IN HIGHER EDUCATION POLICY:
An Exploratory Foray
Sharon M. Samson
B.S., State University College of New York, 1968
M.S., University of Colorado, 1990
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
This thesis for the Doctor of Philosophy
Sharon M. Samson
has been approved
Samson, Sharon M. (Ph.D., Public Administration)
Information Complexity in Higher Education Policy
Thesis directed by Professor Peter deLeon
The postmodern policy movement recognizes that policy is a complex
phenomenon, involving multiple constituencies, interdependencies, degrees
of choice, and behavioral change. In essence, policy is a dynamical system.
This study asserts that existing models are insufficient for studying policy
systems in the same way that linear models are insufficient for studying
natural systems. Instead, the study proposes a nonlinear model that
captures policy relationships in a few simple rules.
Connecting complexity theory and information theory to policy design
is plausible because the nominal differences between the hard sciences and
the soft sciences are narrowing. There is no logical reason for the intrinsic
complexity of public policy to be greater than natural science. A broad
literature review of these three theories created a framework for envisioning
policy dynamics, including:
Understanding information as policys structure.
Defining information as order emerging from data.
Perceiving the connection between information and perturbations in
Extending these information concepts into the policy realm by
proposing that all policy components, traditionally viewed as abstract
elements or events, can be understood as information flows.
Consistent with complexity theory, the research followed a three-step
process study the structure, trace the information flows, and model the
policy dynamics. The experiment demonstrates how dynamic modeling can
expand understanding of possible policy outcomes. The findings suggest
that an anticipatory policy approach may be an effective way to examine the
interaction between external pressure and internal tension rather than
analyzing individual elements.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
This dissertation is dedicated to:
Sam, whose fascination with information and strange attractors created
the initial conditions,
Peter, whose critical eye fostered self-organized criticality,
Jeff, whose intuitive sense of design stimulated the interesting questions
about policy, information, and complexity.
1. INTRODUCTION........................................... 1
Problem Statement and Rationale For The Study .... 3
Theoretical Implications: Public Policy Theory. 11
Policy Design and Modeling Tools................ 13
Organization of the Thesis...................... 19
2. LITERATURE REVIEW......................................22
Policy Design Theory........................... 24
Complexity Theory............................... 51
Emergent Behavior ......................... 56
Unpredictability .......................... 59
Summary of the Literature Review.................63
3. METHODOLOGY FOR MODELING
IN HIGHER EDUCATION POLICY............................ 71
Research Questions.............................. 74
Research Goal................................... 76
Research Design ................................ 77
Data Description................................ 82
Validity, Reliability, and Limitations ......... 83
Face Validity ............................. 84
Content Validity .............................89
Data Reliability .............................89
4. THE POLICY MODEL..........................................90
Model Creation..................................... 90
Step 1: Observe the Admission
and Enrollment Patterns ..............95
Step 2: Identify the Key Variables............96
Step 3: Trace the Energy Flows
Through the System ...................98
Step 4: Represent the Interdependencies
Between Policy Spaces............... 107
Step 5: Calculate Values for Constants
and Functions....................... 122
Step 6: Identify Information Flows
Between Resource Groups............. 125
Step 7: Test the Models Validity........... 133
Step 8: Test the Models Robustness........ 137
5. FINDINGS................................................ 141
Findings Related to the Model:
Dynamic Modeling ................................. 142
Findings That Related to Higher Education Policy . 146
Findings Relevant to Complexity,
information, and Policy Design Theories........... 151
Summary of Findings............................... 156
6. Future Research Directions ............................. 158
A. GLOSSARY OF STELLA MODELING TERMINOLOGY ... 161
B. SUB-MODEL OF ENROLLMENT SECTOR.......... 163
C. MODEL, VERSION #1 ...................... 164
D. PLOT OF APPLICANTS WITH
ADMISSION INDEX SCORE................... 165
2.1 Elements of a Policy System...................................... 27
2.2 Structural View of Design ....................................... 33
2.3 Process View of Design .......................................... 33
2.4 Energy Fields for Incremental and Radical Change ................ 36
2.5 Topology of Policy Change........................................ 37
2.6 Schematic of General Communication System ................ 43
2.7 Information Continuum............................................ 44
2.8 Information Lag ................................................. 45
2.9 Stability ....................................................... 58
2.10 Oscillation .................................................... 58
2.11 Chaos .......................................................... 58
2.12 Gell Manns Model of Adaptation................................. 60
2.13 Emergent Global Structure ...................................... 60
3.1 Methodology for Modeling the Admission Standards Policy.........78
3.2 Admission Standards Policy System Diagram........................ 82
3.3 Policy Near-Equilibrium ......................................... 87
3.4 Far-From-Equilibrium............................................. 88
3.5 Chaos ........................................................... 88
4.1 Linear Policy Structure.......................................... 91
4.2 Basic Loop Structure Within Which All Policies Exist............. 91
4.3 Admission Standards Policy Loop.................................. 91
4.4 System Enrollment by Student Level............................... 92
4.5 Enrollment Change From Previous Year of the Twelve
Four-Year Institutions .......................................... 92
4.6 Number of Colorado High School Graduates ........................93
4.7 Distribution of First-Time Freshmen Applicants...................93
4.8 Goal-Seeking Behavior Occurring in the
Higher Education Environment .................................. 95
4.9 Entity Diagram of the Admission Standards Policy System .........96
4.10 Academic Preparation Loop...................................... 99
4.11 Student Preference Loop....................................... 100
4.12 Academic Success Loop......................................... 101
4.13 Role and Mission Loop ........................................ 102
4.14 Competition Loop.............................................. 105
4.15 Capacity Loop ................................................ 106
4.16 College Preparation Infrastructure............................ 109
4.17 Graph of College Preparation Behavior......................... 111
4.18 Student Preference Infrastructure............................. 112
4.19 Graph of Student Preference Behavior.......................... 113
4.20 Academic Success Infrastructure .............................. 114
4.21 Graph of Academic Success Behavior ........................... 115
4.22 Role and Mission Infrastructure............................... 117
4.23 Graph of Role and Mission Behavior............................ 117
4.24 Competition Infrastructure.................................... 118
4.25 Graph of Competition Behavior................................. 119
4.26 Capacity Infrastructure....................................... 120
4.27 Graph of Capacity Behavior.................................... 121
4.28 Change in Number of High School Graduates..................... 122
4.29 System Diagram of Admission Standards Policy Model ........... 125
4.30 Admission Sector.............................................. 126
4.31 Enrollment Sector............................................. 128
4.32 Policy Sector................................................... 132
4.33 Comparison of Linear Projections to Nonlinear Output...........135
5.1 Graphic Pad of Admission Standards Policy Model................. 150
5.2 Anticipatory Policy Design ...................................... 155
1.1 October 1997 General Fund Enrollment Analysis.................. 14
2.1 Change Scenarios................................................42
2.2 Information Properties..........................................48
2.3 Glossary of Common Concepts and Terms in Complexity
2.4 Surprise Generating Phenomena ..................................60
3.1 Comparison of Linear and Nonlinear Policy Views ............... 73
3.2 Comparison of Information Tools and Methodologies.............. 74
3.3 Organizing Principles for Studying Policy Complexity............76
3.4 Summary of the Stepwise Approach to System Dynamics
Model Creation and Development................................. 80
3.5 Variables Used in the Research Study .......................... 83
4.1 Testing Output for College Preparation Infrastructure......... 113
4.2 Testing Output for Student Preference Infrastructure.......... 115
4.3 Testing Output for Competition Infrastructure................. 119
4.4 Testing Output for Capacity Infrastructure.................... 121
4.5 Average Graduation Rates ..................................... 123
4.6 Undergraduate Enrollment Capacity............................. 124
4.7 High School Graduation Data and Projections................... 136
4.8 Highly Selective College Behavior In Current Policy Environment. 137
4.9 Alternative Policy Scenario When
Number of High School Graduates Decrease..................... 138
4.10 Alternative Policy Scenario For Highly Selective Colleges
When a Policy Parameter Changes .............................. 139
4.11 Alternative Policy Scenario For Selective Colleges
When a Policy Parameters Changes .............................. 140
5.1 Output of Factoring Admission Variables ....................... 143
5.2 Introduction of Qualitative Measures .......................... 145
Complexity theory is the study of complex behavior, especially the
large scale changes that occur from small changes in initial parameters
(Nicolis and Prigogine, 1989). Understanding the implications of complexity
theory is important to many audiences, ranging from theoretical physicists to
public administrators. By focusing on a specific class of irregular behaviors,
complexity theory cuts across traditional scientific disciplines. Turbulence
falls into this behavior class. While physicists and biologists struggle to
understand turbulent behavior in physical and natural systems, social
scientists seek better ways to understand social turbulence and behavioral
dynamics (Dror, 1986).
Higher education is not immune to turbulence and complexity. If
complex behavior is defined as an interesting pattern, higher educations
enrollment, revenue, and migration patterns continue to exhibit a particularly
high level of complexity. Value and interest conflicts, continuous information
exchanges, uncertainty, and complex relationships between people and
institutions characterize the educational culture (Benveniste, 1989; Zemsky,
1994). As the education environment becomes more complex and
interdependent, change becomes increasingly nonlinear, discontinuous, and
unpredictable (Zemsky, 1995). For example, some state higher education
coordinating boards have diffused their statutory power by decentralizing
their policies. This, in turn, altered the underlying governance structure, the
roles of the constituent colleges, and left administrators uncertain about how
of the constituent colleges, and left administrators uncertain about howto
handle new responsibilities. In this environment of diffuse power and loose
control, higher education becomes an underbounded system, that is, an
open invitation to turbulence.
The nonlinear paradigm of complexity theory is well suited for
examining an underbounded system. Nonlinear systems exhibit periods of
stability followed by periods of turbulence after which the system undergoes
structural change. Lasswell (1956: 9) observed that policy systems display
linear patterns, sometimes approaching equilibrium, during stable periods,
but fluctuate structurally during dynamic periods when new patterns of
interaction emerge (Baumgartner & Jones, 1991). Policy cycles rarely, if
ever, return policy conditions to the original state of affairs. His observations
of policy behavior predate complexity theory but presciently recognize the
existence of emergent behavior, a defining property of complex systems.
While several individuals are using complexity to examine change in
public management (Kiel, 1994; Kiel and Elliot, 1995; Wheatley, 1992), little
rigorous research has been conducted in policy design. Complexity theory
may provide some insight on the unpredictable nature of agenda setting and
the potential for self-regulating, self-organizing policies. The theory suggests
that non-equilibrium systems achieve a degree of freedom, autonomy, and
consequently, can adapt creatively to turbulence. Recognizing and
understanding patterns of non-equilibrium systems may augment policy
Problem Statement and Rationale
For The Study
Historically, change in higher education has been characterized as
slow, idiosyncratic, and externally driven (Kerr, 1994a). These attributes are
generally acceptable when (1) the economy is stable, (2) society's
expectations are low; (3) the knowledge base is stable; or (4) growth in one
sector of the system can nurture change without threatening other sectors or
constituencies (Berman, 1980). Most of the conditions that support slow,
incremental change are no longer prevalent in higher education (Atwell,
1996). Opposing forces are pulling higher education apart, in the
technology-driven world, information becomes a global resource. At the
same time, the experiences and aspirations of individuals are highly valued
and knowledge is considered to be an intellectual property. This value
conflict makes higher education a nodal point of pressure between
governmental regulations, commercial interests, and the expectations of
parents and students.
The first national study on higher education proposed that state
coordination was a way to alleviate the pressure points of the public higher
education system. It is important that the place which each institution is to
occupy and the policy, which it is to pursue, be determined in the light of the
whole state problem of higher education (Leonard, 1923: 3). Speaking for
the state, government agencies: (1) set priorities, (2) target resources, (3)
innovate when old policies have failed, (4) coordinate conflicting objectives
into a coherent whole, (5) impose losses on powerful groups, (6) ensure
equity, (7) ensure effective implementation, (8) ensure political stability, (9)
maintain commitments, and (10) manage political cleavages to ensure that
society does not denigrate into civil war (Weaver and Rockman, 1993).
When states established their governance structure for higher
education, they designated a governing board or a coordinating board to
assume this list of responsibilities. Typically, a governing board possesses
greater authority than a coordinating board for administrative decisions that
determine how institutions operate. Coordinating boards, on the other hand,
have been delegated broad authority at the system level to ensure the
orderly growth in higher education that is free of political or parochial interests
(Callan, 1991). Their primary agenda includes shaping state priorities by
initiating public debates on higher education, allocating resources, and
establishing policies. They exercise limited, but significant, authority over
colleges and universities by issuing regulations and rendering decision that
have wide-ranging impact on institutions or governing boards (Task Force on
Coordination, Governance, and Structure of Postsecondary Education,
1973). The citizen members and executive directors of these boards seek to
balance the needs of society with those of educational institutions. They
have the final word in many decisions affecting colleges and universities, yet
their authority sometimes faces practical limitations as it intersects with the
aspirations of major external stakeholders, such as governors and legislators,
or internal stakeholders such as university presidents and faculty (Novak,
1996). The need to resolve the conflicting pressures the forces of
institutional self-interest and those of the public interest has placed heavy
* demands on higher education coordinating boards (Berdahl, 1990).
Overall, state higher education coordinating agencies have been most
effective when they serve as arbitrators between the interests of large,
competing state universities or between the interests of the research
university sector and the other higher education sectors within their state
(Mingle, 1995). For example, most states have a variety of competing
sectors within their public higher education system, including research
universities, specialty schools, comprehensive universities, liberal arts
colleges, and community colleges. Empowered by legislatures to arbitrate
differences among these sectors, coordinating boards rely on policy to
develop an overall plan for the system. Higher education policies express
the operating goals, principles, and processes of the public higher education
system. Policies thus provide a framework for the members of a system, to
live with uncertainty, structural tension, and complexity (Handy, 1996).
