexing issues facing US cities, substantial increases in the mode share of walking and cycling must be achieved. In 1991, the Intermodal Surface Transportati on Equity Act (ISTEA) was signed into law as a reaction to the concerns regarding the harmful effects of automobile reliance in the US. Over the next fifteen years, a series of federal legislation preceded ISTEA, culminating with the passage of the Safe, A ccountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA LU) in 2005. As part of SAFETEA LU, the Non motorized Transportation Pilot Program (NTPP) was created. The NTPP is unique in United States history for its scale and goal s. Four pilot cities were awarded funds for promoting bicycling and walking, with the stated goal: "to demonstrate the extent to which the bicycling and walking can carry a significant part of the transportation load, and represent a major portion of the t ransportation solution." As the first of its kind in the


9 US in both scale and goals, the impacts of the NTPP have far reaching implications for both research and practice. Lessons from the NTPP can be highly instructive if the US is to move forward with i ts goals of reducing automobile reliance. This chapter employs in depth interviews with key informants in each of the NTPP cities to identify and explore program impacts. The goal of the NTPP is to impact city scale transportation systems and behavior, thu s this study explores macro level impacts and directions for further study. Specific attention is paid to context or surrounding environment within which the NTPP interventions were implemented in each city. Similar interventions may have very differen t impacts in different contexts, and this research aims to identify distinctions between each city studied. The dissertation is informed by two sources of secondary data. The first data source is community level survey of the four NTPP "treatment" cities and one "control" city hereafter referred to as the community surveys. The community surveys were conducted in 2006 and again in 2010. The second data source is manual bicycle/pedestrian counts conducted in each of the cities. These data sources, as wel l as NTPP goals and applicable empirical research inform the exploratory research goals of this chapter: theme identification for open ended interviews. The goal s of this chapter are to first, provide background and context of the NTPP and the NTPP citie s, and second, present an exploratory evaluation of the NTPP. Within


10 these broad goal s specific attention is paid to understanding the context and setting of each city. This includes identifying specific goals, objectives, or challenges each community fac es in promoting NMT. Additionally, this chapter explores what types of interventions were put in place as part of the NTPP, paying special attention to interventions completed prior to the 2010 community survey. The perspective of key informants also infor m s possible impacts of the NTPP, for whom, and possible explanations for such impacts. Findings from the control community are compared to the treatment cities to provide preliminary support for impacts of the NTPP distinct from any external factors. Taken together, these research goals will provide an overview and background of the NTPP, its possible impacts in each of the treatment cities, and directions for the research conducted in the following dissertation chapters Methods Key informant interviews are used to understand the context and identify impacts of the NTPP This section details how participants were identified and the data collection and analysis process. The distinction between data collection, theme identification, and analysis was not cle arly defined; while many themes were pre determined based on applicable literature, the NTPP goals, and existing NTPP data sources, semi directed interviews allowed for emergent themes. This approach is in contrast to that of grounded theory (Glaser & Stra uss, 1967), and was intentionally used to identify actionable future research directions.


11 Study Respondents As part of the NTPP, each of the four treatment cities allocated oversight of NTPP funds to committees of local stakeholders and applicable government agencies, typically overseen by a few government employees in public works or planning d epartments. The treatment city also participated in coordinated meetings overseen by national transportation and advocacy organizations. Respondents were con venience sampled from this population of local and national advocates, planners, and engineers with expert knowledge of both the NTPP cities and interventions put in place during the course of the NTPP. Final sample population included 16 respondents: 10 p ractitioners, 4 advocates, and 2 researchers. Data Collection and Analysis It is common in qualitative research for data collection and analysis to occur simultaneously (Miles & Huberman, 1994). This process begins with determining a coding system, built from emergent or pre determined themes. There are two generally agreed upon method of creating codes, the first is an inductive, or "grounded" approach in which coding is not completed until all data is collected (Glaser & Strauss, 1967). The second, and that preferred by Miles and Huberman (1994) involves creating a "provisional start list" of codes prior to fieldwork. The list can be developed from a variety of sources, including a theoretical or conceptual framework, questions or hypotheses, and key var iables for study (Basit, 2003) This chapter employs the second approach, developing initial codes from literature reviews and content analysis of existing NTPP documents. Codes were then incorporated into a semi structured interview scripts.


12 Existing e valuations of the NTPP have used results from m anual bicycle/pedestrian counts to support the conclusion that the NTPP has positively impacted NMT behavior (FHWA, 2012). Such counts are useful for determining bike/ped traffic at a specific count site, but cannot shed light on who these users are. Ecological theories and conceptual models describe individual, interpersonal, and environmental characteristics each impacting behavior (Sallis, et al., 2006) Theoretical applications of this perspective to travel behavior further support the role of individual characteristics (Krizek & Handy, 2009) specifically demographic and socio economic factors (Schneider, 2013) on travel behavior. With this theoretical framework in mind, participants in each city were asked if NMT interventions were targeted at specific demographic groups, populations, or neighborhoods. The NTPP community surveys were instructive in identifying interview themes In addition to travel behavior questions, the community surveys included a seri es of attitudinal questions; specifically, attitudes toward walking and bicycling, stated willingness to use NMT, and perceptions of neighborhood quality for NMT. Travel behavior literature has identified positive correlations between attitudes toward NMT and behavior (Anable, 2005; Gatersleben & Appleton, 2007; Heinen, et al. 2011; McEachan, et al. 2010) As such, one exploratory theme in this analysis was the degree to which interventions may have targeted changing attitudes toward NMT, and if so, among what populations.


13 The interview script addressed a variety of topical areas (Appendix A). An externa l research professor familiar with the NTPP program, as well as a professional qualitative researcher unaffiliated with the NTPP each vetted the interview protocols. A transportation planner unaffiliated with the NTPP also vetted the question list. The scr ipt was based on open ended questions to frame a discussion about the NTPP, the local context within which the NTPP was enacted, impacts of the NTPP, and quantitative findings from the NTPP surveys. Interviews were conducted in the fall of 2012 and spring of 2013. The majority of interviews were conducted in person individually or in small groups at locations chosen by the participant (frequently an office or a local cafe). Due to scheduling conflicts, two interviews were conducted via telephone. All interv iews were recorded using a digital voice recorder. The interview script had two overlapping purposes: (1) shed light on predetermined themes, and (2) maintain flexibility and allow for emergent themes and individual participant insights. A "saturation" c riterion was used to determine when an interview was complete (Mosquera et al., 2012) Specifically, when the interview scripts had been exhausted and when opinions or insights were repeated. Most interviews lasted between 45 and 90 minutes, and in rare c ases continued during community walking or windshield tours. The analytic process occurred simultaneously with data collection. Notes (i.e., "memos" (Miles & Huberman, 1994) were used to track existing and emerging themes. Emerging themes were then compa red across the complete dataset, in a data matrix. The


14 data matrix process was chosen for its widespread acceptance in interview analysis (Miles & Huberman, 1994), particularly as an analytic tool for thematic network identification (Attride Stirling, 2001 ; Mosquera et al., 2012) Background, Context, and Goals Qualitative results are bolstered with existing data on the NTPP, as well as national data for each city. Taken together, the goal is to provide background, overview and context of each city. Back ground for each city informs NTPP goals, as well as expected impacts of NTPP interventions. For example, the addition of a bicycle lane in a large, relatively compact city with extensive existing on street bicycle infrastructure is likely very different fr om it's impact on a small, rural community with very little existing infrastructure. Similarly, the goals of these diverse communities may be very different given projected growth and local and regional travel patterns. Overview of the NTPP Cities The NTPP provided funding to four US cities the "treatment" cities (Columbia, Missouri; Marin County, CA; Minneapolis, MN; and Sheboygan County; Wisconsin) to support policies or infrastructure improvements meant to promote NMT. The four cities were surveyed in 2006 and 2010, in addition to a fifth city the "control" city (Spokane, Washington). The five NTPP survey cities form the basis of site selection for this research.


15 Table 1 NTPP Community Summary Statistics Columbia, MO Marin County, CA Minneapolis, MN Sheboygan, WI Spokane, WA Land Area (sq. mi.)* 63.08 520.31 53.97 511.27 59.25 Population Density (persons/sq mi) 1,720.1 485.1 7,088.3 225.9 3,526.2 Age 27 41 31 37 35 Total Population* 108,500 252,409 382,578 115,507 208,916 Median HH Income ** $43,102 $89,605 $47,478 $52,993 $41,466 Persons per HH** 2.29 2.34 2.17 2.42 2.30 Persons Below Poverty Level* 22.9% 7.2% 22.3% 8.2% 18.6 Commute: Alone by vehicle ** 74.8% 67% 61.4% 82.6% 74.8% Commute: Carpool ** 11.8% 8.7% 8.5% 9.1% 10.8% Commute: Public T r ansit 0.7% 8.5% 14% 0.5% 4.3% Commute: Walk ** 6.5% 3.3% 6.4% 2.9% 3.4% Commute: Other ** 2.6% 2.6% 4.8% 1.6% 2.2% Work from home 3.5% 9.9% 4.9% 3.3% 4.5% Mean Travel Time to work (minutes)** 16.6 28.1 22.1 18.6 19.9 = 2010 US Census ** = American Community Survey (ACS) estimates: 2007 2011 HH = Household Table 1 present's summary statistics drawn from the 2010 US Census as well 2007 2011 American Community Survey (ACS) estimates from each of the five communities. The communities vary widely in population, geographic area, and population density, as well as by median age, median household income, and commute mode shares.


16 Intervention Accounting Interventions completed as part of the NTPP in each community were catalogued in 2007 during the initial report to Congress ( Appendix A ). Programs and infrastructure projects ar e listed as "funded" but it is unclear what has been completed by 2010 (2010 is the year the second community survey was conducted). Because of a lack of clear accounting for specific projects, it is extremely challenging to link interventions to outcomes of the NTPP; however, it is possible to at least confirm key interventions that were completed by 2010 and identify why there is such inconsistency in intervention accounting. Interviews confirmed inconsistent intervention accounting in each city due to a variety of factors. First, interventions were administered by different groups or authorities (i.e., Public Works may be tasked with striping bicycle lanes, while advocacy organizations may spearhead outreach programs). Second, the process for selecting a nd appropriating NTPP funds varied drastically from community to community, in some cases requiring years of planning and public meetings prior to allocation of funds. Third, it is difficult to determine what exactly constitutes an NTPP intervention, as fu nds were frequently used for planning or leveraged against other funding sources (e.g., the Cal Park Hill Tunnel in Marin cost over $27 million, but only $2.5 million was NTPP funds). Despite the limitations on identifying precisely what was put in place i n the NTPP cities between 2006 and 2010 (or later), key projects and programs identified in the final NTPP report to Congress (FHWA, 2012) are instructive in understanding the types of projects in


17 each community. An overview of key projects is included wit h description s of findings regarding context of each of the NTPP cities. Columbia, MO Columbia, MO has the lowest median age of the five communities (27 years) and is home to a major university. It is geographically isolated in the region, and has a population of just over 100,000 people. The city has the highest rate of walk to work trips of any of the communities and has an existing multi use trail and bike lane network spread across the city (Krizek, et. al., 2007). There is a well organized bicycl e and pedestrian advocacy organization that has overseen a number of NTPP interventions ( ) and in addition to numerous social and educational programs the city has funded a variety of new infrastructure, ranging from additional bike pa rking, to pedestrian underpasses, path connections, and intersection improvements (see Table 18 ). The city of Columbia has a long standing history of supporting walking and bicycling through advocacy and infrastructure. Columbia led efforts to support one of the longest rail to trail projects in the US in the 1980s (the Katy Trail), and put in place a "complete streets" policy in 2004; the city also collaborated with non profit partners to draft both trails and bikeway plans prior to the NTPP (FHWA, 2012). A city official and a department of parks and recreation official each described the extent and history of NMT in Columbia as building on existing interest, support, and infrastructure for recreational travel among residents. Residents have access to nume rous off street NMT


18 facilities, but interview subjects described the primary goal of the NTPP in Columbia as shifting behavior from recreational walking and bicycling to utilitarian walking and bicycling. To achieve this goal, the city first focused extens ively on public awareness and educational campaigns (relative to the other NTPP cities (FHWA, 2012), before NTPP capital projects were completed. Capital projects focused on improving on street system connectivity, as well as connections to the recreationa l system. Innovative and context dependent projects were also completed. Taken together, Columbia's primary NTPP goal is to move from recreational walking and cycling to NMT. In terms of physical infrastructure, Columbia was able to drastically increase the extent of their on street bicycle network striping over 50 miles of bicycle lanes and designating an additional 30 miles of bicycle routes from 2007 2010 (Columbia Bike/Ped Planner). Additionally, NTPP funds were focused on key intersection improveme nts and specific off street investments (Table 2 ). With little awareness or interest in NMT among the general population prior to the NTPP, a city official described a four step approach to promoting utilitarian walking and bicycling: 1. Awareness 2. Understand ing 3. "Just try it" 4. "Institutionalize it [in the individual]" This approach specifically addresses the need to introduce individuals to the idea of utilitarian walking and bicycling. It was employed to promote interest and attitude


19 change among residents pr ior to the completion of transportation focused infrastructure improvements. Table 2 Columbia, MO: Key NTPP Interventions Type of Intervention Project Name Description Completion Date Cost Intersection or Sidewalk Improvement Providence Road, Stewart Road Intersection Remodeled intersection geometry, crosswalk construction, signals, sidewalks, striping, and marking to improve pedestrian and bicycle safety, enhanced trail access May 2009 $400,000 Stadium Boulevard Pedway Eight foot wide pedway along north side of Stadium Drive, with 10 foot portion closer to stadium. Designed to accommodate pedestrians and bicyclists August 2010 $726,800 Experimental Infrastructure Windsor/Ash Bicycle Boulevard New lane striping, signage, shared lane pavement markings, construction of new median which provides safe crossing for non motorized users Summer 2010 $28,000 Promotional or Educational Program Bicycle Skills and Safety Class A variety of bicycle skills and safety classes a imed at teaching safe and confident riding on city streets 2008 2010 $200,000 Walking School Bus Program through which children walk to school under adult supervision 2008 2009 $100,000 Source: FHWA, 2012 Marin County, CA North of San Francisco, Marin County consists of a collection of communities localized in the eastern portion of a vast land area. The population is generally older (median age 41) and wealthier (median HH income is $89,605) than the other Pilot Communitie s. While the auto commute mode share is low, that is partially explained by a nearly 9% work from home rate (US Census, 2000). The average travel time is greatest in Marin, which may be explained by regional commute travel (e.g., to Berkeley, Oakland, or S an Francisco) The county boasts over 30 miles of multi use paths (Krizek,


20 et al., 2007) and has focused much of their NTPP funding on projects improving connections across its existing bicycle/pedestrian network (FHWA, 2012) ( see Table 19 ). Marin County has a strong history of recreational cycling as well as walking and cycling advocacy. The sport of Mountain Biking began in Marin County in the 1970s ( Marin is also the home to the national Safe Routes to Schools program, which began in 2000 ( All interview subjects describe d the population of Marin as generally pro environmental, and pro walking and bicycling. Social outreach under the NTPP has focused on a variety of educational campaigns targeted to spe cific populations and issues, including: Bicycle repair lessons in English and Spanish Riding with Youth: bicycling classes for parents and children Street skills courses League of American Bicyclists Certificate Bicycle Instructor training Engineer's bik e/ped facility design training courses "Way to Go" household level social marketing campaign (FHWA 2012) The County has a unique geography and history that has dictated urban form and land uses, leading to complex issues for completing a comprehensive bicycle and pedestrian system. Many of the communities were originally built around the streetcar (Public Works representative), and are relatively high density and walkable; however,


21 they are separated from one another by mountain ridges that pose physic al barriers to NMT use between communities. Approximately 20% of Marin County residents commute into San Francisco for work, with the majority of residents commuting elsewhere, within the county, or working from home (Public Works representative). Addition ally, there is a great deal of congestion on the county's few through roads and very little physical road space or adjacent land for additional infrastructure. Public works officials described road space as a consistent problem; frequently it is impossible to separate on street cyclists from vehicular traffic with on street infrastructure These complex issues have meant that Marin has focused on complex, connectivity oriented projects to bridge critical gaps in an otherwise extensive NMT system (Table 3). Table 3 Marin County, CA: Key NTPP Interventions Type of Intervention Project Name Description Completion Date Cost Significant Regional Infrastructure Cal Park Hill Tunnel A 1,100 foot rail with trail tunnel providing nonmotorized access between San Rafael and Larkspur December 2010 $27,700,000 Network Gap Closure Alameda del Prado Added bicycle lanes and improved sidewalks within the existing right of way along Alameda del Prado in Novato July 2010 $2,947,358 Pedestrian Safety Improvements Medway Road Improvements Shared lane markings (sharrows), widened sidewalks, and installed new transit shelters and street furniture. Utilities were undergrounded through a separate project October 2008 $1,665,300 Source: FHWA, 2012


22 Minneapolis, MN Minneapolis is the most populous, diverse, and densely populated of the fi ve communities surveyed (Table 1 ). The median age is relatively young compared to the other communities, and it is the most ethnically diverse community (US Census, 2010). Minneapolis also has the highest share of non vehicular commuting and the most extensive on and off street bicycle/pedestrian network (Krizek, et al., 2007). The city has planned 25 infrastructure projects and 3 promotional cam paigns to be completed during the treatment period ( see Table 20 ). Minneapolis has a long history of NMT promotion (Krizek & Barnes, 2007 NTPP infrastructure goals include strengthening connectivity with the downtown area to encourage NMT, but also focus ing on innovative infrastructure investments (many of which are unique among NTPP communities). Such investments include bicycle boulevards, road diets, bike sharing, and "bike boxes" at intersections (Table 4). One Public Works official described the city 's infrastructure needs and strategy as "quality over quantity." That is, there is an existing (and extensive) NMT system, but they are moving beyond simply striping bicycle lanes to focus on more innovative (and presumably more effective) interventions su ch as bicycle boulevards.


23 Table 4 Minneapolis, MN: Key NTPP Interventions Type of Intervention Project Name Description Completion Date Cost Network Gap Closures Marshall Avenue, St. Paul Key linkage between on street bicycle facilities along Marshall Ave in Saint Paul and the Grand Rounds Trail system and Midtown Greenway terminus in Minneapolis October 2010 $495,000 Reallocating Roadway Capacity Franklin Ave, Minneapolis Conversion of four lane road to three with a center turn lane and bike lanes on both sides August 2010 $50,000 20 th /Minnehana, Minneapolis Conversion from four lanes to three with a center turn lane and bike lanes on both sides, bicycle left turn lane, enhanced trail crossing October 2010 $150,000 Increasing Access to Bicycles Nice Ride Bicycle Sharing Public bicycle sharing program in Minneapolis and Saint Paul 2010, ongoing $3,629,047 Sibley Community Partners Bike Library Community bike library providing 6 month bicycle loans, classes, and support for low income residents to acquire a bicycle 2010, 2011 $201,000 Source: FHWA, 2012 Like many of the NTPP cities, a primary focus in Minneapolis is to motivate recreational walkers and cyclists to engage in NMT. While the extent of travel behavior change is unclear in Minneapolis, the city has received national attention for its efforts at promoting walking and bicycling. Between 2006 and 2010, the city was named a silver level bicycle friendly city in 2008 and attained gold level in 2011 b y the League of American Bicyclists ( The city was also named top bicycling city in the US by Bicycling magazine in 2010 ( Whether or not participating in the NTPP directly led to national recognition is unclear, but the city has experienced numerous changes during the time period of the NTPP.


24 Sheboygan County, WI Sheboygan County is spread across a vast area, and consists of over a dozen small communities (in cluding the City of Sheboygan). The county has the lowest rate of NMT (ACS, 2011), and is the least ethnically diverse of the five communities (US Census, 2010). The County has reported at least 20 infrastructure projects that have been funded by the NTPP, and 5 p romotional/education projects ( see Table 21 ). Sheboygan has also implemented a walk to school initiative, a bike/walk to school event, and updated their bike/ped comprehensive plan (Krizek, et al., 2007). The region has a long history of recreational cy cling and has been implementing modest, off street, multi use trails since the 1970s (FHWA, 2012), but there has been little focus on transportation walking or cycling prior to the NTPP. At the start of the NTPP, Sheboygan County's on street bicycle networ k was virtually non existent (County official), while the sidewalk network was relatively complete. A multi use path constructed on a derelict rail corridor that runs through the heart of the city of Sheboygan is currently under construction, and taken tog ether, Sheboygan is, like the other pilots, focusing its efforts on moving from recreational walking and bicycling to NMT. The combination of a lack of basic on street NMT infrastructure and an abundance of street space and adjacent land, Sheboygan is in a unique situation among pilot communities : there is a great deal of "low hanging fruit" available in the form of inexpensive and uncontroversial physical interventions to create a connected NMT system (Table 5). While jurisdiction and funding varies by t own, city, and county, road


25 space is generally ample T hus it has been a simple process to approve and add bicycle lanes throughout the region (the county has funded over 60 miles of on street lanes alone with NTPP funds (the majority of lanes were striped by 2011, but it is unclear how many were completed by 2010 (FHWA 2012)). Table 5 Sheboygan County, WI: Key NTPP Interventions Type of Intervention Project Name Description Completion Date Cost Community wide Transportation Network Village of Cedar Grove Sidewalks and Bike Lanes 2,100 foot bicycle lane and new sidewalk along South Main Street Fall 2008 $859,300 Infrastructure Improvements at Schools Howards Grove High School Pathways 1,450 feet pathway for walking and bicycling to and from Howards Grove High School and Athletic Complex Spring 2009 $104,369 Promotional or Educational Program Bike and Walk to Work Week Partnership with the Bike Federation of Wisconsin to encourage bicycling and walking to work and other activities Annual, 2008 2011 variable Source: FHWA, 2012 NTPP social programs have focused on education for all road users (e.g., bike/walk to work week; Table 5), as well as served as an opportunity for education among policy makers, and collaboration to create a bicycle/pedestrian master plan (FHWA). Interview s with both County officials and local advocates suggest that most residents are unaware of the possibility of walking or cycling for transportation purposes. Interviews suggest that "Road rage" against cyclists is not an issue (a local advocate posited th is is likely due to the lack of traffic in the region and the ample road space). Unlike in Columbia, where educational programs were introduced prior to infrastructure investments, officials in Sheboygan chose to implement education and outreach alongside infrastructure. However, implementation has taken much longer than expected and most programed interventions were not completed by 2010 (County Official). Some on street


26 striping and sharrows were implemented prior to 2010, and County officials stated that the public was generally confused by the additional on street markings (i.e., they didn't know what they were for, and had not considered bicycling on the street). Spokane, WA Pilot Program researchers selected Spokane as a control community reasonin g that it has a number of characteristics that make it representative of the treatment communities ; specifically, a significant student population, regional geographic isolation and a reasonable rate of walking and cycling (Krizek et. al., 2007). The city's population is also of average age and income (compared to the other four communities) but lacks some ethnic diversity (US Census, 2010). While the other communities received NTPP funding, Spokane did not; however, the city did receive modest American Recovery and Reinvestment Act funding that went to expanding bicycle lanes and multi use paths (less than $5 million for such projects in total according the City Bike/Ped Coordinator) Additionally, the city has an active bicycle a dvisory board, and recently hired a bike/ped coordinator ( see Table 22 ). In 2009, the city was rated a "Bronze Level" Bicycle Friendly Community by the League of American Bicyclists ( ). Similar to the treatment cities, Spokane has a recreational trail system a number of on street bicycle facilities and a connected sidewalk network. Their goals are to increase bicycling and walking for all purposes (this is distinct from the NTPP, in which promoting transportation walking and cycling was a specific goal). The city has completed extensive planning and prioritization efforts for increased NMT infrastructure.


27 Physical infrastructure goals include improv ing connectivity between the existing off street recreational system and on street fac ilities. In some ways Spokane is similar to the other Pilot Communities; focusing social programming to encourage recreational walkers and cyclists to utilitarian NMT users. However, Spokane has also developed local partnerships between city officials, t he regional health district, and community groups to create social campaigns aimed specifically at road safety for all users. The "Stick Man Knows" campaign was developed as a comprehensive awareness campaign for all users funded by a State Department of T ransportation grant ( and is one example of such targeted social interventions. Unintended Impacts Spokane, WA In experimental research design, an observation is conducted before an intervention (or treatment) and after the interventi on. A control group is measured alongside treatment groups to test the null hypothesis, but studies should be carefully designed to avoid the observation acting as a treatment for the control group. It appears that just such a testing effect" effect, in w hich the act of being measured can have an impact ( Singleton & Strai ts, 2009) occurred in Spokane. One senior planner in Spokane described a "latent group" of motivated officials that "needed motivation" to closely consider and plan for NMT. Exclusion fro m NTPP funding resources motivated inter agency collaboration and led to the creation of a bike/ped coordinator position, TIGER grants for bicycle infrastructure, and "Bronze" bicycle friendly community status.


28 Inclusion as a control community did impact S pokane by helping to build capacity and inter agency collaboration; however, city officials have stressed that without NTPP funds, capacity building has been limited to planning, education, and outreach, with little infrastructure. Spokane has been unable to alter the physical context and lacks resources that would enable newly developed collaborations to challenge the "bureaucratic inertia" of auto centric planning and development. The Moderating Effect of the NTPP The key informant interviews shed ligh t on a number of themes to establish background, context, and directions for further evaluation of the NTPP. Pre defined themes such as behavior change among NMT users, and overall impacts on city scale NMT use or perceptions were examined. One overarching theme that provides a conceptual basis for approaching possible impacts of the NTPP is understanding the degree to which the NTPP served as a moderating factor in NMT promotion for each of the treatment cities. In research design, the presence of a mode rating variable is not necessary for a relationship between two variables to exist, but the presence of the moderator impacts the strength of that relationship (Singleton & Straights, 2009). The conceptualization of the "NTPP as moderator" hypothesizes tha t the NTPP amplified NMT promotion over time to an extent that would not have occurred in it s absence (see Figure 1 ).


29 Compared to highway and road funds, the $25 million awarded to each of the treatment cities is an extremely small amount of money (Field s and Hull); however, this influx of pilot funds expressly for NMT in each city is unprecedented nationally. It was hypothesized that such a funding increase may have a moderating impact on the NTPP cities. As one Public Works official stated, "the pilot f unds really allowed us to raise the bar on the types of bike/ped projects we were doing." This official in turn sketched a step function, which I have adapted ( Figure 1 ) to illustrate the moderating effect of the NTPP. Figure 1 The Moderating Effect of the NTPP Building on the context established in the previous section, "raising the bar" can mean a variety of possible impacts in each community. Impacts of the NTPP grouped


30 under the heading of moderating effects' extended to the local community, infrastructure quality, and institutional capacity. Behavior Change Quantitative data sources travel behavior theories and conceptual models informed specific interview questions regarding behavior change in the NTPP communities. Behavior change in each community was the primary goal of the NTPP: impacting mode shifts from driving to walking and cycling. To quantify this, each community completed a number of independent data collection exercises, in addition to the survey data analyzed in this dissertation (FHWA, 2012). All five communities conducted bicycle counts at various times and locations over the years of the NTPP, and some communities conducted additional surveys. The bicycle counts in each community generally trended upward (i.e., more bicycles each year), but this effect was not consistent over each year of the NTPP in each community. Independent surveys (surveys conducted by individual communities) also found similarly positive findings about satisfaction with NTPP programs and reported increased NMT (FHWA, 2012). A necdotal statements by key informants in each community suggest that there appeared to be more NMT happening over the course of the NTPP (interviewees in Spokane made more cautious claims of increases in NMT). In contrast to the evidence supporting behavior change in the NTPP cities, the community surveys do not show any statistically significant beha vior changes in any of the cities, or over the entire population. These surveys probability sampled the overall


31 population of each city, whereas all of the independent data collection efforts in each city used non probabilistic sampling frames. The conflic ting stories these data sources tell could have a methodological explanation (Gotschi, et. al., 2012) on the one hand, probability sampling is inappropriate for accessing poorly represented populations. On the other hand, non probabilistic sampling frame s lack generalizability. Qualitative data is largely inconclusive in this case, as it is unclear if anecdotal evidence and convenience sampling captures existing NMT users who may be altering route choices or trip types, and travel patterns, or new NMT use rs. Attitude Change Interventions aimed at promoting NMT are also, implicitly, aimed at altering attitudes and perceptions toward NMT. The social and educational programs detailed above target attitudes and perceptions to change behaviors. The NTPP comm unity surveys included a series of questions regarding attitudes and perceptions of NMT generally, as well as neighborhood quality for NMT use (i.e., does an individual perceive their neighborhood as supportive of NMT, or a hindrance to NMT?). Because of t his, the possibility of attitudinal impacts to the NTPP was explored in interviews. Two types of possible attitude changes were identified: perception of NMT on roads, and willing/interest in NMT among the public. Rather than providing evidence of actual i mpacts of the NTPP in these cities, findings from this section represent possible directions for further study.