Recent articles on higher education governance confirm the existence
of serious disturbances in the higher education environment (Johnstone,
1992; Breneman, 1994; Finney, 1994; Davies, 1997). Some argue that
higher education is experiencing an identity crisis that occurs when structural
tension interrupts a systems stability (Coe, 1997). Others observe that the
decisions, actions, and traditions sometimes give the impression that the
state coordinating board serves no important function except to stifle and
impede educational innovation (Benjamin and Carroll, 1993; Epperand
The tense relationship between state government and higher
education has continued to evolve in uncertain and unpredictable ways
(McGuinness, 1994). Some states like Massachusetts, New Jersey, Illinois,
West Virginia, and South Carolina are moving toward decentralization. Other
states, like Maryland, Minnesota, Nebraska, and Texas, are centralizing
higher education's operations and decision making. The decentralization
trend responds to college presidents perception that a unified system
approach to higher education has eroded individual institutions ability to
respond to changing market forces (Fisher, 1995). In contrast, other
governmental leaders fear that the decentralized approach will result in
institutions developing political connections and alliances to the detriment of
other institutions and system unity (Hollander, 1994).
In 1965, the Colorado General Assembly statutorily designated the
Colorado Commission on Higher Education (CCHE) as the policy
coordinating board of the States public higher education system. In the first
twenty years, Colorados relatively small public higher education system grew
as the University of Colorado outreach centers became three, independent
universities while comprehensive colleges, which formerly offered both two-
year and four-year degrees, divided into a separate community college and
baccalaureate institution. With the growth, the General Assembly was
subjected to intensive lobbying by the advocates of its twenty-eight
institutions and heard negative things about the way the Commission
exercised its role from a variety of sources. Not surprisingly, the institutions
favored the elimination of the Commission. Instead, the General Assembly
reorganized the coordinating board in 1985 and expanded its powers.
The introductory words to the authorizing legislation implied that
Colorado lacked a coherent policy system. Wanting broader perspectives
and greater coherence in policy advice and development, the General
Assembly empowered the Commission to create a set of policies to address
the complex issues present in the educational environment (C.R.S. 23-1-
101). The legislation explicitly outlined an aggressive agenda for the
Commission, listing mandates for policies that link planning, accountability,
and budgeting, ones that differentiate graduate education among the
research universities, ones to provide graduated access to various degree
levels, and others to coordinate the transfer of credit between two-year and
four-year and among four-year public colleges.
It is now twelve years since the reorganization of Colorados higher
education system. Some people have observed that higher educations
governance system is poorly attuned to the current higher education
environment (Benjamin and Carroll, 1993). The 1985 statutory goals of
quality, efficiency, and accountability have been usurped by another set of
values (McGuinness, 1997). The political environment favors
entrepreneurship that assumes individuals will take the initiative, academic
competence that assumes an environment of collaboration, and innovation
(Wingspread, 1993; Callan, et al., 1997). Colorados higher education
policies fail to recognize the diversity of the State's education providers and
their clientele a one size fits all mentality. More important, its policies
focus on solving specific problems of the past and ignore the central values
that the Commission supports (Jones, 1995).
The Colorado Commission itself admitted that its policy approach was
inadequate for the current environment. This perception is evident in the
June 14, 1995 Commission meeting minutes that detailed a discussion of the
Statewide Admission Standards Policy. Legislators began the discussion by
challenging the Commission:
The statutes explicitly let you know what the legislative goals
are. If the statutory language is causing problems, ask us to
change the law. If the statutory language is still valid, change
A college president, whose institution complied with the current admission
standards policy, emphasized the dual needs for statewide policy
consistency and stability when he stated:
A policy doesnt work if some colleges can ignore the admission
standards. What's good for one should be good for all. Let's
strengthen the policy, but don't change the funding distribution.
Responding to the inference that some institutions were operating outside
the admission standards policy, another president advocated changing the
policy to meet his particular institutions goals and practices:
You need to look at the students we serve, not just arbitrary
numbers. Diversity within the system is good. Trust the
institutions to make the right decisions by eliminating or
minimizing policy regulation.
After listening to the debate, members of the Commissions advisory
committee expressed their concerns about system efficiency, speculating:
is Colorado utilizing the higher education system at full
capacity? If some institutions can accommodate additional
students, investing dollars in bricks and mortar seems foolish,
especially when changing the existing admission standard
policy parameters may provide additional access to public
Members of the audience articulated their perceptions about the current
policy status and the role of a policy coordinating board in their public
The current policy allows institutions to abuse the system. As
individual agendas supersede state policy goals, the policy
becomes increasingly dysfunctional.
Turf protection dominates the policy environment.
Perhaps higher education needs to look outside the public
system to the private sector.
The Commissioners asked themselves several value-laden questions:
Policy and regulation are the only mechanisms the Commission
has for guiding the systems actions. Is it possible to adjust the
Statewide Admission Standards policy to achieve the
institutional goals and simultaneously protect the state and
student interests? First we want to know what the unintended
consequences of eliminating the standards, enlarging the
window, or strengthening the standards are. Where is the
balance point between state standards and institutional
flexibility? (Official CCHE Minutes of Commission Meetings,
While the words and the positions were very different, the comments
of each stakeholder group demonstrated the perceived disjointed nature of
Colorados current higher education admission policy. Perhaps justifiably, the
desire for immediate enrollment growth preoccupies individual institutions.
Growth aspirations appear incompatible with the existing policy base that
emphasizes stability, control, and homeostasis.
If the perception of a policy misfit was isolated to Colorados
Admission Standards Policy, modifying the policy language would resolve the
perceived problems. Instead, the discussion exemplifies the tension between
the external demand for change and the internal desire for stability present in
Colorado and other states. Recently, governing boards and governors began
appointing business leaders as state higher education executive officers
(e.g., Washington) and college presidents (e.g., University of Northern
Colorado), explaining that higher education needs to replace its reactive
posture with a more dynamic approach to the emerging issues. To operate
effectively in turbulent times, higher education must be make changes (Kerr,
1994b). Yet, as the former chairperson of Wisconsins Higher Education
Board of Regents notes in Howto Move a Battleship With Your Bare
Hands, a change agenda conflicts with the stability goals of the institutions
Colorados funding experiences of 1994-95, 1995-96, and 1996-97
exemplify the conflict between the legislative and educational perspectives.
In the first year, the legislature rewarded higher education by funding $14
million of new money in three designated policy areas student
achievement, academic quality, and productivity. To the legislature, funding
these specified policy priorities symbolized that the system was committed to
change. In the second year, after reviewing the supporting documentation,
the legislature appropriated $8.6 million for the funding priorities, but
caveated the appropriation with their disappointment about the lack of
significant change. In the final year, the legislature chose not to fund the
policy areas, commenting that higher education just doesnt get it. The
institutions, on the other hand, felt betrayed by the lack of funding. In short,
the documentation of higher education practices, no matter how persuasive
the evidence once was, no longer adds up to a compelling whole. Society,
and legislatures in particular, are holding higher education to higher
expectations (MacTaggart, etal., 1996).
These experiences made the Colorado Commission aware that the
higher education system preferred stability to change, even when change
means positive growth. It also realized that defending the status quo was not
an acceptable political strategy. The Commission announced that it would
focus on the assumptions and values driving its policy system rather than the
regulations and exemptions. It began to search for a new policy approach
and tools to communicate the values and directions of the public higher
Public Policy Theory
The purpose of this research is to explore policy dynamics and
change. Policy design theories continue to gloss over the complexity
inherent in policy systems. Since complexity is endemic to policy design,
studying policy design under complexity may provide a needed theoretic
perspective of social change and information dynamics.
The current policy approaches include prospective policy analysis,
retrospective policy analysis, and integrative policy analysis. While
examining policy from different ends of the spectrum, the prospective and
retrospective approaches use a static information perspective to understand
the policy system. Prospective policy analysis is concerned with predicting
and recommending actions before they happen. Its strengths come from its
normative theory base and its capacity to construct an evaluation of potential
policy alternatives. As commonly implemented, it focuses only on ten
percent of policy activities and has failed to recognize that goals and
conditions change as policies are implemented (Allison, 1971). In contrast,
the main concern of retrospective policy analysis is to understand a problem
after it has happened. It relies on descriptive decision theory to identify
cause and consequence. Its inherent weakness lies in the fact that hindsight
provides intellectual knowledge about the past and ignores the
discontinuities of a changing environment (Weiss, 1980). Integrated policy
analysis blends the prospective and retrospective policy approaches, and by
that introduces more information into the analysis. The merger of the
normative and descriptive theories implies the importance of policy
relationships. It also incorporates the control thinking of prospective and
retrospective related theories, including the emphasis on problem finding and
problem solving. According to Wildavsky (1993), effective policy design
focuses on human behavior rather than the resolution of problems or singular
More recently, scientists and practitioners are questioning the validity
of control-based theories for understanding a world that operates under
uncertainty (Kaufman, 1991; Kauffman, 1995). Notable thinkers have moved
away from the problem-solving paradigm and gravitated toward theoretical
concepts contained in complexity, competition, creativity, and network
theories (Hamel, 1996; Bennis, 1997; Senge, 1997). This outlook is
consistent with trends in public administration. Theorists in this field
historically have doubted the degree of control operating in relatively stable
environments (March and Simon, 1958; Katz and Kahn, 1978). They
identified cognitive limits on rationality due to distorted information flows,
policy environments, feedback, and the complexity of joint action (Pressman
and Wildavsky, 1984).
Complexity theory provides a framework to examine the emergence of
different behaviors that occur under conditions of uncertainty. It offers a
normative, theoretical approach that is interested in patterns created by
competition, diversity, and choice. It probes assumptions, principles, and
values. While several individuals are applying this theory to the social
science, the work primarily focuses on management issues (Kiel, 1993; Kiel
and Elliot, 1996). The exploration of complexity theory may identify if policy
systems adapt to change in the same ways as other systems do and how
policy design helps systems evolve.
The search for patterns and order links policy science and the study
of complex adaptive systems. Policy design inherently deals with the
turbulence created by conflicting values, obscure goals, problems not
amenable to interrelated problems, uncertainty, and the rate of change in an
information society (Schon, 1983). Complexity theory searches for
irregularities in patterns and the strange patterns that exhibit periodic chaotic
behavior but global system stability. Self-organized criticality may enlighten
policy theory about the interrelationships between policy behavior and policy
structure. Applying this theory to public policy tests complexity theorys
premise of global emergent behavior. Public policy research in this area may
provide insights to other disciplines exploring complexity.
Practical Rationale: Policy Design And
Policy is sometimes judged by the tools or methodologies used to
create it. Policy decisions often communicate the systems story through
analyses and data. For instance, In the 1980s, CCHE implemented several
academic policies whose primary purpose was to redirect enrollment within
the systems. Because the Commission was uncertain about the effect the
policies would have on enrollment, it used a two-year enrollment lag funding
model, holding harmless any institutions that experienced a decline in
enrollment an example of retrospective policy analysis. CCHEs policy
interests shifted in 1990 to demographic factors. Along with other states in
the west, it anticipated rapid enrollment growth and developed a model to
project ten-year enrollments. Based on the assumption that increases in the
number of students graduating from high school translates into higher
education enrollment growth, the methodology projected exponential growth.
Funding allocations and capital construction planning were based on the
During the past three years, Colorados actual full-time equivalent
student enrollment (FTE) did not reach CCHEs projected FTE numbers.
Even with annual corrections to the enrollment model, the disparity was so
significant in spring 1997 that the State Legislatures Joint Budget Committee
recommended reallocating $2.5 million among the States colleges and
universities (Table 1.1). The community college governing board and the
University of Northern Colorado were hit the hardest by the 1997
Table 1.1 October 1997 General Fund Enrollment Analysis. Source:
Joint Budget Committee ____________________________
Governing Board 1996-97 Enrollment Surplus (Deficit) 1997-98 Enrollment Surplus (Deficit)
State Colleges ($673,620) ($704,933)
State Board of Agriculture ($682,239) $377,797
University of Colorado $306,270 ($2,116,000)
School of Mines ($265,800) ($303,716)
University of Northern Colorado $1,502,553 $495,830
Community Colleges $2,262,087 ($178,822)
Higher Education System Overfunded (Underfunded) FTE $2,449,250 ($2,429,844)
reallocation. By fall 1997, it appears that the situation has swung $5 million
in the opposite direction. The University of Colorado alone accounts for $2.5
million as it oscillated between an enrollment funding surplus to an
enrollment funding deficit. The fluctuating pattern suggests that higher
education enrollment is in a far-from-equilibrium state and that the linear
enrollment methodology currently used is unable to replicate the enrollment
Policy tools that do not provide useful information prevent both an
executive director and a coordinating board from executing their mandated
roles. Executive directors are particularly sensitive to the fact that their
political credibility rests on anticipating change (Kerr, 1994a). In the July
issue of The Chronicle of Higher Education, CCHEs Executive Director
Dwayne Nuzum acknowledged that the enrollment projected for Colorado
had not materialize as his state had anticipated (Healy, 1997). The article
concluded that unforeseen trends and mistaken projections have raised
doubts about the reliability and relevance of current forecasting methods.
Because many states are experiencing a gap between enrollment projections
and actual enrollment levels, the Western Interstate Commission of Higher
Education (WICHE) convened a meeting of higher education executive
officers to discuss the impact of changing enrollment trends on state
planning. The attendees concluded that effective leadership relies on
reliable information, especially when advocating for state funds.
The Commission, on the other hand, perceives that effective policies
stimulate change. It relies on information about the leading and lagging
indicators of the system to make decisions. By nature, a coordinating board
is future-oriented. It is less interested in the past than it is in unrealized
potential. Therefore, information is interesting to the coordinating board only
if the information helps to envision the future. With a strong policy
commitment, a coordinating board takes over the policy space and creates
the future environment in which public colleges and universities play, The
linear-based methodologies and policy tools are inadequate for constructing
images of the future (Hamel, 1996).
The current evidence suggests that it is not possible to predict higher
education enrollment patterns accurately even at the aggregate level in the
current policy environment. Realizing that predicting the consequence of
certain policy actions with any kind of certainty is not possible, coordinating
boards still have to act. They have to make three-year, five-year, and ten-
year policy decisions (e.g., Affirmative Action goals) and be prepared to
abandon policies that are no longer compatible with the systems goals.
States need a more constructive approach for devising policy if they are to
serve the needs of the system instead of focusing attention on a single
aspect or an individual problem (Bensimon et al., 1989: 65). Therefore,
higher education policymakers arguably need to give more attention to
complex thinking patterns and tools that recognize the existing complexity
(Birbaum, 1987; Chaffee, 1988; Bennis, 1997). Several experts advise
developing tools that model the interrelationships that exist among policies
(Ascher, 1983; Morrison, 1995; Guskin, 1996). Wthoutthe right policy tools,
a coordinating board is severely handicapped.