32 Perception of NMT on Roads Perceptions of NMT on roads applies specifically to bicycling: perceptions of bicyclists (and bicycling) on roads in mixed traffic by either drivers or cyclists. Road rage and safety concerns can be powerful barriers to bicycling (O'Connor & Brown, 2010; Pucher & Dijkstra, 2003; Pucher & Komanoff, 1999; Wegman, et al. 2012) All of the NTPP cities (including Spokane) focused soft interventions on improving acceptance of NMT on roads, educating all users about proper etiquette, and stressing safety (e.g., Columbia, "Let's roll together" campaign; Bike ambassadors in Minneapol is, "Stick man knows" in Spokane). Two communities described changes in perceptions of bicycling on the road: Minneapolis and Columbia. In Minneapolis, one local advocate introduced the idea of a "sea change," in which cyclists in the city felt accepted by vehicles as a viable road user. Similarly, in Columbia, a city official described a "peak of animosity" in about 2009 that may have stemmed from a backlash by some residents to NMT promotion, but that since that time both drivers and bicyclists have be en more courteous to one another. Interviewees in both cities note that it is difficult to gauge the degree of actual animosity from drivers there had existed, citing concerns that "vocal minorities" can create a great deal of noise clouding an otherwise p ositive shifts in acceptance of bicycling on city streets. Willingness/Interest in NMT The second type of attitude changes identified in the NTPP cities is willingness/interest in NMT by the general public. City officials in Sheboygan noted they


33 are "sta rting from scratch" with NMT promotion and described initial successes in educating residents. At the start of the NTPP, there was a great deal of local coverage of pilot program activities and goals, and that led to a great deal of initial support and int erest in walking and cycling. However, the majority of NTPP funded projects were not completed by the 2010 survey, and city officials describe a "plateau" of interest or excitement about NMT. In contrast to Sheboygan, officials and advocates in Marin not ed broad acceptance and interest in NMT. Interviews with local advocates identif ied the Safe Routes to Schools program in Marin as being influential in altering perceptions of NMT among children and youth in the area. There are young people in Marin who ha ve been at SRTS schools for their entire k 12 education. Empirical studies in Marin support the effectiveness of SRTS on target populations (Boarnet, et al. 2005; Staunton, et al. 2003) Whereas in Sheboygan, basic education and promotion may influence a ttitudes regarding willingness to engage in NMT, Marin may have found itself at a saturation point of positive attitudes. If attitude change exists on a spectrum from lack of knowledge to a strong opinion, Sheboygan and Marin represent opposite ends of tha t spectrum, and it is possible that the correlation between attitude and behavior weakens once attitudes are generally highly positive. Institutional Capacity Existing data sources evaluating the NTPP focus on measuring impacts on the general population but research in Minneapolis suggests that the NTPP may have played


34 a role in "altering the bureaucratic inertia' and institutional capacity of the communities (Fields & Hull, 2012). Specifically, there are myriad policies and regulations in the US that support and promote auto oriented development (Levine, 2008). The NTPP as a program may have induced capacity building and inertia of culture change away from such policies and regulations. Rather than focusing on population level changes, Fields & Hull po sit that a modest influx of NMT funding in an existing bureaucratic structure could lead to increased capacity through collaboration, and a culture change toward sustainable transportation policies. Without prompting in interviews, this topic arose, with s ome variation, as an emergent theme in each of the NTPP communities. Capacity building and culture change are distinct yet related themes, each involving impacts of the NTPP on the individuals and institutions charged with administering and implementing the NTPP. Interviews in Minneapolis and Marin illustrate the impact of the NTPP on building capacity by empowering the advocacy community and bringing planners and engineers together to create lasting impacts on the bureaucratic structure in these communities. In Minneapolis, a new department has been created in public works to administer NTPP projects. City officials in Marin described how, prior to the NTPP, local agencies did not make a practice of working together, but the NTPP "brought us to the same table" (Public Works official). Advocacy directors in Marin described ho w "we [the advocate community] were taken seriously [by officials] now that we had money." Similarly, Marin Public Works officials described enhanced communication across local and regional authorities to promote NMT.


35 Capacity building strengthening co nnections within local government and perhaps more importantly, between planners and the local advocacy community has led to extensive infrastructure interventions in Marin and innovative interventions in Minneapolis. Participants in Marin described how connections between local and regional authorities and advocates were directly responsible for the leveraging of NTPP funds for a rail and bike/ped tunnel project (see Table 3: The Cal Park Hill Tunnel), as well as numerous highway and bike/ped path improv ements. Similarly, in Minneapolis, the creation of a central authority for bicycle and pedestrian transportation has allowed public works officials to specialize in "quality over quantity." That is, interviews in Minneapolis described how NMT promotion has shifted toward fewer, innovative, higher impact interventions rather than "just striping a bike lane on every street" (Public Works official). Minneapolis is one of the first cities in the US to pilot street treatments such as bicycle advisory lanes, gr een lanes, and bicycle boxes (FHWA, 2012). Increased institutional capacity and culture changes due to the NTPP may, in turn, lead to reduced barriers to implementing NMT interventions in the future. Conclusion This chapter provides an overview of the cities involved in the NTPP and an exploratory examination of possible impacts of the NTPP. The research employed semi directed key informant interviews informed by secondary, quantitative data The NTPP is a unique program in US history, representing a na tional interest in reducing reliance on the automobile. As such, the NTPP also represents a unique opportunity to assess the impacts of a relatively modest (in terms of city scale transportation funding) influx of funds on


36 the treatment cities. To put the NTPP funding in perspective, consider that all NMT projects in Minneapolis and St. Paul over the time of the NTPP accounted for about 4% of total federal transportation funds (this included SRTS, Recovery and Reinvestment Act, and non NTPP bike/ped enhance ment funds) Without accounting for NTPP funds, these cities typically spend about 3% of federal funds on NMT, while the national average is only about 1.5% (Fields & Hull). Thus, NTPP funding represents a massive increase in local NMT funds, but is dwarfe d by total federal transportation spending The interviews shed light on two areas of the NTPP establishing context for each city and identifying possible impacts for future study The context, background, and goals of the NTPP in each community is important as it guides expectations regarding impacts, and the role of the control community in the NTPP research design. Identifying possible impacts of the NTPP in turn will allow for focused conceptualization and evaluation of specific elements of the N TPP. Each of the NTPP cities faced specific issues and challenges in unique context s but generally the goal of NMT promotion was capitalizing on existing recreational NMT use and improving NMT infrastructure. In communities with virtually no infrastruct ure, such as Sheboygan, this meant a focus on striping bicycle lanes and completing sidewalk improvements. In cities with more extensive existing NMT systems, the focus was on costly and complex connectivity issues (Marin County), or new and innovative int erventions (Minneapolis). Identifying specific interventions put in place in each city between 2006 and 2010 presents numerous practical challenges. Additionally focusing


37 on an exact accounting may be misguided, as similar interventions may have different impacts depending on context. Conceptualizing the NTPP as a moderating factor in NMT promotion offers a helpful lens for identifying and distilling impacts into a useful form. This lens does not require specifying parameters or quantifying start and end points, rather a comparison to the control community. Interviews subjects in the control city support the assertion that inclusion in the measurement of the NTPP did positively impact institutional capacity and lead to small scale education and outreach c ampaigns. However, no infrastructure could be implemented without external funds. In the treatment communities, a wide variety of infrastructure could be implemented, and in many cases, modest NTPP funds were leveraged for high cost infrastructure interven tions. In cities like Sheboygan and Columbia, extensive on street striping programs were completed, clearly "raising the bar" for basic infrastructure. The outstanding question, however, is how the NTPP influenced the general population. It is outside of the scope of this chapter to quantify population level impacts of the NTPP; however, this analysis suggests that NMT interventions have likely impacted both attitudes and behavior, but it is unclear how much, and for whom? Another way to consider the NTPP 's impact is if there is any evidence of negative effects. None of the interviews suggest that an influx of funds led to bureaucratic gridlock or large scale, anti NMT campaigns. At the very least, the NTPP has increased the multi modal capacity of transpo rtation systems, and increased communication and collaboration within


38 institutions and between institutions and advocacy groups. A modest investment in national transportation goals may have far reaching impacts on city scale transportation networks, and s uch investments could be the first step to supporting significant mode shifts away from automobile reliance.


39 CHAPTER I I I ATTITUDES, NEIGHBORHOOD PERCEPTIONS, AND NON MOTORIZED TRANSPORTATION: EXAMINING CHANGES IN ATTITUDES AND PERCEPTIONS OF BICYCLING AND WALKING IN FIVE US CITIES Introduction Bicycling and walking are increasingly being considered as viable options to address congestion, pollution, livability, and health. Despite wide ranging interest in promoting these forms of non moto rized transportation (NMT), widespread increases in rates of NMT have been elusive to identify. Measuring NMT use is challenging in the United States as only a small subset of the population engages in regular bicycling or walking, and survey methods often fail to adequately capture this subset of the population, or the extent of their NMT use. This study described in this chapter posits that a critical "first step" to increased NMT use is an increasingly positive and accepting attitude toward such modes as viable, safe, and acceptable. Such a proposition is drawn from psychological theories of behavior change and the link between favorable attitudes toward NMT, and increased use. Using data from the Non motorized Transportation Pilot Program (NTPP), I show that while there is not statistically significant evidence of behavior change from 2006 (Phase 1 of the study) to 2010 (Phase 2), attitudes toward NMT have become more positive at the same time that interventions have been put in place to promote NMT However, this effect is complicated by interaction terms included in the statistical analysis. Improved attitudes correspond to bicycle traffic counts conducted in each of the


40 cities studied While these correlations fail to prove a direct relationship b etween attitudes and behaviors, they do present compelling evidence that one component of increasing NMT use is improving attitudes and perceptions of these modes as viable, safe, and acceptable among the general population. This chapter informs future inv estments in NMT promotion by providing evidence from the NTPP community surveys of positive attitudes changes in specific communities and among specific populations that are correlated over time with specific interventions. Literature Review Traditionally, mode choice has been predicted based on utility theory: the theory that individuals will make rational mode choice decisions by weighing various options and choosing the option that maximizes utility (Levinson & Krizek, 2008). Utility theory has been criticized for its limited scope (Levinson & Krizek, 2008); for example, utility alone is unable to account for why otherwise similar individuals may make very different mode choice decisions (Handy, 2005). The framework offered by utility theory also specifies that an individual's context (the local environment, the requirements of a specific trip, available mode choices, appropriate infrastructure, etc.) directly influence that individual's behavior. Drawing on theories of behavior change, this research conceptualizes an additional step in the "causal chain" implied by utility theory: context (and any changes to context) first impact attitudes or perceptions regarding NMT behavior, which impact a utility assessment (e.g., "travel liking" (Ory & M okhtarian, 2005) and finally impacting actual behavior (Figure 2 : Conceptual Model).


41 Figure 2 Conceptual Model There is both a conceptual and a pragmatic rationale for considering attitude change as an impact of interventions aimed at promoting N MT that precedes behavior change. There is a vast body of literature measuring NMT behavior: mode choice decisions, frequency and trip type of NMT travel, etc. (Heath et al., 2006; Heinen,, 2010; Ogilvie, Egan, Hamilton, & Petticrew, 2004; Pucher, e, 2010) but much of this work is plagued by either the challenge of measuring small behavior changes among a subset of travelers (i.e, frequent NMT users) in a population based survey (Gotschi, et. al., 2011), or measuring the behaviors of a non repr esentative sample population (as in non probabilistic surveys). The 2009 National Household Travel Survey (NHTS), reports that about 1% of trips in the US are by bicycle, and about 0.5% of work trips are by bicycle (Pucher,, 2011) while all NMT t rips in 2009 account for about 12% of travel in the US. Measuring attitude change may circumvent both of these practical challenges, as all members of a


42 population will likely have some attitude (positive or negative) towards NMT, while only a small subset may engage in that behavior. Changing Attitudes to Change Behaviors Empirical literature supports the relationship between attitudes/perceptions of NMT and behavior that is, "the degree to which a person has a favorable or unfavorable evaluation or a ppraisal of the behavior in question" is correlated with the behavior (Ajzen, 1991). Attitudes and perceptions have been shown to be significantly positively correlated with NMT use (Dill, 2010; Gatersleben & Appleton, 2007; Heinen, et. al., 20 11; Rose & M arfurt, 2007 ; Handy 2005). The majority of studies exploring the attitude behavior link employ the Theory of Planned Behavior (or TPB) (Ajzen, 1991) The TPB has been applied to NMT use (Heinen et al., 2011) as well as the impact of specific interventions aimed at promoting NMT (Dill, 2010) The TPB theorizes that behavior change is influenced by three primary factors: attitudes, subjective norms, and perceived behavioral control. This research hypothesize s that shifts in attitudes regarding NMT encourage increased NMT use due to significant correlations identified in the transportation literature (Heinen et al., 2011; Joh,, 2011; Lorenc, et. al., 2008 ) This notion has also been "field tested" repea tedly in a variety of settings; for example, the public health literature provides numerous examples of behavior change programs that target attitudes to change behaviors (drunk driving and safe sex campaigns are just two common examples (Xing, 2012)). Xi ng (2012 ) suggests that planners draw inspiration from anti


43 smoking campaigns which, despite their goal of discouraging a behavior through attitude change (rather than encouraging a behavior like bicycling), use the same conceptual framework: attitude chan ge is a precursor to behavior change. The automobile is the primary method of transportation for most travelers in the US. In much of the US, utility theory dictates that auto travel is the most appropriate choice for travelers living and working in all b ut the densest urban environments. Transportation infrastructure is designed primarily to facilitate vehicle flow, and public policy prioritizes auto infrastructure funding and subsidized parking. This "context" (i.e., a political and physical environment prioritizing auto accessibility), in turn marginalizes walking and cycling. It may appear that, in the US at least, attempting to drastically alter attitudes toward NMT among most average Americans may be a zero sum game, but there is an historical precedence for just such an about face. Prior to the popularization of automobile travel in the US, walking, cycling, and transit were the preferred and prioritized modes. Norton (2008) offers an account of how American streets in the 1920s and 1930s were "socially reconstructed as places where motorists unquestionably belonged" (pg. 1): "Motorists arrived in American city streets as intruders, and had to fight to win a rightful place thereMotorists who ventured into city streets in the first quarter of th e 20 th century were expected to conform to the street as it was: a place chiefly for pedestrians, horse drawn vehicles,


44 and street cars. But in the 1920s, motorists threw off such constraints and fought for a new kind of city street a place chiefly for m otor vehicles. With their success came a new kind of city a city that conforms to the needs of motorists." (Norton, 2008, pg. 7) The social reconstruction (i.e., attitude change) of the automobile in American culture was a necessary precursor to public s upport for investment in automobile infrastructure and a set of policies founded on the idea that automobiles are the sole and rightful users of roads. It is important to move toward a social reconstruction of the American street that includes bicyclists a nd pedestrians. While impacting national attitudes toward bicycling and walking has proven challenging in the US, Denmark, The Netherlands, and Germany have all managed to overcome a post WWII philosophy of auto dependence and become international examples of best practices in encouraging NMT (Pucher & Buehler, 2008) Study Background This study builds on earlier studies employing psychological theories of behavior change by first proposing that attitude change is an important impact of interventions aime d at promoting NMT an impact that, to our knowledge, has been overlooked in NMT evaluation research. I then propose that improved attitudes toward NMT are vital to increasing NMT use, road safety for all users, livability, and a decreased reliance on exp ensive, de dicated NMT facilities (Figure 3 ).


45 Figure 1 Model of Positive Attitudes The conceptual model of positive attitudes (Figure 3 ) provides a framework for understanding the impact of positive attitudes in a commun ity directly (through primary impacts) and indirectly (through secondary and long term impacts). These impacts compound to create a positive feedback loop in which attitudes toward NMT play a critical role). For example, as prevailing attitudes in a commun ity become more positive towards (or accepting of) NMT, there will be decreased hostility towards NMT users and a reduction in perceived barriers to use. In turn, rates of NMT use will increase, as will investments in NMT supportive policies and infrastruc ture. Leading eventually to a more equitable transportation system, safer and more livable streets, and improvements in individual and environmental health. The converse to the positive feedback loop (Figure 4 ), is an equal and opposite feedback loop ill ustrated by Figure 4 (Model of Negative Attitudes). In this case negative


46 attitudes toward NMT lead to reduced acceptance (or increased marginalization) of NMT, lack of support for NMT policies or infrastructure, continued reliance on the automobile, and accompanying deleterious impacts of auto reliance. Figure 2 Model of Negative Attitudes As recent studies (cited above) have shown, favorable (or more favorable) attitudes toward NMT are significantly and positively corr elated with use. While the precise causal pathways, timeframe, and interactions between attitudes, behaviors and the built environment is uncertain (Handy, et. al., 2006) the relationship exists and should therefore be considered an important aspect of any policy meant to promote NMT. This is not to say that attitude change has been ignored in NMT promotional activities, it is in fact a crucial component of most social marketing campaigns and many event based activities (e.g., bike to work day, etc.), but i t is unmeasured as an impact of such activities.


47 Non motorized Transportation Interventions I nterventions to promote NMT either explicitly or implicitly influenc e attitudes to change behaviors. Interventions aimed at promoting walking and cycling have been categorized as either "hard" (infrastructure interventions such as bike lanes) or "soft" (policy programs or social marketing campaigns) (Krizek, et al., 2009) Essentially, all interventions aimed at promoting walking and cycling that are not strict ly hard measures include, either explicitly or implicitly, an element of attitude change to encourage behavior change. Bike to Work day programs are one example of such campaigns that aim to promote bicycling by changing attitudes (Rose & Marfurt, 2007) There is also a burgeoning literature evaluating the impacts of a variety of voluntary behavior change (VTBC) measures on car use and greenhouse gas emissions (Bršg,, 2009; Stopher, et. al., 2009) Evaluation efforts of NMT promotional activities l argely overlook attitude change. In VTBC studies, for example the attempt is to impact behavior change by influencing attitudes; however, bike to work day studies and VTBC researchers have not incorporated attitude measurement into their evaluations. E valuation efforts frequently focus on mode share shifts or facility use, and this is understandable since these are the end goals of most NMT promotional activities. But attitude change is a prerequisite, and precursor, to voluntary behavior change. Behavi or change theories are adept at describing this process, but are less straightforward about the time horizon by which attitude change occurs and mediating factors that may cloud the link between attitude and behavior change (e.g., aspects of the built or s ocial


48 environment). This is further complicated considering that promotional activities may impact different populations very differently, both in terms of attitudes and behaviors. Attitude change may also be more widespread and rapid than behavior change (and also may be more pronounced when mode choice options are limited). There is also likely a "lag effect" between an intervention, a change in attitudes, and a change in behaviors. Lag Effects Between Interventions and Outcomes Reviews of the NMT intervention literature are frequently characterized by their breadth. W hile such reviews offer valuable guidance regarding substantive empirical findings regarding a wide variety of interventions lag effects are frequently unstudied or unreported Becaus e of these factors, this review relies on two complementary published literature reviews to infer existing evidence of lag effects. Literature reviews by Yang, et al. (2010), and Ogilvie, et al. (2007) used rigorous criteria for study inclusion, evaluating only those with experimental (or quasi experimental) control (both random and non random), before and after (or some type of time series measure greater than 1 day). Ogilvie, et al. (2007) took the added step of specifying minimum response and attrition r ates. Neither review excluded studies based on sample size, outcome measures, or analysis method In both cases disparate measures precluded a meta analysis (Yang, et al., 2010), and all outcome measures were of behavior (i.e., none of the studies measure d attitude change). These restrictions, in particular requirements for control groups, also limited the reviews to evaluating predominantly (but not solely) "soft" interventions (policies and educational campaigns) delivered at either the population, group or


49 individual levels O nly one intervention study evaluated the impact of a "one time" or "event based" intervention (in this case, a "walk to school week" event). Yang, et al. (2010), and Ogilvie, et al.'s (2007) findings offer few insights into the l ag effect of interventions aimed at promoting walking and cycling. Studies cited by Yang, et al., and Ogilvie, et al. collected follow up data from as little as 3 weeks after the intervention to as much as 10 years, but the majority of studies followed up within 1 year. Of the 25 cycling intervention studies identified by Yang, et al., two reported negative results (Cervero, et al., 2002) or insignificant findings (Groesz, 2007), with follow up periods of nine and five months respectively. Of the 29 walking intervention studies identified by Ogilvie, et al., no statistically significant negative results were reported, but 12 studies found insignificant changes in time spent walking/week between follow ups of six weeks and one year. Six studies reported in th ese two literature reviews employed follow ups between one and three years, and half of these studies either found no statistically significant evidence of behavior change, or did not report statistical significance. It makes intuitive sense to allow for a lag effect prior to measuring interventions to expect lasting and measurable behavior change resulting from specific NMT interventions. Despite a dearth of empirical literature on lag effects Yang, et al. (2010), and Ogilvie, et al. (2007)'s reviews pr ovide insight into standard evaluation practices among NMT researchers. Further, Ogilvie, et al., (2007) notes that evidence of intervention efficacy is skewed towards individually focused interventions with short


50 follow up times utilizing convenience or v olunteer samples. It is reasoned that population level interventions are less frequently implemented, are more challenging to measure with adequate controls, and their modest impacts may be difficult to measure in large population samples (Ogilvie, et al. 2007). Yang, et al. (2010) echoes this sentiment and stresses the need for consensus on appropriate measures of NMT behavior. Neither study is able to provide recommendations for accounting for the lag effect. In the absence of empirical evidence to supp ort or refute a timeframe for lag effects, just such an effect may be at play in the NTPP cities (Gotschi, et al., 2011). As described in Chapter 2, the extent of interventions put in place in the NTPP cities between 2006 and 2010 is unclear. Some level of both hard and soft measures were implemented by 2010, but isolating the strength and lag effect of specific individual interventions is beyond the scope of this work. If any impacts of NTPP interventions are to be identified, then they would likely be ide ntified as attitude changes. Thus, this research tests the hypothesis that attitudes toward walking and cycling have improved over the same time period community level interventions were put in place to promote attitude and behavior change, and these attit udes changes likely precede behavior changes for some subset of the population. Quantitative Analysis This chapter aims to measure attitude change in four cities in the United States using data from the NTPP. While the conflicting results of the NTPP dat a analyzed in this study and the bicycle counts and intercept surveys may not be entirely surprising based


51 on the sampling methods employed, this chapter explores population level impacts of interventions aimed at promoting bicycling and walking. By analyz ing possible attitude change in the NTPP communities, it can be determined if there have been measurable changes in attitudes toward NMT and whether population based surveys can be used to capture such changes. It is beyond the scope of this research to link attitude change to specific infrastructure investments or policy changes in individual communities. It is possible to broadly determine the extent and direction of attitudes toward NMT in four cities which received considerable federal funding to prom ote these modes, compared to a control city that did not receive additional funds for promotional activities. For a complete description of the NTPP including specific interventions completed in each community see Chapter 2. Data The NTPP community survey s are employed to identify attitude change in the NTPP cities between 2006 and 2010 The surveys use a quasi experimental research design and probability based sampling frame, with control S urveys were administered in each of the four pilot communities (as well as in Spokane, Washington, the control city ) in 2006 (Phase 1), at the start of the program, and then again in 2010 (Phase 2). Repeat cross sectional data is rare but frequently called for in non motorized research ( Forsyth, et al., 2009) It pres ents a unique opportunity to assess trends and possible impacts of


52 NMT interventions across time, rather than through much more widely utilized cross sectional data sources. It is hypothesized that while behavior change may not have been observed in the NTPP surveys (for a variety of practical reasons, described in Gotschi, et al., 2011), attitude change should be assessed to determine whether or not any critical impacts of NT PP activities to promote bicycling and walking have gone unnoticed. This chapter consists of a statistical analysis of the attitude variables that measure perceptions regarding: (1) barriers to bicycling and walking, (2) infrastructure and land use charact eristics that may support or d iscourage bicycling and walking to answer the question: have attitudes toward bicycling and walking changed in the NTPP cities between phase 1 and phase 2? The NTPP Community Survey This section describes the sample populati on's demographic characteristics and sample sizes across NTPP cities and phases of data collection. The community surveys employed a probability based sampling frame to administer telephone surveys. Certain neighborhoods were oversampled to reach a minimum number of NMT users in areas designated for NMT interventions The oversampling led to some geographic bias (as most of the oversampled census tracts were located near downtown centers or universities) as well as a sample that is not entirely representati ve of the overall populations (Krizek, et al., 2007). The sample is generally higher income, less racially diverse and wealthier than census population data (Krizek, et. al., 2007) would suggest.


53 Responses have not been weighted to reflect actual populati on characteristics as it is geographically representative of those intended to be reached by NMT interventions. The NTPP communities are each unique and approached allocation of NTPP funds with specific issues, goals, and objectives in mind (described at length in Chapter 2). This quantitative evaluation is the first step at beginning to link NTPP interventions to population level outcomes. This research is aimed at informing future interventions within each of the NTPP communities, as well as offering br oad recommendations for other communities in the US. Sample Population Characteristics T he NTPP community surveys include a variety of demographic questions at the individual and household level. Sample population is 43.6% female (and 56.4% male), and 9 2.4% white; 38.9% of the population is 65 years of age or older (and 82.5% of the population is 45 or older), and 68.3% of the population reported a n annual household income of $35,000 or more ( Table 6 ) Census data on overall population descriptives in ea ch city is presented in Chapter 2. Binary variables (i.e., gender and race) were included in Table 6 to identify significant differences between communities but mean scores for these variables are not meant to be interprete d directly (see Table 23 for variable coding scheme).


54 Table 1 Descriptive Statistics by Phase of Data Collection Phase All Columbi a Marin Minneapolis Sheboygan Spokane Sample size 1 1514 313 272 343 297 289 2 1807 342 317 406 349 393 Mean income a 1 4.33 3.96 5.65 4.29 4.23 3.82 2 4.38 3.97* 5.41 4.44 4.27 4 Gender (% female) b 1 52.5 55 51.1 49.9 51.5 55.4 2 43.6*** 46.6* 47.9 42.4* 46.8 36*** Mean age a 1 51.77 48.57 56.48 49.34 53.29 52.19 2 58.36* 54.85 62 58.18 58.29 58.84 Mean household size a 1 2.27 2.31 2.25 2.04 2.54 2.25 2 2.18*** 2.18 2.16 1.99 2.25*** 2.32 Mean number of children a 1 0.55 0.63 0.5 0.44 0.65 0.52 2 0.44*** 0.48* 0.39* 0.34* 0.46*** 0.53 Mean years at current residence a 1 25.73 18.7 23.94 24.56 35.32 26.59 2 32.97* 24.25 30.14 33.93 42.7 33.29 Race (% White) b 1 94.3 92.4 95.5 91.8 97.6 94.6 2 92.4* 85.5* 93.8 91.1 97.6 94.2 a = Significance test: Independent samples t test b = Significance test: Pearson's chi square = p < .05; *** = p < .001 The NTPP community surveys measured perceptions during an NMT trip. Respondents answered nine questions measuring attitudes and perceptions regarding a specific NMT t rip that they completed (Table 7 ). Also included are nineteen questions measurin g general perceptions of a variety of neighborhood characteristics ( Table 8 ) and twenty variables measuring general factors that may promote or discourage bicycling and walking (Table 9 ) Reference trip responses were coded as binary variables, while general attitudes and perceptions were measured on a 4 point Likert scale (1=not likely 4 very likely).


55 Table 2 Reference Trip: Attitude/Perceptions Variable Phase All Columbia Marin Minnea polis Sheboygan Spokane Walking reference trip (causes for concern; % "yes") Not enough crosswalks 1 11 21.7 4.2 2 10.3 17.3 2 10.5 14.1 9.8 7.1 5.9 15.8 Not enough sidewalks or paths 1 16.7 30.9 15.8 3 16.5 18 2 12.9 10.9*** 13 8.2 9.8 22.9 Not enough lighting 1 17.6 18.1 11 16.7 22.1 20 2 15.1 18 11 13.8 16.3 16.3 Drivers not stopping for you to cross 1 24.3 29 13.7 22.4 26.8 29.3 2 20.9 18.5 18.5 25.3 16.8 25.3 Bicycling reference trip (causes for concern; % "yes") Motorist behavior 1 50.3 57.1 54.9 49.2 41.4 50 2 41.3* 41.7 50.7 39 37.1 41.1 Roads too narrow or too much traffic 1 44.6 45.1 60 38.3 37.1 46.8 2 33.6* 35.7 41.2* 23.2* 33.3 37.8 Roads or bike paths not well maintained 1 25.6 31.4 30 16.7 22.1 28.9 2 24.7 23.8 27.9 21.2 28.7 22.7 Particularly problematic intersection 1 28.9 34.3 31.4 29.5 24.6 23.4 2 28 25 36.8 26.3 25 29.2 Route runs through unsightly or unsafe place 1 11.1 4.3 5.9 11.5 13.2 23.4 2 9 10.8 8.8 6.1 6.2 13.6 Significance test: Pearson's chi square = p < .05; *** = p < .001 Methods and Findings Measuring changes in attitudes began by first examining descriptives of all attitude variables in the community surveys. Second, statistically significant differences in mean values in all five communities (and the total sample population) were tested using Pearson's chi square test and independent sample t tests ( r esults reported in A ppendix B ). Difference in mean tests allowed for preliminary exploration of attitudinal data across phases, but presented two challenges (1) it is difficult to identify trends across numerous survey locations, dozens of variables, and direction of change over time, and


56 (2) it is not possible to account for influence of covariat es. To account for these two challenges, dimensional reduction techniques as well as linear, logistic, and ordered logistic regression models were employed (based on the measurement scale of the dependent variables). The methods used in this research are e xploratory and conservative in nature to reflect the objective of this research: offering empirical evidence to support the inclusion of attitude measures as impacts of interventions aimed at promoting NMT. Difference in Means and Regression The nine va riables measuring attitudes and perceptions of a reference NMT trip were tested for changes from phase 1 to phase 2, as were the 39 variables measuring general attitudes and perceptions of NMT. Differences in reference trip variables (measuring on a binary scale) were found to be largely insignific ant with few exceptions (Table 7 ); however, it is instructive to note that responses to barriers generally trended toward "no" in phase 2, indicating the possibility of a modest trend toward reduced negative perce ptions of specific trips by NMT travelers. Significant attitude changes were first tested on the 39 variables measuring neighborhood perceptions and attitudes toward factors that may support or discourage bicycling and walking. Results of independent sam ple t tests of the 39 variables measuring general attitudes and perceptions of NMT provide a diverse picture of possible attitude change across the entire sample and within each city ( see Table 24 ). These tests yielded a number of statistically significant shifts in individual attitudinal variables, but


57 the number of tests and variety of findings precludes their usefulness in addressing the primary research question. Linear regression was then used to account for the influence of each city of residence, p hase of data collection, socio demographic variables, and interaction effects. The 39 attitudinal variables were examined as dependent variables in linear regression models, the linear models are of the form (coding scheme for independent variables can be found in Table 23 ): Attitude Variable = Constant (Spokane) + Columbia + Marin + Minneapolis + Sheboygan + Phase + Income + Gender + Age + #Children + Race + Phase City of residence + Phase*Income + Phase Gender + Phase Age + Phase*#Children + Phase*Race Respondents were identified by their city of residence through binary indicator variables, and the remaining independent variables in the model are e ither binary or ordered categories. The impact of phase of data collection in each city was operationalized as interaction effects of each city phase. Methodologically, the interaction specifically addresses the possible effect of the NTPP on attitudes i n each city individually. The city phase interaction represents a moderating term (Baron & Kenny, 1986) conceptualized in Chapter 2.


58 Using four indicator variables for the five cities represented in the NTPP data also allows each city to have a differ ent mean value for each dependent variable, in every analysis. This approach is not quite as flexible as a fully nested model, but is appropriate given the exploratory goal of this work: to identify measurable changes in attitudes during the same time peri od as the NTPP. A nested approach would be appropriate for future work aimed at identifying within group attitude change differences, but statistically significant attitude changes should first be identified. [ Note: Given the relatively low R 2 values in th e models described in detail in the Findings section this is the conservative approach.] Interaction effects between phase and city of residence, phase and demographic variables, as well as gender and demographic variables were tested in the above mod el. The interaction terms were included to test the hypothesis that NMT promotional activities in some (or all) of the NTPP cities could have impacted various populations differently, possibly working to improve attitudes toward NMT among some populations while marginalizing other populations. The regression models were refined through a manual backwards stepwise process (i.e, rather than allowing the statistical software to automatically reduce the model). The resulting models were largely a poor fit (R 2 values < .09), and it was difficult to detect a pattern in the results (see Table 25 for individual model details). Because the attitude dependent variables are measured on a Likert scale, 39 ordered logistic regression models of the same form as the line ar models were run to crosscheck results.


59 The ordered logistic regression models yielded similarly inconclusive findings (i.e., poor R 2 values and lack of discernable pattern in results). Taken together these efforts confirmed the need for a dimensional re duction technique to decrease the amount of noise' in the data. Dimension Reduction Two Principal Component Analyses (PCA) were then conducted (in SPSS v.20): one on each subset of the attitude variables using an orthogonal (i.e., varimax ) rotation to minimize the complexity of any identified components. The first PCA (neighborhood perception variables) yielded a single factor explaining about 1/3 of the variance in data (29.9% variance explained) with all other factors accounting for less than 10% of variance. This factor represents positive perceptions of neighborhoods bicycling/walking accessibility factor (Table 8 : Neighborhood Perception factor).


60 Table 3 Neighborhood Perception Factor Statement Loading Stores are within easy walking distance of my home .562 There are many places to go within easy walking distance of my home .627 It is easy to walk to a transit stop (bus, train) from my home .465 It is easy to bicycle to a transit stop (bus, train) from my home 616 The streets in my neighborhood are hilly, making it difficult to walk* .323 There are sidewalks on most of the streets in my neighborhood 663 The sidewalks in my neighborhood are well maintained 601 There are pedestrian trails in or near my neighborhood that are easy to get to 661 There are crosswalks and pedestrian signals to help walkers cross busy streets in my neighborhood 692 My neighborhood streets are well lit at night .536 The crime rate in my neighborhood ma kes it unsafe to go on walks during the day* .063 Major streets have bike lanes .535 The city has a network of off street bicycle paths .564 Streets without bike lanes are generally wide enough to bike on .474 There are bike lanes, paths or routes that connect my home to places that I would like to ride to 696 The bike route network has big gaps* .232 Bike lanes and paths are free of obstacles 673 Stores and other destinations have bike racks .406 Intersections have push buttons or sensors for bicycles and pedestrians .534 variable re coded to account for reverse worded items Loadings greater than .6 in bold The second PCA (factors impacting bicycling and walking use) was completed on two sets of similar questions one set regarding walking, then other cycling focusing on how likely a variety of factors would be to "get you to walk/bike more than you curren tly do." The resulting PCA found that one factor explained nearly half the variance in the data (45.4%) and all other factors accounted for less than 10% of variance individually. All questions relating to bicycling loaded heavily on the first factor, whil e walking variables consistently loaded less highly, thus creating a "cycling barriers" factor (Table 9 : Cycling Barriers factor). While such a finding would initially suggest that the variance in the data could be best explained by two factors (one loadin g heavily for each mode, bicycling and walking), it instead appears that respondents generally reported stronger feelings toward factors impacting bicycling.


61 Table 4 Cycling Barriers Factor Statement Loading "How likely are the following factors to get you to walk more often": More sidewalks .578 Better conditions on sidewalks .603 Safer intersections .676 Areas free from crime .648 More lights in walking area .637 Areas free from fast moving traffic .674 The cost of parking and driving increased .533 More destinations close to home .628 More destinations close to work .555 If I had to pay to park my vehicle .511 If parking was hard to find .558 "How likely are the following factors to get you to bike more often": More marked bike lanes on existing streets .764 More off street bike paths .749 More lights on existing bicycle facilities .764 Safer intersections (with regard to motorists) .807 Safer or better bike parking .789 Showers available at my destinations .524 Motorists who obey traffic laws .772 Areas free from crime .762 Areas free from fast moving traffic .797 Loadings greater than .7 in bold Variables were then created from the primary factors identified in each of the above PCAs (substituting means for missing values). Difference in mean were tested prior to further analysis yielding statistically significant differences in each factor from p hase 1 to phase 2 (Table 10 ), and the direction of the difference supports the hypothesis that attitudes shifted in expected directions over time.


62 Table 5 Attitude Variable Descriptives Results of Independent Samples T tests Latent Variable Mean (SD) t df Phase 1 Phase 2 Neighborhood Perception Factor 0.0367 (0.937) 0.0308 (1.041) 1.949* 3319 Cycling Barriers Factor 0.0377 (1.127) 0.0315 (0.836) 2.029* 3319 = p < .05; *** = p < .001 The factors created through the two PCAs each require distinct interpretations. The "neighborhood perception factor" can be interpreted as higher values equally more positive attitudes and perceptions of a neighborhood for NMT, as it is created from questi ons measuring the degree to which respondents agree with positive statements about their community. In contrast, the "cycling barriers factor" requires an opposite interpretation. This factor is drawn from variables measuring the likelihood that an individ ual would engage in NMT, given an improvement in a variety of factors. Thus, the more negative a score is for this factor, the less a respondent feels there are immediate barriers to their use of NMT. Each attitude factor was then operationalized as a de pendent variable in a linear regression model identical to the linear model presented above, except that the dependent variables in the subsequent analyses are latent variables created through the two PCAs. The final regression models reported below were a lso refined using a manual backwards stepwise process. The resulting regression models could easily substitute a measure of attitude for a self reported or observed behavior measure, and are meant to provide an example of how the analysis of attitude chang e could be undertaken given similar data.


63 Results Table 6 Model 1: Neighborhood Perception Factor Regression Unstandardized Coefficients Standardized Coefficients t statistic p SE (Constant) .060 .074 .815 .415 Columbia .114 .072 .046 1.586 .113 Marin County .432 .060 .160 7.217 .000 Minneapolis .879 .053 .370 16.561 .000 Sheboygan .022 .055 .009 .398 .691 Phase .279 .091 .139 3.053 .002 Income .048 .014 .092 3.366 .001 Phase x Income .060 .018 .158 3.262 .001 Phase x Columbia .298 .088 .092 3.382 .001 R 2 = .13 Statistical Model 1 Each of the regression models tests the correlation between phase of data collection and a latent measure of attitudes, while controlling for the effect of a variety of other factors. The dependent variable in model 1 measures an individual's perceptions of the quality of their neighborhood for NMT, and perceptions are more positive as the score increases. Results from the regression analysis of the neighborhood perception factor ( Table 11 ) show that perceptions of NMT use, when controlling for all other variables, may be generally more positive in Minneapolis and Marin County than the other NTPP cities, which did not differ significantly from the co ntrol (i.e., "constant"). However, the interaction of Phase*Columbia suggests a significant and positive correlation with neighborhood perception from phase 1 to phase 2. Generally higher perceptions can be interpreted as, all else being equal, respondents from these two


64 communities on average scoring their community as more amenable to NMT than the other communities surveyed. The primary variable of interest, phase, appears to be negatively correlated with the dependent variable. That is, phase 2 respons es are negatively correlated with positive attitudes towards NMT (phase unstandardized coefficient: .190). The effect of income is similarly negative; as income levels rise, there is also a negative correlation with NMT (Income unstandardized coefficient: .044). These findings, while initially suggesting that any NMT promotional activities had the opposite of their intended effect, are complicated by the interaction term. When examined alone, both phase and income suggest that any promotional activities occurring in the NTPP communities negatively impacted attitudes toward NMT; however, the direction of the interaction term (phase income) is the opposite of the direction of the main effect of each variable individually. The graph below ( Figure 4) illus trates this interaction and points to some interesting interpretations. One such interpretation is that there may be an equity issue associated with promoting NMT; interventions aimed at promoting walking and cycling, while possibly resulting in a positive impact on attitudes toward neighborhood perceptions, may affect different income groups differently. For example, the interaction effect shows that for respondents in households earning greater than approximately $30,000 annually, interventions appear cor related with improved perceptions. This effect is then inverted for those earning less than $30,000 annually. However, given the overall low R 2 value and the fact that the


65 dependent variable is a latent factor drawn from a survey not specifically designed to measure hypothesized latent attitude measures, such conclusions should be treated cautiously. Figure 3 Phase x Income in Model 1


66 Statistical Model 2 Table 7 Model 2: Cycling Barrier Factor Regression Unstandardized Coefficients Standardized Coefficients t statistic p SE (Constant) 1.198 .117 10.196 .000 Columbia .110 .057 .045 1.936 .053 Marin County .016 .063 .006 .258 .796 Minneapolis .098 .055 .041 1.765 .078 Sheboygan .110 .057 .045 1.939 .053 Phase .054 .147 .027 .370 .712 Income .056 .010 .109 5.625 .000 Age .020 .002 .316 11.431 .000 Race .130 .031 .130 4.201 .000 Phase x Age .004 .002 .133 1.788 .074 Phase x Race .102 .038 .105 2.658 .008 R 2 = .10 The dependent variable in Model 2 measures attitudes toward barrier's of cycling. Lower values on the dependent variable should be interpreted as fewer barrier's to cycling. Thus, decreases in the mean value of the dependent variable (Table 12 ) and negativ e coefficients of covariates may support the hypothesis that attitudes toward NMT improved over time (i.e., perceived barriers to NMT use decreased). Model results suggest that cycling attitudes across the NTPP communities do not significantly differ from the control city (at the p < .05 level); however, three communities differ significantly at the p <.1 levels. In contrast to the analysis of the neighborhood perception factor (Model 1), Model 2 suggests that attitudes specific to bicycling do not exhibit a


67 significant correlation with phase of data collection; however, similar to findings in Model 1, the interaction effects offer insights into possible impacts of phase across specific populations. Findings confirm the hypothesis that different population s may hold different attitudes toward NMT. Because Phase of data collection is of primary interest to this study, and a number of socio demographic variables (income, age, race) are highly significant in the model, interaction terms were modeled to test th e moderating effects of individual characteristics on phase. The interaction between phase of data collection and race appears to be significant in model 2 (Figure 5 ). This interaction, while significant, illustrates a relatively modest correlation between phase of data collection and a reduction in perceived barriers toward cycling among white respondents. In contrast, reduced barriers toward cycling from phase 1 to phase 2 among non white respondents may be greater, but perceived barriers toward cycling a re still stronger among non white populations. Caution should be used when interpreting this interaction: first, these findings are limited by the small sample populations of non white racial/ethnic groups ( Table 6 ). Second, modeling diverse populations ac ross diverse survey locations with binary indicators over simplifies numerous complicating factors that may impact attitudes (and change in attitudes over time).


68 Figure 4 Phase x Race in Model 2 Discussion Including attitude change as a new metric of NMT intervention evaluation may be a valuable tool for understanding intervention impacts, but there are some important qualifications and limitations of this research. Impacts of interventions on attitudes and perceptions may be highly susceptible to individual context and socio demographic characteristics, and only a limited number of these factors could be accounted for in the data. Also, while theory dictates that attitude change precedes behavior change, the


69 attitude behavior relationship could also operate as a feedback loop in which attempting a behavior informs attitude creation. Further complicating this is the unknown lag effect between intervention implementation, attitude change, and behavior change. It is not possible to link specific interventions in any of the NTPP communities to specific changes in attitudes or perceptions regarding NMT because it is unclear exactly what interventions were put in place in each of the communities between phase 1 and phase 2 surveys. The choice to operationalize city of residence as indicator variables, while increasing parsimony in regression, treats each city as equivalent. In the absence of precise and comparable data on the quality, quantity, and type of interv entions put in place in each city between phase 1 and phase 2, this was the pragmatic approach to identifying general trends in attitudes across the sample population. Additionally, if a complete inventory of infrastructure improvements, policy changes, an d social programs was available, then it would be possible to better address the lag effect issue. Conclusions and Future Directions This paper proposes that attitude changes be considered and measured as impacts of interventions aimed at promoting walki ng and cycling. Hypothesizing that population based surveys, prized for their ability to provide generalizable results from representative populations may be well suited to capture attitude shifts (as opposed to changes in behavior among small sub sets of the population). The methods section in this chapter provides an overview of how such an analysis of attitude change at the population could be conducted, and the results of this research offer preliminary support for the hypothesis.


70 This research demonstr ates that it is possible to measure attitude change in a general population over a relatively modest time period (4 years, in this case). In accordance with theories such as the TPB and trans theoretical model, attitude change precedes behavior, and this r esearch is the first step in testing that relationship for promoting NMT use. Despite limiting factors, there are numerous avenues for future research. First, NMT intervention research using robust research designs is limited (Ogilvie et al., 2007; Yang,, 2010) This analysis of the NTPP data confirms that non probabilistic methods may be better suited for behavior change analysis, but population based, probability sampled surveys should be studied further for their abil ity to measure changes in attitudes. Such studies could confirm that panel data, while ideal, may not be required to measure certain impacts over time. Second, attitude measurement may serve as a proxy for undetectable behavior change, a measure of latent demand, or a predecessor to behavior change. Each of these possibilities deserves further attention. Accurately assessing latent demand for NMT use could prove invaluable for procuring support and funding for bicycle and pedestrian projects in communities. Understanding the strength of the relationship between attitudes and behaviors, as well as the lag effect, could inform future time series evaluation. Individual socio demographic characteristics may be spatially correlated and interventions in each of the communities surveyed may also be geographically biased. Geography, neighborhood demographics, and the degree of empowerment in a community are inextricably linked to where interventions are put in place, and can in turn


71 compound feelings of marginaliza tion among already marginalized populations. Further study of this phenomena is essential to communities seeking to promote bicycling and walking across a city, not only among populations already inclined to use these modes. Identifying measurable behav ior changes in NMT use is fraught with practical challenges (Krizek et al., 2009) thus this chapter argues that measuring changes in attitudes may precede behavior change, or serve as a proxy measure of behavior change. The research is informed by previou s literature operationalizing elements of the TPB as correlated with NMT use, but is novel because it considers individual attitude change as an impact unto itself of interventions aimed at promoting NMT. The quantitative analysis indicates that, in the ab sence of measurable changes in behavior in the cities surveyed, there is a modest and statistically significant correlation between attitudes regarding bicycle/pedestrian accessibility and the time period of the Non motorized Transportation Pilot Program.


72 CHAPTER IV CARROTS, STICKS, AND NON MOTORIZED TRANSPORTATION: TESTING THE MEDIATING EFFECT OF DISCOURAGING DRIVING ON PROMOTING WALKING AND BICYCLING Introduction Practitioners and researchers in the United States have been striving to identify appropriate and cost effective interventions (programs and infrastructure investments) to promote walking and bicycling non motorized transportation (NMT) Evaluation research frequently focuses on the efficacy of specific interventions that pro mote walking and cycling (e.g., a high quality bike/ped crossing or a "bike to work day" event). Modest interventions have modest impacts on NMT use, and encouraging NMT without discouraging the primary alternative driving may be of limited impact. R es earch offers some guidance into the efficacy of policies that discourage driving but is much more limited in addressing the impact of combined policies that encourage one behavior while discouraging another It is possible that promotional interventions "carrots" may be inherently limited in their effectiveness because they only offer encouragement for the targeted mode choice, without increasing barriers to alternatives. In this chapter, I hypothesize that truly effective NMT interventions should incl ude "sticks" that dis incentive driving (the most common alternative to NMT in the US), either as stand alone interventio ns or in concert with promotional carrot interventions. Together, targeted interventions that utilize carrots to promote NMT use an d sticks to discourage driving may be the most effective at inducing behavior change.


73 Public health and psychological theories support the notion that individual behavior change is predicated on a number of factors, including perceived barriers and benef its (Bauman, et al. 2002) Thus, leveraging carrots to promote one activity in concert with sticks to discourage alternatives may be most effective at encouraging NMT use (Krizek, et. al. 2009) There are numerous examples of such combined interventions from the Community Based Social Marketing (CBSM) field, but these examples lack rigorous methods and peer review. Researchers have successfully evaluated combined interventions in Safe Routes to Schools (SRTS) programs largely supporting the notion that combining carrots, such as safer crossings, bicycle lanes, and social programs, with sticks such as reduced speed limits can significantly increase walking and cycling to school (Levin Martin, et al. 2009) T he political climate in the US can be hostile to any sticks that may limit or discourage the convenience, affordability, and accessibility of the automobile in the general population A few cities and regions, due to a combination of public support, political will, and various land use decisi ons have enacted stick policies that explicitly discourage auto use ( e.g ., the parking "cash out" program in Los Angeles CA (Shoup, 1997) ) or may implicitly impact travel behavior and physical activity (e.g., urban containment policies (Aytur, et al. 200 8; Rodriguez, et al. 2006) but these communities are outliers. For the majority of the US, there are few sticks in place to discourage driving, and a lack of barriers to driving may hinder the growth of non motorized mode share among the general populatio n.


74 In the absence of representative cases to test the impact of carrot and stick interventions on travel behavior in the general population this research tests the interaction between attitudes toward mode choice options Specifically examining perceive d mode choice options walk, bike, or car to gauge the degree to which individuals would consider using NMT based in part on their perceptions of the cost and convenience of driving ; that is, does a perceived convenience of driving mediate perceived bar riers to walking and cycling? Attitudes and perceptions of NMT have been shown to be positively correlated with NMT use, and unlike NMT behavior, can be more easily measured in a general population (see Chapter 3) By examining attitudes and perceptions of NMT, and it s primary alternative, driving, the research presented in this chapter can shed light on attitude formation and the types of attitudes and perceptions interventions should target to best promote NMT as a viable mode choice. This research assu mes that p erceptions of mode options are formed through a combination of where you live, who you are (i.e., socio demographics), and your perceptions of al ternative mode choices. The primary research question and hypotheses are: Question : Is there an interrelationship between attitudes regarding driving and attitudes toward walking and cycling? a. Hypothesis 1 : Testing a mediation effect perceptions of driving have a mediating effect on how socio demographic factors impact your perceptions of NMT as via ble transportation options.


75 b. Hypothesis 2 : The direction of the effect the perceived inconvenience of driving predicts reduced perceived barriers of walking and cycling The above question and hypotheses are tested using data from the Non motorized Transp ortation Pilot Program (NTPP) First, the literature on carrot, stick, and combined interventions are reviewed and a theoretical rationale for a mediation relationship is presented. Second, the mediating effect of perceptions of the convenience of driving on perceived barriers to walking and cycling is tested Third, recommendations are offered for targeting future NMT interventions, and the relative importance of discouraging driving in increasing the perceived viability of walking and cycling. Literatur e Review While research has identified correlations between attitudes and NMT there is a lack of studies examining the impacts of how perceptions of the convenience of auto use may impact perceptions of NMT. NMT literature reports the impacts of individual interventions on NMT attitudes or behaviors ( Krizek et al., 2009; Ogilvie et al., 2007; Pucher, et al. 2010; Yang, et al. 2010) Environmental behavior literature and theory suggests that interventions are most effective when they include both carrots and sticks incentives as well as disincentives (Geller, 2002). This chapter aims to provide empirical support for the impact of combined interventions on the general population (rather than targeted sub sets of the population). Empirical findings will a lso have implications for theory testing as numerous theories from psychology, public health, and transportation


76 hypothesize a mediating relationship (or similar interrelationship) of perceived barriers to one mode impacting the decision to use alternativ e transportation modes. Non motorized Transportation Interventions Carrots Existing research on interventions that promote NMT can be categorized as either micro or macro studies. Micro studies typically assess the impact of a specific intervention be it a policy or an infrastructure intervention. Interventions can be categorized as "hard" (i.e., infrastructure) measures, or "soft" (i.e., policy) measures (Krizek et al., 2009) Examples include analyses of the impact of a new bicycle path a hard measure on cyclist behavior (Krizek & Johnson, 2005) or the effects of a bike to work day event a soft measure on bicycle commuting (Rose & Marfurt, 2007) Micro studies of spe cific policies, programs, or infrastructure investments find consistent significantly positive correlations between such inte rventions and NMT behavior (Pucher et al., 2010) but such findings are commonly based on non probability sampled populations. In general, small, individual interventions likely have little impact on overall NMT, but this remains an outstanding question for researchers ( Krizek et al., 2009) Macro studies, in contrast, tend to take a system based or city scale approach to factors (i.e., interventions), such as measures of bike/ped accessibility, or density. Many of these studies also focus entirely on promotional aspects of interventions to guide future work; this is likely due to the proliferation of carrots to be studied, a nd the relative lack of sticks. Metrics and outcomes of these studies focus on impact on auto use rather than their impact on alternative mode choices Neighborhood design and accessibility


77 measures have been used to assess mode choice decisions (Joh et al., 20 08) encouraging increased walking, (Boer, et al. 2007) and bicycling (Barnes & Krizek, 2005; Plaut, 2005) A variety of density measures have also been applied to mode choice and vehicle miles traveled, with moderately positive results for reducing car u se (Ewing & Cervero, 2010) and encouraging NMT (Saelens, et al. 2003) Interventions Aimed at Dis incentivizing Driving Sticks Another relevant thread in the NMT literature evaluates the impact of a variety of policy measures as sticks (Pucher, et al., 2011 ; Bax, 2011 ). S tick policies, such as costs associated with owning or driving a car, are difficult to accurately quantify. Sticks can be operationalized through proxy measures, such as computing costs of driving as a function of gas mileage and ga s prices at the pump ( Buehler, 2010) Parking policies have long been cited as appropriat e sticks to discourage driving (Shoup, 2004; Willson & Shoup, 1990) and free parking is frequently associated with increased odds of driving (Buehler, 2012; Piatkowsk i & Marshall 2013). California's "cash out" parking subsidy was found to reduce single occupancy vehicle work trips by 17% (Shoup, 1997) ; however, the "cash out" program incentivized not driving by offering cash in lieu of a parking subsidy and thus funct ion as a carrot This review found no examples of parking policies used as sticks in the literature. Tolling schemes to mitigate congestion, as with other proposed stick interventions, are rare and controversial due to issues of social equity and politic al controversy (Ben Elia & Ettema, 2009) Urban containment policies (e.g., growth


78 boundaries) may also be effective policy sticks to alter land use and promote physical activity through NMT (Aytur, et al. 2008) Sticks alone may have a significant influe nce on mode choice, but the literature also suggests that combining carrots and sticks may be most effective ( Krizek et al., 2009; Pucher et al., 2011) Empirical evaluation of combined approaches in the transportation literature is rare, with the exception of evaluation Safe Routes to Schools (SRTS) programs. SRTS programs began in 2005 with the passing of national legislation funding states to support safer and healthier school travel (Levin Martin et al., 2009) SRTS funds may be used for hard o r soft measures and can include infrastructure improvements carrots such as sidewalk connections, street crossing, bicycle lanes, as well as safety patrols, walking school busses, and positive social messaging (Levin Martin et al., 2009) Sticks employ ed in SRTS programs are typically modest in their actual impact on dis incentivizing driving. They frequently involve traffic control measures around schools, speed zones, crossing guards, etc. (Chriqui et al., 2012) interventions focused on traffic cal ming and improving safety in and around neighborhood schools. Sticks as part of SRTS programs can include state level traffic calming legislation (Chriqui et al., 2012) There are many possible reasons why carrot and stick studies in the NMT literature are limited to SRTS evaluations. First, while sticks to discourage driving are extremely contentious, when framed as improvements to children's safety and health, such measu res are much more acceptable to the general public and elected officials. Second, policy sticks "soft" measures require minimal monetary investment


79 compared to infrastructure improvements to local bike/ped networks, and may be more attractive options t o SRTS administrators working with limited resources Third, there are fewer methodological barriers to SRTS evaluation: programs are geo graphically constrained, target populations are readily identifiable and accessible, and these populations can be obse rved multiple times with minimal attrition rates. Taken together, SRTS evaluations are instructive examples of the combined impact of car rot and stick interventions. When NMT interventions are examined through the lens of carrots or sticks, a significant gap in the literature eme rges. Carrots have consistently modest positive impacts on certain populations, sticks are largely untried and untested, and combining carrots and sticks, while promising, has not been rigorously examined empirically. The research presented in this chapter presents one method of addressing this gap in the literature by testing interrelationships between carrots and sticks. Mode Choice and Decision Theory When considering an individual's attitudes toward driving as a mediating fa ctor in their attitudes toward walking or cycling, it is instructive to consider to mode choices in simple terms of u tility. Utility Theory is a foundational mode choice theory that posits that individuals make rational decisions about their mode choices based on the relative utility of that option compared to other available options. This uses a rational, utility focused framework to draw conclusions about the role of competing attitudes and perceptions of various modes in impacting perceptions of walking and cycling (Handy,


80 1996) The utility of individual mode choice options may change based on carrots or sticks in place meant to promote one mode while discouraging another. A utility based perspective, while helpful in conceptualizing a straight forward and bounded decision between mode choices for a specific trip is limited by this simplicity it does not account for individual or environmental factors factors that in turn impact relative perceptions of barriers to specific activities (Handy, 1996) To address the limitations of utility theory and provide a theoretical basis for a mediation analysis of perceptions of competing mode choices, this work is informed by public health and psychological theories whose relevance to NMT has been demonstrated e mpirically (Sallis et al., 2006) ; specifically, the Health Belief Model (HBM), The Theory of Planned Behavior (TPB), and the Transtheoretical Model (TTM). Each of these theories stresses that behavior change (or behavior intention as in the case of the HB M) is predicated through some a combination of knowledge, norms, intention, attitudes, and perceived barriers and benefits (Bauman et al., 2002) The HBM in particular incorporates a number of factors, including perceived susceptibility to a disease, the perceived severity of the issue/condition, perceived barriers, and (most relevant to this work) perceived benefits (Janz & Becker, 1984; Nisbet & Gick, 2008) The HBM has been operationalized and tested against the Theory of Planned Behavior and Locus of C ontrol, finding specifically that "perceived barriers" a factor explicitly lacking from the TPB and Locus of Control, were strong predictors of the intention to wear a bicycle helmet (Lajunen & RŠsŠnen, 2004)


81 In recent years, the TPB and TTM have been shown to provide a helpful explanatory framework for attitude behavior relationships in travel behavior (Dill, 2010; Heinen, et al. 2011) and physical activity research (Bauman et al., 2002; Jewson, et al. 2008; Lemieux & Godin, 2009) The TPB focuse s on the relative impact of social norms, attitudes, and perceived behavior control, mediated by intention to act (Ajzen, 1991) The TTM has been applied to event based behavior change interventions (i.e., Bike to Work Day events) due to its emphasis the o n process, contemplation, and maintenance (Rose & Marfurt, 2007) Schneider (2013) has proposed a Theory of Routine Mode Choice Decisions to fill a gap left by the TPB and TTM. "The TPB and TTM tend to represent a thought process related to performing a particular behavior (typically a normative goal) rather than choosing between alternatives" (pg. 130). Schneider's theory proposes five steps to mode choice decisions, which interact with individual socio economic factors to inform habit, and in turn, mode choice ( Figure 7 ). Step three of the Schneider model Convenience and Cost presents a trade off between relative barriers and benefits of mode choice options. This step in th e t heory comports with the HBM's hypothesized role of perceived barriers and benefits. The mediating role of convenience and cost has yet to be evaluated qua ntitatively. Taken together, each of the theories described points to an interaction effect or decision making process by which rational actors weigh options based on perceptions.