Political decisions, time pressures, and accountability make public
policy systems more difficult to model than private sector systems (Newcomer
and Caudle, 1991, Wnston, 1997). Policy problems are really systems of
interdependent problems or messes (Ackoff, 1994). Most important policy
problems are ill-structured because they are really complex systems of
problems that involve high levels of conflict among competing stakeholders.
Because models are the principal vehicles for carrying decision information
into the political arena, legislators often challenge model information. To
approach policy arguments and debates in a value-critical way means that
models need to represent the operating ideology of the policy environment
and the behavioral assumptions about a particular target population
(Schneider and Ingram, 1990a).
Another methodological constraint is timing. Despite tremendous
advances in modeling software, model development remains a difficult, labor
intensive, and costly process. This is especially true when attempting to
model complex problems. These type of models that often rely on large
amounts of input data continue to demand the technical skills of a computer
expert and often provide static findings. Moreover, legislators do not have
the background, patience, or opportunity to wait for policy design
recommendations obtained from complicated, data-intensive methods (Dunn,
1994). Bobrow and Dryzek (1987) argue that a better basis for making
choices or managing diversity in the policy environment may exist.
Simple descriptive statistics cannot cope with complex, interactive
problems that involve dealing simultaneously with a sizeable number of
factors (Shannon and Weaver, 1948: 539; Casti, 1995). Descriptive statistics
represent past patterns and ignore the impact of human choice. The physical
sciences have used thought experiments to investigate complex concepts
when the available technology is unable to fit the theoretical propositions
(Brown, 1991). In thought experiments, a theoretical concept is correlated to
the experimenters knowledge (Einstein, 1954; Kuhn, 1977). Mental
manipulations, however, are difficult to introduce into discussions with
policymakers and lack accountability. Higher education policymakers require
policy tools that correlate policy values and human behavior patterns if they
are to catalyze discussions with multiple stakeholders and reach informed
decisions in what is an increasingly complex policy environment (Rohrbaugh,
1986; Lindquist, 1991).
Peter Senge (1997) contends that giving up control under any
circumstance is difficult, but it is virtually impossible if a system has no idea
about what will take its place. In a decentralized environment, leaders
require policy tools that provide some indication of possible future system
behavior. If demonstrating possibilities to divergent audiences is crucially
important, a model that simulates the relationship between enrollment
changes and policy possibilities would provide a higher education
coordinating board with a valuable tool.
Nonlinear models emerging from complexity research appear to be
prime candidates for representing system behavior under changing
conditions (Zabusky et al., 1993; Lumley, 1997). Nonlinear modeling
techniques are compatible with issues that are ambiguous, vague, and ill-
defined (Casti, 1995). They provide a communication tool for groups to think
creatively about what they value in a holistic way (Senge, 1997). To acquire
a deeper understanding of the dynamics of policy, this research study
proposes to develop a policy model that captures tension between competing
values and the energy generated by the tension. Complex policy models
may identify situations when a policy change attracts so much energy that a
policy "stalls," a situation often called policy failure (Ingram and Mann, 1980).
It is necessary to frame an effective strategy for communicating policy
recommendations. In this sophisticated media age, a commission must
discover technological tools to package its message. This implies that a
commission takes on a political task: finding an effective way to reach an
audience, to resonate with the public mood, and to build a consensus around
solutions (El-Khawa, 1997). Dynamic models enable a coordinating board to
host real-time discussions of what appears to be small and perhaps
insignificant changes in policy parameters. For example, a legislative
committee couid see potential system changes as the members consider new
initiatives rather than waiting for the next scheduled meeting. Policy experts
traditionally select and validate the choice of methodologies for a policy
issue. While this is an efficient way to build a consensus, more often than
not it uses a black box approach to policy. Dynamic modeling reveals
If, as Harold Lasswell suggested more than twenty-five years ago, the
purpose of the policy sciences is to improve the quality of governance by
improving the quality of information provided policy makers, then dynamic
modeling may easily fit into the policy sciences tradition (Lasswell, 1971).
A complexity-based policy model may assist Colorado higher education
policy makers to rethink their policy assumptions and move beyond the
current technical approaches.
Organization of The Thesis
This study concerns policy design. It is not interested in describing
how things are, but with conceiving how they might become.
The first chapter documents the need for rethinking policy design. To
illustrate the gap between higher education policy and its environment, this
chapter references several events that concern Colorado Statewide
Admission Standards Policy. It suggests some theoretical and practical
implications for researching policy theory in conjunction with information
theory and complexity theory.
Using system dimensions as a frame of reference, the second chapter
maps the similarities and dissimilarities present in three disciplines: public
policy, information and complexity theories. It underscores the common
transformation nature of policy and information while introducing adaptation
as the defining characteristic of complex systems. The interdisciplinary
nature of three theories catalyzed a nonlinear review of the bodies of
literature; however, the organization of the dissertation somewhat
camouflages the nature of the search by dividing the theories into separate
sections. Despite this arrangement, ideas overlap and references cross the
section boundaries, as will be emphasized in the summary section of chapter
Beginning with the assumptions and research questions, chapter three
presents a methodology for linking the three theories to policy modeling. It
discusses the general validity of the modeling approach. Unlike hypothetical
policy modeling studies, this research used a real policy in its experiment.
Colorados Statewide Admission Standards Policy served as a reference
point for observing the information flows of a specific policy. The chapter,
therefore, lists the actual admission standards policy elements available in
the database that measure the policys reference behavior patterns.
Chapter four begins by documenting the model construction process,
relying on diagrams of the proposed mental model and actual images
imported from the software package. The chapter documents the validity of
the proposed model by displaying the results of the mechanical, robustness,
and policy tests. The chapter concludes with a series of graphs that simulate
the admission standards policy environment under changing policy
conditions. Modeling itself is an abstraction of a priori theory and a posteriori
data. During some periods, the literature review drove the model. Other
times, the model development motivated additional exploration of the
Chapter five summarizes the findings of the experimentation of the
admission standards policy model. As an exploratory study, the findings are
tentative rather than confirming. The policy insights emerging from the twin
lens of complexity and information theories redefine the relationship between
competition, requisite variety, and tension. The insights on information,
anticipation, and innovation may enhance policy design theory and connect
the policy orientation to complexity research occurring in natural, physical,
and social sciences.
Because this is an exploratory study, chapter six concludes with
intuitive observations of what is interesting about policy design in this new
policy environment. The chapter focuses on the future directions of this
Throughout their relatively brief history, the policy sciences have
acknowledged their interdisciplinary nature, creative potential, and
information-rich domain (Lasswell, 1951). In fact, some policy proponents
advocate studying man and his search for order to increase the
understanding of policy design (Simon, 1992:159). Following Lassweils and
Simons leads, this dissertation examines two theoretical bases outside the
social science boundaries to explore policy design theory.
Complexity theory, information theory, and policy design theory all
share critical similarities. While complexity theory studies special attractors,
policy design focuses on the attractiveness of different policy options.
Information theory examines the way a message affects the receiver. Policy
design theory, in turn, probes the dynamics that affect human behavior.
Each theory emphasizes interconnectivity. The theories recognize
that isolating one element of a system provides incomplete and possibly
inaccurate perceptions of behavior (Ackoff, 1994). Connectivity means that
behavior cannot be decomposed into parts. Information theory investigates
the communication connections (Wiener, 1950). Complexity theory is built
on the premise that all aspects of a system affect all others within that system
to an extraordinary degree (Gell-Mann, 1994). The most important aspect of
the performance of social systems is how the parts interact with one another
(Bailey, 1992; White, 1992). More to the point of this dissertation, the
postmodern movement recognizes the interrelationship between information
and change (Dery, 1990; Dobuzinskis, 1992; Overman, 1996).
The theories are different ways of talking about the relationships of
systems. Regardless if the system is natural or artificial, a proper system
substantially meets the following criteria (Sutherland, 1975: 99):
(1) A system must be in, or capable of obtaining a state of
integration sufficient to separate it from its milieu.
(2) A system must encompass two or more morphologically
determinable subsystems where the differentiation may be
structural, functional, or spatial.
(3) A system must be capable of constrained animation among its
subsystems so that their behavior is not entirely autonomous.
A portion of the subsystem energy must be coopted by the
system for the larger mission or the maintenance of integrity.
The first property encompasses the systems immediate environment,
external interchanges, and its relations with other systems that may somehow
affect or be affected by it (von Bertalanffy, 1968). The second property deals
with the internal behavior patterns of the subsystems. The third property
captures the dynamics of multiple parts acting on the whole.
Policy, information, and complexity provide three congruent
approaches to study a systems interaction with the environment, including
entropy, information, selection, tension, and adaptation. Inherent in each
are the concepts that a systems level of responsiveness relies on the
amount of information and disturbance present in the system. Policy design,
for example, looks for behavior patterns, external pressure points, internal
friction, mutual and competing values.
Each theory studies the relationship between space and time. The
tradeoff between time and space is a well-known computer science problem.
A major aspect of policy design is the tradeoff between the time and
resources required by various policy alternatives to accomplish similar goals
(Ostrom, 1989). Complexity theory examines behavior under changing
spatial and temporal conditions (Casti, 1995).
The way each theory frames selection differentiates the three from
each other. Policy design theory has a normative dimension that the others
lack. Policy selection identifies the temporal point between the state when a
policy schema is emerging and the state when an organization chooses a
specific direction. Complexity theory, on the other hand, frames choice as
the intervention of short scale randomness, generating innovation, and long
range order, enabling the system to sustain its domain within spatial regions
and time intervals (Nicolis and Prigogine, 1989). Predicated on the concept
that a message carries information whenever it conveys something not
already known, information theory defines choice as the difference between
two states of uncertainty (Shannon and Weaver, 1948). In all three theories,
the act of selection implies that time is irreversible. The system is moving in
a certain direction and cannot return to its prior state.
Policy Design Theory
Policy design has often been criticized for its fact/value dichotomy, the
insufficient attention paid to uncertainty, and its reliance on the control
paradigm to analyze contemporary policy dilemmas (Dobuzinskis, 1992;
Rein, 1983). In the real world, values and facts are often blurred and
overlapping evidence of what Simon calls "limited rationality." The fuzzy
information components and the speed of feedback compound the
accelerating rate of change and the increasing complexity generated by the
processes of change (Dror, 1992).
The perception that public administrators face a dynamic and fluid web
of interconnected problems spans the last century (Wilson, 1887; Benveniste,
1989; Dunn, 1994). Taylors one best way, Lindbloms muddling through,
Simons satisficing, and Minnowbrooks search fora new, socially relevant
public administration are different attempts to deal with the complexity of the
policy environment. The complexity from dealing simultaneously with a
sizeable number of factors makes it difficult to predict outcomes, as
witnessed by the generalized failures in forecasting (Argyris, 1973; Ascher,
1978; Blalock, 1982). Data about the future predictions are commonly
the weakest points of policy design. They require either a theoretical
understanding of the phenomena to be predicted or phenomena that are
sufficiently regular that they can simply be extrapolated. Since data about
human affairs are seldom regular, predictions will only be as good as our
theories (Simon, 1992: 170). Lasswell (1935) observed that if events appear
predictable, it is due to the limits of human knowledge and the methodologies
While the acknowledgment of complexity in the policy environment is
not new, the growing awareness of the nonlinear nature of policy is recent
(Daneke, 1990; Overman, 1996). Nonlinear dynamics is the study of the
temporal evolution of nonlinear systems. Nonlinear systems reveal
dynamical behavior such that the relationship between variables is unstable.
Furthermore, changes in these relationships are subject to positive feedback
in which changes are amplified, breaking up existing structures and behavior
and creating unexpected outcomes in the generation of new structure and
behavior (Kiel and Elliott, 1996: 1). In the social sciences, a nonlinear
system is one characterized by overlapping interactions, unstable
boundaries, relationships, and continuous change (Landau, 1973: 535;
Stone, 1988: 309). Currents of change are rolling through every domain of
society, shaking the stable state (Schon, 1971). This perception does not
merely question the linearity and predictability of social systems, but also
suggests that the stable state is a special case of the unstable. Douglas Kiel
(1993:152) contends that understanding nonlinearity will facilitate public
administration researchers understanding of societal turbulence that is
created by conflicting values, obscure goals, environmental uncertainty, the
rate and volume of information in the policy design domain.
One way to understand complexity is to look broadly at the various
dimensions that describe a system: purpose, space-time, structure, and
process. Figure 2.1 organizes policy under these four dimensions. Policy
systems exist and political events happen in the time-space continuum. The
structural dimension contains information about the system itself, its
environment, and the boundaries which data cross. Processes are the
internal activities that test new data against current information, thereby
changing a systems knowledge base. Purpose is the difference between
what a system is and what it wishes to become (Swieringa and Wierdsma,
1992: 93). If the internal differentiation of structure and process, which is
frequently accompanied by great interdependence of parts, defines
complexity, then systems are marked by a tendency toward greater
complexity (Kaufman, 1991: 104).
PURPOSE SPACE/TIME STRUCTURE PROCESS
broadest sense, [value-embedded] [playing field] [information] [energy flows]
policy systems are RULES PRINCIPLES ENVIRONMENT PRESSURES DESIGNING
both rules of motions FACTS SELECTING
and patterns of SYSTEM RELATIONSHIPS
behavior. Rules TENSIONS IMPLEMENTING
indicate how a BEHAVIOR BOUNDARIES FEEDBACK EMERGING ADAPTING
system should DECAYING
behave and include Figure 2.1 Elements of a Policy System
all the explicit and
implicit instructions affecting the desired behavior (Ostrom, 1986; Stone,
1988). They are based on insights -the explicit theories, concepts, and
opinions operating in the policy space (Weiss, 1980). Insights represent
what the members of a system know and understand about their goals, the
system goals, and the systems environment. Principles, consisting of shared
beliefs and assumptions, constitute the systems identity and mitigate
ambiguity; they maintain cohesion between the prevailing insights and the
rules of motion. They reflect the culture and values of society itself. All
policies contain designs that establish an explicit relationship between the
prevailing insights and principles and the rules of motion that determine what
entities are allowed to do.
Three goal factors contribute to a policys complexity. Most policies
have multiple goals that compete for attention. In some situations, what
seems to be conflicting goals are not mutually exclusive. Choosing among
them is a matter of balance (Coe, 1997). The degree of ambiguity associated
with the systems goals introduces different levels of uncertainty (Dery, 1984;
Dunn, 1988). Public values are situational and change in relation to what is
possible to achieve (deLeon, 1988). Therefore, policy designs continue to
evolve, change, expand, contract, or even disappear as part of the overall
dynamic process (Berman, 1980).