82 Figure 1 Theory of Routine Mode Choice De cisions (source: Schneider, 2013) Intervention Theories While studies have tested the impact of sticks empirically, this work has been largely atheoretical in nature The environmental psychology literature as well as practitioners of Community Based Social Marketing (CBSM) ( offer a number of suggestions regarding carrots and sticks to promote environmentally friendly behavior. This literature cautions against sticks/dis incentives/penalties and emphasizes the importance of context on in tervention effectiveness. Specifically, the environmental psychology literature supports the use of combined carrot and stick interventions that are context dependent and informed by the individuals or groups they are meant to impact. regionssuchasPortland,OR,NewYork,NY,andWashington,DC, buttherehasbeenlessgrowthinbicyclinginsuburbanpartsof theseregionswhereactivitylocationsaremoredispersed( Pucher etal.,2011 ).Certaininterventionsmayalsobemoreeffectivefor peopleinspecictravelbehaviorsegments( Anable,2005 ; Steg, 2005 ).Forexample,somepeoplemaybemalcontentedmotorists'' whoarefrustratedwiththeirhighlevelofautomobileuseand desiretodriveless.Thesepeoplemaybemuchmorereceptiveto interventionsencouragingbicyclingandwalkingthanpeoplewho arecomplacentcaraddicts''whothinkitisdifculttochangetheir travelbehavioranddonotseeamoralobligationtodriveless ( Anable,2005 ).Broadmodeshiftsrequireaclearerunderstandingof thebarrierstochoosingwalkingan dbicyclingfordifferenttypesof peopleindifferentcommunities. Thepurposeofthispaperistoproposeanoperationaltheory ofthemodechoicedecisionprocessandsupportitwithin-depth, qualitativeinterviewsfromtheSanFranciscoBayArea.This informationisintendedforplanners,designers,engineers,and othertransportationprofessionalswhoarechargedwiththetask ofachievingmodeshiftpolicygoals.Manystrategieshavebeen proposedtochangetravelbehavior,butselectingtheoptimalset ofactionstopursueinaparticularcommunityischallenging. Practitionerscanusetheoperationaltheoryasaguidetounderstandthemodechoiceprocessandidentifyactionsthatmayhave themostpotentialtoincreasewalkingandbicyclingintheirlocal socialandgeographiccontexts. 2.Proposedmodechoicedecisiontheory Thissectionproposesanoperationaltheory,calledtheTheoryof RoutineModeChoiceDecisions,todescribehowpeoplechoose transportationmodesforroutinetravelpurposes,suchaslocal shoppingorothererrands.Thistheorysuggeststhatthereareve stepsinthemodechoicedecisionprocess( Fig.1 ).Therstpart,(1) awarenessandavailability,determineswhichmodesareviewedas possiblechoicesforroutinetravel.Thenextthreeelements,(2)basic safetyandsecurity,(3)convenienceandcost,and(4)enjoyment, assesssituationaltradeoffsbetweenmodesinthechoiceset.These middlethreestepsmaybeconsideredsimultaneouslyorinvarious sequences.Thenalpart,(5)habit, reinforcespreviouschoicesand closesthedecisionprocessloop.Socioeconomiccharacteristics explaindifferencesinhowindividualsvieweachpartoftheprocess. OperationaltheoriesliketheTheoryofRoutineModeChoice Decisionsareusefulbecausetheyc anprovideconcise,understandableframeworkstosummarizepreviousresearchforpractical application.Thistheorydrawsonotherstudiesthatprovideclues tohowpeoplechoosebetweenautomo bile,publictransit,bicycling, andwalkingincertainsituations.Itcombinesndingsfromthetravel behaviorandpsychologyelds,assuggestedby VanAckeretal. (2010) .Thetravelbehavioreldhastraditionallyfocusedontime, cost,andsocioeconomicfactors buthasmorerecentlyevaluated perceptionsofthelocalenvironmentandattitudestowardsspecic modes.Thepsychologyeldhasdescribedthethoughtprocessused toselectatravelmode,includingintentionsandhabits. 2.1.Modechoiceinsightsfromthetravelbehavioreld Walkingandbicyclingtendtobelesstime-competitivewith motorizedmodesoverlongerdistances( CerveroandDuncan,2003 ; Purvis,2003 ; KimandUlfarsson,2008 ),andthesemodesmayhave muchhighertraveltimesthanautomobilesfortripchains(tours)to multiple,dispersedactivitylocations( BowmanandBen-Akiva,2001 ). Plentifulautomobileparkingandlowoperatingcostsalsobenet driving( Rodriguezetal.,2008 ; Krizeketal.,2009 ).Othertravel barrierstowalkingandbicyclingincludetravelingwithotherpeople, heavypackages,hills,andbadweather( CerveroandDuncan,2003 ; Mackett,2003 ; KimandUlfarsson,2008 ).Localenvironmentbarriers topedestrianandbicycleactivityarerelatedtoalackoffacilities(e.g., sidewalks,bicyclelanes,ormulti-usetrails)( DillandCarr,2003 ; CliftonandDill,2005 ; DoumaandCleaveland,2008 ; Handyetal., 2010 ),roadwaycharacteristics(e.g.,fasterautomobilespeeds,higher automobilevolumes,anddifcultstreetcrossings)( Ewingand Cervero,2001 ; Gehl,2002 ),andpublicspacecharacteristics(e.g., sterilebuildingfacades,poorlighting,noise,andfewstreettrees) ( Appleyard,1980 ; Landisetal.,2001 ; Gehl,2002 ; Southworth,2005 ; Ewingetal.,2006 ). Individualandsocialfactorsareal soimportant.Individualfactors associatedwithdrivingratherthanwalkingorbicyclinginclude socioeconomiccharacteristics(e.g.,greaterautomobileownership, physicaldisabilities)( MeyerandMiller,2001 ; CerveroandDuncan, 2003 ),concernsabouttrafcsafety(e.g.,riskofbeingstruckbya vehicle)andpersonalsecurity(e.g.,riskofbeingavictimofcrime) ( Saelensetal.,2003 ; McMillanetal.,2006 ; Handyetal.,2010 ),lack ofawarenessofothertravelmodes( RoseandMarfurt,2007 ),and habitualdriving( LoukopoulosandG arling,2005 ).Somecommunitiesmayperceivepedestriansandbicycliststohavelowersocial statusthandrivers( MokhtarianandSalomon,2001 ; Dugundjiand Walker,2005 ).Yet,therearealsoindivi dualandsocialfactorsthat motivatepeopletowalkandbicycle,suchaspersonalenjoyment (e.g.,physicalexercise,freshair,timetobealone)( Handyetal., 2010 )andconcernfortheenvironment( Kitamuraetal.,1997 ; MokhtarianandSalomon,2001 ).Notethattheinuenceofthese factorsonwalkingversusbicyclingmayvarygreatlydueto differencesintravelspeed,roadw aypositioning,andothercharacteristics( Krizeketal.,2009 ). Althoughtravelbehaviorresearchhasidentiedmanyfactors associatedwithwalkingorbicycling,itisnotclearfromthis literaturehow,when,orinwhatorderthesefactorsareconsideredbyindividualsduringthemodechoicedecisionprocess. Apsychologicallensisneededtounderstandthethoughtprocess involvedinchoosingaparticulartravelmode. 2.2.Modechoicetheoriesfromthepsychologyeld Psychologicaltheoriesfocusonthecognitiveprocessinvolvedin selectingatravelmode.Forexample,theTheoryofInterpersonal Behavior(TIB)suggeststhatmod echoicesdependonindividual attitudestowardsavailablemodesandsocialinuences(similarto enjoyment),habits,andfacilitatin gconditions(e.g.,traveltimeand cost;individualsocioeconomiccharacteristics)( Galdamesetal., 2011 ).TIBcontainsseveralcomponentsoftheproposedTheoryof RoutineModeChoiceDecisions. Fig.1. ProposedTheoryofRoutineModeChoiceDecisions. R.J.Schneider/TransportPolicy25(2013)128137 129


83 The issues most com monly associated with stick interventions address the strength (or severity) of the consequence (the stick), and the impact of those consequences. Geller (2002 ) reviews the efficacy of rewards versus penalties, building on Skinner 's ( 1987 ) claim that behav ior is determined by consequences, but this depends on individual's actually experiencing said consequences ( i.e., rather than perceptions of the consequences ) The "sticks" literature cited above uses policy levers to disincentivize driving, but Geller no tes that such sticks require both extensive promotion and extensive enforcement. Geller references Skinner, (1971) and Brehm (1972)'s asserting that in their most severe forms sticks can be perceived as threats to an individual's freedom and can lead to contrary and opposite behavior. In contrast to the drawbacks of sticks, carrots require minimal oversight, are associated with fostering new social norms, and can be used to both motivate initial behavior change and help to maintain pro environmental behavior. Carrot interventions support positive attitudes, and this positive association with the desired behavior can increase the associated social norm ( Geller, 2002 ). Increase d social norms and incentive based interventions can then in turn foster and support voluntary behavior change. CBSM, utilizing incentives to promote behavior change, has shown promising results, particularly in waste reduction and in home energy use reduc tions ( McKenzie Mohr, 2011 ). While these examples support the effectiveness of carrot interventions, the effectiveness of both waste reduction and energy use schemes (employed in the US and Canada) can be partially accounted for by the carrots put in place as well as the context in which they are carried out. That is, modest decreases in energy use or solid waste can

PAGE 100

84 be attained with very little physical effort, and can result in money savings the existing context already supports (or does not actively d iscourage) this behavior. But Stern et al. ( 1999 ) cautions that understanding the complexities of decision making "is to elaborate on the truism that behavior is a function of the organism and its environment" (pg. 415). The impact of interventions depends greatly on context are the interventions appropriate for the situation and needs of the target population? Stern et al. ( 1999 ) illustrates this by describing an inverse U shaped curve in which, at the top of the U, the attitude behavior relation ship is strongest as contextual factors are neutral, but drops dramatically as contextual forces either positively or negatively compel or prohibit a behavior. The decision to bicycle may be based largely on an individual's attitude in a neutral context, but could vary dramatically when tha t same individual is placed in an auto centric US suburb or compact and bicycle friendly city in Northern Europe When implementing interventions aimed at impacting NMT, understanding context will determine how an inte rvention(s) influences behavior change, as a carrot or a stick ( as the definition of carrot and stick also depends on context ) CBSM terms this a "two pronged approach" to understanding context, articulating intervention goals, and specifying and targeting barriers and benefits ( McKenzie Mohr, 2011 ). McKenzie Mohr describes presents a "two pronged," best practice approach to promoting bicycling to work: bike lanes, bike parking and showers at work may act to reduce barriers to cycling, but commuters may sti ll be enticed by the perceived convenience of driving. Increasing gas

PAGE 101

85 taxes, reducing parking availability, or increasing parking cost could in turn make cycling more attractive a carrot and stick approach on a city scale ( McKenzie Mohr, 2011 ). Recent theoretical advancements in travel behavior support existing public health theories and common practices of CBSM. Namely, decisions are arrived upon based on a variety of factors, including weighing alternatives. This forms the theoretical basis for an em pirical evaluation of the impact of carrots alone, or in addition to sticks. Existing evidence from SRTS programs supports the assertion that NMT promotion is more successful when applied alongside sticks, but lacks empirical analysis of the interaction be tween carrots and sticks. The preceding analysis aims to fill this gap in the literature by determining the degree to which sticks and carrots interrelate. Data This research builds on previous work identifying attitude change as impacts of interventions aimed at promoting NMT (see Chapter 3 for an overview of the sampling frame, descriptive statistics, and earlier analysis of the NTPP data). This chapter employs the same data source for analysis, drawn from evaluation efforts of the Non motorized Transpo rtation Pilot Program. Previous research identified measurable changes in attitudes from 2006 to 2010 as a useful metric that may precede behavior changes in a population, inform estimates of latent demand for NMT, and help to target future interventions. The goal of this research is to inform the latter point: how best to target interventions for maximum impact on attitudes toward NMT.

PAGE 102

86 To inform future targeted interventions that may impact perceptions of carrots or sticks regarding mode choice decisions this research uses a cross sectional analysis of the Phase 2 (2010) NTPP survey data. The Phase 2 data was chosen for a variety of reasons. First, the Phase 2 survey administration protocol had been refined after the Phase 1 (2006) surveys and employed o nly Computer Assisted Telephone Interviewing (CATI) (as opposed to a self mailer survey as well as a CATI survey as used in Phase 1). Second, Phase 2 surveys were not collected late into the winter (as was the case with the Phase 1 survey). Finally, all co mmunities (including the control community) received some form of treatment between Phase 1 and Phase 2 (ranging from outreach and education to policy and infrastructure investment), which is hypothesized to help foster a more informed Phase 2 survey popul ation. That is, by Phase 2 survey respondents are more likely to have formed opinions toward NMT and other mode choice options (note: exploratory analyses were conducted on phase 1 and 2 datasets, with full results reported in Appendix B, Tables 1 and 2). This research aims to uncover how attitudes toward different modes interrelate, thus an informed population provides a more appropriate sample for analysis. Methods This analysis uses structural equation models (SEM) to test a mediating relationship betw een latent factors relating to encouraging walking and cycling, and dis incentivizing driving. SEM is a flexible multivariate statistical method used in travel behavior research since the early 1980s (Golob, 2003) SEM can be used to model a variety of complex travel behavior relationships, including the impact of the built environment (Aditjandra, et al. 2012; Rutt & Coleman, 2005) socio demographics

PAGE 103

87 (Bamberg, et al. 2007; Lu & Pas, 1999; Ren & Kwan, 2009) res idential self selection (Cao, et al. 2009) personality and attitudes (McEachan, et al. 2010; Xing, et al. 2010) intentions ( Yang W allentin, 2004) and perceptions (Ory & Mokhtarian, 2009) SEM is applied herein to model a mediation relationship betwee n latent factors. Exploratory analysis is used to identify latent factors, the factor structure is then confirmed, and finally operationalized in a structural mediation model. This is a novel application of attitudinal data and a SEM model to gain insights into NMT interventions. Factor Analysis The first step in this analysis is to examine the attitude variables for any latent factor constructs (Table 1 3 ) In contrast to the Principal Components Analysis (PCA) first employed in Chapter 3 (which was used as a dimension reduction technique to identify a linear combination of original variables), Exploratory Factor Analysis (EFA) is used in this chapter to parse the shared variability in the data and formulate hypothesized latent variables (Byrne, 2012) A Confirmatory Factor Analysis (CFA) was then conducted to ensure the appropriateness of the latent variables (" barriers to walking ," barriers to cycling ," and convenience of driving") in the SEM testing mediation. To ensure robust factor structure identification, the EFA and CFA were each conducted on unique randomly generated subsets of approximately 50% of the NTPP dataset (a ll analysis was completed using Mplus version 7 )

PAGE 104

88 Table 1 Attitude/Perception Variables Variable Code "How likely are the following factors to get you to walk more often": More sidewalks c7a Better conditions on sidewalks c7b Safer intersections c7c Areas free from crime c7d More lights in walking area c7e Areas free from fast moving traffic c7f The cost of parking and driving increased c7g More destinations close to home c7h More destinations close to work c7i If I had to pay to park my vehicle c7j If parking was hard to find c7k "How likely are the following factors to get you to bike more often": More marked bike lanes on existing streets c8a More off street bike paths c8b More lights on existing bicycle facilities c8c Safer intersections (with regard to motorists) c8d Safer or better bike parking c8e Showers available at my destinations c8f Motorists who obey traffic laws c8g Areas free from crime c8h Areas free from fast moving traffic c8i Kaiser's criterion (select # of factors based on Eigenvalues > 1.0) suggests that there sh ould be no more than 3 factors (Table 14 ). However, latent factor structure identification should be approached with caution to avoid over or under specification thus, Kaiser's criterion provides only a suggested f actor structure Because of this, EFA was completed on 1 4 factor models and each model was considered based on a combination of Kaiser's criterion, model fit indices (fit indices in Table 14 ), and careful review of latent variable constructs in each model (note: factor specification should be guided by the following parameters: SRMR < .05; RMSEA good fit < .08, best fit <.05, unacceptable fit > 1.0; CFI good fit >.90, best fit >.95).

PAGE 105

89 Table 2 Exploratory Factor Analysis: Resu lts and Model Fit Indices The three and four factor models each meet required model fit indices. Upon closer inspection, the four factor model was considered an over specification as the fourth factor identified was made up of two variables ("I would walk more if there were areas f ree from crime", and "I would bike more if there were areas free from crime"), and both variables cross loaded strongly on existing latent walking related and bicycling related factors, respectively. Factors Eigenvalues Chi Square SRMR RMSEA CFI 2 df p 1 10.640 2235.001 170 0.00 0.109 0.087 0.943 2 1.853 1174.290 151 0.00 0.081 0.065 0.972 3 1.677 537.528 133 0.00 0.044 0.044 0.989 4 0.924 314.008 116 0.00 0.034 0.033 0.995 SRMR = Standardized Root Mean Square Residual RMSEA = Root Mean Square Error of Approximation CFI = Comparative Fit Index

PAGE 106

90 Table 3 Exploratory and Confirmatory Factor Analysis Results EFA Model Results CFA Model Results** Variable Rotated Loadings Standardized Coefficients 1 2 3 1 2 3 c7a 0. 834 0.067 0.005 0.660 c7b 0.924 0.088 0. 051 0.721 c7c 0.769 0.012 0.156 0.790 c7d 0.564 0.110 0.222 0.669 c7e 0.646 0.126 0.014 0.699 c7f 0.637 0.066 0.192 0.731 c7g 0.078 0.041 0.702 0.740 c7h* 0.364 0.113 0.398 c7i* 0.369 0.014 0.430 c7j 0.013 0.031 0. 887 0.747 c7k 0.104 0.043 0.814 0.792 c8a 0.112 0.979 0.001 0.831 c8b 0.113 0.936 0.023 0.798 c8c 0.121 0.755 0.043 0.785 c8d 0.094 0.936 0.167 0.857 c8e 0.036 0.816 0.055 0.818 c8f 0.168 0.630 0.205 0.499 c8g 0.005 0.949 0.104 0.828 c8h 0.300 0.586 0.053 0.748 c8i 0.010 0.950 0.046 0.871 omitted from CFA due to cross loading on multiple factors ** all estimates significant at p < .001 Rotated loadings greater than .55 are in bold CFA fit indices can be interpreted identical ly to EFA, and results of the CFA support a three factor model (! 2 = 506.121; df = 132; p < 0.00; SRMR = 0.044; RMSEA = 0.042; CFI = 0.954) (Table 15 ) The three latent factors can be generally interpreted as barriers to walking ," barriers to cycling ," and convenience of driving That is, two factors relating to promotional effects of walking and cycling, and one factor relating to the convenience (alternately, inconvenience) of driving.

PAGE 107

91 Mediation Analysis Overview This research asks the question, d o perceived barriers to driving increase the perceptions of reduced barriers to walking and cycling? SEM is used to conduct a mediation analysis of both the direct effects of independent variables on latent NMT attitude factors, as well as the indirect effec t of independent variables mediated by the latent factor: dis incentivizing driving. The independent variables were chosen based on theory and available data. Ecological models suggest that where you live (i.e., city of residence), as well as who you are ( i.e, socio demographics) impact travel behavior. Additionally, the role of current/regular travel behavior (i.e., habit) also informs your perception of available mode choice options, and willingness to engage in NMT. The research question places the la tent factor, perceived convenience of driving, as a variable that may mediate the relationship between city of residence and socio demographic variables, and the latent factors, attitudes toward NMT. The mediating relationship is ill ustrated in Figure 8

PAGE 108

92 Figure 2 Mediation Conceptual Model. Baron & Kenny (1986 ) state that: "in general, a given variable may be said to function as a mediator to the extent that it accounts for the relation between predictor and criterion" (pg. 1176). The single mediator path diagram (Baron & Kenny, 1986) is the model most commonly applied in psychological research (MacKinnon, et al. 2007) and illustrated in Figure 9 Figure 3 Single Mediator Path Diagram (Source: Baron and Kenny, 1986). Baron and Kenny (1986) describe three criteria by which a variable can be said to function as a mediator:

PAGE 109

93 1) Variations in levels of the independent variable significantly account for variations in the presumed mediator (Path a) 2) Variations in the mediator significantly account for variations in the dependent variable (Path b) 3) When Paths a and b are controlled, a previously significant relation between the independent and dependent variables is no longer significant, with the strongest demonstration of mediation occurring when Path c is zero In th e present analysis, step 1 (above) would test the degree to which socio demographics predicts perceived convenience of driving. Step 2 tests the predictive relationship of convenience of driving on the latent variables, barriers to walking, and barriers to cycling, individually. Step 3 tests the extent to which a socio demographic variable's impact on perceived barriers to walking or cycling depends on (i.e., is mediated by ) perceived convenience of driving. Consider the example of income impacting barriers to walking and cycling. The mediation model first tests if income is shown to significantly predict perceived convenience of driving (path a), as well as perceived barriers to walking and cycling (path c). Then, if convenience of driving significantly predicts perceived barriers to walking and cycling (path b), and the mediating path of income > driving > walking/cycling is significant, then there is evidence of partial mediation. Full mediation

PAGE 110

94 (i.e., in which the relationship between income and walking/cycling is only significant through a meditational pathway), would be present if path c were no longer significant. The mediation path diagram (Figure 9 ) is a theoretical mediation model. Mackinnon, et al (2007) present s an empirical mediation model that corresponds to an analytic approach to mediation (Figure 10 ). MacKinnon's empirical model illustrates the most popular method of mediation analysis, the "causal steps" approach that Baron & Kenny (1986) d escribe in theoretical terms. Figure 4: Empirical Mediation Model Figure 4 Mediation Conceptual Model. Empirically testing mediation using the causal steps approach require using information from three regression equati ons: 1) Y = i 1 + c X + e 1 2) M = i 2 + a X + e 2 3) Y = i 3 + c' X + b M + e 3

PAGE 111

95 Where i 1 i 2 i 3 = intercepts Y = dependent variable X = independent variable M = mediating variable a = coefficient of the dependent variable on the potential mediator b = coefficient of the potential mediator to the dependent variable c = coefficient of the dependent variable to the independent variable c' = coefficient of the dependent variable to the independent variable adjusted for the mediator e 1 e 2 e 3 = residuals (Soure: MacKinnon, 2004) From the above regression equations, four steps are used to determine mediation (Baron & Kenny, 1986; MacKinnon et al., 2007) : 1) The independent variable predicts the dependent variable (Equation 1) 2) The independent variable predicts the potential mediator (Equation 2) 3) The mediator predicts the dependent variable controlling for the independent variable (Equation 3) 4) [optional] If the effect of X on Y is not significant with the inclusion of M, full mediation is said to have occurred, but if X on Y is significant, partial mediation is said to have occurred

PAGE 112

96 In social science research, there are few rigid guidelines when identifying and testing mediation through statistical methods (MacKinnon et al., 2007) "Mediation is not defined sta tistically; rather statistics can be used to evaluate a presumed meditational model" (Kenny, 2013). Baron & Kenny caution that while a complete mediation effect requires the c pathway be reduced to zero when a mediator is included in the model, the pragmat ic approach is to seek mediators that significantly decrease the c pathway. Noting, "a significant reduction demonstrates that a given mediator is indeed potent, albeit not both a necessary and sufficient condition for an effect to occur" (Baron & Kenny, 1 986) The literature goes further to cite practical concerns with following the causal steps exactly. Step four is not essential unless complete mediation is expected. Additionally, Kenny, et al. (1998) states that "most contemporary analysts" consider o nly steps two and three to be absolutely essential, as "a path from the initial variable to the outcome is implied if steps two and three are met" an example of inconsistent mediation. Incomplete mediation could occur when a mediator has a suppressing ef fect, for example, Kenny cites the relationship between stress and mood mediated by coping: "Presumably, the direct effect is negative: more stress, the worse the mood. However, likely the effect of stress on coping is positive (more stress, more coping) a nd the effect of coping on mood is positive (more coping, better mood), making the indirect effect positive. The total effect of stress on mood then is likely to be very small because the direct and indirect effects will tend to cancel each other out." (Ke nny, 2013)

PAGE 113

97 Due to inconsistencies in operationalizing the causal steps approach for social science research, and the reliance on steps two and three to infer mediation, the significance of those steps should be verified. Sobel's test (1982) provides a tes t of the null hypothesis that paths a and b equal zero. This test has been widely applied in mediation analysis (MacKinnon et al., 2007) This is the default test of indirect paths employed by Mplus, and results are reported in Table 2 6 Testing Mediation with Structural Equation Models The steps to testing mediation (as outlined in the previous section) can be completed individually in sequential regression analysis or simultaneously using SEM. In contrast, SEM is the most appropriate method in this case because it allows for both the modeling of latent factors and correlations among the dependent variables (Byrne, 2012). Additionally, it is adept at dealing with large amounts of randomly missing data. Modeling latent factors in regression requir es that factor scores be computed for each case based on mean responses to the corresponding observed variables that make up the latent factors. This creates problems when applied to the NTPP data, because attitudinal questions were administered randomly. The survey administration technique maximized responses on a variety of variables but left the data with large amounts of random missing data (i.e., no complete cases on all attitude variables). To deal with this, SEM employs Full Information Maximum Likel ihood (FIML) for missing data estimation and thus allows latent variables to be modeled in the mediation analysis (Enders & Bandalos, 2001) SEMs also accounts for measurement error in both the observed and latent

PAGE 114

98 variables (the "e"s and "d"s in the model) as well as correlations between all latent factors (specifically, the SEM models the residual correlations the correlations between what is not being predicted by the outcome covariates (Byrne, 2012). Mediation Structural Equation Model Figure 8 is a simplified model of the SEM created to test the mediation effect of the latent factor dis incentiv iz ing driving, on the latent factors, barriers to walking and cycling encouragement. The "complete" mediation model SEM (Figure 11 ) simultaneously models a, b, and c' pathways (these individual pathways are not explicitly drawn in the mediation model). Figure 5 Mediation SEM Diagram.