Public policy attempts to motivate individuals to engage in policy-
preferred behavior (Deutsch, etal., 1986; Schneider and Ingram, 1990a;
Rapoport, 1985). Policy generates energy for action from the structural
tension that emerges when a policy goal differs from the current situation
relative to the goal (Fritz, 1989). Because rules reflect a preference for a
desirable system state, they promote movement from a less favorable to
more favorable system state (Sutherland, 1975: 241). From a design
perspective, feedback and behavioral assumptions become important
features for assessing policy patterns of behavior (Schneider and Ingram,
A system manifests itself in space and separates itself from its
environment by imposing boundary conditions. The type of boundary
separation determines a systems ability to gain or lose energy. In the
context of policy, impenetrable boundaries isolate a system from the external
environment, protect it from external pressure, but limit the amount of energy
that can be generated. In contrast, an open systems permeable boundaries
allow it to replenish its energy supply. Boundaries become the places where
the inner policy world meets the real, outer world, defining Simons (1992)
sciences of the artificial." Closed policy spaces are simple to define. More
often, policies operate as open systems and continually exchange energy
and knowledge across their boundaries (Cleveland, 1985; Tushman, 1977).
Disturbances introduce noise and uncertainty into the policy arena
(Bobrow and Dryzek, 1987). Organizations are averse to uncertainty and
generally tend to respond to its presence by either expanding their
boundaries or contracting the system size (Kaufman, 1985:117). If the
uncertainty is inside the boundaries, the system contracts, (f it is outside the
boundaries, the system expands. Since there are always new sources of
uncertainty outside the boundaries, systems tend to expand. Since every
expansion introduces new uncertainty within the boundaries, systems tend to
contract. Consequently, policy boundaries are inherently unstable (Stone,
Time and energy flows are interrelated. Einstein (1952) proposed that
space and time are not separate entities but a time-space continuum. He
observed that a three-dimensional happening can be conceived as an
existence in the four-dimensional world. The geometry of the time-space
continuum generates both the form and path of the object of interest. Within
the time-space continuum, behavior patterns emerge, depicting the direction
and shape of a system transitioning from one state to another.
A systems structure contains the information that differentiates it from
its environment and makes it coherent. Policy structure refers to the internal
arrangement of elements, linkages, and intersections (Majone, 1989). This
arrangement is captured in information. Information density determines the
degree of granularity in a system. Coarse granularity signals an open
information structure with many degrees of freedom (Bronowski, 1978: 88).
Fine granularity denotates precise instructions, numerous interrelationships,
and limited degrees of freedom.
Information density indicates a systems degree of flexibility and
capacity to change. The amount of information present is measured by its
granularity. Systems with coarse information granularity can accommodate
new information without rearranging its existing information structure. When
new information comes into systems with a highly-developed structure, the
system experiences high levels of tension or constrained animation among
its subsystems (Drucker, 1995). Tension and information flows counter a
systems natural tendency to move toward entropy (Kaufman, 1991;
Human systems interact with, and make sense of, their environment by
communicating. The environment transmits an infinite array of event data.
By imposing meaning, structure, and utility upon these data, the mind
produces information (Kim, 1985). According to Jantsch (1980: 163), the
human mind mirrors an outer reality that it rebuilds. The mirror image
emerges from an exchange process between sensory impressions and the
mental models that the mind projects. This exchange process breaks the
symmetry or equilibrium between the inner and outer worlds. It allows
systems to act in the present and anticipate the future (Dunn, 1986). In
simple terms, imagination, judgment, and purpose invent the policy future
(Lasswell, 1974: 177). But some information is accessible only through
cooperation and exchange (Dunn, 1986). Participatory design processes
open the policy system to greater involvement and facilitate the discovery of
shared meaning among the competing sets of knowledge (deLeon, 1992).
Design introduces new information, new ideas, and new choices into a
system (Loewy, 1979). People traditionally associate design with the visual
arts, i.e., the creation of two-dimensional graphics or three-dimensional
products. But a more accurate perspective recognizes that design applies to
a wide range of human activities (Ingraham, 1987). The patterning of any act
toward a desirable, foreseeable outcome constitutes design (Caplan, 1982;
Loewy, 1979; Papanek, 1971). While the latter definition represents
industrial design's design orientation," it is similar to Lasswell's policy
Design takes place at two levels in the mind of the designer and in
reality. The first level entails internalizing information and the second entails
externalizing information (Diesing, 1991: 291). Internalizing involves several
forms of thinking sensing, intuition, reasoning, and feeling. While sensing
alerts the designer to external elements of reality, intuition is the artful
application of knowledge (Sergiovanni, 1984). Reasoning or logic compares
two or more things or discovers a relationship existing between two things
(Bronowski, 1978). In the design process, logic weighs the intuitive
expectations against the external facts. Feeling determines the value or
importance associated with a situation. Although each is an independent
process, individually they are insufficient to adequately test reality.
Interlinked, the four forms of internal thinking become a unified means to
generate ideas (Stacey, 1988).
While internal design processes involve conceptualization, external
processes communicate ideas (Ruben, 1985). Designers externalize their
concepts by translating them into concrete forms or artifacts. Every artifact
that man makes represents a transfer of information from the human mind
into a tangible, sharable, inanimate object for others to discuss and modify
(Issar, 1995). Externalization moves ideas from the infinite set of possibilities
to a discrete set of probabilities that is the core activity of policy design or
policy formulation (Kingdon, 1984).
After testing ideas, the designer improves the remaining themes by
creating new combinations of elements until a concept emerges that fits" the
clients specifications (Weinberg, 1988; Loewy, 1979). To gain acceptance
for policy initiatives, policy design follows a recursive cycle of inventing and
winnowing (Polsby, 1984). During the policy design process, certain ideas
are selected and others discarded (Kingdon, 1984: 210).
Policy researchers often refer to the internalizing and externalizing
cycle as forming, storming, norming, and performing (Tuchman, 1984).
Forming tentatively tests what is and is not allowed by scanning the
environment. Information about human needs and environmental constraints
collide during the policy storm until a set of criteria emerge from the
collision. These become the basis for shared views, values, and direction or
the norms. The policy design evolves as alternatives are measured
against the norms. Conceiving design as a dynamic phenomenon
recognizes the importance of the relationship between turbulence and
change (Brewer and deLeon, 1983; Torgerson, 1986).
The design process can be expressed in terms of S specifications, C
constraints, d design, p parameters, and f features (Young, 1987). The
designer identifies a set of specifications composed of functional
requirements and constraints that describe the physical limits. Design
alternatives that meet the specifications and constraints exist within these
bounds. Each design is expressed in a set of features that provide a
distinctive identity. Each feature is expressed in a set of parameters that
qualify the features. Figure 2.3 reframes the structural view of design (Figure
2.2) into an open process that recognizes the human dimension and
environmental constraints. Together they form the operating criteria. As
information flows into the policy system, the design continuously evolves.
Design involves deliberate, focused actions. Focus relates to the
degree that the design specifications are known in advance. If the goals are
known a priori, then the design focuses on a known set of preferences.
When design goals are identified a posteriori, it is said to be unfocused and
unstructured. Corporate research centers often follow the unfocused,
creative thinking approach to explore complex issues (McLaughlin, 1990).
Figure 2.2 Structural View of Design
Figure 2.3 Process View of Design
For example, the gooey operating system that emerged from Xeroxs
research activities is a product of an unfocused approach. Only in a careful
retrospect did the research goals become apparent, when Apple Computer
capitalized on Xeroxs invention.
Public policy seldom uses the unfocused approach. Typically, policy
analysts draw causal links between a problem and a desired behavioral
outcome. By anticipating social behavior patterns, policy designers can find
the levers to create change (Schneider and Ingram, 1990a; Weimer, 1995).
They understand, however, that complex systems rarely exhibit clear links
that translate into simple, casual behavior patterns (LaPorte, 1975; Argyris,
1993; Lumley, 1997).
Design thinking is the ergonomics of a social system. Engineers make
things work; design makes them workable (Caplan, 1982). The social
engineering approach to policy analysis involves test and conjecture about
improving a social condition. Policy design involves human debate about the
appropriateness of the approach and the importance of the social issue to
the individuals affected (Bobrow and Dryzek, 1987). Supporting postmodern
thinking, the operating design paradigm conceives policy as the process of
humans learning to generate and adapt models for human beings to use
(Kim, 1992). Figure 2.3 proposes a schema that incorporates the human and
transformational nature of design processes. Diverse information entering
the policy field refines policy knowledge. Figure 2.2 espouses a convergent
perspective while Figure 2.3 represents a divergent view of policy design.
Uncertainty concerns present or future conditions, the consequences
of actions, and individual and social preferences for those actions (March,
1978: 589). The internal need to act and the external pressure to decide
generates tension and conflict in the policy process. Crozer (1956)
distinguished between conflict as a means to an end and conflict as an end
in itself. As an end in itself, conflict can lead to polarization, diminished
communication, and system deterioration. As a means, conflict can stimulate
action and broaden perspectives.
Some policies contribute to system deterioration by stifling information
flows. When a policy attempts to circumscribe a system with rules, it erects
information barriers (Jenkins-Smith, 1988: 69). By limiting access to existing
information, it affects the ability of policy stakeholders to collaborate and
potentially create new information (Schon, 1983). Higher education policies,
for instance, historically have supported conserving energy. Whenever they
protected institutions turf, resisted new ideas, or maintained level funding,
state commissions have stifled information flows to and from the external
environment and invited regulation (Jones, 1995).
Conflict also promotes system evolution or transformation (Kaufman,
1991; Torgerson, 1986: 52). When the future is unpredictable, sources of
difference may become occasions of convergence. Differences can increase
the number of policy options by generating new points of view. According to
the postmodern policy movement, diversity is essential for challenging the
status quo in the face of continuous change and contentious issues (deLeon,
1992; Pascale, 1990; Stacey, 1992; White, 1992). Fluctuation and
disturbances promote information exchanges. Through constant exchanges,
a system clarifies its values and priorities, information increases, and the
policys efficacy increases correspondingly. The members of information-rich
systems interact in a more coherent manner than information-poor systems
Activities that create movement are potential creators of new
information (Casti, 1978; Meltsner, 1983). When information barriers are
low, ideas spillover into other policy arenas (Polsby, 1984: 202). The
spillovers may disrupt grand lines of policy" when fundamentally new policy
values challenge and displace the old ones (Baumgartner and Jones, 1991).
Jumping outside the system, paradigm shifts, and lateral thinking refer to
this creative phenomenon (Imparato and Harrari, 1994; Kuhn, 1970; de Bono,
In the policy context, information expresses the physical
measurements, knowledge limitations, and time of the system (Stone, 1988).
In essence, information embeds the space, time, purpose, process, and
structure dimensions. In the rationale paradigm, information connotes a
means to control fragmented policies; the postmodern perspective perceives
an evolutionary process in which ideas to fuse and information diffuses into
the policy environment (Weiss and Gruber, 1986).
Information flows generate the energy that generates order (Rhodes,
1991; Wheatley, 1992: 116). Design itself can be conceptualized as an
energy field that forms around an idea to propel it through the phases of
innovation (deBono, 1992; Roberts and King, 1996). Innovative ideas that
provoke little resistance need relatively small energy fields to move them
forward. On the other hand, radical ideas that engender large amounts of
resistance require larger amounts of social energy. Figure 2.4 diagrams the
energy-resistance relationship proposed by Roberts and King (1996: 225).
Feedback within the system amplifies and reduces the tension that
surrounds new ideas (Bozeman, 1993). Information feeding back on itself
can increase the number of competing signals circulating in the policy space
(Wheatley, 1992). It also alleviates some internal tension by providing
feedback about past choices, which the system uses to make future choices
(Bobrow and Dryzek, 1987: 203). In complex, open-ended situations,
feedback provides a stream of data that reduces ambiguity and uncertainty
by mobilizing the social energy, i.e., information, involved in change.
Figure 2.4 Energy Fields for Incremental and Radical Change
Uncertainty means the existence or threat of novelty. Conversely,
creativity, which is synonymous with novelty, variation, invention, and
discovery, provides a means to address uncertainty. In this sense, creativity
requires cutting across information patterns to invent or discover an unknown
form (deBono, 1992). But even policies that break with past practices are
rarely invented (Kingdon, 1984:135). On a broader plane, any thinking
process that requires modifying or rejecting previously accepted ideas is
creative (Simon, 1979: 145; Schumpeter, 1939: 88). The coupling of familiar
elements is more common than the appearance of wholly new forms in the
policy arena (Kingdon, 1984: 210). In either case, creativity opens the
system to new connections and introduces change into the system
F 0 Redesign Rediscovery
C Policy Policy Expansion
s Policy Streams &
E Policy Windows of
D implementation Opportunity
Policy pinching Garbage Can
Figure 2.5 Topology of Policy Change
offer explanations of
policys creative capacity to
change behavior. The most
common policy innovation
terms are captured in a
topology of policy change,
using focus and type of
change as the topologys
dimensions (Figure 2.5).
Focus denotes the degree
of ambiguity associated with
goals. Linear change is a
smooth transition while nonlinear change is characterized by fluctuations.
The linear / focused quadrant includes theories that explain slight
policy variations. Minimal policy modifications allow a system to adapt to
discrete changes in its environment. Policy redesign, incremental change,
single loop learning refer to intentionally modifying an existing policy system
to create small amounts of change (Lindblom, 1959; Argyris and Schon,
1978). A policy environment where policies evolve as they are implemented
is conducive to this type of change (Wildavsky, 1978; Majone, 1989).
The nonlinear/ focused quadrant of Figure 2.5 represents policy
processes that are not exclusively logical, although they are neither random
nor arbitrary. Policy pinching (Schneider and Ingram, 1988), unself-
conscious choice (Dryzek, 1983), and advocacy coalitions (Sabatier and
Jenkins-Smith, 1993) are means of maintaining a policy fit to the
perturbations in the environment. Policy pinching innovation originates not
from the source of ideas but from using exogenous ideas in unfamiliar
contexts. By coupling common solutions to uncommon problems, ideas
become the primary vehicle for promoting behavioral changes. Policy-
oriented learning within coalitions identifies performance gaps or
demonstrates deficiencies in anothers set of core beliefs (Sabatier and
Jenkins-Smith, 1993). The energy exchanges between stakeholders are the
principal vehicles of change. In the nonlinear / focused approach, innovation
often springs up at the boundaries of disciplines rather than in the middle
(Kanter, 1988). Since innovation involves fusing two dissimilar ideas, entities
operating on the boundaries are more likely to be exposed to variety and
difference (Roberts and King, 1996).
The nonlinear/ unfocused quadrant of Figure 2.5 includes the policy
innovation theories that explain change by chance. Structure and context are
immaterial in this setting. Ideas incubate in the mind of the designer to
emerge spontaneously when a problem presents itself (Meltsner, 1991).
Under assumptions, dramatic change may occur even during a time when no
new policy is intended. They offer an explanation for the chaotic activities
that may occur when dramatic shifts in the social-political context alter
constraints and opportunities and create.
The concept exemplifying the linear / focused approach of the upper
right quadrant of Figure 2.5 is Kingdons (1984) policy streams. Policies are
generated in a policy primaeval soup regardless of whether they are solving
a problem. Policies expand in an unplanned fashion by randomly reaching
into the policy stream for an alternative that can be reasonably seen as a
solution (Kingdon, 1984). This view captures the fluid dynamic process but
downplays the self-interests that do not necessarily account for collective
Complex systems are prone to circular relationships that require
addressing several different levels of associated behavior simultaneously
(Ahl and Allen, 1996). If it is difficult to develop coherent policy within a
single policy area, it is harder to coordinate multiple policies. Fuhrman
(1990) notes that opportunities to multiply policy effects through coherent
approaches are often missed. In the worst cases, contradictory policies
cancel out each other. Higher education policy often has less effect than
policymakers desire because the interplay of policy and practice may
stimulate or repress certain behavior patterns (Fuhrman, 1990: 9).
There is increasing evidence that collective behavior, such as social
choice, is related to complexity and nonlinear dynamics (Richards and Hays,
1996; Lumley, 1997). For instance, Brian Arthur (1990) contends that
positive feedback occurs in economic systems in which the price equilibria
are unstable. His theory contrasts with the Adam Smith's view of economic
processes that posits prices tend toward a level at which supply equals
demand. Policies are highly susceptible to surprising behavioral outcomes
merely because they involve humans making choices (Bobrow and Dryzek,
1987:19). If the policy environment is dynamic and systems are prone to
nonlinear behavior, the interesting questions revolve around the ways policy
systems respond to change (Brewer and deLeon, 1983). The way policy
makers respond depends in part on their perception of change:
(1) Things will continue the way they are.
(2) Things will evolve in the same way they always have.
(3) Things will react to the current state of affairs.
(4) Things will change in ways that people can anticipate.
(5) Things will change due to events that cannot be anticipated.
A tentative analysis of the five change scenarios is presented in Table 2.1.
They represent five different perceptions of time: reversible, past, present,
future, and chaotic.
The traditional policy modes are captured in the first and second
scenarios that things will stay the way they are or evolve in the same way
they always have (Lindblom, 1959; Wildavsky, 1978). Recently, policy
studies have proposed that scenario one and two are limited and unrealistic
(Kingdon, 1984; Kiel and Elliott, 1995). The first assumes a closed policy
system. Change only indicates individual deviations from the single state.
The second perspective ignores the possibility of nonlinear or discontinuous
At the other end of the spectrum, scenario five assumes that all
change is discontinuous and cannot be predicted with any confidence.
Policy in the presence of a lesser degree of uncertainty is a fruitless effort.
The policy coin-tossing metaphor fits this mental model of change
(Lindblom and Cohen, 1979). Fast change implies that the world cannot wait
and varied change implies that the system cannot plan (Cook, et al., 1997).
Agreements among special interest groups drive temporary changes
in the third scenario. Sometimes policy makers operate in this venue
because they are uninformed, misinformed, or over informed. More often,
policy leaders knowingly make over-optimistic promises in order to satisfy
critics or quiet discontent (Kaufman, 1991: 53). Change is a reflexive motion
to power plays and the system returns to its original state when the
politicians attention is distracted or another special interest group gains
influence. The system plays the waiting game (Bardach, 1974).
In contrast to these four views, contemporary policy scholars advocate
that systems need to deal with change by: (1) developing a holistic view, (2)
embracing contention, i.e., tension, as a source of energy, (3) using energy
to achieve goals, and (4) developing a sensitivity to external realities by
communicating with constituents and stakeholders (Kouzes and Posner,
1995; Schrange, 1990; Torgerson, 1986). To counter the deficiencies of the
rational, economic approach of policy design that often produces solutions
that are out of touch with the public, the postmodern movement proposes
expanding both information and citizen involvement (deLeon, 1992). The
pace and diversity of the expanded environment predispose a system to
operate dynamically (Schon, 1971).
The theoretical question then is: What are valid policy assumptions in
a dynamic environment? This research study will explore this question by
further studying two related theories: information theory and chaos theory.
Table 2.1 Change Scenarios
SCENARIOS POLICY COMPONENTS Continue As They Are (D Evolve the Same Way As Past (2) React to the Current State of Affairs (3) Change Can Be Anticipated (4) Change Cannot Be Anticipated (5)
VALUES Efficiency Stability Accountability Self-Interest Diversity Order Freedom
PRESSURES Closed system: minimal or no external pressure Semi-closed: internal preferences predominate Semi-open: External interests predominate Open system: combination of internal & external Open system
ALTERNATIVES Single Means to end Limited Determined by stakes and number of actors Diverse Infinite degrees of freedom
PREMISE Status Quo Predictable Power Risk-tolerant Uncontrolled
CHOICE Emphasis on structure Emphasis on outcomes: maximizing benefit Emphasis on process: coalition and compromise Emphasis on interrelationships Emphasis on context: No choice
IMPLEMENTATION Maintenance Incremental Reactive Adaptive Spontaneous
TIME Static Reversible Discontinuous Irreversible Discontinuous and Irreversible
BEHAVIORAL CHANGE Reinforce existing behavior Deterministic, intended con- sequences, potential long term change Shift in power but minimal behavior change; return to prior state Deterministic, consequences, potential long term change Temporary shift in behavior
INFORMATION CHARACTERISTICS Data; controls Causal Anecdote Patterns Ad hoc
ENVIRONMENT Stable Stable Stable Dynamic Dynamic
Information is a basic descriptive concept used in business,
communication, information science, law, natural sciences, and social
sciences. While dictionaries provide an array of definitions, including facts,
details, intelligence, knowledge, material, news, these synonyms describe
different phenomena. The fact that the word information" is used
interchangeably with data and knowledge illustrates its ambiguity and the
To mitigate this ambiguity, information scientists developed a
schematic of general communication (Figure 2.6) and a basic taxonomy to
differentiate the levels of information. Syntactical information is the
physical transmission of signals from a transmitter to a receiver. While
syntactical information has no content, it is still prone to distortion or noise.
External influences on a channel can change the sequence or combination of
At the semantical level, grammar and vocabulary rules provide a
common frame for transmitting content. Semantic disturbances occur when
an information source and a destination use different sets of rules and are
unable to translate data even though the physical transmission is reliable.
Signal Received Signal
| i: Transmitter!-!^ Receiver |>j
| Noise |
i Source |
Figure 2.6 Schematic of General Communication System
Figure 2.7 Information Continuum
The element of surprise differentiates data from information (Klapp,
1986). At the symbolic level, a message contains information when it
surprises the receiver with previously unknown facts, knowledge, or
imagery (Wiener, 1950:132, Casti, 1995: 210). Its value is its capacity to
provoke a response. As illustrated in Figure 2.7, the amount of information
present in a system is proportional to the degree of surprise. Redundancy
protects a system against error, but it also clutters the communication system
with repetitive patterns that are monotonous, trivial, and lacking in surprise
(Weaver, 1967: 210). At the other extreme of the information continuum,
random data signals generate noise (Pierce, 1961). The information
continuum exhibits entropy as the amount of redundancy or noise present
increases. The second law of thermodynamics specifies that energy
degrades to a more probable, less informative state. Systems tend to move
toward entropy by eliminating variances.
Claude Shannon formulated a mathematical measure of information in
1948. Shannons entropy S is defined in terms of a question Q and
knowledge X about Q. Information / is the difference between the entropy
associated with knowledge X before the message and the knowledge X after
the message (Shannon and Weaver, 1948).
I = S( QIX ) S( Q IX ) (2.1)
Although Shannons measure of information was developed to reduce
the noise in a telephone line, its significance lies in the fact that information
may be measured independent of the means to generate information. Policy
information may be measured by the difference between the present state of
the system and a future state. When the difference tells something unknown,
it produces a change in policy knowledge. Shannons formula implies that
inquiry is essential for creating knowledge, which in turn promotes changes
in system behavior.
Feedback, load, and lag are the characteristics that measure a
systems response (Von Bertalanffy, 1949). Load is the amount of
disturbance present in a system. The system lacks sufficient energy to act
when only small disturbances are present. In large amounts, however, load
creates conflicts, uncertainty, delay, and disjointed actions. The amount of
disturbance also affects the time a system
Data accumulates quickly while
meaning develops at a relatively steady
rate (Figure 2.8). Synthesizing large
amounts of data into meaningful patterns
requires time. Data becomes noise when
the volume exceeds processing capacity.
If a channel can reliably transmit
information at a certain rate, then information overload occurs when the
transmission rate of data exceeds the capacity of the channel (Shannon and
Weaver, 1948). Policy systems inevitably operate in information overload
with many problems and solutions competing for attention (Kingdon,
1984:185). Policy lags occur when people are unable to agree on larger
patterns despite the fact they share the same facts and data (Weiss, 1980).
needs to respond, i.e., the lag.
Figure 2.8 Information Lag
By dissipating random data, noisy systems can move toward a desired
state of information (Bennett, 1988). To counter the overload and lag effect,
systems use filters to moderate the rate of data entering the system. A
system using a short term filter, i.e., accepting only recent data, is highly
responsive and conducive to ad hoc policy approaches. A long-term filter
accepts signals from the past, but is relatively unresponsive to short term
Consequently, filters determine a systems volatility. All filters
potentially destabilize a system because they intervene between an
information source and an intended destination. Filters also may stabilize a
system by separating noise from substance (Tufte, 1995:78). By creating
temporal gaps, they allow the system to assimilate, select, and organize data
into new patterns. Organized data becomes information. Information
absorbed, understood, and accepted, becomes knowledge (Lindblom and
Cohen, 1979; Holznerand Marx, 1979; Cooley, 1987).
Initially dealing with communication concepts, information theory
spawned cybernetic theory that linked communication and control (Wiener,
1961). Although many of its concepts system boundaries, subsystems,
circular causality, feedback are relevant, the social science community
generally considered cybernetics too simple, linear, and mechanistic to apply
to human behavior.
In the 1970s, classical cybernetics evolved into a second-order
cybernetic theory (Laszlo, 1972). Second-order cybernetics stressed live
intelligence and interaction rather than the negative feedback loops and
control orientation of earlier cybernetics theory. The new perspective
expanded the systems nomenclature to include autonomous systems,
positive feedback, evolution, and self-organization (Umpleby, 1990). Its
| principles center around a systems interaction with the environment
| (Holland, 1975):
A system is a set of interacting and interrelated parts.
Open systems have permeable boundaries and are continually
engaged in importing, transforming, and exporting energy, information,
Systems are capable of increasing their order. They are able to
survive and grow if they work out a mutually beneficial relationship
| with their environment.
A system is more than the sum of its parts. Its properties emerge from
the relationship among its parts and from the system's relationship to
Systems tend to maintain steady states, i.e., states of dynamic
equilibrium in which diverse forces are approximately balanced.
To maintain a steady state, systems need feedback loops that enable
the systems to sense relevant changes in their internal and external
environment and to adjust appropriately.
Its propositions challenged some of the control and homeostatic
concepts of the original cybernetic theory (Holland, 1975). It implicitly
redefined information as a self-referential system one that contains
information and knowledge about itself, its own state, structure, and
processes (Geyer, 1994).
A more recent contribution to information science is the set of
j information properties defined by Gell-Mann and Lloyd (1996). These
properties (Table 2.2) offer a framework for identifying discrepancies in
physical, social, political, and economic patterns (Holland, 1996). Because
irregularities occur when a system does not respond in expected ways,
Table 2.2 Information Properties
1. I (A) Â£ 0 Non-negativity
2. I (A, B) = I (B, A) Symmetry
3. I (A, B) > I (A) Accumulation
4. I (A) + I (B) Â£ I (A, B) Convexity
where I = Information
anomalies call into question the fundamental assumptions underlying
paradigms (Kuhn, 1970). Policy scientists refer to irregularities or anomalies
as unintended consequences or rogue outcomes (Demchek, 1992).
In summary, modern information theory deals with the dynamic
qualities of information. Information's value depends, not on its quantity or
mass, but on a systems response to its presence. As data moves across
system boundaries, it becomes the energy that stimulates change. Entropy,
i.e., negative information, is a measure of energy dissipation. While entropy
measures arbitrariness, disorganization, or the unknown, information
measures order. Increased order corresponds to decreased entropy, often
referred to as an increase in negative entropy or negentropy.
Information is an evolving element that captures both pattern and
transformation (Oyama, 1985; Kaufman, 1991). Information can be
considered a construct of the mind operating on raw data while knowledge is
the union of a number of sets of ordered information (Issar, 1995). Energy
flows manifest themselves in the reorganization of data and structures
(Jantsch, 1980: 35). As information evolves, it creates more connections
than may be needed at a time (Morgan, 1986: 96). Redundant relationships
endow systems with the flexibility to self-organize and replace unnecessary
connections. Acquiring and sharing knowledge simultaneously creates
information and erodes structures (Weaver, 1967; Cleveland, 1997).
Inserting policy into these concepts, information can be described as
the energy reorganizing a policy system. As data moves across policy
boundaries, it becomes the energy that stimulates change in a social system
(Holznerand Marx, 1979, Cleveland, 1997). The intersection of policy
problems, alternatives, and desired outcomes may be stated as the point
where policy data of what is (facts) meets information (policy action) by
testing policy knowledge of what is right (values) (Dunn, 1981). Anticipating
policy outcomes or envisioning future states that are likely to occur if the
system pursues a particular policy alternative challenges the assumptions of
policy knowledge (Feldman, 1989).
The expected utility of knowledge varies with the composition, scope,
and effect of knowledge (Rich, 1981; Dery, 1984; Weiss, 1991). High-level
knowledge facts and formal rules is used to inform the public or make
policy decisions that allocate resources or benefits. Heuristics are the high-
level skills that support the conceptual or behavioral use of information.