PAGE 115

99 Results The SEM mediation analysis provides evidence for statistically significant mediatio n effects on the dependent variables. All standardized and unstandardized coefficients (significant at the p < .05 level), are presented in Table 16 The SEM model simultaneously tests a, b, and c' pathways (Figure 10 ). Residual correlations for the mediation model (i.e., the correlations between what is not being predicted in the mod el), are reported in Figure 22

PAGE 116

100 Table 4 Phase 2 Model Results Mediation SEM Model Fit Indices Chi Square SRMR RMSEA CFI 2 df p 727.273 297 0.00 0.041 0.032 0.930 SEM Results Dependent Variable Independent Variable Unstandardized Standardized p Barriers Walk Convenience Drive 0.629 0.539 *** Barriers Bike Convenience Drive 0.528 0.450 *** Convenience Drive Columbia, MO 0.168 0.062 Marin, CA 0.027 0.009 Minneapolis, MN 0.277 0.110 Sheboygan, WI 0.096 0.036 Income 0.076 0.137 *** Gender 0.062 0.043 Age 0.012 0.178 *** Walk Regularly 0.309 0.104 Bike Regularly 0.243 0.099 # Children 0.002 0.012 Race 0.396 0.112 Barriers Walk Columbia, MO 0.036 0.011 Marin, CA 0.154 0.045 Minneapolis, MN 0.298 0.101 Sheboygan, WI 0.111 0.035 Income 0.054 0.082 Gender 0.179 0.109 *** Age 0.004 0.060 Walk Regularly 0.270 0.078 Bike Regularly 0.067 0.023 # Children 0.004 0.027 Race 0.117 0.027 Barriers Bike Columbia, MO 0.283 0.089 Marin, CA 0.029 0.008 Minneapolis, MN 0.089 0.030 Sheboygan, WI 0.042 0.013 Income 0.026 0.040 Gender 0.107 0.064 Age 0.013 0.167 *** Walk Regularly 0.140 0.040 Bike Regularly 0.586 0.203 *** # Children 0.005 0.030 Race 0.027 0.006 Barriers Walk R 2 = .35 Barriers Bike R 2 = .36 Convenience Drive R 2 = .12 = p < .05 *** = p < .001

PAGE 117

101 The SEM shows a significant mediation effect of the latent factor: convenience of driving. Individuals from Minneapolis, MN are the only population whose scores on the convenience of driving factor differed significantly from the control population (i.e., Spokane, WA). Minneapolis is the only city variable that is a significant predictor in the mediation model, while income, age, and race are also each significant predictors. Overall results of the mediation SEM identify the convenience of driving latent fa ctor as significantly and positively correlated with both the barriers to walking latent factors as well as barriers to cycling latent factor. Significant mediation paths hold across both barriers to bicycling and barriers to walking. This is unsurprising as these variables are significantly and strongly correlated ( r = .506, p < .001). In Figures 12, 13, and 14 standardized coefficients from the SEM model are illustrated in terms of a, b, and c' paths. The causal steps approach provides tests of individ ual pathways, but does not provide any information about the significance of a mediation effect (i.e., the pathway from X to M (path a) as well as the pathway from M to Y (path b)). Sobel's test of mediation provides a conservative estimate of the signific ance of the mediation pathway (MacKinnon et al., 2007) the results of which are also included in Figures 12, 13, and 14

PAGE 118

102 Figure 6 Mediation Effects City of Residence. Figure 7 Mediation Effects Socio demographics.

PAGE 119

103 Figure 8 Mediation Effects Regular NMT Use. Latent Factor Mediation The SEM demonstrates a significantly positive mediation relationship between latent factors, but this result should be interpreted carefully as the latent factors are essentially artifacts of the data. The mediation effects should be considered as explorat ory. Literal interpretations of the coefficients and their direction should be conducted cautiously: while the latent factor coefficients have been normalized (mean = 0, variance = 1), a "unit" increase in a latent factor cannot be equated to a meaningful quantity. Similarly, it is important to recall the relationship between the direction of coefficients and measured survey responses. The SEM model tests the nature and degree

PAGE 120

104 to which perceptions of the convenience of driving mediates the relationship be tween the independent variables and the latent factors: perceived barriers to walking and perceived barriers to cycling. Convenience of driving is a measure of the degree to which, as barriers to driving increase, the reported likelihood of walking instead also increases. Barriers to walking and cycling should be interpreted as when values increase, perceived barriers are stronger. An interpretation of the relationship between latent variables would be that increased willingness to walk as driving becomes less convenient, is associated with increased perceptions of barriers to NMT use in one's neighborhood; that is, an individual's likelihood to consider walking as driving is considered inconvenient positively predicts perceived built environment barriers to NMT. Alternately, individuals who report that the inconvenience of driving would not impact their decision to walk are likely to report few perceived barriers in their neighborhood to NMT use. On the one hand, the inconvenience of driving compounds pe rceived barriers to walking and cycling. On the other, decreased willingness to walk if driving were inconvenient predicts low perceived barriers to NMT. The interpretations are somewhat counter intuitive; why would an individual's decision to consider wal king instead of driving (if driving were more inconvenient) predict increased perceptions of neighborhood barriers to walking and cycling, and vice versa? Residential location may be partially responsible for this result. That is, if an individual lives in a non NMT supportive environment, sticks put in place to discourage driving may simply highlight a

PAGE 121

105 lack of available transportation alternatives. Alternately, the convenience of driving for an individual in an NMT supportive environment may have little be aring on their perceptions of neighborhood quality for walking and cycling. Indirect Effects Figures 12 13, and 14 illustrate all identified significant indirect (i.e., mediation) pathways. The standardized coefficients for paths a are identical across all walking and cycling mediation models, respectively, as they report a given independent variable's impact on the mediator (path a). The b paths are identical within walking and cycling models as they report the mediators' impact on the dependent variab le (path b). To interpret possible mediation effects, a, b and c' paths should be evaluated simultaneously. When all paths and the results of Sobel's test are accounted for, there is preliminary evidence for complete mediation effects when the indirect pat hway is significant (path a and b), but the direct pathway (path c') is not. Significant direct and indirect pathways may offer preliminary support for partial (or incomplete) mediation. Interpreting a mediation effect is not a clear process; consider t he example of the variable "regular cyclist" predicting "barriers to bicycling" as mediated by "convenience of driving." In this case, a regular cyclist is more likely to report a willingness to walk if driving were less convenient (path a). The mediation effect then amplifies perceived barriers to bicycling in a neighborhood (path b). The direct path shows a similar relationship between regular cycling and perceived barriers to bicycling. The alternative interpretation of this effect is that an individual who does not regularly cycle, likely

PAGE 122

106 would not be willing to walk if driving were more expensive, and perceives fewer (or less severe) barriers to bicycling. The model demonstrates that each of these relationships is significant, what might explain the dir ection of the partial coefficients? This is one example of a counter intuitive mediation effect that will be discussed in detail in the next section. Discussion The mediation effect of perceptions of barriers to driving impacting perceived barriers to walking and cycling has been identified using the SEM. This effect, and the accompanying significant indirect pathways, however, should not be overstated or over int erpreted. It is preliminary evidence to support future work into the relationships between relative perceptions of barriers to various mode choice options, which is particularly significant in a few relationships between city of residence, socio demographi cs, regular NMT use, and perceptions. The SEM mediation analysis does point to specific directions for future research into confounding variables impacting perceptions of mode choices, and intervention evaluation. The theoretical foundation for this work (described in the literature review section) specifies that the consideration of mode choice alternatives informs final mode choice decisions. From this foundation, findings address the impact of specific intervention options: carrots, sticks, or carrots and sticks. A direct path SEM is employed to understand how city of residence, socio demographics, and current behaviors impact carrots or sticks, but a mediation model is best suited for testing the interrelationship of

PAGE 123

107 carrots and sticks. Empirical resul ts support this modeling structure; R 2 values are higher, and residual correlations between latent factors are reduced in the mediation model (see Table 16 and for comparison, Table 28 for R 2 values, and Appendix C, Figures 22 and 2 3 for residual correlati ons) Interpretation Mediation effects are generally challenging to interpret and frequently rely heavily on theory to inform possible mechanisms. This is especially true when the mediating variable and the dependent variables are latent, rather than observed, variables. Conve nience of driving, for example, is a latent factor drawn from responses to questions regarding a stated willingness to consider walking more if various aspects of driving and parking were more onerous. In contrast, the barriers to walking and cycling varia bles are drawn from responses to questions about the degree to which individuals agree or disagree that various aspects of their neighborhood would impact their decision to walk or bicycle. Thus, the mediating variable indicates a scale of willingness to c onsider walking if driving conditions were worse (higher scores equally greater willingness), and the dependent variables indicate a scale of perceived barriers to NMT (higher scores equally a perception that barriers are more numerous). These factors are linked to carrots and sticks because the latent factors are indicative of perceptions formed by carrot interventions to reduce barriers to NMT and stick interventions that reduce the convenience of driving.

PAGE 124

108 Socio demographics The literature supports a s ignificant relationship between various socio demographic factors and attitudes, perceptions, and use of NMT (Dill & Voros, 2007; Xing et al., 2010) Age, and income are frequently positively correlated with NMT use, but this relationship may break down a cross utilitarian versus recreational travel (Dill & Voros, 2007) Older, more affluent populations may be less inclined to see monetary disincentives as particularly impactful regardless of trip type. Such populations may alternately be more mobile and si mply alter destination choices based on auto related disincentives. However, older populations are more likely to suffer from physical limitations that could impact NMT perceptions or use (Dannenberg et al., 2003) Regardless of age, it may be easier for high income populations to alter destination choices to fit the most convenient mode, and they may be more susceptible to congestion and parking availability than increases in the costs of driving and parking. Socio demographic factors are frequently spati ally correlated (Aytur et al., 2008) and this relationship requires further study. The mediation effect of race requires the most cautious interpretation, as these respondents make up a small portion of the NTPP sample. Taken together, socio demographic factors of significance suggest numerous possible interpretations that deserve future consideration. Alternative interpretations for mediation effects for low income younger, non white populations are myriad, and require future research to elucidate. Fo r example, it is possible that low income or transit dependent populations could confound results, as barriers to driving are already in place for these individuals. 13.4% of the NTPP

PAGE 125

109 population stated they have a health condition that "has lasted more tha n six months that has made it difficult to travel outside the home" and 8.6% of the NTPP population states they have zero "automobiles, vans or trucks kept at home for use by members of the family." In the NTPP sample, only 6.5% of the population reports t heir race/ethnicity as non white. Data constraints limit the ability to assess impacts across specific racial/ethnic populations, transit dependence versus an inability to afford an automobile, or physical condition that specifically impacts NMT use. Exist ing findings that significant mediation effects exist across socio demographic variables are indicative of the need for future research. City of Residence Unlike the socio demographic variables described above, Minneapolis is the only city variable that was insignificant in previous time series analysis of this subset of the survey data (see Chapter 3), but is a significant predictor on the dependent variables when mediated by the convenience of driving factor. Among the NTPP cities surveyed, Minneapolis is unique in its generalizability. Minneapolis is the largest survey city, as well as the most diverse in terms of socio demographics (specifically in terms of race/ethnicity, and income) as well as land uses. The city itself has a relatively dense urban c ore surrounded by high density, transit and NMT accessible inner ring suburbs, as well as much lower density suburban developments expanding outwards. As is the case with the significant socio demographic variables, conclusions regarding mediation effect s in Minneapolis must be cautious and taken as exploratory

PAGE 126

110 evidence for future study. The Minneapolis sample may not be large enough (n = 406), considering both the number of exogenous variables in the model and the extent of missing data. It is also impos sible to link specific carrot and stick interventions to individual perceptions. SEM results, however, should be used to guide more detailed evaluations of the ways in which specific carrot and stick interventions impact specific populations and communitie s. Regular NMT Use Perceptions of walking and bicycling, for any purpose, likely depend on past experience and use of such modes. The specific variables for walk and bike regularly are binary responses to the question: "in a typical week do you walk/bike for at least 10 minu tes?" 83% of the sample reported regular walking and 23% reported regular bicycling. The NTPP survey also included follow up questions regarding number of minutes engaging in each mode weekly, but these variables were omitted because in the case of both wa lking and bicycling, the vast majority of responses were "30 minutes." These variables were chosen because they do not specify trip type, but simply demonstrates regular use, which in turn likely influences the latent perception variables. One interpreta tion of regular cycling predicting barriers to bicycling, as mediated by convenience of driving was described in the previous section. When entering into this research it was hypothesized that regular NMT users would be more likely to consider walking if d riving was less convenient (i.e., a positive coefficient in path a). Similarly, regular NMT users would report fewer neighborhood barriers as a result of their current

PAGE 127

111 reported walking and bicycling (i.e., negative coefficients on paths b and c'). The firs t hypothesis is supported by the analysis, while the second is not. It may be that regular NMT users have formed opinions about their neighborhood through their past behavior and experience, and that opinion is negative. Alternately, if an individual does not engage in an activity, how can they be expected to express informed opinions about barriers to that activity? This effect deserves further study now that there is clear evidence of an interrelationship between perceived carrots, sticks, and existing be haviors. Conclusion The goal of the mediation analysis is to provide empirical support for the assertion that promoting NMT is mediated by barriers to driving. That is, perceptions of NMT are mediated by perceptions of the primary alternative to walking and cycling, driving. Factor analysis is employed to first identify latent factors associated with the perceived convenience of driving as well as perceived barriers to walking and cycling. These factors are modeled in an SEM framework with city of residen ce and socio demographic variables to model a hypothesized mediation effect. Results confirm the existence of a significant mediating relationship in which perceptions of auto use impact perceptions of walking and cycling in the general population. Percept ions of various mode choices can be directly impacted by interventions targeting behavior change, as psychological theories dictate that attitude change precedes and informs behavior change. This research considers the implications of carrot and stick inte rventions in terms of their impact on perceptions in the general population rather than a subset of NMT users. The strength of this approach is that it provides an indication of possible behavior change

PAGE 128

112 among a broad population, rather than among the sma ll subset of individuals who regularly engage in NMT. This study is primarily limited by data constraints. Secondary data allows for exploratory analysis to identify the existence of latent factors suggesting perceptions of the convenience of walking as well as perceived barriers to walking and cycling may in teract, but these are not ideal measures of barriers and benefits or of carrots and sticks. The latent factors are drawn from observed variables not explicitly created to infer carrot and stick identification and mediation analysis. Finally, sample size do es not support a disaggregate (in terms of neighborhood or specific population subset) analysis of carrot and stick interaction effects. Each of these limitations is, in turn, an opportunity to expand on this ultimately exploratory analysis that finds empi rical support for a mediating effect of the convenience of auto use on barriers to walking and cycling. The results of this analysis present significant contributions to the literature. First, this is a unique application of attitudinal data, linking per ceptions to specific types of interventions. Second, while SEM is used extensively in travel behavior literature, informing intervention effectiveness through an SEM framework is novel and may be a promising future research direction. Third, this study is the first of its kind to find an empirical relationship linking carrots and sticks; suggesting that carrots may be more effective with sticks in place, but this relationship is likely quite complex. The exploratory nature of this work has demonstrated a nu mber of significant yet counter intuitive interrelationships in the data, each offering directions for future study. This

PAGE 129

113 chapter lays the groundwork for further evaluation research and provides evidence that interventions aimed at promoting walking and bi cycling may be most effective when implemented in concert with one another.

PAGE 130

114 CHAPTER V CARROTS VERSUS STICKS: ASSESSING INTERVENTION EFFECTIVENESS AND IMPLEMENTATION CHALENGES FOR NON MOTORIZED TRANSPORTATION Introduction The myriad facto rs that can impact travel behavior can generally categorized as either "carrots" or "sticks." Carrots are any policy, infrastructure investment, or social program that encourages non motorized transportation (NMT). Sticks, on the other hand, are any interv entions that discourage driving. For cities focusing on increasing NMT mode share as an alternative to driving, the decision to focus on carrots or sticks should be based on their relative efficacy. This chapter asks the questions: (1) are carrots or stick s more effective at influencing NMT, and (2) what is the difference in terms of ease of implementation between carrots and sticks? In the United States, carrots to promote NMT are widely implemented and evaluated, but how effective are they at influencing travel behavior? It is possible that, for the most part, carrots in US cities create a more pleasant climate for existing NMT users and those already strongly predisposed to walk or bicycle. Examples of sticks are rare in the US, but examples of carrots a nd sticks in combination can be found internationally. In many European countries with high levels of NMT, carrots are in place to support a safe and pleasant environment for walking and cycling, but so are strong sticks discouraging driving.

PAGE 131

115 In communit ies lacking basic NMT infrastructure, such as sidewalks and bike routes/lanes, implementing stick policies that discourage driving may leave vulnerable populations with no other transportation options. Carrots serve an important purpose by providing option s for travelers, but encouragement alone may not be a strong enough force to impact widespread, population level behavior change. Sticks may offer important leverage to discourage driving, but in the US, infrastructure and policies for much of the last cen tury have favored automobile use. When land use, infrastructure, and policies favor auto use, stick policies alone may disproportionately impact vulnerable populations. The research reported in this chapter uses a multiple methods approach to two related but distinct research questions. A quantitative evaluation of the Non motorized Transportation Pilot Program (NTPP) employs structural equation models to assess the impact of carrots, sticks, or a combination of carrots and sticks on mode choice perceptio ns (research question 1). A qualitative approach, using in depth, key informant interviews in the NTPP cities is presented to understand the challenges of implementing carrot or stick interventions (research question 2). The combined findings will shed lig ht on what types of interventions are most effective at influencing NMT behavior, and how challenging such interventions are to implement. Literature Review This literature review addresses applicable research regarding impact and implementation issues a ssociated with various NMT interventions and theoretical foundations underpinning the study. The review is subdivided into applicable literatures

PAGE 132

116 regarding carrot interventions, stick interventions, combined (carrot and stick) interventions, and relevant t heory. Carrot Interventions Interventions can function as carrots or sticks, depending on perspective. Traffic calming measures can be seen by NMT users as carrots, and by drivers as sticks. The NMT literature focuses almost exclusively on the carrot aspect of interventions (i.e., th e ability of specific, often localized interventions to support or promote NMT ) Numerous literature reviews address the extent to which infrastructure interventions or policies meant to promote NMT may impact bicycling specifically (Pucher, et al. 2010) physical activity (Heath et al., 2006, 2012; Sallis et al. 1998) and NMT (Krizek, et al. 2009) These reviews focus on infrastructure investments that may improve the local environment, as well as policies that are meant to alter social norms. A m uch less common group of carrot policies those that provide rewards for not driving may be powerful carrots (one study found a 39% increase in NMT given cash incentives not to drive to work (Shoup, 1997) while another identified a strong willingn ess to avoid peak hour driving (Ben Elia & Ettema, 2009) Unfortunately I h ave only identified two examples in the empirical literature of such carrots. It is challenging to draw conclusions concerning the magnitude of impacts of specific interventions, becaus e sampling frames, analysis strategies, and outcome variables vary widely (Yang, et al. 2010) Systematic literature reviews of studies with

PAGE 133

117 prospective or controlled retrospective designs (Ogilvie, et al. 2004) and controlled before and after designs ( Ogilvie et al., 2007; Yang et al., 2010) find that interventions aimed at NMT promotion carrots have modest (Yang et al., 2010) often short term impacts (Ogilvie et al., 2007) resulting in, at best, 5% mode shifts to NMT at the population level (O gilvie et al., 2004) Modest empirical findings regarding short term impacts of carrot interventions may not be telling the full story of their importance in promoting NMT. In the long term, carrots may help to establish an equitable transportation syst em One in which stick and combined interventions may then be applied for maximum impact on travel behavior Theories of environmentally significant behavior suggest that context (i.e., environment) must support targeted behavior change ( Stern, et al., 199 9 ). When promoting NMT, a key predecessor to behavior change is creating a physical and social environment that is safe and friendly towards walking and bicycling. Establishing a groundwork to support future NMT use can be considered a long term impact of carrots, and one that is understudied in the travel behavior literature Long term impacts of carrots are especially critical from a transportation equity perspective. If sticks to discourage driving are put in place prior to establishing an NMT supporti ve context, those most impacted by the sticks will be low income auto dependent populations. Increased "personal immobility" (TCRP 1998 ) among disadvantaged populations further marginalizes already disadvantaged group s This in turn can increase social ex clusion an in ability to fully participate in numerous aspects of society due to

PAGE 134

118 socio economic statu s (Dodson, et al. 2010) T he literature has linked transit access to social exclusion; specifically linking transit availability for low income populations to employment, education, health care, and childcare opportunities (TCRP, 1998) There is a lack of literature explor ing the ability of NMT carrots to support behavior change among vulnerable populations, for whom sticks limiting auto use are l ikely already in place. NMT literature focuses on modest behavior changes that may be attributable to carrots, but long term impacts are unclear. This raises the question: are carrots simply creating a more pleasant environment for existing NMT users, an d possibly a small population of individuals who were highly likely to start using NMT, or can long term impact of carrots be identified ? Non probabilistic surveys cannot shed light on this question, because without representative samples it is impossible to tell if behavior change is occurring on a population level, or simply within a non representative sub group (Krizek et al., 2009) F or population level changes in the US, national data sources may provide some insights. The National Household Travel Survey (NHTS) has surveyed travel behavior in the US since 2001, and builds on the earlier Nationwide Personal Transportation Survey (NPTS) collected in the US since 1969 ( www. From 1990 to 2010, walking trips in the NHTS increased from 7.9% to 10.9 % of all trips reported, however, survey design changes in the 2001 NHT S may have impacts this figure (Clifton & Krizek, 2004) Bicycling rates in that same time period have gone from 0.7% of all trips to 1.0% (Krizek et al., 2009) There is also a regional aspect to national mode share ; bicycling

PAGE 135

119 rates in the P acific N orthwest are approaching 2% of all trips, while they are as low as .1% in the South (Pucher, et al. 2011) Pacific N orthwest cities ( Portland, Oregon, in particular ) are known for progressive policies and infrastructure investments that support NMT, but correlation between NMT interventions and mode shifts does not equal causation (Krizek & Barnes, 2009) The existing literature suggests that the null hypothesis, that NMT carrots are limited to modestly impacting the behavior of non representative populations, cannot be disproven. That is, carrots may only support the behavior of existing NMT users, or th ose predisposed to use NMT. Stick Interventions Compared to the exhaustive literature on carrots, there is little literature linking stick interventions designed to curb auto use to changes in NMT behavior. Mode choice models that include measures of ba rriers to driving find that the availability of free parking is associated with increased likelihood of driving, instead of NMT or transit use when the free parking is available at work (Buehler, 2012 ; Piatkowski & Marshall, 2013 ) or at home (Weinberger, 2012) Studies of the effectiveness of parking policies to curb auto use date back nearly 25 years (Willson & Shoup, 1990) The lack of literature on stick policies is likely a direct result of a lack of such policies in the US. Despite the fact that stick policies such as market based instruments to reduce work travel by car as well as peak hour congestion have long been advocated for (Rye & Ison, 2005) there is significant political opposition to these policies (Ben Elia & Ettema, 2009; Eliasson & Mattss on, 2006; GŠrling & Schuitema, 2007)

PAGE 136

120 Resistance to stick policies comes from a variety of sources for a number of reasons. Governments, TDM practitioners, and economists have advocated congestion pricing and toll road schemes as a method of dealing with congestion in urban areas (Harrington, et al. 2001; Hensher & Puckett, 2007) B ut there are very few examples of such policies (Hensher & Puckett, 2007) and they are chiefly attacked for their equity impacts on low income populations (Ben Elia & Ettema, 2009; Eliasson & Mattsson, 2006) TDM scholars have recognized that specific attention should be paid to impacts of stick interventions across different socio economic groups (GŠrling & Schuitema, 2007) In the US, congestion pricing is often considered coercive by motorists because there is no perceived alternative to paying a tax, and such a tax is rarely directly linked to any benefit to the driver (Harrington et al., 2001) Toll roads are seen as a less coercive alternative, yet still carry equity concerns (Eliasson & Mattsson, 2006; Verhoef, et al. 1996) E ducation al campaigns that support policies focus ing on substantial return of tax revenue to road users or other posit ive policy impacts may see 7% 22% increases in public support (Harrington et al., 2001; Kunchornrat, et al. 2008) Despite these findings, market based stick policies, and their impact on NMT are rare. Combined Approaches: Carrots and Sticks Combining carrots and sticks may be a powerful strategy to impact NMT behavior; however, the distinction between carrots and sticks is flexible, as a carrot to an NMT user can also serve as a stick to a driver. The Community Based Social Marketing (CBSM) approach (w is instructive in identifying interventions as either

PAGE 137

121 carrots or sticks. That is, does a specific intervention, from an individual's perspective encourage the uptake of a desired behavior, or discourage alternative undesirable behaviors? Foste ring specific behaviors involves careful consideration of barriers and benefits to best focus interventions (Steg & Vlek, 2009) CBSM advocates for combining carrots and sticks, and this approach has been successfully applied to mitigating household greenh ouse gas emissions (Dietz & Gardner, 2009) The CBSM approach has also been applied widely to transportation, waste, water, and numerous other environmentally significant behaviors (MacKenzie Mohr, 2011), but such applications are largely not evaluated in the literature. C ombining carrots and sticks may be an effective approach to increasing NMT in US cities, but there is a gap in the literature evaluating combined approaches. Combined approaches are also evidenced in the Travel Demand Management (TDM) literature. TDM is broadly defined as an action or combination of actions that influence travel behavior in such a way that reduces automobile demand/use and supports non automobile travel (TCRP, 2010). The TDM literature conceptualizes interventions as ca rrots to promote desired behaviors and sticks to discourage undesirable behaviors, but tends to categorize such measures as regulatory (i.e, policies), mobility improvements (by mode), traffic operations and flow improvements, and market based mechanisms ( Meyer, 1999) Interventions within these categories can be perceived by users as either carrots or sticks. TDM literature focuses on qualitative case studies exemplifying a particular strategy, as well as how and why it was implemented, but lacks quantitat ive evaluations

PAGE 138

122 at the population level, as well as comparisons of effectiveness and implementation challenges. I n the US, combined carrot and stick transportation interventions are rare, and the literature evaluating the impact of combined approaches is rarer still The "Complete Streets Movement" aims to enact policies that ensure projects are designed and planned to meet the needs of all types of community members and travelers (Seskin & McCann, 2012) In practice, such policies can involve both carrots that create an NMT supportive environment as well as sticks in the form of parking restrictions, lane width reductions, speed limits, and traffic calming measures (LaPlante & McCann, 2008) Such combina tions of carrots and sticks, however, may be more palatable to the general public, as traffic calming measures are modest sticks (compared to pricing schemes, for example). The degree of difficulty in implementing many combined approaches may rest on the s everity of the sticks included, but existing literature cannot confirm or refute this. Perhaps because of the modest sticks and focus on inclusion, c omplete streets initiatives are becoming more common in municipalities in the US, but empirical research fo cuses on implementation challenges and strategies (Heinrich, et al. 2011) rather than intervention effectiveness. Safe Routes to Schools (SRTS) programs, frequently combine carrot s and sticks to incentive children's and parent's use of NMT (Chriqui et al., 2012) but "sticks" are typically extremely modest (e.g., reduced speed zones) Despite only the mildest of sticks incorporated into combined approaches, SRTS programs are instructive for their

PAGE 139

123 implementation strategy. SRTS implementation is focused o n a near universally accepted positive: children's safety, but there is little literature on this aspect of SRTS. SRTS evaluation studies point to success among students (Boarnet, et al. 2005; Staunton, et al. 2003) but relative impacts of carrots or st icks on children surveyed, or on NMT use among parents or the general population are outside of the scope of these studies W hile SRTS and Complete Streets may be promising avenues for future study, they currently offer little guidance into how to evaluate combined interventions, or the effectiveness of such interventions on the general population. Assertions in the NMT literature that combined carrot and stick approaches are most effective at promoting NMT are grounded in international examples, particul arly case studies of Northern European cities. The TDM literature supports the inclusion of incentives and disincentives to significantly impact auto use (in the US), but auto reliance is so high in the US that implementing combined strategies is largely i mpossible (Meyer, 1999) Indeed, percent of total trips completed by bicycle in the Netherlands is 27%, Denmark, 18%, and 10% in Germany and Sweden (Pucher & Buehler, 2008) compared to about 1% in the US (Bassett, et al. 2008) At the city level, only a few German cities have cycling mode shares below 5%, while the majority of US cities have cycling rates well below 5% (Pucher & Buehler, 2008) Boulder, Colorado, and Portland Oregon (two outlier cities known for supporting cycling) have bicycle commute mo de shares of about 10%, 6%, respectively ( ACS, 2010) Several factors have been identified which explain why walking, cycling, and transit use are more common in European countries: 1. Compact, dense cities with mixed land uses that generate short trips

PAGE 140

124 2. Restr ictions on car use such as car free zones, low speed limits, and prohibitions of through traffic 3. Extensive, safe, and convenient facilities for walking and cycling 4. Traffic calming of residential neighborhoods 5. Coordination of public transit with walking and cycling to transit stations and stops, including bike parking, as well as safe sidewalks and bikeways 6. Traffic regulations and enforcement policies that favor pedestrians and cyclists over motorists 7. High cost of owning and operating a car resulting from ex pensive driver licensing, high gasoline prices, and high taxes on car purchases ( Source: Bassett et al., 2008) The above list can be easily divided into carrots that encourage NMT (1, 3, 5) and sticks that discourage auto use (2, 4, 6, 7). Combinations of regulatory, policy, land use, and infrastructure constitute powerful examples of the effectiveness of combined approaches to dramatically impact city scale and national travel behavior. European case studies of combined approaches confirm the effectiveness of carrot and stick approaches in concert at the national ( e.g., Denmark, The Netherlands, and Germany ( Pucher & Buehler, 2008 ) ) and city ( e.g., Freiburg ( Buehler & Pucher, 2011 ) ) scales. Such studies, however, utilize aggregate data or case study methods (respectively) and cannot inform impacts of individual carrot or stick interventions or address the hypothesis that carrots should be in place prior to implementing sticks.

PAGE 141

125 The transportation literature evaluating carrots and sticks provides a few key insights. First, individual carrots (policies or infrastructure interventions) have been extensively evaluated i n the literature and tend to have modest positive impacts on targeted populations; however, these effects are only observed in the short term. Second, it is unclear if modest carrot interventions are improving the social and environmental context for exist ing users, or influencing mode shifts to NMT. Third, there is a dearth of literature on stick interventions, and the available studies focus on implementation strategy rather than evaluation. Fourth, stick policies in isolation may exacerbate equity issues and social exclusion among already marginalized populations, so carrots may be a pre requisite to support effective combined policies. Finally, International comparisons suggest that combined approaches are highly effective at influencing NMT use, but do not speak to the timeframe by which carrots or sticks should be implemented. These key insights reveal a need to better understand the long term role of carrots in supporting sticks and combined approaches, as well as better guidance on implementation stra tegies and evaluation of sticks and combined approaches. Relevant Theory There are numerous relevant theories of behavior change informing this chapter. Steg and Vlek (2009) identify two lines of theoretically informed research into individual, motivatio nal factors. (1) weighing costs and benefits: this line assumes reasoned actors considering options and is strongly influenced by the Theory of Planned Behavior (TPB) (Ajzen, 1991) The TPB has been proven helpful in explaining mode choice decisions (Bambe rg, et al. 2003; Heinen, et al. 2011) (2) Morals and norms: The

PAGE 142

126 Value Belief Norm theory (Stern, et al. 1999) has been employed to explain pro environmental behavior and illustrates the frequently limited impact of values and beliefs on behavior when c ontext strongly opposes the specified behavior (Steg & Vlek, 2009) Contextual factors are important as they can strongly facilitate or constrain behaviors (Steg & Vlek, 2009; Stern et al., 1999) Context can directly impact behavior (e.g., if transit is not available, one cannot use transit even if they want to), or mediate individual factors (e.g., the presence of bike lanes may reduce stigma against bicycling). Individual and contextual perspectives each assume rational decision making (Steg & Vlek, 200 9) A parallel line of theoretical research examines the impact of habit on mode choice decisions. Individuals make many decisions based on habit, and particularly in the US, habitual auto use is the norm (Bamberg & Schmidt, 2003) Habits may involve "misperceptions and selective attention" (Steg & Vlek, 2009) and have been found to be impacted by behavior change interventions (Thgersen & Mller, 2008; Thgersen, 2009; Verplanken, et al. 2008) However, the overall impact of hab it on behavior may be marginal as travel behavior is primarily a reasoned choice (Bamberg et al., 2003) While habit is an important theoretical consideration, for the purposes of this study it is sufficient to know that habit is vulnerable to interventio ns, and may only play a modest role in travel behavior decisions. Underpinning the discussion of the relative efficacy of carrots, sticks, and combined approaches to promoting NMT are two key theoretical perspectives built on the assumptions of (1) mode choice as a (mostly) reasoned action, and (2) the importance of

PAGE 143

127 cons idering both individual and contextual factors. The first theoretical underpinning to mode choice decisions is utility theory. Utility theory is rooted in economic conceptions of rational actors making transportation mode choice decisions based on weighing pros and cons of various options (Maat, et al. 2005) Utility is typically operationalized as time and distance, but can include accessibility, destination preferences, daily activity patterns, and mode preferences. This theoretical perspective is releva nt because it provides a straightforward conceptual approach to understanding the impacts of carrots and sticks. Carrots serve to increase the utility of NMT by creating safe, high quality routes that may in turn improve NMT accessibility. Carrots may also improve individual attitudes and perceptions of NMT as well as broader social norms. In contrast, sticks work to decrease the utility of automobile use. The examples above focus on economic disincentives and travel time but can also operate on preference s and norms to decrease the utility of alternative to NMT use. The second theoretical perspective underlying this research is Stern's Attitude Behavior Context (ABC) Model (Stern, 2000). The ABC model is a model of environmentally significant behavior b uilt on the fundamental assumption that behavior is a function of the organism and its environment (Stern, 2000); "behavior (B) is an interactive product of personal sphere attitudes (A) and contextual factors (C)." In this framework, attitudes can include a wide variety of factors including beliefs, predispositions, and individual norms. Similarly, context can include physical environmental influences, social norms, policies, and monetary costs or incentives. The example of recycling is frequently used to explain the interplay between context and

PAGE 144

128 attitudes on a given behavior (Figure 1 5 ). On one hand, attitudes have the greatest influence on behavior in the absence of contextual factors; on the other hand, attitudes are insignificant in the presence of stro ng contextual factors. Thus, if it is very hard or very easy to recycle, attitudes do not matter, as the former case no one recycles, while in the latter case, everyone recycles. This formulation may help explain the role of carrots and sticks in promoting NMT behavior. Consider the proposition that currently in the US, carrots to promote NMT are reducing the negative effects of context that is, the "norm" in the US is an unsupportive context, and carrots are creating a less un supportive (or more suppo rtive) context for NMT use. If the role of context in the ABC model is neutral (or less negative), then NMT use will be highly correlated with pro NMT attitudes. In a neutral context, NMT may not be the most attractive (or highest utility) option, but NMT use is not hindered by context. This formulation would explain the modest impact of carrots in US cities; carrots are simply decreasing the negative impact of external context on behavior. This in turn supports the assertion that carrots support the behavi or of individuals who currently use NMT (have strong pro NMT attitudes) and those who would likely strongly prefer NMT anyway.