Common sense includes the manual, perceptual, and cultural skills that
facilitate data collection (Collins, 1990). By promoting the sharing of policy-
relevant knowledge, policy design considers information a social commodity
(Meehan, 1994). Policy information can be considered a construct that
creates order from raw data and a some heuristics or rules while knowledge
is the union of a number of sets of ordered information.
Humans have a deep faith in the efficacy of organized information and
the ability to produce it on demand (Holzner and Marx, 1979). Most
instances of ineffective use can be traced to communication breakdowns
(Rich, 1981). When communication breaks down, a system tends to move
toward entropy (Kuhn, 1970; Ewell and Chaffee, 1984). For example, when a
coordinating board or legislative committee receives irrelevant information on
a particular issue, they typically react by requesting additional data. If the
response involves providing available data without the appropriate context,
the influx of new data may disorganize the policy system, consume energy,
and move the system toward entropy. When people observe that information
is being supplanted by data that does not represent reality (Dery, 1990) or
that data confuses and obfuscates policy issues (Kraemer et al., 1987), they
are commenting on the failure of information to communicate (Ewell, 1989).
Information signals that a system considers a change occurring in its
environment significant (Feldman and March, 1981). In the policy
environment, information is used to detect anomalies, make connections,
systemize experiences, assess consequences of alternative courses of
actions, and establish credibility (Ewell, 1989). Credibility involves continual
testing to see if different systems share similar values (Kuhn, 1970: 200).
Higher education administrators rely on information to establish external
credibility with the political system and to identify what is important.
Higher education policy makers are less interested in using
information for credibility purposes and more interested in using it to
understand potential policy consequences (Baldridge, 1990). Using
contextual information to understand possibilities contributes to its credibility.
Contextual information places an issue in its proper framework by defining
the ways it affects other parts of the system (Churchman, 1975). Because
context captures multiple experiences, it represents global patterns and
interesting data relationships (Tufte, 1997). No order can be created without
information (Bateson, 1979: 47).
| A paradigm is a model or pattern of the way things are. Newton gave
j the world a paradigm of order, logic, and sequence. Over time, this paradigm
! has influenced not only the way that scientists approach science, but also the
j way that social scientists study public administration and the way
| policymakers design policy. The rational policy model, the policy version of
the Newtonian paradigm, is orderly and predictable. It can be understood by
; studying each of its parts (Allison, 1971; Nagel, 1990, Ackoff, 1996).
Scientists studying complex systems, however, found a world less orderly
| and less sequential than the Newtonian paradigm permits (Lorenz, 1995).
From complexity science, a new paradigm is emerging for the social
Complexity theory is the qualitative study of behavior of nonlinear
dynamical systems (Kellert, 1993: 2). It posits that a simple set of rules
; exists at the root of all complex systems (Gell-Mann, 1994). The interest in
! the global nature of systems offers a cross-disciplinary understanding of
choice and change. For example, social scientists are relating the new
assumptions originating from studying physical systems to human behavior.
Natural and physical scientists are overlaying the human scale of social
science on the physical world. Both are seeking connections between
| system behavior and different patterns of irregularity.
Consequently, complexity theory is saturated with terminology
| originating from a variety of diverse scientific disciplines. The language of
! complexity ranges from scientifically and mathematically precise definitions to
metaphors. But all concepts deal with pattern, change, or information. Table
2.3 presents a list of simple definitions used in this paper.
The Web Dictionary of Cybernetics and Systems(www. comp Iexity. htm)
defines change as a difference in the values of a variable over time" or the
difference in a systems state observed at different times. It differentiates
quantitative change changes in numeric or mass variables such as
volume, wealth, or transmission rate" from qualitative change
differences in structure, pattern or level.
In some cases, the complexity definitions differ from policy
connotations. For example, chaos connotates negative policy images
turbulence, disorder, anarchy, or lawlessness. Complexity theory merely
postulates that chaos is an irregularity. The two meanings of chaos reveal
the difference between the assumptions of the Newtonian paradigm and the
complexity paradigm. In contrast to the Newtonian assumption that any
realistic theory needs to account for randomness by including a noise factor,
complexity scientists assume that interesting patterns emerge due to
information gained or lost.
The terminology is supplemented with graphic imagery that represents
behavior and change. Different attractors points, periodic, strange
describe different dynamic states equilibrium, near-equilibrium, far-from-
equilibrium, chaos (Wolfram, 1988; Merry, 1995). As information moves the
subsystems to the global system, an attractor amplifies small uncertainties.
At the same time, feedback processes reinforce a stable state or stabilize
unstable periodic orbits on strange chaotic attractors (Ott, Sauer, Yorke,
Table 2.3 Glossary of Common Concepts and Terms in Complexity Theory
bifurcation a directional shift; a moment of choice in a systems evolution.
change a difference in variables over time.
chaos irregular, unpredictable forms.
complexity from Greek pleko, meaning to to plait or twine.
dissipative structures open systems that maintain coherence by reconfiguring at higher levels.
dynamic rule about system behavior.
entropy tendency of any system to slide toward a state of increasing disorder.
equilibrium system without energy exchanges.
far-from-equilibrium system with high energy flows from the environment.
information capacity to surprise or create a new form.
mutual information degree that one part of the system contains information present in other parts.
periodicity time between periods of chaos and order.
perturbations fluctuations that agitate a system.
self-organization continual renewal and learning.
state essential information about a system.
The most elementary attractors are points and lines. Whatever the
initial conditions, certain systems will move toward a point attractor or along a
line. In contrast, certain systems will exhibit deterministic chaos, that is
irregular and unpredictable evolution. Deterministic chaos is considered a
strange attractor because there is no volume in the phase space where all
nearby trajectories converge (Baker and Gollub, 1990).
Using these classes of behavior as a frame of reference, scientists
discovered that complex systems exhibit certain universal features,
independent of the details of the system (Stein, 1989: xiii-xv), including:
Nonreducibility. Complexity is embedded in the numerous ways that
the components of a system interact.
Emergent behavior. Complex systems evolve toward greater
Unpredictability. Complex systems exhibit surprising behavior from
small changes in initial conditions.
Social scientists and physical scientists agree that a dynamical system
is more than a collection of parts. A large number of components, great
differentiation, and tight internal coupling among the components
characterize a complex system (La Porte, 1975). Complexity tends to
increase as a dynamical system adapts to break through limitations, handle
exceptional circumstances, or adapt to a more complex world (Arthur, 1993).
The complexity of a deterministic, nonlinear, dynamic system is embedded
in the numerous ways that the components interact (Waldrop, 1992: 11).
Because a systems individual components are coupled, a change in one
element of the system is coupled to changes in other components. The
circularity or self-referentiality allows a system to maintain a stable set or
relations under changing conditions (Manturana and Varela, 1992).
Dynamical systems fall into two major categories: conservative
systems and dissipative systems (von Bertalanffy, 1968). Conservative
systems neither export nor import energy. All their interdependencies are
internal. Dissipative systems are open to the environment, continually
dissipating and consuming energy (Mainzer, 1996). Since energy crosses
dissipative system boundaries, dissipative systems have numerous internal
and external interdependencies.
The structure of a dynamical system consists of the interplay of two
parts: a state the essential information about a system and a dynamic
a rule about how a system changes with time (Crutchfield, et al., 1986: 49).
A rule determines when a system transitions from an ordered to a less
orderly state, and conversely, from disorder to a more orderly state. A
classic example of the interplay of state and dynamics is Craig Reynolds
computer simulation of a flock of boids. Each boid followed three simple
rules of behavior using information about the state of the flock:
Maintain a minimum distance from other objects in the
environment, including other boids.
Match velocities with boids in its neighborhood.
Move toward the perceived center of mass of boids in its
The simulation demonstrated the dense coupling among a dynamical
systems parts and the simple rules that determine how a flock transitions
from an orderly state to a less orderly one and back to a more orderly state.
Design also involves the interplay of information and rules. It is the
conscious effort to impose a meaningful order (Papanek, 1971: 4). A primary
rule of design specifies that form originates from function, but evolves as
human behavior changes (Loewy, 1979). This design dynamic is consistent
with the more global rule regarding dissipative systems. The rule specifies
that the structure of dissipative systems originate and evolve from the free
exchange of energy between a system and its environment (Prigogine,
The notion of emergence that is antithetical to much of modern
science is a principal message of complexity science (Lewin, 1992: 191).
Two significant ideas emanate from emergence: (1) change is not dominated
by the action-reaction mechanical laws, but a creative phenomenon; and (2)
the emergence of information drives a system toward greater complexity
(Gleick, 1987 :5). The emergence of a drive toward greater complexity and
greater information processing implies increased order (Lewin, 1992: 149).
Order can be spontaneous, externally imposed, or self-generated
(Keller, 1985: 132; Gell-Mann, 1994). Many complex systems will evolve
spontaneously to the critical edge between order and chaos (Bak, 1996).
This phenomenon is referred to as self-organized criticality. In simpler
terms, self-organized means a system is capable of reaching a steady state
without external controls. Criticality means that the system is barely stable.
Dynamical systems tend to maintain a barely stable state by accommodating
external pressure and internal tension. Semi-autonomous agents literally
reconstruct themselves as they co-evolve (Kauffman, 1995). By operating in
a barely stable state, a system experiences large disturbances less
frequently than small ones (Bak and Chen, 1991). If a system is not on the
edge of chaos, learning and evolution will move it in that direction. Learning
and evolution will pull a system back from the edge of chaos if it should drift.
In other words, learning and evolution will make the edge of chaos stable
(Waldrop, 1992: 295).
Although many systems enjoy periods of stable fluctuations, other
systems experience periods of increasing tension and turbulence. Complex
systems require small amounts of tension to disrupt them (Landau, 1973:
535). The rule of requisite variety states that a dynamical system will self-
organize when its experiences small perturbations (Wiener, 1961). Individual
entities self-organize based on information transmitted from the system and
the environment. When the energy level of small fluctuations exceeds a
systems adaptive capacity, the fluctuations begin to perturb a system. This
produces disorder and instability. As the perturbations accumulate, a system
will adjust to the volatility of its environment by altering its structure (Monod,
1971). It bifurcates or shifts direction. A system that fails to adjust is prone
to die (Tainter, 1988; Kaufman, 1991). A dynamical system resists entropy
by becoming more ordered and more complex.
The behavior of nonlinear systems differs from that of linear systems.
A nonlinear system may be stable under several behaviors. Near-equilibrium
systems have a single attractor and tend to move toward a steady state
(Figure 2.9). But a system may drift into a far-from-equilibrium behavior
pattern. If increased energy and matter flow into the system, the attractor
that dominated the behavior of the system may become unstable. The
system is driven away from equilibrium. New attractors appear (e.g., cyclic
patterns). Energy flows often create an oscillating pattern, a condition which
is characterized by repetitive oscillations with regular periods and amplitudes
(Figure 2.10). An oscillating pattern may be a steady state for a system.
Deterministic chaos, as diagramed in Figure 2.11, is characterized by
non-repetitive oscillations with variable periods and amplitudes (Ayres,
1994). Its divergent pattern may look disorderly, but underneath the
seemingly random behavior is a sense of order and pattern. The processes
of chaotic systems that appear to proceed by chance, in fact, follow a precise
set of rules or dynamics. A chaotic system could be considered stable if its
particular pattern of irregularity persists or if its global behavior is near-
Figure 2.9 Figure 2.10 Figure 2.11
Global behavior that can not be predicted from individual component
behavior emerges from the interaction of the components. The emergent
patterns are a joint property of the agents, their interactions with each other,
and their interactions with the ambient environment (Casti, 1997: 34).
Global, emergent behavior, in turn, influences the behavior of the individual
components that produced it. The continual process of change creates new
and more evolved order (Toffler, 1971).
It follows then that the evolution of complex systems is irreversible
because the only alternatives available to the system are those of increasing
complexity or extinction (Laszlo: 1969; Waldrop, 1992).
Modern science recognizes a deterministic system as one whose
future states are determined by its information state and its rules of dynamic
motion (Casti, 1995: 87). This contrasts with the Newtonian view that the
future of a deterministic system can be predicted, which requires problems
amenable to logic, perfect information, and a single solution. However,
agents have limited access to the existing information in a dynamical system
(Forrester, 1994: 62) and open systems experience divergent problems that
do not lend themselves to a single solution (Pascale, 1990). Contemporary
scientists conclude that a complete set of past and present data is insufficient
to predict the future (Field and Golubitsky, 1992).
Peter Senge (1990) draws a distinction between detailed complexity
and dynamic complexity. Detailed complexity requires perfect information
because it relies on identifying all the variables that could influence a
problem. Dynamic complexity occurs when cause and effect are not close in
time and space. Obvious interventions do not produce expected outcomes
because other unplanned factors dynamically interfere.
By acquiring data about its environment and its own interaction with
that environment, an adaptive system identifies regularities, condenses those
regularities into a schema, and reorganizes itself according to its internal
schema (Gell-Mann, 1994:17). The data acquired communicates the present
state of the environment; information is the difference between the data and
the systems internal schema (Bateson, 1972: 317). Figure 2.12 represents
Consequences (real world)
Description, prediction, behavior
SCHEMA that summarizes and is capable
of predicting (one of competing variants)
Identification of regularities
Previous Data, including
Behavior and Effects
Figure 2.12 Gell-Manns Model of Adaptation
Gell-Manns model of how a system grasps data and gives them meaning.
Christopher Langston framed
information as a double loop process
(Figure 2.13). All arrows represent
information flows. In the local
feedback loops, each entity is
acquiring information about its
environment and its own interactions
with other parts of the system. They
are individually identifying
regularities and condensing them
into schema. Local interaction and
the global structure are linked in a
tight feedback loop (Lewin, 1992).
Shared information enables a system to modify the interactions among
its subsystems (Senge, 1997). When subsystems compete for information,
the resulting internal tension generates irregular behavior (Monod, 1971).
Prisoners dilemma and MITs Beer Game simulate situations where
communication patterns break down and crises ensue (Senge, 1990). In
contrast, cooperation is fostered in high-stakes, high-tension situations when
feedback is present (Wiggins, 1997: 7). The intensive exchange of
information facilitates cooperation among agents (Merry, 1995: 46).
In the Newtonian paradigm, time is perfectly reversible. If time were
reversible, the past and future are symmetrical and prediction is probable. If
time were reversible, humans would have the capability to explain future
phenomena. Because all antecedents would be part of human knowledge
and experience, nothing new can ever happen. Irreversible time implies that
something new may always happen and the future is not predictable. John
Casti (1997) compiled a list of surprise effects that have broad, behavioral
implications (Table 2.4). To deal with the surprise generating phenomena,
complexity science acknowledges the importance of all information, including
the nonsense essential to creativity (Zukav, 1979; Thompson, 1997: 94).
Table 2.3 Surprise Generating Phenomena
source: J. Cast', 1994, 263
Discontnuity from smoothness
Output transcends rules
Behavior cannot be decomposed
The direction and stability of a system may be investigated by
considering the surprise effects of even small disturbances (Froyland, 1991).
Scientists studying small disturbances observed that systems are likely to
change in the direction that allows energy to flow at the maximum rate. A
system in a far-from-equilibrium state will use all the possibilities open to it,
but will be attracted toward the one with the maximum information flow rate.
Order emerges from a competition among flow efficiencies.
Complexity studies link chaos with the emergence of new information
(Gleick, 1987: 258). A system evolves by alternating between periods of
turbulence that are characterized by inventions and mutations and periods of
relatively stable behavior that are characterized by incremental adaptation.
The influx of information forms new relationships. Systems are constantly
adapting to change through mutual accommodation and mutual rivalry.
In summary, complexity is a different way of looking at the world.
While the Newtonian paradigm decomposes hard problems into causes and
effects and parts, complexity is concerned with patterns and the whole.
Whereas conventional science has dismissed irregularities, complexity
theory suggests that small irregularities grow to major ones over time. The
defining features of complex systems include:
Irreducibility, making it impossible to segment the system without
losing the very information that makes a complex system a system.
Emergent behavior that transforms the system from one state to
Feedback and feedforward loops to enable the system to restructure
the interaction pattern among its variables, opening up the possibility
for a wider range of behaviors and surprising consequences.
Summary of the Literature Review
The literature review of complexity theory, information theory, and
policy design indicate a paradigm shift is occurring in the social sciences.
The Newtonian method is losing its place as the dominant perspective.
Because modern policy design, information theory, and complexity theory are
oriented toward interaction, innovation, and transformation, the theories
appear to share similar constructs, including information, communication, and
adaptation. This summary draws certain connections between the three
Design is the process of inventing new things that display new physical
order, organization, form, or function (Alexander, 1966). Policy design
creates new social order or organization. As most commonly used,
organization implies a complex, complementary dependency in behavior
(e.g., an ecological system or a social organization like higher education)
whose members follow conventional rules of conduct. The second meaning
applies to operationally closed systems. An autopoietic organization is an
autonomous and self-maintaining entity in which the interaction of the
components form patterns without apparent inputs and outputs (Morgan,
1986). The third meaning, the relationships and processes of communication
among the components of a system that determine its dynamics and the
transformations it may undergo, best captures the concepts inherent in
complexity theory, information theory, and policy design.
Each theory values dynamic flows of information rather than static
control. Literally, information is that which forms. Theoretically, information is
the difference between two forms of order or between two states of
uncertainty. A message carries information to the extent it conveys
something not already known. Ambiguity is the coexistence of more than one
meaning or interpretation. In political discourse, ambiguity is often intended
to convince or attract a diverse audience. Information reduces ambiguity and
uncertainty. The degree of uncertainty removed by a message exemplifies
the efficacy of the communication occurring in the system.
Communication and use of information are central to the practice and
theory of policy design. Policy information involves a dynamic between an
end that is factual and a means that is valuative (MacRae, 1976). But it also
involves a dynamic between the creation of policy knowledge creation and
the use of information. Carol Weiss (1980) has noted that knowledge-driven
policy models influence decisions, information-driven policy models create
new knowledge, and social-driven policy models create temporal links
between policymakers and constituents. The growing complexity,
interdependencies, and pace of change of contemporary society make
existing knowledge obsolete, thus validating the need for information-driven
Information allows systems to self-organize. The adaptive self-
organization of a system is a precondition of its ability to function under
change (Laszlo, 1969). The concept of negative feedback is self-stabilizing.
It presupposes an ongoing flow of information between the components,
controlled by a code or standards that tends to perpetuate itself. If the
standard is not fixed but adaptable to the information flowing within the
system, a system possesses the ability to search out factors of invariance.
Inadequate information leads to exploratory responses, not the absence of
response. In essence, a system that functioned adequately at one time
ceases to function. When that happens, positive feedback exploration
locates new codes and negative feedback stabilizes its flow until further
changes produce a mismatch, calling for renewed adaptive self-organization.
Thus, information theory proposes that a system continually maps its
Complexity theory identifies three fundamental dimensions of self-
organizing systems: identity, information, and relationships that apply to
policy (Wheatley and Kellner-Rogers, 1996). Identity includes vision,
mission, and values, as well as the systems interpretation of what it has been
in the past and what it wants to be in the future. Information that is created
and transformed by relationships keeps the system dynamic. It demonstrates
intelligence of the system. Open systems predominately operate in an non-
equilibrium state, keeping off-balance so that they can change and grow.
As individual actions and events transform society, policy systems
move toward increasing internal complexity and external interdependence
(De Greene, 1993). The past policy approaches have effectively responded
to technological changes, but current conditions require adaptive change
(MacTaggart, 1996). The failure to provide the dynamic perspective stems
from the fact that the traditional policy approaches accent parts. The
postmodern view of policy and systems, with its emphasis on the whole and
behavior patterns challenges the historical reliance on retrospective data
(Drucker, 1995; Handy, 1997).
A nonlinear system with distinct feedback loops and strange attractors
may serve as a metaphor for policy. As a policy system evolves in time, it
shifts directions or bifurcates. Effectively, the system jumps from one
attractor to another. Each attractor corresponds to a particular policy
For instance, the debate surrounding affirmative action in higher
education exemplifies a policy bifurcation resulting from changes in the
environmental data. Until recently, the dominant policy ioop for states that
wished to increase higher education access to underserved student
populations involved measuring the percentage of minority graduates against
statewide goals a point attractor. The graduation data of minority students
indicated that the policy goal is no longer operating as intended. Recently,
Colorado and Wisconsin have adopted a more individualized set of goals and
activities for their affirmative action policies. As a result, the policy systems
are exhibiting signs of chaotic determinism with no data points converging on
the same space. The systems jumped to a new attractor.
A small change in a policy parameter may also trigger a bifurcation
within the policy domain. Under the 1992 Taxpayers Bill of Rights (TABOR
Amendment), which limits the amount of tax revenue that government may
spend, Colorados state government appears to display the early signs of
bifurcation. Paradoxically, the tax limitation was intended to control spending
and maintain a stable state. But the excess, unspendable tax revenues
generated in 1997 moved Colorado into a far-from-equilibrium state. The
discontinuous change in revenues may lead to annual tax rebates, decreased
tax rates, or special elections. In all cases, energy in the form of dollars is
dissipating from the state policy system.
Complexity theory view of system behavior is echoed in policy
sciences view of system behavior. Jam's (1982) coined groupthink to
describe a policy system that converged to a steady state. Tuchman (1984)
provided examples of how well-educated groups end up mutually reinforcing
their biases all the way to self-destruction. A policy system that resists
change has a high probability that it will disintegrate through a major
upheaval (Jantsch, 1980). A measure of success of a policy system is its
search for new forms of order within the framework of disorder (Kiel, 1994:
135). To further illustrate this point, parallels are drawn between equilibrium,
near-equilibrium, far-from-equilibrium, chaos, and entropy behavior and the
five policy scenarios described in Table 2.1 (p 42):
(1) Repeating former behavior in the same way.
(2) Varying behavior slightly and predictably.
(3) Accommodating competing pressures.
(4) Deterministic chaos, leading to a more complex state or entropy.
The first scenario represents a deterministic system whose future
states are completely fixed by its current state and its rules of motion. In this
case, the rules only recognize equilibrium. Policy serves to protect the
collective public interest, but ignores the individual interests and demands.
The second scenario represents a deterministic system in near-
equilibrium. When its subsystems are minimally coupled, a system will
behave predictably and support first-order change, the type that occurs in a
stable system making modest adjustments and muddling through. By
slowing down, the system is able to absorb data and circumvent the need to
change radically (Merry, 1995). Tractable policy problems foster near-
equilibrium behavior. They require modest amounts of behavioral change or
affect a minority of the constituents (Mazmanian and Sabatier, 1989: 24).
Typically, this type of behavior is seen in the financial policy setting where
small changes are negotiated through consensus building (Sabatier and
The third scenario represents a system that moves forward in a
determined manner, then moves away from the policy target, oscillating, and
ultimately giving up its goal (Fritz, 1989). It behaves like a stochastic system
whose future states are completely contingent on entities in the external
environment (Bateson, 1980; Kingdon, 1984).
The fourth scenario represents a deterministic system with emergent
behavior patterns. It involves the interaction of semi-autonomous entities and
multiple information flows. The systems permeable boundaries allow
feedback to enter and exit the system and flow within the system. The
behavior of this system echos the assumptions of the bounded rationality
interpretation of policy in which decisions are based on limited information
and action is directed toward local goals (Simon, 1979; Morecroft, 1994).
Policy dynamics seek balance points by accommodating everything in the
environment at a particular moment (Follett, 1940).
The fifth scenario represents an indeterministic system. It involves
numerous closed systems operating without feedback loops. Competition
and controversies among varied and clashing perceptions, values and
predispositions create chaotic behavior. In a severely stochastic system,
every event is its own data point and progression from earlier to later states is
primarily affected by the randomness present in the system. Some policy
systems operate as a loose collection of ideas and engaged in random
actions (Cohen, March and Olsen, 1972).
The most interesting dynamical systems are those that are locally
unpredictable and globally stable (Lorenz, 1995). Most policy systems exhibit
this type of global behavior while experiencing periods of local instability. The
unstable behavior during the recent faculty tenure discussions before the
Colorado State Legislature provides a graphic example. Prior to the bill,
tradition was operating as the sole policy attractor. When the bill was
introduced, the system oscillated between action and relief as the legislature
requested and received a series of studies and reports. During the testimony
before the Senate Education committee, the perturbations became stronger
and reached deeper into the system. More perturbations were introduced as
the media amplified the data and faculty coalitions began to split. Even
though Governor Romer vetoed the bill, the influx of data created instability at
the local level. All governing boards reorganized their existing tenure
procedures after the session ended and created a new policy order. The
higher education system remained globally stable during this period of high
One of the overarching goals of complexity is to discover a set of
principles that will make it possible to understand the behavior of a wide
variety of seemingly dissimilar complex systems. In particular, it is interested
in a dynamical systems sensitivity to small changes in initial conditions from
which new information and unexpected outcomes emerge (Crutchfield et al.,
1986). Public policy likewise is concerned with surprising behavior (Schon,
1971), paradoxes (Stone, 1988), and emergence the unexpected outcomes
that arise when individual entities of a complex system operate according to
their own preferences (Ingram & Schneider, 1991). As Bobrow and Dryzek
(1987) suggest, there is no logical reason for the intrinsic complexity of public
policy to be greater than the complexity found in natural sciences.
Complexity scientists are seeking ways to identify patterns that are
applicable to a wide range of phenomena. Some of the common threads
uncovered in the literature review that appear relevant to this search include:
Dynamical systems self-organize by breaking symmetry and creating
Non-equilibrium states enable a system to avoid thermal disorder and
to transform the data communicated from the environment into energy.
Dynamical systems exhibit signs of mutual cooperation and rivalry
adaptation and natural selection.
Adaptation creates new information and enables a system to evolve to
a higher level of complexity.
Information and evolution make the edge of chaos stable.
The general dissatisfaction with logical positivism (deLeon, 1992;
1997) as a theory of knowledge indicates that policy science is also searching
for tools that accommodate new ideas and patterns of behavior. In keeping
with the design model diagramed in Figure 2.3 (p. 33), policy design involves
the acquisition of new information and the ability to analyze that information
creatively, learn from it, and apply that learning in useful ways. As noted,
design processes involve confusion, curiosity, experimentation, discovery,
practice, and feedback. Modeling the study and construction of
relationships may be a powerful tool to apply the concepts of complexity
theory and information theory to policy design. A policy model may be a
means to acquire new information, analyze information creatively, and apply
that learning to discover ways a policy system may change as policy
METHODOLOGY FOR MODELING INFORMATION COMPLEXITY IN
HIGHER EDUCATION POLICY1
Until recently, it was possible for higher education policy boards to
wait for and react to demands from the educational environment. The
environment reflected a near-equilibrium, if not stable, state. Consequently,
linear methods of policy estimation adequately satisfied a policy board's
information needs (Brewer and deLeon, 1983).
Today the situation has changed. Higher educations environment is
more turbulent, less predictable, and more susceptible to external influences.
The current higher education policy system is decentralized, which implies
that the institutions act as semi-autonomous entities and the system is more
flexible. The environment favors local decisions. Table 3.1 contrasts a linear
policy perspective with a nonlinear view of higher education policy. A linear
approach emphasizes control and seeks to maintain a stable, known system.
Higher education policies that are predominately procedural and use
penalties to enforce compliant behavior to maintain the present state of the
system are indicative of a linear approach and a higher education system
that wishes to minimize risks.
A non-linear policy approach strives for adaptive and self-organizing
behavior. Tension is no longer viewed as a negative condition but a positive
This chapter describes the research approach and the proposed models
validity and reliability. Refer to Chapter 4 for the details of the specific model construction.
force that catalyzes action and change (Wilson, 1998). The system operates
in a discovery mode, seeking different ways to achieve evolving roles or
address emerging issues. Risk taking, therefore, is viewed as an acceptable
practice and sometimes encouraged with policy incentives. Colorados
Admission Standards Policy illustrates the non-linear policy approach. While
the policy purports to address the higher education systems enrollment
goals, the interaction between student decisions (e.g., a students decision to
accept an admission offer) and institutional decisions (e.g., the decision to
offer admission to a particular student) that influence the enrollment
The enrollment patterns emerge from the individual student
acceptance decisions, based on social, economic, or academic information,
and individual institution admission decisions, based on political and
Social, economic, and political dynamics call for approaches that are
not provided by the conventional equilibrium models (Shubik, 1996). The
policy sciences has long recognized this dilemma. Statistical approaches are
inappropriate for examining organized complexity (La Porte, 1975). Humans
have limited ability to deal with the dynamics of information-feedback
systems (Simon, 1977).