PAGE 145

129 Figure 1 The ABC Model ( adapted from : Stern, 2000) The ABC model can be extended beyond c arrots to explaining impacts of combined approaches, such as those used in many European countries. Combined approaches diminish the power of individual factors (i.e., attitudes) by fostering a context that strongly favors NMT. Context improves the utility of NMT by reducing barriers to walking and bicycling while simultaneously reducing benefits of auto use. The power of individual factors (attitudes) is reduced, and a larger subset of the population engages in regular NMT use. Whereas in the US, reducing negative context supports the behavior of those who would prefer to use NMT, in the European example, the context overwhelms individual factors to increase NMT mode share.

PAGE 146

130 Study 1: Quantitative Evidence for "Carrots or Sticks" vs. "Carrots and Sticks" to Encourage Non motorized Transportation Background and Data The motivation for this study into the impact of carrots versus sticks and the challenges for implementation grows out of earlier work examining the relationship between carrots and st icks (Chapter 4). Chapter 4 describes in detail the empirical test of what types of interventions are most effective at impacting NMT, and applicable results are summarized below. Using the NTPP community surveys (described in detail in Chapters 3 and 4), changes in attitudes and perceptions were identified during the time period of the NTPP (Chapter 3). While it is impossible to link attitude changes to specific interventions, this data offers a unique opportunity to explore population level changes in NMT attitudes over the same time period as community level interventions were put in place in US cities. A cross sectional analysis of the 2010 NTPP data was used to identify latent attitudes regarding perceived mode choices. An exploratory factor analysis revealed that attitudes in the NTPP could be explained by three latent variables. The first two latent factors included (1) perceptions of neighborhood quality for walking, and (2) a factor relating to neighborhood quality for bicycling (i.e., carrots). Th e third factor included stated willingness to switch to NMT if driving were more inconvenient (i.e., sticks). These findings were confirmed through a confirmatory factor analysis. The underlying factor structure allows for a unique opportunity in NMT resea rch, to test the existence of

PAGE 147

131 a significant interrelationship between carrots and sticks. Specifically testing the hypothesis that the impact of carrots is strengthened in the presence of sticks. Methods The "convenience of driving" factor was modeled i n a structural equation model (SEM) as a mediator between independent variables (including city of residence, demographics, and existing travel behavior) and perceptions of walking and bicycling. A mediation SEM was chosen because it parsimoniously represe nted the theorized relationship of perceptions of alternatives informing perceptions of walking and bicycling, and for its frequent application to social science research (see Chapter 4 for a complete explanation of methods) The final model is presented b elow (Figure 16 ) Figure 2 Mediation Structural Equation Model (Source, Chapter 4) In the above model (Figure 16 ), the outcome measures are "carrots," perceptions of barriers in an individual's neighborhood to walking and cycling. City of residence and

PAGE 148

132 individual socio demographics are the independent variables. "Sticks" (perceived convenience of driving) is operati onalized as a mediator to test the degree to which the relationship between independent and dependent variables depends, in part of in full, on the perception of driving (as the primary alternative to NMT). If significant mediation effects are identified i n the model, then it can be confirmed that combining carrots and sticks may be more effective than carrots or sticks alone at influencing perceptions of NMT. Results The mediation SEM identified a significant interrelationship between perceptions of stic ks and perceptions of carrots. Specifically, stated willingness to use NMT if driving were less convenient significantly mediates the effects of the independent variables Minneapolis, Income, Age, and Race on perceived barriers to walking and cycling (see Chapter 4 for full mediation results). The mediator increases perceptions of neighborhood barriers to walking and cycling. This means that as individual's state that they are less willing to consider NMT as driving becomes more inconvenient, they are also likely to report increased barriers to NMT in their neighborhood. The reverse of this interpretation is that individual's reporting a strong willingness to switch to NMT if driving becomes more inconvenient are also likely to report fewer barriers to walki ng and cycling in their neighborhood. The direction of the mediation effect is somewhat counter intuitive. It appears that the inconvenience of driving compounds perceived barriers to walking and cycling. This

PAGE 149

133 could be due to a number of factors and illu strates a limitation of the data. It is possible that an individual's residential location may help to explain these results; if an individual lives in a non NMT supportive environment, sticks put in place to discourage driving may simply highlight a lack of available transportation alternatives. This relationship may be further complicated by the addition of independent (exogenous) variables as predictors, mediated by convenience of driving. Possible explanations for the direction of partial coefficients a re myriad, but what is important is that there is a significant interrelationship between carrots and sticks, that offers greater explanatory power than SEM models of direct paths between independent variables and carrots or sticks separately (Chapter 4). The role of sticks in the SEM a stated willingness to shift to NMT as driving is dis incentivized confirms behavioral theories, suggesting that consideration of NMT use is significantly influenced by considering barriers put in place to discourage al ternatives. This finding is largely exploratory for three reasons. First, measuring attitudes and perceptions of mode choices is not the same as measuring perceptions of specific interventions. Second, the direction and size of the impact of sticks as a me diator is difficult to interpret as the SEM confirms an interrelationship between latent constructs drawn from a survey not designed specifically to capture perceptions of carrots or sticks. Third, relating perceptions to behavior using this dataset may be of limited value because of its aggregate nature (theories described herein suggest that this relationship is strongly context dependent, and context varies dramatically across the sample population).

PAGE 150

13 4 Despite these limitations, empirically validating the combination of carrots and sticks has broad implications for practice and research. Study 2: Assessing Implementation Challenges for Various Types of Interventions Aimed at Promoting Non motorized Transportation Background and Data Study 2 complements th e quantitative analysis by identifying the barriers to implementation for various types of interventions, and can serve as a guide for practitioners and advocates seeking appropriate interventions for specific goals. The second study uses in depth, key inf ormant interviews with policy makers, planners, public works officials, and advocates in each of the NTPP cities. The NTPP was guided collectively in each community by a combination of local officials, citizens, and advocates, and interview subjects were s ampled from this population in each community. At least one local advocate and two officials (either planners, engineers, or policy makers) were interviewed in each city. 17 interviews were completed: 14 in person interviews in the respective communities, and 3 over the phone interviews with national research and advocacy professionals. Follow up questions and clarifications were addressed via email or by phone as well during the data analysis process. Interviews were targeted at identifying and exploring issues surrounding carrot, stick, and combined interventions. Subjects were asked to identify their roles' in the

PAGE 151

135 NTPP, motivations and expected impacts of specific interventions, and examples of carrot, sticks, and combined interventions. All interviews w ere recorded and reviewed for two purposes: (1) identify interventions in each community and assess the degree to which specific interventions functioned as carrots, sticks, or combinations of the two. (2) Understand the implementation process for each typ e of intervention. The planners experience was prioritized in the interviews because of their expert status in understanding implementation issues and challenges in their respective city. Advocates were also surveyed to gain an understanding of their speci fic goals and experiences in influencing public opinion to implement various interventions. Findings Focus on Carrots The interventions put in place in the NTPP cities were overwhelmingly carrots focused on NMT promotion or building basic infrastructure. This is to be expected because, ultimately, planners and local officials serve the interests of the public, and canno t be expected to allocate public funds to unpopular interventions. However, there are examples in the NTPP cities of unique and innovative interventions that are rare or non existent in much of the US. Cities like Columbia, MO and Sheboygan County, WI, foc used primarily on building transportation centered infrastructure because that was identified as a basic need in the community. In contrast, Marin County, CA, and Minneapolis, MN, had more extensive on street infrastructure in place at the start of the NTP P. In Marin and Minneapolis, the overwhelming focus was also on carrots, but with

PAGE 152

136 an emphasis on dealing with complex connectivity challenges (Marin), or "quality over quantity" (Minneapolis). The focus on carrots, to some extent, conflicts with the stated goals of the NTPP: "to demonstrate the extent to which the bicycling and walking can carry a significant part of the transportation load, and represent a major portion of the transportation solution, within selected communities." This in turn highlights t he conflict between programmatic goals and actual implementation within unique communities. "Providing Options" Officials in each of the cities responded to the question "what is the goal of NMT promotion in your city?" with some variation of the stateme nt "providing transportation options." Providing multi modal options is a complementary goal to that of the NTPP, but distinct in its focus. "Representing a major portion of the transportation solution" requires an explicit shift from automobile use to NMT and implies that transportation infrastructure and planning should be oriented to increasing the utility of non motorized modes instead of the automobile. The disconnect in stated goals of planners within communities and those of the NTPP is not a fail ure of local officials, but likely a symptom of operating in a context which overwhelming prioritizes auto travel. Officials in each city all supported the assertion that combining carrots and sticks is likely to be the most effective at promoting mode shi ft, and carrots alone have a limited impact. These same officials also confirmed that sticks disincentivizing driving face extreme public resistance in most cases. Opposition can be so extreme that modest projects that may even only appear to have

PAGE 153

137 "stick" components can be blocked for years. An advocacy director in one city described how an approximately one mile re surfacing project encountered severe resistance when parking removal became a possibility: "[It] is a relatively minor project i n the bigger scheme of things The reason it got "exciting" is because there were a number of re striping regimes being proposed to make t he road safer for cyclists. One of the regimes removed 3 out of so mething like 40 parking spaces Because of the high drama, the to wn chose to segment the project into two smaller ones. The controversial one did not yet go forward. It will be delayed 2 to 3 years. (Local Advocacy Director) In this case the public process is ongoing to work with local business owners concerned with reduced parking. One bike/ped planner described the public outcry at a city council meeting when a bike lane (that would prohibit on street parking) was proposed on a street which few people parked on. Residents were concerned about "losing their right" to park and were concerned that "if they had a party, where would their guests park?" Because of these concerns, in this case, the city council would not approve the project. In a political context that is so highly sensitive to even perceived sticks, in whi ch relatively benign projects may be cancelled or delayed for years, it is understandable that local officials focus on carrots.

PAGE 154

138 Carrots Informing Social Norms There are significant challenges to implementing sticks or combined approaches in the US. I n contrast, the only significant barriers to carrots (e.g., off street NMT interventions and pedestrian investments) that interview subjects identified were funding and available land. However, on street NMT facilities that do not explicitly prioritize aut omobiles are often perceived as sticks. Interviews revealed powerful public concerns that even modest improvements to on street NMT infrastructure could be perceived as sticks against the automobile. To assuage public fears of strong anti auto sticks leadi ng to drastic changes in their community, NTPP officials leveraged NTPP funds and "pilot project" status to implement creative treatments. Such treatments are meant to strike a balance between planners' goals of impacting city scale travel behavior and pub lic concerns of limited auto mobility. This section details two examples of creative treatments in the NTPP communities are described in this section. These creative treatments deserve attention because while they are carrots aimed at promoting NMT, the y faced increased opposition to implementation due to parking concerns and limited road width for standard on road bicycle infrastructure treatments. Each of the NTPP communities has abundant on and off street parking, and in many of the communities there is an overabundance of parking. Yet as described above, parking is a contentious issue for drivers. Parking can also pose a safety risk for cyclists concerned about being "doored" (when a driver parked on street unexpectedly opens their door into a bikelan e). For these reasons, bicycle lanes are not

PAGE 155

139 recommended on roads with parking, or they are striped between parking lanes and vehicle lanes. Accommodating each of these uses requires minimum road widths and on streets lacking the necessary space, bike lane s become difficult to implement because the obvious solution is to remove parking. Two examples of creative on street NMT interventions were identified that do not limit parking or vehicle right of way, but serve to educate drivers that bicycles have a pla ce in city streets: "BLIPs" and "Advisory Bike Lanes." Bicycle Lane with Intermittent Parking (BLIPS) Columbia, MO and Sheboygan County, WI have installed BLIPS on certain roads. A BLIP is a bike lane striped on a road without removing on street parking (Figure 17 ). The BLIP is a compromise between residents' concerns about reduced neighborhood parking whi le still implementing a connected on street bicycle system. The rationale behind a BLIP is that on streets with infrequent on street parking, it is safer for novice cyclists to have a BLIP, as opposed to a shared lane marking (i.e., "sharrow"). BLIPS are w idespread in Columbia, MO, with BLIPS accounting for an estimated 24% of the 126 miles of striped bike lanes (Ted Curtis, Bike/Ped Coordinator, City of Columbia). Despite safety concerns for cyclists using BLIPs, officials in Columbia and Sheboygan have not identified any increases in accidents. Officials in each city have described the influence of BLIPs on parking behavior "[BLIPs are] perceived as hazardous for the vehicle: it [the vehicle] may get clipped by a passing vehicle" (Ted Curtis, City of Co lumbia). "It makes drivers feel like they shouldn't be parking there"

PAGE 156

140 (Planner in Sheboygan County). An official in Columbia also explained that BLIPs are meant to foster better cycling skills by forcing cyclists to look behind them when they approach a pa rked vehicle prior to venturing out of the bicycle lane and into the vehicle lane (Figure 17 ). No studies have examined the actual impact of BLIPs on vehicle or cyclist behavior. Figure 3 Ideal Bicyclist Behavior in BLI P (Photo Credit: Ted Curtis) The BLIP deserves further study. Safety concerns should be evaluated prior to continued implementation of this treatment. Despite this, BLIPS may serve to connect bicycle facilities and lead to a more extensive on street bic ycle network, while avoiding public concerns of reduced parking.

PAGE 157

141 Bicycle Advisory Lane Bicycle advisory lanes involve removing a centerline and striping bike lanes on either side of the street, with the traffic adjacent b ike lane stripe dashed (Figure 18 ). The rationale for this treatment is that the centerline removal provides less guidance to a car, encouraging driver awareness and reducing vehicle speeds (Minneapolis Public Works Official. Bicyclists behave as they would on a street with shar rows or no treatment, and cars are permitted to drive in the bicycle lanes when no cyclists are present, but yield to bicyclists and oncoming traffic. They are termed "advisory lanes" because vehicles are permitted to enter them when no bicycles are presen t, as opposed to solid striped bicycle lanes. Minneapolis has been a national leader in supporting NMT. The city was awarded silver level bicycle friendly city status in 2008 and attained gold level status in 2011 by the League of American Bicyclists (ww Minneapolis was also named top bicycling city in the US by Bicycling magazine in 2010 ( One way in which Minneapolis has achieved this status is through innovative street treatments that are new to the US. The bicycle advisory lane is one such intervention whose purpose is to promote safety, visibility, and a sense of belonging on streets for bicyclists while encouraging drivers to yield. The bicycle advisory lanes function as a combined approach, because they limit a uto mobility by prioritizing bicycles and vehicles equally. However because of their design specifics, bicycle advisory lanes do not require policy changes or parking

PAGE 158

142 restrictions for implementation on many roads. In this way, bicycle advisory lanes are ea sier to implement than more extensive combined approaches to roadway treatments. The advisory bicycle lane can be implemented in Minneapolis without revising street design guidelines. Minneapolis adheres to the Manual on Uniform Traffic Control Devices ( MUTCD). On roads with less than 6,000 vehicles per day, centerlines are not required. Additionally, because vehicles are permitted to use the advisory lane when no bicycles are present, road width minimums are unaffected by the treatment. According to Stev e Clark, the Bicycling and Walking Program Manager at Bike Walk Twin Cities, advisory bike lanes can increase comfort and safety for cyclists trying to avoid the risk of being "doored" by assigning more on street space to bicycles, suggests that for driver s, "perceiving that you have a bit less space can foster a bit more prudence." Advisory bike lanes remain unstudied in the literature, due to their rarity in the US. But, as an intervention, they act as a carrot for bicyclists who can utilize designated on road space, and as sticks for drivers who must yield to both cyclists in the advisory lanes and to oncoming vehicles. This treatment did require modest education and outreach (signage was put in place to explain the lanes in Minneapolis), but because it conforms to requisite design guidelines does not face the implementation challenges of sticks.

PAGE 159

143 Figure 4 Bicycle Advisory Lane (Photo Credit: Simon Blensky) The BLIP and Bicycle Advisory lane offer promising examples of carrot interventions that include an educational component and do not require policy changes to implement. That is, they may do more than promote NMT, but not face the implementation challenges of interventions with stick components. The Bicycle Advisory Lane, in particular could be very promising in re allocating road space for multiple modes and forcing increased driver awareness and responsibility when operating on such treatments. I have not identified relevant empirical research on the effectiveness of BLIPs or the extent or appropriateness of the intervention in various settings; literature is lacking on their application and impact in the US.

PAGE 160

144 Limited Examples of Sticks There were few examples of sticks in the NTPP cities. Officials in all of the cities confirmed that their focus was on "break[ing] down the us' versus them' mentality" (Official in Columbia) of drivers vs. NMT users (particularly cyclists) and it was feared that sticks would be counter productive in this context. Columbia provides two examples of how policy changes that can be construed as sticks by drivers require public and bureaucratic support, and then required additional evidence to support implementation and conti nued education during implementation to avoid reactive behavior. Columbia implemented two ordnances in 2008 and 2009. One ordnance lowered residential speed limits to 25 miles per hour (mph), and the other was an anti harassment ordnance to curb driver h arassment of NMT users. The speed limit ordnance enjoyed widespread community support, as well as support of the Mayor, but was resisted by the Public Works Department, who felt that an ordnance would not impact actual traffic speed. Researchers at the Uni versity of Missouri found that neighborhoods with posted 25 mph speed limits had reduced traffic speeds. Findings of local researchers and community support led to adoption of the speed limit ordnance (Pednet, 2008a). The anti harassment ordnance in Colum bia came about because, prior to the ordnance, police were required to charge an offender with Third Degree Assault a charge police were reluctant to use. The proposed ordnance made a variety of forms of harassment punishable as a misdemeanor offence (Pe dnet, 2008b). Local advocates recognized that the proposed ordnance "ha[d] triggered a negative reaction in the

PAGE 161

145 community." Community members felt that excessive resources were being spent on NMT promotion, and that "dangerous and illegal behavior of some cyclists" was the issue, rather than harassment by motorists (Pednet, 2008c). The local advocacy coalition, Pednet, used educational classes, and a variety of public media outlets to educate confirmed citizens and work with the local police department to e nsure enforcement of laws for both bicyclists and drivers. The examples described above illustrate that even modest stick policies that enjoy multi level support face powerful implementation challenges and may incite backlash in the community. From the a uto user's perspective, 5 mph speed limit reductions in neighborhoods and a harassment ordnance do not appreciably limit or disincentive driving from a mobility, accessibility, or financial perspective. Despite the modest nature and extensive support for t hese policies, a lengthy public process and impact assessment were required prior to implementation. Additionally, interviews suggest that these ordnances also may have contributed to a backlash in the community against NMT. Given these barriers to impleme ntation, focusing on sticks may be very challenging in terms of implementation, but also may depend heavily on the type of sticks proposed, and exactly they work to dis incentivize driving. Combining Carrots and Sticks Com bined approaches to NMT interve ntions involve both carrots to promote or incentivize walking and cycling (the desired behaviors) and sticks to discourage or limit driving (the alternative) in the intervention area. Combined approaches can take the form

PAGE 162

146 of micro (e.g., a single street tr eatment) or, macro (e.g., policy levers to impact mode choice) treatments. European examples described in the literature review work on a national level as a combination of carrots and sticks. In the US, a combined approach can involve a natural experiment in which carrots to encourage NMT are already in place (e.g., a high quality, connected bike/ped infrastructure system), and an "external" stick, such as gasoline price spikes or congestion increases, can create a combination that may be more effective th an a carrot or stick alone. External Sticks Gas prices are determined on the national level and vary only moderately across the US. In the NTPP cities, gas prices did not significantly increase from 2006 to 2010, but in many cases actually decreased (Table 1 7 ). The impact of gas price as an external stick, however, depends much more on the spike in gas prices in 2008, in which gas prices rose from approximately two dollars per gallon in 2006 to a spike of about four dollars per gallon in the summer of 2008 (before then dropping dramatically in the fall of 2008). During the gas price peak, members of PedNet, the primary bike/ped advocacy organization in Columbia, reported an increase in calls about an "earn a bike" program (now defunct); members of the community were interested in earning a bike to ride because driving was becoming too expensive. Unfortunately, this is only anecdotal evidence suggesting that an external stick can increase interest or involvement in carrot type activities or interventions

PAGE 163

147 Table 1 Gas Prices in NTPP Cities (1 Gallon in US Dollars) 2006 2010 Lowest price* Highest price** Difference low to high price Columbia, MO 2.94 2.55 1.38 3.98 2.60 Marin County, CA 3.18 2.98 1.71 4.66 2.95 Minneapolis, MN 3.02 2.78 1.56 4.00 2.44 Sheboygan, WI 3.13 2.88 1.65 4.12 2.47 Spokane, WA 3.02 3.02 1.72 4.32 2.60 *Lowest gas prices occurred in all cities in Fall 2008 **Highest (peak) gas prices occurred in all cities in Summer 2008 (source: Congestion in the San Francisco Bay Area is a persistent issue, and Marin County, CA is an example of a possible future natural experiment in external sticks. To incentive NMT in Marin, the community constructed the Cal Park Hill Tunnel ; a one mile long dedicated bike/ped facility connecting the towns of San Rafael and Larkspur (including a 1,100 foot dedicated tunnel). The tunnel is also a vital connection to the 25 mile Marin North South Greenway that provides an almost entirely off st reet route from the Golden Gate Bridge (at the south end) to Novato in the north. The tunnel offers a faster than rush hour route between San Rafael and Larkspur than by car ( Advocates and city officials in Marin County noted the import ance of providing NMT options as congestion continues to increase. Micro scale Combined Approaches Combined carrot and stick interventions in the NTPP cities can be considered "small scale" as they consist of street treatments along a single street or r oute, or neighborhood scale interventions around local schools. Minneapolis has successful implemented "bicycle boulevards," and Marin County has implemented a variety of

PAGE 164

148 carrot and stick measures as part of Safe Routes to Schools (SRTS) campaigns. Both He nnepin County (which includes Minneapolis) and Marin County have in place Complete Streets policies, which one advocacy director described as a "critical first step" in implementing combined carrot and stick interventions. Complete streets policies are mea nt to encourage streets that "are designed and operated to enable safe access for all users, including pedestrians, bicyclists, motorists, and transit riders of all ages and abilities" ( ). A national advocacy director interviewed described complete streets legislation critical as it "forces a conversation" about multi modal options. Lessons from the NTPP cities suggest that when considering combined interventions, implementation must first involve garnering public support through education and outreach. SRTS and bicycle boulevard interventions include components that may be construed by drivers as sticks because they often limit vehicle speeds (SRTS), and reduce auto mobility on certain neighborhood streets (bicycle boulevards), b ut are at most very modest dis incentives to driving. Despite this, such interventions require education and outreach components beyond those required for carrot type interventions. Crucial to successful implementation is framing these interventions not in terms of the limitations they impose on drivers, but on likely benefits for the local community. The SRTS program goal is to promote safe physical activity among children (McDonald & Aalborg, 2009) not punish driving, and bicycle boulevards may improve n eighborhood livability and increase home prices (Walker, et al., 2009).

PAGE 165

149 Officials and advocates in Minneapolis and Marin described that in addition to reframing interventions in a more positive light, educating community members with examples of proposed interventions was an important step to implementation. A public works representative in Minneapolis described the relative ease with which bicycle boulevards can now be implemented once "pilot" bicycle boulevards were installed. The bicycle boulevard has been particularly popular in Minneapolis, as the focus of NMT promotion in the city has moved from "quantity to quality" (Public Works Official). Minneapolis boasts an extensive on and off street NMT network, but this official described how the city is foc using on strategic infrastructure rather than ensuring that as many streets as possible have bike lanes or sharrows. It is clear from interviews with planners and advocates that there exists a scale of ease of implementation from carrots to sticks, with combined approaches in the center of the scale. The critical factor underlying ease of implementation for all interventions, however, is funding. The NTPP cities are a unique case study in that the primary barrier to all interventions was significantly red uced with the influx of $25 million in federal funds. Once funding is removed as a barrier to implementation, the challenges and difficulties of implementing carrots, sticks, or combined approaches becomes clear. The distinction between carrots, sticks, a nd combined approaches is blurred in the NTPP examples, as it depends context and the user perspective. From the perspective of an existing NMT user, there is no distinction between carrots and combined approaches. For drivers, however, sticks and combined approaches are have similar impacts on the

PAGE 166

150 driver's perception of his current mode choice, but combined approaches offer an alternative. The degree to which combined approaches, on the small scale (e.g., bicycle boulevards) or natural experiments of combi ned approaches (e.g., external sticks) may induce mode shift or negative and reactive behaviors from drivers is unclear. The example of modest sticks in Columbia suggests that implementing any intervention that may be perceived by automobile users (i.e., t he general public) as a stick must be accompanied by education and outreach. Discussion and Conclusion This chapter addresses two related but distinct questions, the first asks what types of interventions may be most effective, and the second, what is the ease of implementation for various interventions. Using multiple methods, I have identified a significa nt challenge for cities interested in encouraging mode shifts from driving to NMT: the interventions which may be most effective face significant barriers to implementation. The structural equation model provides empirical support for the assertion that co mbined approaches may be more effective than carrots alone. However, in terms of ease of implementation, combining carrots and sticks measures are more difficult to implement than carrots alone.