Studies of complex systems rely on computer modeling tools.
Dynamical modeling differs fundamentally from the other computer-based
tools listed in Table 3.2. Dynamic models focus on modeling in-the-large,
capture the configuration aspects of the phenomenon of interest, and
communicate asynchronously. Relational databases and object-orientated
programming are more suited for organizing and controlling situations.
Expert systems focus on modeling in-the-small, inferring solutions to specific
Table 3.1 Comparison of Linear and Nonlinear Policy Views
LINEAR POLICY APPROACH
Based on 19th century
stability, efficiency, power.
Institutions are treated
Issues are isolated and
No real dynamics in the sense
that each policy strives for
stability and maintaining the
status quo. Tension is
perceived as noise and
Elements are FTE and
Sees issues as structurally
simple, either/or situations.
Policies correct mistakes,
problems, and control
Operates in a risk
NONLINEAR POLICY APPROACH
Based on 20th century
physics structure, pattern,
Institutions are treated as
Issues are intertwined and
A policy system is a dynamical
system with new patterns
emerging and existing
patterns disappearing. Policy
tension creates movement
Elements are innovation and
Sees issues as inherently
complex with multiple
Policies anticipate choices
and facilitate creative action.
Operates in a transformational
Table 3.2 Comparison of Information Tools and Methodologies
Relational database Database schema populated database query language
Expert systems Rule base rules and fact base feference engine
Programming Object-oriented or boolean logic program code and Input data computer
Spreadsheets Cells with formulae formulae and values expression evaluator
Dynamic models Mathematical relationships equations and starting parameters graphing calculator or computer
cases from pre-defined objects and object classes. Optimization routines in
spreadsheets can replicate dynamic relationships but offer single solutions.
Nonlinear models emerging from complexity research appear to be prime
candidates for modeling dynamic behavior under changing conditions.
To understand the policy information flows, the research will search for
significant enrollment patterns, directions and magnitudes of enrollment
change, and information interconnections. The following research questions
frame an exploration of higher education policy under complexity and
Can a few simple rules model higher education policy dynamics?
What is the relationship between the amount of information present in
the policy system and enrollment stability?
How sensitive is policy to slight changes in initial parameters?
What is the relationship between energy flows and enrollment
Is there a simple principle behind pattern-forming policy phenomena?
Complexity theory proposes a set of assumptions that explain
behavioral dynamics. Table 3.3 lists the key assumptions and a parallel set
of propositions that will guide the admission standards policy research study.
The first proposition builds on the assumption that information is energy.
Data may record a particular point in time, but information not only conveys,
but creates movement. Policy systems create movement. By extension,
policy systems are information or energy flows within a particular policy
environment. This proposition defines information, and consequently, policy
as a visible set of flows or transformations. The second bullet assumes that
information flows perturb or excite the environment. The third proposition
acknowledges the intertwined system of policies of a complex system. The
fourth proposition addresses the sensitivity of complex systems to small
changes. Because this makes systems unpredictable, chaotic behavior may
disguise goal-seeking behavior.
Table 3.3 Organizing Principles for Studying Policy Complexity
1. Information is energy.
2. Information flows perturb the
3. Complexity is embedded in
the numerous ways that the
components of a system
4. Complex systems exhibit
unexpected behavior from
small changes in initial
5. Complex systems exhibit
unpredictable behavior within
patterns of regularity.
Policy systems are information
Information flows will introduce
perturbations into the higher
Higher education system behavior
emerges from the local behavior
occurring at the various
The higher education system may
exhibit unexpected enrollment
fluctuations from small policy
The higher education system may
exhibit unpredictable behavior
within patterns of regularity.
The purpose of this study is to create a policy model that captures all
the salient features of policy dynamics in the simplest and most compact
form. The modeling process will focus on admission and enrollment
dynamics by examining the relationships between student behavior patterns,
institutional behavioral patterns, and enrollment changes in Colorados public
higher education system. The unit of analysis is Colorados Admission
Standards Policy as implemented by the Colorado Commission on Higher
The research approach used to examine policy behavior is similar to
the one complexity scientists are using to study natural and physical complex
systems studying a systems structure, investigating its energy flows,
modeling the system, and observing the systems dynamics under different
For the policy scientist, modeling implies fitting a mathematical pattern
to a body of empirical data to represent human behavior. As diagramed in
Figure 3.1, the study proposes to: observe the applicant admission and
student enrollment patterns , identify the key variables , trace the energy
flows through the system , represent the interdependencies between
different policy spaces , calculate values for constants and functions ,
and test the models validity and robustness by simulating different policy
scenarios 2. In short, the research plan involves three specific tasks:
creating the model, testing the model, and comparing the model output to the
attributes of a complex system.
2Brackets refers to diagram numbers on Figure 3.1.
" Identify key ;
variables and 1 -<
closing loops j ____
( Compare 1
against theory j +
Figure 3.1 Methodology for Modeling the Admission Standards Policy
To account for a systems dynamics, scientists developed modeling
tools that incorporated feedback structures. One of these tools is STELLA, a
system dynamics simulation application. This research study used this
software package to model the dynamics created by a higher educations
admission standards policy.
The softwares language is built around structure and process. Stocks
(representing the conditions of the system) and flows (representing the
changes) are its fundamental building elements. Infrastructures are built up
from combinations of stocks and flows. Feedback loops link stocks to
constraints. Appendix A provides a glossary of the modeling terms used in
the STELLA Research software.
One way to find out how a system behaves is to simulate each step in
its evolution explicitly (Wolfram, 1988). The modeling process followed
Wolstenholmes (1994) step-wise approach for creating a system dynamics
model (Table 3.4). The beginning stages required perceptual skills for
envisioning a mental model and visualizing relationships in abstract terms.
Step 1 begins by examining the basic relationships between different
components, especially in terms of the direction of change. Positive polarity
means that an element changes in the same direction as its associated
element. Negative polarity means that when an element (e.g., retention rate),
its associated element (e.g., institutional physical capacity) moves in the
opposite direction. Step 4 involves analyzing the critical relationships to
select aggregate data values for initial parameters. While steps 1 through 7
describe the basic modeling process for a single subsystem, step 8 captures
the dynamics occurring within the system and between the system and its
environment. It is the critical design activity of the modeling process and
captures one of the guiding propositions of this study complex policy
Table 3.4 Summary of the Stepwise Approach to System Dynamics
Model Creation and Development.
Source: E. F. Wolstenholme (1994).
1. If possible, identify a reference mode of behavior for the system
under study and sketch the skeleton feedback loops responsible
for this mode. Try to identify the polarity of such loops.
2. Identify the key variables associated with the skeleton diagram.
3. Identify some of the initial system resources associated with the
4. Identify some of the initial states (levels) of each resource.
These initial states should be defined at a reasonably high level
5. Construct resource flows for each resource, containing the
identified states and their associated rates of conversion.
Include any significant process delays. A resource flow must
contain at least one resource state and one rate.
6. If more than one state of each resource is involved, cascade the
resource flows together to produce a chain of resource
conversions, alternating the levels and the rates.
7. Within each resource flow identify organizational boundaries,
behavioral/information flows, and strategies by which the levels
influence the rates. Indicate any significant delays in the
8. Identify similar organizational boundaries, behavioral/information
flows, and strategies between resource groups. This linking of
resource flows should recreate any feedback loops identified in
step 1. 9
9. Reiterate as necessary for each feedback loop.
behavior occurs due to the interaction among various policies and policy
Population size and participation rate are the external factors that
shape the aggregate demand for higher education (Zemsky and Oedel,
1983). The decision processes of the policy system iterate between the
student and the admitting institution. The admission standards dynamics
occur as students apply, institutions select, and students enroll. These
processes operate in three different regions. The first region contains all
possible high school graduates who wish to enter Colorado colleges. The
second region contains the students colleges admitted into the system. In
the third region, enrolled students progress until they complete a degree or
otherwise exit. A skeleton model of the relationships between these three
regions (Figure 3.2) indicated that college applicants, accepted applicants,
and enrolled students provided only a partial picture of the admission
standards policy system. While they illustrated the energy flowing into the
system, the system was also losing energy as institutions denied admission
or students declined offers, exited before graduating, or completed a degree
program. The open loops (i.e., the arrows) needed greater definition.
The points where students aggregate in different policy spaces
became the stocks of the model. The present CCHE database captures
term data on aggregate numbers of applicants, admitted applicants, and
enrolled students. As students move from one policy space to another, they
create energy flows. To indicate the movement, the researcher characterized
each energy flow with qualitative verbs, such as increasing, exiting, or
refusing. These represent the periods of change between the stocks.
Converters capture the rates of change. Closed loops depicted the initial
model assumptions about how each policy element related to the other
Figure 3.2 Admission Standards Policy System Diagram
elements. STELLAS loop tracing function allows the modeler to check the
The next step created the models information structure calculating
the values and functions. The initial parameter values were extracted from
the database by generating SPSS crosstab reports.
This study relies on CCHEs cohort database as its primary data
source. It contains ten years of student enrollment and college applicant
data, which equates to approximately 1.3 million student records. Table 3.5
clusters the significant variables into several data categories.
CCHE FTE reports and Colorado Department of Educations high
school graduation data supplemented the primary data source.
Table 3.5 Variables Used in the Research Study
data category variables model values
student high school GPA probability of college
preparation high school rank ACT test scores admission
student choice submitted applications percent of multiple
college enrollment applications
student enrollment headcount
performance college success attrition rates
college enrollment time-to-degree
number of college graduates graduation rates
institutional acceptance acceptance rates
choice enrollment yield
environmental number of high school graduates [initial value]
policy enrollment capacity boundary condition; number of students an institution can physically serve.
Validity. Reliability, and Limitations
The research design stipulates that the policy model is not
predetermined, but that it will emerge through experimentation. The
exploratory nature of the Investigation indicates that a valid approach to
modeling policy dynamics searches for the model in the information rather
than using an existing set of equations.
Face validity is a test of the models adequacy for the specified
research purpose, it applies to the selection of the modeling tool and the
mental model of the admission policy environment.
The primary criterion for testing the adequacy of the model considers
how the model approach relates to the research purpose. If the research
includes generating new and unexpected contingencies, then a loose
method might be more valid than a precise one (Deutsch, 1963: 81). The
systems dynamic model offers a loose method for generating future
scenarios. It allows the researcher to model policy situations that do not
currently exist and identify information flows and feedback loops(Forrester,
According to Lane (1994), the system dynamical simulation tools offer
advantages that appear to meet the research agenda of this study. The
Explicitly captures ideas, qualitative relationships, and system
Allows risk-free experimentation of policy alternatives.
Indicates the areas that will most affect the system performance when
coupled with sensitivity analysis.
Fosters inclusive participatory policy making by representing
Another test for face validity assesses the compatibility of the
modeling tool with the underlying assumptions of the research study. This
study adopts the following assumptions:
A perception that the higher education environment is a set of
interconnected policies and practices.
Acceptance of the definition of information as form and consequently
as an energy phenomenon.
Extension of this definition into the policy realm by proposing that all
policy information components, traditionally viewed as abstract
elements or events, can be understood as energy flows.
Recognition that a policy change in the system will affect the systems
performance over time in varied and complex ways.
Acknowledgment that change will have unintended as well as intended
The system diagram of Chapter 4 demonstrates the face validity of this
modeling approach. Incorporating the stated model assumptions, it includes
six interconnected policies and practices, defines policy relationships in
terms of change, and provides a simulation platform to identify shifts in
policy directions rather than project the present trend.
To keep the model simple, the system boundaries include only the
entities directly affected by the admission standards policy, i.e., Colorados
public colleges and universities. Even though it represents abstract
relationships, the system diagram (Figure 3.2) emphasized the movement
crossing policy boundaries. This indicated that the methodology need to
incorporate a means to represent this phenomena. I
The theoretical constructs provide a framework for testing the policy
models validity (Senge and Steman, 1994: 200). It involves testing if the
concepts are clear, significant, and correspond with empirical data, the
relational terms are clear and qualified with adequate values, and the
descriptions are accurate and adequate. Secondly, the concepts, relational
terms, and descriptions of a valid model must be linked to the theory
(Meehan, 1994: 114).
The testing for construct validity occurred as the model was
developed. Two concepts policy and nonlinear system are key to
this research. Policy is regarded as the knowledge, combined with
appropriate means, to transform behavior, energy, or information from less
desirable to more desirable forms." The concept of a nonlinear system is that
the relationship between the relevant institutional actors and other
exogenous forces reveals a spectrum of behavior from seeming stability to
apparent instability (Kiel and Elliott, 1995:154). The operational definition of
a nonlinear, dynamical system includes:
(1) Relationships between internal agents and external forces.
(2) Exhibiting far-from-equilibrium behavior.
(3) Characterized by energy exchanges and transformations.
(4) Self-organizing (Casti, 1995).
Self-organizing means that an entity strives to capture the available
resources by competing against other systems. The search for greater mass
causes the systems components to reorganize and change. In the context of
this research study, four-year institutions compete for qualified freshmen
applicants. The institutional enrollment levels change as students are
attracted to one institution over the another. Likewise, the institutions
themselves change with the ebb and flow of their student bodies.
The existence of different patterns in the simulation output is another
evidence of the models validity. Complexity theory differentiates between
equilibrium, near-equilibrium, far-from-equilibrium, and chaos. Evidence of
one or more of these behavioral patterns in the policy experimentation
supports the models validity. Figures 3.3, 3.4, and 3.5 depict near
equilibrium, far-from-equilibrium, and chaotic states. Conceptually, policy
becomes the rules of motion and the x axis represents time. The unshaded
shape represents the policy system in the original phase state while the
shaded shape shows it as it evolves. The wavy lines at the top and bottom of
the box represent the perturbations caused by internal and external system
pressures, e.g., institutional capacity and demand for admission.
The equilibrium state is missing from this set of diagrams because its
form is self-evident. If a policys purpose is to change behavior, equilibrium
is a policy anomaly. The total freshmen enrollment would equal the total
number of applicants. The enrollment would remain constant at the various
institutions. There would be no need for policy.
Figure 3.3 illustrates a policy system in a
near-equilibrium state. The external forces
(e.g., applicant and legislative demands) are
relatively minimal. As the perturbations die on
the right side, the system returns to a stable
state. The relatively small change in system
shape represents the relatively small amount of
new information introduced into the near
equilibrium system. The admission standards policy environment may be in
Figure 3.3 Policy Near-