PAGE 167

151 Figure 5 Summary of Multiple Methods Findings Figure 19 summarizes the multiple methods findings described in this chapter. Sticks are described as "high" for both challenges to implement and impact, but a critical caveat is that while sticks alone may have a strong im pact, that impact may be inequitably distribute; future research is needed to confirm this assertion. Planners and policy makers looking to encourage mode shifts to NMT with minimal challenges for implementation, in the absence of external sticks, should s trive for combined interventions. The general public is supportive of carrots that promote NMT, assuming funding is available and NMT infrastructure does not impede existing auto rights of way. The combined approaches identified in the NTPP communities o nly modestly limit or discourage driving along certain corridors, and were completed with NTPP funds (i.e.,

PAGE 168

152 auto infrastructure funding was not used for NMT facilities); however, public resistance to treatments such as bicycle boulevards can be extreme and delay or derail interventions, which can improve neighborhood safety and livability. It is currently unclear whether public resistance comes from a vocal minority, but regardless, such resistance is a real barrier to innovative NMT policies and infrastruc ture. The ABC model is an informative lens to consider the impacts of interventions in US communities. First, assume that even the most supportive contexts for NMT in the US are at best creating a neutral setting. Then, in this context, multiple mode cho ices are available, and attitudes will more strongly correlate with behavior (moderate congestion, ubiquitous free parking, and high auto mode share in the NTPP cities support this assertion). So NMT interventions may be only supporting the behavior of tho se currently pre disposed to walk or cycle. But this is the extent of the ABC model's explanatory power in this context; to understand city scale mode shifts, the role of time in the behavior change process should be taken into account. Figure 6 Transtheoretical Model ( Prochaska & Diclemente, 1983 ) The trans theoretical model (TTM) introduces temporal aspects to behavior change, and considers a continuum ranging from pre contemplation and contemplation,

PAGE 169

153 to behavior chan ge, an d finally, maintenance (Figure 20 ). The TTM is useful to inform future research into the impact of NMT interventions and their ease of implementation. Consider the hypothesis that, to realize significant mode shifts, cities must make significant chan ges to prioritize NMT instead of driving (Meyer, 1999) Currently, significant changes face extreme public resistance. The first step is building public support for NMT interventions that may include modest dis incentives to driving (combined approaches). As US cities implement carrots that promote a neutral context that supports multiple modes, the attitude behavior link is strengthened (according to the ABC Model). Thus, influencing attitudes becomes crucial to building public support for combined inter ventions, which reduces barriers to implementation. The cyclical process of public perceptions informing the built environment, which in turn leads to further changes in public perceptions (Krizek & Handy, 2009) functions as steps similar to those posite d in the TTM. Carrots to promote NMT may spur contemplation of the activity. Combined interventions or situations that act as carrots for NMT and sticks against driving may in turn spur NMT use. Occasional or even one time NMT use can impact attitudes, lea ding to public support (or reduced public resistance) to combined interventions. For example, the cost and difficulty of parking to reach a holiday parade downtown, and the availability of bike routes, may influence an individual's decision to bicycle to a n event. This experience, and the knowledge that bicycling downtown is a possibility, may improve individual perceptions of bicycling as a viable transportation mode. This individual may then support a proposed bicycle boulevard in their neighborhood. If t he bicycle boulevard successfully reduces vehicle through traffic,

PAGE 170

154 increases safety, and enhances neighborhood quality, the neighborhood will likely support further NMT interventions. There is evidence that combined approaches are most effective, but imple menting them takes time and requires building public support. US cities face a choice, and that choice depends on goals. Implementing carrots alone may have the long term benefit of preparing a city for external sticks, whereas combining carrots and stic ks is more likely to spur mode shifts. In Marin County, for example, external sticks may manifest sooner and more strongly than in other cities, and this strategy may be appropriate. But preparing for external sticks assumes a dynamic environment. In a She boygan County, for example, such a strategy may be much less effective in eventually supporting mode shifts. Sheboygan County has less growth and development than the other NTPP cities, and no shortage of available land. In this context, congestion or park ing cost and availability will not likely become powerful external sticks. Significant increases in auto related costs may be the only powerful external stick that, when combined with the extensive NMT network completed under the NTPP, can impact travel be havior. A key finding of this chapter is a dearth of NMT evaluation research. Research and evaluation provides visibility and credibility (Bosch & Nanda, 2011) to practice. Collecting data provides visibility for interventions, and bike/ped counts in the NTPP cities provide that. Credibility, however, requires high quality data that can support generalizable findings, and such data is lacking in the NTPP communities. Pursuing further population level surveys can help identify the extent to which NMT interv entions

PAGE 171

155 are supporting the behavior of existing NMT users or changing behavior. The timeline between attitude change and behavior change is unclear and the moderating effect of the environment further complicates this issue, but without time series data, t hese relationships remain speculative. The NTPP community surveys provide a baseline measure of attitudes and behavior in each of the cities, as well as an observation after some interventions were put in place. All of the NTPP cities are still in the proc ess of completing interventions that were funded through the pilot program, so context continues to change in these cities. Population surveys were conducted in 2006 and 2010, but a third observation would provide valuable data for evaluation and a possibi lity to shed further light on the application of the ABC theory to travel behavior. Ultimately, while this chapter suggests that some combination of carrots and sticks incentives and dis incentives is most effective at promoting NMT, carrots alone ap pear to serve an important role. Without basic infrastructure to enable options, sticks have numerous drawbacks, and carrots alone may work to impact social norms and ease implementation of more effective options. Once the context in US cities reaches a po int that supports multiple modes, then policymakers must weigh the importance of sustainability issues and TDM goals, and make choices accordingly. Transportation and auto reliance is a polarizing issue in the US because of land use practices and regulator y policies reaching back decades (Levine, 2008). While methods for achieving significant mode shifts are availab l e and evidence based, institutional capacity and public support to act on this evidence is unclear.

PAGE 172

156 CHAPTER VI CONCLUSIONS AND CONTRIBUTIONS Introduction The dissertation research was approached with two key goals in mind: Identifying impacts of interventions aimed at promoting walking and bicycling, and applying these findings to inform future NMT promotion and evaluation effort s. Multiple approaches were used to address these goals, qualitative and quantitative. The background and context of the NTPP in each city was established using key informant interviews. Impacts of interventions were identified in the NTPP cities through a quantitative analysis of repeat cross sectional community surveys. An exploratory quantitative analysis found that interventions impact individual perceptions in distinct ways, and this relationship was in turn tested to provide insights for intervention effectiveness. Finally, a multiple methods evaluation of the NTPP provides directions for intervention implementation. This final chapter reviews the methods, findings, and implications of the research then closes with a discussion of future research direc tions (see Figure 2 1).

PAGE 173

157 Figure 1 Overview of Research Completed Figure 2 1 provides an overview and summary of the four research efforts completed and reported within the dissertation. The first research effort, Identify ing impacts of the NTPP, introduces the nuanced nature of the NTPP across four treatment cities. Describing the focus, goals, and challenges in each community as well as possible impacts of the NTPP on the treatment community. The remaining research effort s use both qualitative and quantitative method s to identify previously unmeasured metrics as impacts of NTPP interventions, and provide lessons for intervention implementation and efficacy. Applications of this work to research and practice are described b elow.

PAGE 174

158 Expanding Definitions of Interventions Impacts NMT interventions are typically evaluated on the basis of their ability to induce behavior change, but this evaluation framework carries certain assumptions and methodological challenges. The stated go al of the NTPP is to impact significant mode shifts in the population through NMT interventions. Earlier attempts to quantify the impact of NTPP interventions focused solely on behavior change as a metric for effectiveness. But focusing on this singular ou tcome may be overlooking key impacts, particularly given that both affecting and measuring population mode shifts can be both difficult and costly. As a result, intervention effectiveness may be underestimated and the case for future interventions may be w eakened. I argue in this dissertation for the expansion of outcome measures used to evaluate NMT interventions. This argument is first grounded in interview data reported in Chapter Two, which i dentifies a "moderating effect" of the NTPP. Conceptualizing the NTPP as "raising the bar" across a number of metrics, including type and extent of interventions put in place, and institutional capacity increases due to a relatively modest (in terms of cit y scale transportation funding) influx of funds. Context matters, and evaluating the NTPP based on quantity of interventions is not appropriate. In cities lacking basic NMT infrastructure, extensive additions were completed, and in the cities whose needs h ad moved beyond basic infrastructure, the combination of funding and institutional capacity led to innovative and unique interventions. The exploratory findings in chapter two inform background and context for the dissertation, but it is chapters three,

PAGE 175

159 fo ur, and five that provide important additions to research and practice by empirically testing and evaluating impacts of interventions that have not previously been assessed in the literature. In chapter three, I posit that it may be appropriate to measur e changes in attitudes regarding NMT in the NTPP cities as evidence of intervention effectiveness. This proposition is first grounded in theories of voluntary behavior change, particularly the Theory of Planned Behavior (TPB) and the Trans Theoretical Mode l (TTM). The work presented in this chapter is a novel addition to the literature because it demonstrates statistically significant changes in attitudes regarding NMT over the same time period as the NTPP. Previous studies either focus on modest changes in behavior over time, or have find positive correlations between attitudes and behaviors, but this work is the first to identify attitude change over time (as opposed to behavior change over time). Attitude change may be a useful metric for practitioners in terested in identifying population level impacts in NMT a behavior notoriously difficult to capture using probability based survey methods. Future research is needed regarding the attitude behavior connection in travel behavior, and this work offers prel iminary insights into this issue. Carrots versus Sticks in the United States Further exploration of the attitude data included in the NTPP community surveys identified two broad categories underlying the data. One category related to perceived neighborhood quality for NMT a measure of carrots, or NMT promotion in a neighborhood. The second category was related to stated willingness to switch to NMT as

PAGE 176

160 driving becomes less convenient sticks that discourage the primary alternative to driving. NMT intervention literature is lacking evaluations of sticks, and most importantly, how s ticks and carrots interact Chapter four proposes that by examining the interrelationship between these latent constructs, travel behavior researchers can provide evidence to support the implementation of so called combined approaches to NMT promotion. C hapter four operationalizes and tests the interrelationship between carrots and sticks in a novel application of structural equation models (SEM) in the travel behavior literature. Operationalizing a decision making process, such as a mode choice, is compl icated, but at its core involves weighing options. In terms of carrots and sticks, an individual's consideration of carrots, or the relative quality and feasibility of using NMT, may depend in part on the attractiveness of driving instead. This type of rel ationship specifies sticks regarding driving as a mediator and was tested as such in a structural equation framework. Findings illustrate the potential of this application of data and methods, and inform policy recommendations arrived at separately in the qualitative analysis reported in chapter five. Chapter five builds on quantitative findings that support the assertion that combining carrots and sticks may be most effective and uses Stern's Attitude Belief Context Model of behavior as a framework for p ostulating the impacts of various types of interventions. For example, in a context that is highly unsupportive of NMT (as is the case in most US cities), carrots alone can have three possible impacts: (1) provide mode

PAGE 177

161 choice options, (2) enable individual s predisposed to NMT to begin walking and bicycling, or (3) increase existing rates of walking and bicycling. Once cities reach a saturation point in which transportation infrastructure supports multiple mode choices, the impact of carrots may weaken. The question then becomes, what can cities do to induce significant behavior change among those not previously inclined to use NMT? The results of the SEM presented in chapter four show that combining carrots and sticks may be most effective at influencing mode ch oice decisions. Chapter five incorporates the results of key informant interviews with chapter four's SEM to provide policy relevant directions for promoting NMT; weighing intervention effectiveness against ease of implementation. On the one hand, c arrot interventions face the fewest barriers to implementation, but cities reach a point of diminishing returns once multi modal infrastructure is established. On the other hand, sticks alone carry powerful barriers to implementation and their impacts may be inequitable. This research suggests that practitioners should focus on "quality over quantity" and seek opportunities to implement interventions that are primarily carrots but may include modest sticks (or perceived sticks). Such an approach has long be en advocated, but until now has lacked empirical support. Future Research There are three related directions for future research identified in this dissertation. The first involves an expansion of existing quantitative analysis using structural equation models, incorporating changes over time into the analysis. Similarly, the second research

PAGE 178

162 direction would focus on incorporating behavioral outcomes to the structural model. Finally, future intervention research should more closely examine lag effects betw een implementation, attitudinal outcomes, and behavioral outcomes. Behavioral Outcomes and Changes Over Time The structur al equation model presented in c hapters 4 and 5 specifically tests the degree to which carrots are more effective in the presence o f sticks. This model represents the first step in a more extensive quantitative analysis. First, the model presents a cross sectional analysis of the NTPP data, and a logical extension of the degree to which carrots and sticks interrelate, is how that rela tionship has changed from the 2006 to the 2010 observations. Additionally, regular NMT use was included in the existing models as a crude approximation for habitual/current behavior, but another logical extension of the model is to add behavior as an outco me measure. Travel behavior, and the factors that impact such behavior, may change dramatically based on the trip type. This is likely especially true with perceived carrots and sticks, and the NTPP data may provide an opportunity to examine the impact o f carrots and sticks on specific trip types (unfortunately, the data cannot support a tour based analysis). The NTPP data will require extensive manual coding to create appropriate outcome measures. Survey respondents provided origin, destination, and mode choice data on 1 3 reference trips. These reference trips were not coded by trip type, and origins and destinations are listed by address and name of establishment visited. While respondents were asked about their typical mode to work, walking and bicycli ng

PAGE 179

163 made up 4% and 1.6% of responses, respectively, across the combined (phase 1 and 2) dataset. For this reason, the dissertation research focused on the interaction between latent variables and will pursue behavioral outcomes and change over time measures in later research. Lag Effects One proposition in the quantitative analysis presented in Chapter 2 is that attitude change precedes behavior change, and that there exists a "lag effect" between attitude change and behavior change. There is a dearth of literature on this lag effect, but an increased intere st in longitudinal data sources and time series analysis in NMT. Unfortunately, shedding further light on the lag effects of interventions using the NTPP community surveys would require a "Phase 3" surv ey, but due to funding constraints this is highly unlikely. Instead, opportunities for natural experiments still exist within the NTPP communities because most of the cities have not (at the time of writing) completed all NTPP funded interventions. Without the capacity for costly, probability sampled data, research should focus on panel studies with modest sample sizes to assess impacts of interventions over time and shed light on lag effects.

PAGE 180

164 REFERENCES Aditjandra, P. T., Cao, X. (Jason), & Mulley, C. (2012). Understanding neighbourhood design impact on travel behaviour: An application of structural equations model to a British metropolitan data. Transportation Research Part A: Policy and Practice 46 (1), 22 32. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision 50 179 211. American Community Survey (ACS, 2010). 2010 Population Estimates. Retrieved from: Ana ble, J. (2005). "Complacent Car Addicts" or "Aspiring Environmentalists"? Identifying travel behaviour segments using attitude theory. Transport Policy 12 (1), 65 78. Attride Stirling, J. (2001). Thematic networks: an analytic tool for qualitative researc h. Qualitative Research 1 (3), 385 405. Aytur, S., Rodriguez, D. a, Evenson, K. R., & Catellier, D. J. (2008). Urban containment policies and physical activity. A time series analysis of metropolitan areas, 1990 2002. American journal of preventive medici ne 34 (4), 320 32.

PAGE 181

165 Aytur, S., Rodriguez, D. a, Evenson, K. R., Catellier, D. J., & Rosamond, W. D. (2008). The sociodemographics of land use planning: relationships to physical activity, accessibility, and equity. Health & place 14 (3), 367 85. Bamberg, S., Ajzen, I., & Schmidt, P. (2003). Choice of travel mode in the theory of planned behavior: The roles of past behavior, habit, and reasoned action. Basic and Applied Social Psychology 25(3), 175 187. Bamberg, S., & Schmidt, P. (2003). Incentives, Moral ity, Or Habit?: Predicting Students' Car Use for University Routes With the Models of Ajzen, Schwartz, and Triandis. Environment & Behavior 35 (2), 264 285. Bamberg, S., Hunecke, M., & Blšbaum, A. (2007). Social context, personal norms and the use of publ ic transportation: Two field studies. Journal of Environmental Psychology 27 (3), 190 203. Barnes, G., & Krizek, K. (2005). Tools for Predicting Usage and Benefits of Urban Bicycle Network Improvements. Minnesota Department of Transportation, Report NO: 2 005 50. Retrieved from: Baron, R., & Kenny, D. (1986). The moderator mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51 1173 1182. Basit, T. (2003). Manual or electronic? The role of coding in qualitative data analysis. Educational Research 45 (2), 143 154. doi:10.1080/0013188032000133548

PAGE 182

166 Bassett, D. R., Pucher, J., Buehler, R., Thompson, D. L., & Crouter, S. E. (2008). Walking, cycling, and obesity rates in Europe, North America, and Australia. Journal of Physical Activity & Health 5 (6), 795 814. Bax, C. (2011). Policy Instruments for managing EU Road Safety Targets: Carots, sticks, or sermons? SWOV Institute for Road Safety Research, Report NO: R 2011 15. Retrieved from: 2011 15.pdf. Bauman, A. E., Sallis, J. F., Dzewaltowski, D. a, & Owen, N. (2002). Toward a better understanding of the influences on physical activi ty. American Journal of Preventive Medicine 23 (2), 5 14. Ben Elia, E., & Ettema, D. (2009). Carrots versus sticks: Rewarding commuters for avoiding the rush hour a study of willingness to participate. Transport Policy 16 (2), 68 76. Boarnet, M. G., Anderson, C. L., Day, K., McMillan, T., & Alfonzo, M. (2005). Evaluation of the California Safe Routes to School legislation: urban form changes and children's active transportation to school. American journal of preventive medicine 28 (2 S uppl 2), 134 40. Boer, R., Zheng, Y., Overton, A., Ridgeway, G. K., & Cohen, D. a. (2007). Neighborhood design and walking trips in ten U.S. metropolitan areas. American journal of preventive medicine 32 (4), 298 304.

PAGE 183

167 Bosch, S., & Nanda, U. (2011). Outsi de the ivory tower: The role of healthcare design researchers in practice. Journal of Interior Design 36 (2). Bršg, W., Erl, E., Ker, I., Ryle, J., & Wall, R. (2009). Evaluation of voluntary travel behaviour change: Experiences from three continents. Tran sport Policy 16 (6), 281 292. Buehler, R. (2010). Transport Policies, Automobile Use, and Sustainable Transport: A Comparison of Germany and the United States. Journal of Planning Education and Research 30 (1), 76 93. Buehler, Ralph. (2012). Determinants of bicycle commuting in the Washington, DC region: The role of bicycle parking, cyclist showers, and free car parking at work. Transportation Research Part D: Transport and Environment 17 (7), 525 531. Cao, X., Mokhtarian, P. L., & Handy, S. L. (2009). E xamining the Impacts of Residential Self Selection on Travel Behaviour: A Focus on Empirical Findings Transport Reviews (Vol. 29, pp. 359 395). Chriqui, J. F., Taber, D. R., Slater, S. J., Turner, L., Lowrey, K. M., & Chaloupka, F. J. (2012). The impact of state safe routes to school related laws on active travel to school policies and practices in U.S. elementary schools. Health & Place 18 (1), 8 15. Clifton, K., & Krizek, K. (2004). The Utility of the NHTS in understanding bicycle and pedestrian travel Proceedings of the National Household Travel Survey

PAGE 184

168 Conference: Understanding our Nation's Travel Retrieved from Dannenberg, A. L., Jackson, R. J., Frumkin, H., Schieber, R. a, Pra tt, M., Kochtitzky, C., & Tilson, H. H. (2003). The impact of community design and land use choices on public health: a scientific research agenda. American journal of public health 93 (9), 1500 8. Dietz, T., & Gardner, G. (2009). Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Retrieved from Dill, J. (2010). Long Term Evaluation of Individualized Marketing Programs for Travel Demand Management, (July). Retrieved from RR 10 08_interior.pdf Dill, Jennifer, & Voros, K. (2007). Factors Affecting Bicycling Demand: Initial Survey Findings from the Portland, Oregon, Region. Transportation Research Record 2031 ( 1), 9 17. Dodson, J., Burke, M., Evans, R., Gleeson, B., & Sipe, N. (2010). Travel Behavior Patterns of Different Socially Disadvantaged Groups. Transportation Research Record: Journal of the Transportation Research Board 2163 ( 1), 24 31.

PAGE 185

169 Eliasson, J. & Mattsson, L. G. (2006). Equity effects of congestion pricing. Transportation Research Part A: Policy and Practice 40 (7), 602 620. Enders, C., & Bandalos, D. (2001). The Relative Performance of Full Information Maximum Likelihood Estimation for Missin g Data in Structural Equation Models. Structural Equation Modeling: A Multidisciplinary Journal 8 (3), 430 457. Ewing, R., & Cervero, R. (2010). Travel and the Built Environment. Journal of the American Planning Association 76 (3), 265 294. Federal Highw ay Administration (2012). Report to the US Congress on the Outcomes of the Nonmotorized Transportation Pilot Program SAFETEA LU Section 1807. US Department of Transportatoin's Volpe National Transportation Systems Center. Fields, B., & Hull, T. (2013). Po licy implications of the Nonmotorized Transportation Pilot Program: Redefining the transportation solution. In Renne, J.L., & Fields., B. (Eds.), Transport beyond oil; Policy choices for a multimodal future (Chapter 15). Island Press: Washington, DC. Fors yth, A., Krizek, K., & Rodriguez, D. (2009). Hot, Congested, Crowded and Diverse: Emerging research agendas in planning. Progress in Planning 71 153 205. GŠrling, T., & Schuitema, G. (2007). Travel demand management tareting reduced Private Car Use: Eff ectiveness, Public Acceptability and Political Feasibility. Journal of Social Issues 63 (1), 139 153.

PAGE 186

170 Gatersleben, B., & Appleton, K. M. (2007). Contemplating cycling to work: Attitudes and perceptions in different stages of change. Transportation Researc h Part A: Policy and Practice 41 (4), 302 312. Geller, E. (2002). The Challenge of Increasing Proenvironment Behavior. In Bechtel, R., & Churchman, A. (Eds.), Handbook of Environmental Psychology Wiley, New York, NY. Glaser, Barney, G. & Strauss, Anselm L., 1967. The Discovery of Grounded Theory: Strategies for Qualitative Research Aldine Publishing, Chicago, IL. Golob, T. F. (2003). Structural equation modeling for travel behavior research. Transportation Research Part B: Methodological 37 (1), 1 25. Gotschi, T., Krizek, K., McGinnis, L., Lucke, J., Barbeau, J., (2011). Nonmotorized Transportation Pilot Program Evaluation Study Phase 2. Center for Transportation Studies, University of Minnesota Handy, S. L. (1996). Understanding the Link Between Urba n Form and Nonwork Travel Behavior. Journal of Planning Education and Research 15 (3), 183 198. Handy, S. (2005). Critical assessment of the literature on the relationships among transportation, land use, and physical activity. Transportation Research Boa rd and the Institute of Medicine Committee on Physical Activity, Health, Transportation, and Land Use, Washington, DC

PAGE 187

171 Handy, S., Cao, X., & Mokhtarian, P. L. (2006). Self Selection in the Relationship between the Built Environment and Walking: Empirical Evidence from Northern California. Journal of the American Planning Association 72 (1), 55 74. Harrington, W., Krupnick, A. J., & Alberini, A. (2001). Overcoming public aversion to congestion pricing. Transportation Research Part A: Policy and Practice 3 5 (2), 87 105. Heath, G. W., Brownson, R. C., Kruger, J., Miles, R., Powell, K. E., & Ramsey, L. T. (2006). The Effectiveness of Urban Design and Land Use and Transport Policies and Practices to Increase Physical Activity#: A Systematic Review. Heinen, E., Maat, K., & Wee, B. Van. (2011). The role of attitudes toward characteristics of bicycle commuting on the choice to cycle to work over various distances. Transportation Research Part D: Transport and Environment 16 (2), 102 109. Heinen, E., Van Wee, B., & Maat, K. (2010). Commuting by Bicycle: An Overview of the Literature. Transport Reviews 30 (1), 59 96. Heinrich, K. M., Aki, N. N., Hansen Smith, H., Fenton, M., & Maddock, J. (2011). A comprehensive multi level approach for passing safe routes to schoo l and complete streets policies in Hawaii. Journal of physical activity & health 8 Suppl 1 (Suppl 1), S135 40.

PAGE 188

172 Hensher, D. a., & Puckett, S. M. (2007). Congestion and variable user charging as an effective travel demand management instrument. Transportati on Research Part A: Policy and Practice 41 (7), 615 626. Janz, N. K., & Becker, M. H. (1984). The Health Belief Model: a decade later. Health education quarterly 11 (1), 1 47. Retrieved from Jewson, E., Spittle, M., & Casey, M. (2008). A preliminary analysis of barriers, intentions, and attitudes towards moderate physical activity in women who are overweight. Journal of science and medicine in sport / Sports Medicine Australia 11 (6), 558 61. Joh, K., Boarnet, M. G., Nguyen, M. T., Fulton, W., Siembab, W., & Weaver, S. (2008). Accessibility, Travel Behavior, and New Urbanism: Case Study of Mixed Use Centers and Auto Oriented Corridors in the South Bay Region of Los Angeles, California. Transpor tation Research Record: Journal of the Transportation Research Board 2082 ( 1), 81 89. Kenny, D. (1986). Retrieved from: Krizek, K., & Barnes, G. (2009). Analyzing the effect of bicycle facilities on commute mode share over time. Journal of Urban Planning and Development 66 73. Krizek, K.J, Barnes, G., Wilson, R., Johns, R., and McGinnis, L. (2007). Nonmotorized Transportation Pilot Pr ogram Evaluation Study. Federal Highway Administration.

PAGE 189

173 Krizek, K., Forsyth, A., & Baum, L. (2009). Walking and Cycling International Literature Review. Retrieved from and cycling internationa l literature review.pdf Krizek, K., & Handy, S. (2009). Explaining changes in walking and bicycling behavior: challenges for transportation research. Environment and Planning B: 36 725 741. Krizek, K., & Johnson, P. (2005). Proximity to Trails and Cycli ng and Walking; Effects on urban cycling and walking. Journal of the American Planning Association, 27(1), 33 42. Kunchornrat, J., Pairintra, R., & Namprakai, P. (2008). Sustainable energy management in urban transport: The public's response of road conges tion pricing in Thailand. Renewable and Sustainable Energy Reviews 12 (8), 2211 2226. Lajunen, T., & RŠsŠnen, M. (2004). Can social psychological models be used to promote bicycle helmet use among teenagers? A comparison of the Health Belief Model, Theory of Planned Behavior and the Locus of Control. Journal of safety research 35 (1), 115 23. LaPlante, J., & McCann, B. (2008). Complete streets: we can get there from here. ITE JOURNAL (May). Retrieved from content/uploads/2 012/01/Complete Streets We Can Get There from Here.pdf

PAGE 190

174 Lemieux, M., & Godin, G. (2009). How well do cognitive and environmental variables predict active commuting? The international journal of behavioral nutrition and physical activity 6 12. Levin Marti n, S., Moeti, R., & Pullen Seufert, N. (2009). Implementing safe routes to school: application for the socioecological model and issues to consider. Health promotion practice 10 (4), 606 14. Levine, J. (2006). Zoned out: Regulation, markets, and choices i n transportation and metropolitan land use. RFF Press: Washington, DC. Levinson, D., & Krizek, K. (2008). Planning for Place and Plexus: Metropolitan land use and transport. New York, NY: Routledge. Lorenc, T., Brunton, G., Oliver, S., Oliver, K., & Oakle y, a. (2008). Attitudes to walking and cycling among children, young people and parents: a systematic review. Journal of Epidemiology and Community Health 62 (10), 852 7. Lu, X., & Pas, E. I. (1999). Socio demographics, activity participation and travel b ehavior. Transportation Research Part A: Policy and Practice 33 (1), 1 18. MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual review of psychology 58 593 614. Maat, K., Wee, B. Van, & Stead, D. (2005). Land use and travel behaviour: expected effects from the perspective of utility theory and activity based theories. Environment and Planning B: Planning and Design 32 (1), 33 46.

PAGE 191

175 McDonald, N. C., & Aalborg, A. E. (2009). Why Parents Dri ve Children to School: Implications for Safe Routes to School Programs. Journal of the American Planning Association 75 (3), 331 342. McEachan, R. R. C., Sutton, S., & Myers, L. (2010). Mediation of personality influences on physical activity within the t heory of planned behaviour. Journal of health psychology 15 (8), 1170 80. McKenzie Mohr, D. (2011). Fostering Sustainable Behavior; An introduction to Community Based Social Marketing. New Society Publishers, Canada. Meyer, M. (1999). Demand management as an element of transportation policy: using carrots and sticks to influence travel behavior. Transportation Research Part A: Policy and Practice 33 575 599. Miles, M., & Huberman, M. (1994). Qualitative Data Analysis SAGE, Thousand Oaks, CA. Minneapoli s Public Works (MNPW). (2012). Understanding Bicycle Markings in Minneapolis; A guide for motorists and bicyclists. Retrieved from mages/wcms1p 083551.pdf. Mosquera, J., Parra, D. C ., Gomez, L. F., Sarmiento, O., Schmid, T., & Jacoby, E. (2012). An inside look at active transportation in Bogot‡: a qualitative study.

PAGE 192

176 Journal of physical activity & health 9 (6), 776 85. Retrieved from Nisbet, E. K. L., & Gick, M. L. (2008). Can health psychology help the planet? Applying theory and models of health behaviour to environmental actions. Canadian Psychology/Psychologie canadienne 49 (4), 296 303. Norton, P. (2008). Fighting Traffic: The da wn of the motor age in the American city. Cambridge, MA: MIT Press O'Connor, J. P., & Brown, T. D. (2010). Riding with the sharks: serious leisure cyclist's perceptions of sharing the road with motorists. Journal of science and medicine in sport / Sports Medicine Australia 13 (1), 53 8. Ogilvie, D., Egan, M., Hamilton, V., & Petticrew, M. (2004). Promoting walking and cycling as an alternative to using cars: systematic review. British Medical Journal (Clinical research ed.) 329 (7469), 763. Ogilvie, D., Foster, C. E., Rothnie, H., Cavill, N., Hamilton, V., Fitzsimons, C. F., & Mutrie, N. (2007). Interventions to promote walking: systematic review. British Medical Journal (Clinical research ed.) 334 (7605), 1204. Ory, D. T., & Mokhtarian, P. L. (2005). When is getting there half the fun? Modeling the liking for travel. Transportation Research Part A: Policy and Practice 39 (2 3), 97 123.

PAGE 193

177 Ory, D. T., & Mokhtarian, P. L. (2009). Modeling the structural relationships among short distance travel a mounts, perceptions, affections, and desires. Transportation Research Part A: Policy and Practice 43 (1), 26 43. Pednet, Pedestrian and Bicycle Network, Columbia, Missouri. (2008a). Columbia will lower residential street speeds to 25 mph Retrieved from: street speeds lowered.html. Pednet, Pedestrian and Bicycle Network, Columbia, Missouri. (2008b). Bicycle harassment ordnance passes unanimously Retrieved from: es/2009/bicycle harassment ordinance passed.html. Pednet, Pedestrian and Bicycle Network, Columbia, Missouri. (2008c). PedNet Coalition announces official suport for cyclist education, law enforcement, and anti harassment' law Retrieved from: http://www supports anti harassment law.html. Piatkowski, D. & Marshall, W. (2013). "New" versus "Old" Urbanism: A comparative analysis of travel behavior in large scale New Urbanist communities and older, more established n eighborhoods in Denver, Colorado. Forthcoming in Urban Design International

PAGE 194

178 Plaut, P. O. (2005). Non motorized commuting in the US. Transportation Research Part D: Transport and Environment 10 (5), 347 356. Prochaska, J., & DiClemente, C. (1983 ). Stages and Processes of Self Change of Smoking: Toward an integrative model of change Journal of Consulting and Clinical Psychology 51 390 395 Pucher, J., & Buehler, R. (2008). Making Cycling Irresistible: Lessons from The Netherlands, Denmark and Germany. Transport Reviews 28 (4), 495 528. Pucher, J., Buehler, R., & Seinen, M. (2011). Bicycling renaissance in North America? An update and re appraisal of cycling trends and policies. Transportation Research Part A: Policy and Practice 45 (6), 451 475. Pucher, J., & Dijkstra, L. (2003). Promoting safe walking and cycling to improve public health: lessons from The Netherlands and Germany. American journal of public health 93 (9), 1509 16. Retrieved from artid=1448001&tool=pmce ntrez&rendertype=abstract Pucher, J., Dill, J., & Handy, S. (2010). Infrastructure, programs, and policies to increase bicycling: an international review. Preventive medicine 50 Suppl 1 S106 25. Pucher, J., & Komanoff, C. (1999). Bicycling renaissance in North America?:: Recent trends and alternative policies to promote bicycling. Research Part A: Policy and

PAGE 195

179 Practice 33 Retrieved from Ren, F., & Kwan, M. P. (2009) The impact of the Internet on human activity travel patterns: analysis of gender differences using multi group structural equation models. Journal of Transport Geography 17 (6), 440 450. Rodriguez, D., Targa And, F., & Aytur, S. (2006). Transport implic ations of urban containment policies: A study of the largest twenty five US metropolitan areas. Urban Studies 43 (10), 1879 1897. Rose, G., & Marfurt, H. (2007). Travel behaviour change impacts of a major ride to work day event. Transportation Research Pa rt A: Policy and Practice 41 (4), 351 364. Rutt, C. D., & Coleman, K. J. (2005). Examining the relationships among built environment, physical activity, and body mass index in El Paso, TX. Preventive medicine 40 (6), 831 41. Rye, T., & Ison, S. (2005). O vercoming barriers to the implementation of car parking charges at UK workplaces. Transport Policy 12 (1), 57 64. Saelens, B. E., Sallis, J. F., & Frank, L. D. (2003). Environmental correlates of walking and cycling: findings from the transportation, urba n design, and planning literatures. Annals of behavioral medicine!: a publication of the Society of Behavioral Medicine 25 (2), 80 91.

PAGE 196

180 Sallis, J. F., Cervero, R. B., Ascher, W., Henderson, K. a, Kraft, M. K., & Kerr, J. (2006). An ecological approach to c reating active living communities. Annual review of public health 27 297 322. Schneider, R. J. (2013). Theory of routine mode choice decisions: An operational framework to increase sustainable transportation. Transport Policy 25 128 137. Seskin, S., & McCann, B. (2012). Complete Streets Local Policy Workbook. Planning Retrieved from Shoup, D. C. (1997). Evaluating the effects of cashing out employer paid parking: Eight case studies. Transport Policy 4 (4), 201 216. Shoup, D. C. (2004). The ideal source of local public revenue. Regional Science and Urban Economics 34 (6), 753 784. Singleton, R., & Straits, B. (2009). Approaches to Social Research (Fifth Edition) Oxford University Press, USA. Skinn er, B. (1987). Upon Further Reflection Pearson Press, Ann Arbor, MI. Staunton, C. E., Hubsmith, D., & Kallins, W. (2003). Promoting safe walking and biking to school: the Marin County success story. American journal of public health 93 (9), 1431 4. Retri eved from ntrez&rendertype=abstract

PAGE 197

181 Steg, L., & Vlek, C. (2009). Encouraging pro environmental behaviour: An integrative review and research agenda. Journal of Environmental Psychol ogy 29 (3), 309 317. Stern, P. ( 2000 ). Toward a Coherent Theory of Environmentally Significant Behavior. Journal of Social Issues, 56 (3), 407 424 Stern, P. Dietz, T., Abel, T., Guagnano, G. A., & Kalof, L. (1999). A Value Belief Norm Theory of Support for Social Movements#: The Case of Environmentalism, 6 (2), 81 97. Stopher, P., Clifford, E., Swann, N., & Zhang, Y. (2009). Evaluating voluntary travel behaviour change: Suggested guidelines and case studies. Transport Policy 16 (6), 315 324. Wegman, F., Zhang, F., & Dijkstra, A. (2012). How to make more cycling good for road safety? Accident; analysis and prevention 44 (1), 19 29. Transit Cooperative Research Program (TCRP). (2008). Strategies to A ttract Auto Users to Public Transportation. Transportation Research Board Retrieved from: Transit Cooperative Research Program (TCRP). (2010). Employer and Institutional TDM Strategies; Traveler responses to transportation system changes. Transportation Research Board Retrieved from:

PAGE 198

182 Thgersen, J. (2009). Promoting public transport as a subscription service: Effects of a free month tr avel card. Transport Policy 16 (6), 335 343. Thgersen, J., & Mller, B. (2008). Breaking car use habits: The effectiveness of a free one month travelcard. Transportation 35 (3), 329 345. Verhoef, E., Nijkamp, P., & Rietveld, P. (1996). Second Best Conge stion Pricing#: The Case of an Untolled alternative. Journal of Urban Economics 279 302. Verplanken, B., Walker, I., Davis, A., & Jurasek, M. (2008). Context change and travel mode choice: Combining the habit discontinuity and self activation hypotheses. Journal of Environmental Psychology 28 (2), 121 127. Walker, L., Tresidder, M., & Birk, M. (2009). Fundamentals of Bicycle Boulevard Planning & Design. Retrieved from: ok(optimiz ed).pdf Weinberger, R. (2012). Death by a thousand curb cuts: Evidence on the effect of minimum parking requirements on the choice to drive. Transport Policy 20 93 102. Willson, R., & Shoup, D. (1990). Parking subsidies and travel choices: assessing the evidence. Transportation, 17, 141 157 Xing, Y. Contributions of Individual, Physical, and Social Environmental Factors to Bicycling: A structural equations modeling study of six small US cities. Research

PAGE 199

183 Report: UCD ITS RR 12 28. Institute of Transportation Studies, University of California, Davis. Xing, Y., Handy, S. L., & Mokhtarian, P. L. (2010). Factors associated with proportions and miles of bicycling for transportation and recr eation in six small US cities. Transportation Research Part D: Transport and Environment 15 (2), 73 81. Yang, L., Sahlqvist, S., McMinn, a., Griffin, S. J., & Ogilvie, D. (2010). Interventions to promote cycling: systematic review. British Medical Journal 341 (oct18 2), c5293 c5293. Yang wallentin, F. (2004). Is There Any Interaction Effect Between Intention and Perceived Behavioral Control. Methods of Psychological Research Online, 8 (2), 127 157.

PAGE 200

184 APPENDIX A CHAPTER 2 SUPPLEMENTAL DOCUMENTATION This appendix includes supplemental documentation for Chapter 2, including the interview guide and interventions accounting by NTPP city. Interview Guide: NTPP semi structured interviews 1. How would you characterize your community for non motorized travel ( NMT ) ? 2. How would you characterize perceptions in your community regarding NMT ? a. As safe/dangerous? b. As viable for exercise or recreation? both? c. Only for specific people or purposes? 3. Can you please tell me a little bit about your role i n the NTPP in your community? 4. Can you tell me about other "key players" in the NTPP in your community 5. Can you give me a little bit of background on the NTPP in your community? a. How would you characterize your community for NMT prior to the NTPP? b. What were some key goals or initiatives regarding NMT in progress prior to the NTPP? 6. How did involvement in the NTPP impact existing NMT goals or initiatives? a. Did the NTPP allow you to continue working towards existing key goals, change/refocus those goals, or choo se new goals? 7. What were some of the initial goals of the NTPP? a. Have you made progress on those goals? 8. How were decisions made regarding specific projects? 9. How do you think the NTPP has impacted your community? a. Has it changed how people think about NMT ? i. Changing perceptions among NMT users? Among other road users? b. Has it changed how people use nonmotorized modes? i. How much they use NMT? ii. For different purposes/trip types? c. What about how the NTPP has changed how your community is perceived by neighboring c ommunities or nationally ? 10. Can you tell me about some of the most effective interventions in your community? a. Why do you think it/they were so successful? 11. Can you tell about some of the less effective interventions put in place? a. Why would you describe them as less effective? 12. Were there any particular issues that you identified and were looking to address regarding bicycling and walking in your community? a. How were they identified? b. How do you feel the NTPP interventions addressed them? 13. Who do you think are the current bicyclists and peds in your community, and why are they using NMT?

PAGE 201

185 14. Who is not (or is the least likely to be) using NMT in your community, and are they a population you are trying to target for more NMT use? 15. Who (what population) would you most wan t to promote NMT, and why? 16. Are there targeted neighborhoods where you are trying to promote NMT? 17. Are there targeted populations where you are trying to promote NMT? 18. What types of activities have done through from 2006 2010 to promote NMT? 19. What was the spec ific goal (within promoting NMT) of these activities? 20. To what extent do you think the interventions worked as planned and implemented? 21. How long did you find it took interventions to show impacts in your community? (when they were implemented/programmed and when results were seen) a. What were the impacts you saw, and have you continued to see such impacts increase in magnitude level off, or decrease to earlier levels ?

PAGE 202

186 Table 1 8 Intervention Accounting: Columbia, MO (Source: Krizek, et al., 2007) Columbia, MO Project Type Project Status Infrastructure Missouri Kansas Texas Rail Trail (MKT) Hinson Creek and Bear Creek trail projects with six neighborhood connections Acquisition of additional trail ROW for four trails Downtown and University of Missouri Columbia hub/spoke bicycle lanes Demonstration bicycle route project in downtown Three intersection projects Five bridge overpass projects Demonstration grate re placement project Downtown bicycle racks University projects (including shelters, racks, striping, and trail extensions) Neighborhood and school area sidewalks Three pedestrian walkways Deemed highest priority projects Education Bike safety courses Seminars for engineers Targeted group clinics Shift program Adult ed night courses Funded; starting in 2007 Public Awareness Web site Project Office (e.g. storefront) Print and e mail newsletters Social Marketing Media relations Event participation Funded; starting in 2007 Encouragement and Support Errand bikes Earn a bike program Bike, Walk, and Wheel Week Walking School Bus Sunday Street Closings Other experimental or demonstration projects Funded; starting in 2007 Assessments/ Surveys Manual counts Automated counts Funded; starting in 2007 Wayfinding Maps On street markings Funded; starting in 2007

PAGE 203

187 Table 19 Intervention Accounting: Marin County, CA (Source: Krizek, et al., 2007) Marin County, CA Project Type Project Status Infrastructure Gate Six Rd/Bridgeway Intersection Improvements San Rafael Transit Center Improvements Enfrente Road Connector Bridgeway to Ferry Path Puerto Suello to Transit Center connection Mahon Creek Path to Transit Center connection Northgate Gap Closure Reserve funding for Cal Park Tunnel Pathway and Puerto Suello Pathway Alameda del Prado Sir Francis Drake sidewalk and crosswalk improvements in Ross, Fairfax, and San Anselmo Tenness ee Valley Path Doherty Drive Manzanita Connector Medway Road Improvements Terra Linda/Freitas Parkway Multiple site, countywide projects including bicycle racks and lockers, signing/striping, minor intersection improvements, and steps, lanes, and pa thways All projects are funded Planning Central Marin Ferry Connection Alto Tunnel/Mill Valley Corte Madera Divide Access Study San Rafael to Fairfax Corridor Study Bridgeway Path Francisco Blvd. East Improvements Miller Creek Las Gallinas Improvements All projects are funded Education Bicycle education/street skills Riding with Youth workshops Facility Design Seminars for Engineers Safety Campaign development All projects are funded Public Awareness Street S marts program Health promotion, co sponsored with County Health Dept. Share the Road/Share the Path program Informational booths at community events All projects are funded Resources Bicycle repair classes or programs Maps for directional signage Community pathway/ walking maps All projects are funded Incentives Personal travel planning All projects are funded

PAGE 204

188 Table 20 Intervention Accounting: Minneapolis, MN (Source: Krizek, et al., 2007) Minneapolis, MN Project Type Project Status Infrastructure Bicycle parking in Minneapolis (at various locations, including schools, employment centers, recreation facilities, transit stops, and other community activity centers) A construction project which will provide a significant travel connection between high traffic destinations of the University of Minnesota main campus, downtown Minneapolis, and the University of Minnesota St. Paul campus A 3.23 mile project in South Minneapolis that will include a bicycle boulevard treatment, pro viding an alternative for bicyclists to the heavy arterials between several neighborhoods and downtown Minneapolis Funded Planning Pedestrian Plan for the City of Minneapolis A study to develop a Central Corridor Bicycle and Pedestrian Plan. The study will build upon the Central Corridor Development Strategy, to determine where bicycle and pedestrian connections can be created or improved to the anticipated light rail line between downtown Minneapolis and downtown St. Paul The Douglas Drive Corridor E nhancement and Connection to Luce Line Trail Study seeks to provide a safe nonmotorized connection to Minneapolis. This will focus on land use issues and trail and sidewalk improvements, enhancing an important suburban travel route to Minneapolis Funded Educational/ Promotion "Bike Walk Twin Cities" promotional campaign "Bike Walk Ambassadors" "Sibley Bike Depot" Bike library "Smart Trips St. Paul" educational campaign Funded

PAGE 205

189 Table 21 Intervention Accounting: Sheboygan County, WI (Source: Krizek, et al., 2007) Sheboygan County, WI Project Type Project Status Infrastructure Bike rack installation at County facilities Town of Sheboygan bicycle/pedestrian facility on Mueller Road City of Sheboygan bike racks on buses City of Plymouth sidewalk construction on Eastern Avenue and Highland Avenue Village of Howards Grove sidewalk construction and bike lane striping on Millersville Road between Elk Street and Highway 32 Village of Oostburg sidewalk on north side of school district campus from 6th to 8th Street Countywide bike lane striping initiative for urban areas Paved shoulders on County Highway A/J in the Village of Elkhart Lake Paved shoulders on Sunset Drive in the City of Plymouth connecting the City with large employers including Sargento Village of Random Lake/Town of Sherman pathways, paved shoulders, and sidewalks eliminates school busing to surrounding neighborhoods Sidewalks, pathways, and bike lane stri ping on Audubon Road and Mill Street in the Village of Howards Grove Paved shoulders on CTH A connecting the Village of Howards Grove with Lakeland College Paved shoulders on CTH PP connecting the City of Plymouth with new industrial park Village of Ad ell sidewalk network updates CTH O updates to include sidewalks, bike lanes, and paved shoulders City of Sheboygan Falls Comprehensive buildout to include bike lanes, road diets, pathways, sidewalk gap updates, and signage Village of Cedar Grove sidewal ks and bike lanes on South Main Street eliminates school busing to surrounding neighborhoods Village of Cedar Grove pathway between new subdivisions and school campus eliminates school busing to surrounding neighborhoods All projects are funded Education/ Promotion Village of Elkhart Lake Safe Routes to Schools Bike to Work Week focusing on the city of Sheboygan, Sheboygan Falls, Plymouth, and the All projects are funded

PAGE 206

190 village of Kohler Countywide "walk and bike to school days" (two events each year) Association of Pedestrian & Bicycle Professional/League of American Cyclists Bicycle Friendly Community Workshops WE Bike, etc. law enforcement training program Planning/ Research/ Policy Countywide planning for the Safe Routes to School progra m Update of the comprehensive pedestrian and bicycle plan to better enable Sheboygan County to plan for the programs and projects that move forward as part of the NTPP. The plan extends past the end of the NTPP to help the county continue to enhance its pedestrian and bicycle programs well into the future. All projects are funded Table 22 Intervention Accounting: Spokane, WA (Source: Krizek, et al., 2007) Spokane, WA Project Type Project Status Infrastructure Additional ~10 miles of shared use paths Additional ~25 miles striped bike lanes At least three ARRA funded roadway improvements Funded and completed Planning/ Administrative/ Advisory Bike/walk master plan Bicycle and pedestrian coordinator Bicycle advisory board in place Completed

PAGE 207

191 APPENDIX B CHAPTER 3 SUPPLEMENTAL DOCUMENTATION This appendix includes supplemental documentation for Chapter 3, including detailed coding schemes and preliminary statistical analyses. Table 23 Coding Scheme for Household Survey Variables Independent Variable Coding Scheme Constant (i.e., Spokane) Indicator: 0 = No, 1 = Yes Columbia Indicator: 0 = No, 1 = Yes Marin County Indicator: 0 = No, 1 = Yes Minneapolis Indicator: 0 = No, 1 = Yes Sheboygan Indicator: 0 = No, 1 = Yes Phase Binary: 0 = No, 1 = Yes Income 1 = $0 $14,999 2 = $15,000 24,999 3 = $25,000 $34,999 4 = $35,000 $49,999 5 = $50,000 $74,999 6 = $75,000 $99,999 7 = $100,000 or more Gender Binary: 0 = Female, 1 = Male Age Numeric (years) Children Numeric (number of children) Race 0 = All others, 1 = White

PAGE 208

192 Table 1 4 Independent Samples T tests of Attitude Variables by NTPP City Variable code Mean (SD) t df Phase 1 Phase 2 All Cities (Complete Dataset) When do you estimate the last time you rode a bike in your city? 3.66 (.88) 3.87 (1.05) 5.745* 2761 The cost of parking and driving increased 2 (1.14) 1.91 (1.09) 1.804* 1813 More destinations close to work 2.31 (1.21) 2.12 (1.13) 3.008*** 1322 More marked bike lanes on existing streets 2.43 (1.26) 2.33 (1.23) 1.587* 1860 More lights on existing bicycle facilities 2.16 (1.19) 1.95 (1.08) 3.853*** 1802 Safer intersections (with regard to motorists) 2.4 (1.24) 2.34 (1.21) 1.097* 1893 Safer or better bike parking 2.21 (1.20) 2.11 (1.15) 1.866*** 1790 Showers available at my destinations 1.64 (1.06) 1.53 (.96) 2.418*** 1829 Areas free from crime 2.26 (1.27) 2.21 (1.19) 0.708*** 1759 There are bike lanes, paths or routes that connect to my home 2.26 (1.13) 2.41 (1.18) 1.914* 924 The bike route network has big gaps 2.78 (1.07) 2.67 (1.14) 1.395* 837 Intersections have push buttons or sensors for bikes and peds 2.4 (1.15) 2.45 (1.22) 0.518* 797 Columbia, MO The cost of parking and driving increased c7g 2.0 (1.20) 1.75 (.997) 2.255*** 375 More destinations close to work c7i 2.44 (1.17) 2.15 (1.04) 2.178* 274 More marked bike lanes on existing streets c8a 2.37 (1.29) 2.01 (1.12) 2.918*** 382 More lights on existing bicycle facilities c8c 2.11 (1.19) 1.79 (.997) 2.678*** 361 Safer intersections (with regard to motorists) c8d 2.54 (1.23) 2.20 (1.16) 2.749* 387 Areas free from crime c8h 2.25 (1.23) 2.04 (1.13) 1.64* 357 The streets in my neighborhood are hilly, making it difficult to walk c6e 2.21 (.988) 2.27 (1.152 0.005* 0.443 The crime rate in my neighborhood makes it unsafe to go on walks c6k 1.43 (.8) 1.59 (.9) 1.457* 218 There are bike lanes, paths or routes that connect to my home c6o 1.78 (.941) 2.59 (1.09) 5.021* 160 Stores and other destinations have bike racks c6r 2.16 (.929) 2.56 (1.143) 2.466* 161 Marin County, CA The cost of parking and driving increased c7g 1.93 (1.14) 1.83 (1.04) 0.836* 313 Showers available at my destinations c8f 1.64 (1.06) 1.47 (.916) 1.521* 325 Minneapolis, MN Areas free from crime c7d 2.93 (1.14) 2.7 (1.21) 1.96* 415 Showers available at my destinations c8f 1.88 (1.17) 1.64 (1.03) 2.086* 412 There are crosswalks and pedestrian signals in my neighborhood c6i 3.28 (.85) 3.15 (1.03) 1.147* 258 The crime rate in my neighborhood makes it unsafe to go on walks c6k 1.42 (.737) 1.53 (.93) 0.982* 245 The city has a network of off street bicycle paths c6m 3.16 (.757) 3.41 (.83) 2.414* 254 Streets without bike lanes are generally wide enough to bike on c6n 2.6 (.89) 2.64 (1.08) 0.295* 234 Intersections have push buttons or sensors for bikes and peds c6s 2.84 (.98) 2.67 (1.15) 1.093* 176 Sheboygan County, WI More lights in walking area c7e 2.23 (1.20) 2.04 (1.12) 1.536* 337 More lights on existing bicycle facilities c8c 2.21 (1.24) 1.97 (1.09) 1.85* 346 Areas free from crime c8h 2.2 (1.32) 2.17 (1.21) 0.27* 349 Stores are within easy walking distance of my home c6a 2.39 (1.11) 2.24 (1.24) 0.945* 224 It is easy to bicycle to a transit stop (bus, train) from my home c6d 2.61 (1.22) 2.46 (1.38) 0.8* 205 Spokane, WA Areas free from fast moving traffic c7f 2.44 (1.24) 2.41 (1.13) 0.206* 362 More lights on existing bicycle facilities c8c 2.13 (1.16) 1.84 (1.02) 2.502* 370 Safer or better bike parking c8e 2.33 (1.29) 2.15 (1.18) 1.399* 357

PAGE 209

193 Table 24 cont. Areas free from crime c8h 2.32 (1.28) 2.38 (1.19) 0.491* 365 Stores are within easy walking distance of my home c6a 2.49 (1.29) 2.45 (1.18) 0.295* 260 There are many places to go within easy walking distance c6b 2.83 (1.12) 2.65 (1.23) 1.185* 247 It is easy to walk to a transit stop (bus, train) from my home c6c 3.41 (1.17) 3.24 (1.87) 0.817*** 254 It is easy to bicycle to a transit stop (bus, train) from my home c6d 3.19 (1.09) 2.82 (1.30) 2.179* 219 The crime rate in my neighborhood makes it unsafe to go on walks c6k 1.70 (1.01) 1.37 (.78) 2.674*** 209 The bike route network has big gaps c6p 3.0 (.96) 2.84 (1.18) 0.893* 152 Bike lanes and paths are free of obstacles c6q 2.25 (1.07) 2.50 (1.22) 1.493* 183 Significance test: Independent samples t test = p < .05; *** = p < .001

PAGE 210

194 Table 25 Backward Stepwise Linear Regression Attitudinal Variable Models Dependent Variable # of steps R 2 Columbia Marin Minneapolis Sheboygan Phase Income Gender Age #children Race c6a 4 0.080 .147** .095* .162** .083* c6b 2 0.060 .104** .248** c6c 4 0.033 .074* .104** .072* .091** c6d 4 0.093 .101** .160** .181** .093** c6e 6 0.065 .150** .128** .109** .070* c6f 4 0.152 .096** .399** .066* .139** c6g 2 0.081 .297** .087** c6h 4 0.138 .077* .290** .339** .061 c6i 2 0.087 .200** .264** c6j 4 0.048 .080* .147** .122** .078* c6k 3 0.082 .224** .065* .146** c6l 7 0.133 .243** .267** .344** .101** .075* c6m 3 0.128 .196** .111** .385** c6n 1 0.005 .074* c6o 2 0.116 .109** .350** c6p 2 0.031 .163** .122** c6q 3 0.063 .167** .263** .107** c6r 2 0.017 .090* .093** c6s 4 0.107 .252** .243** .111** .108** c7a 5 0.057 .097** .100** .064* .065* .164** c7b 3 0.040 .061* .160** .095** c7c 2 0.045 .156** .148** c7d 4 0.062 .143** .116** .066** .157** c7e 5 0.070 .083** .076** .120** .063** .167** c7f 2 0.036 .166** .079** c7g 4 0.060 .099** .108** .170** .078** c7h 3 0.101 .068** .051* .312** c7i 1 0.077 .278** c7j 4 0.083 .099** .164** .132** .224** c7k 3 0.046 .084** .081** .165** c8a 3 0.065 .116** .069** .234** c8b 3 0.080 .066** .262** .054* c8c 5 0.076 .1** .070** .082** .217** .076** c8d 3 0.061 .054* .226** .062* c8e 5 0.072 .064* .109** .094** .219** .053* c8f 4 0.073 .064* .090** .220** .057* c8g 2 0.064 .243** .050* c8h 4 0.057 .091** .112** .164** .068** c8i 2 0.052 .058* .218** = p < .05; ** = p < 01

PAGE 211

195 APPENDIX C CHAPTER 4 SUPPLEMENTAL DOCUMENTATION This appendix includes supplemental documentation for Chapter 4, including comparative tests of mediation (i.e., Sobel's test), alternate/additional structural equation model results and residual correlations across structural equation models Table 26 Sobel's Test of Mediation: Significant Direct and Indirect Effects Phase 1 Phase 2 Dependent Variable (y) Independent Variable (x) Direct Path (x > y) Indirect Path (x > m > y) Direct Path (x > y) Indirect Path (x > m > y) Barriers Walk Columbia, MO Marin, CA *** Minneapolis, MN *** Sheboygan, WI Income *** Gender *** Age *** Walk Regularly Bike Regularly *** # Children Race Barriers Bike Columbia, MO Marin, CA Minneapolis, MN *** Sheboygan, WI Income *** Gender Age *** *** Walk Regularly Bike Regularly *** *** *** # Children *** Race p < .05 *** p < .001

PAGE 212

196 Table 1 7 Phase 1 Model Results SEM Mediation Models Model Fit Indices Chi Square SRMR RMSEA CFI 2 df p 894.889 351 0.00 0.039 0.041 0.924 SEM Results Dependent Variable Independent Variable Unstandardized Standardized p Barriers Walk Convenience Drive 0.644 0.034 *** Barriers Bike Convenience Drive 0.456 0.034 *** Convenience Drive Columbia, MO 0.041 0.016 Marin, CA 0.293 0.100 Minneapolis, MN 0.470 0.181 *** Sheboygan, WI 0.076 0.029 Income 0.098 0.174 *** Gender 0.050 0.023 Age 0.000 0.049 Walk Regularly 0.363 0.124 *** Bike Regularly 0.398 0.156 *** # Children 0.061 0.062 Race 0.455 0.098 Barriers Walk Columbia, MO 0.071 0.023 Marin, CA 0.570 0.164 *** Minneapolis, MN 0.363 0.118 Sheboygan, WI 0.134 0.043 Income 0.000 0.000 Gender 0.513 0.066 *** Age 0.000 0.064 Walk Regularly 0.153 0.044 Bike Regularly 0.165 0.055 # Children 0.045 0.038 Race 0.405 0.074 Barriers Bike Columbia, MO 0.085 0.029 Marin, CA 0.224 0.068 Minneapolis, MN 0.086 0.030 Sheboygan, WI 0.107 0.036 Income 0.021 0.033 Gender 0.159 0.066 Age 0.000 0.023 Walk Regularly 0.239 0.073 Bike Regularly 0.674 0.236 *** # Children 0.108 0.099 *** Race 0.212 0.041 Barriers Walk R 2 = .38 Barriers Bike R 2 = .30 Convenience Drive R 2 = .12 = p < .05 *** = p < .001

PAGE 213

197 Table 2 8 Phase 2 Model Results Direct Path SEM Model Fit Indices Chi Square SRMR RMSEA CFI 2 df p 727.273 297 0.00 0.041 0.032 0.930 SEM Results Dependent Variable Independent Variable Unstandardized Standardized p Convenience Drive Columbia, MO 0.168 0.062 Marin, CA 0.027 0.009 Minneapolis, MN 0.277 0.110 Sheboygan, WI 0.096 0.036 Income 0.076 0.137 *** Gender 0.062 0.043 Age 0.012 0.178 *** Walk Regularly 0.309 0.104 Bike Regularly 0.243 0.099 # Children 0.002 0.012 Race 0.445 0.112 Barriers Walk Columbia, MO 0.059 0.022 Marin, CA 0.116 0.040 Minneapolis, MN 0.104 0.042 Sheboygan, WI 0.145 0.055 Income 0.086 0.156 *** Gender 0.119 0.084 Age 0.010 0.146 *** Walk Regularly 0.393 0.134 *** Bike Regularly 0.073 0.030 # Children 0.003 0.020 Race 0.232 0.088 Barriers Bike Columbia, MO 0.329 0.117 Marin, CA 0.038 0.013 Minneapolis, MN 0.051 0.019 Sheboygan, WI 0.082 0.029 Income 0.059 0.101 Gender 0.066 0.044 Age 0.017 0.248 *** Walk Regularly 0.268 0.087 Bike Regularly 0.632 0.248 *** # Children 0.005 0.035 Race 0.232 0.056 Barriers Walk R 2 = .09 Barriers Bike R 2 = .18 Convenience Drive R 2 = .12 = p < .05 *** = p < .001

PAGE 214

198 Figure 2 2 Residual Error Mediation SEM

PAGE 215

199 Figure 2 3 Residual Error Direct Path SEM

PAGE 216