Citation
How are migrants selected on health?

Material Information

Title:
How are migrants selected on health? evidence from a Mexico-Colorado labor stream
Creator:
Díaz Pérez, María de Jesús
Publication Date:
Language:
English
Physical Description:
xiv, 242 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Migrant agricultural laborers -- Health and hygiene -- Colorado ( lcsh )
Health status indicators -- Colorado ( lcsh )
Mexicans -- Health and hygiene -- Colorado ( lcsh )
Migrant agricultural laborers -- Selection and appointment -- Colorado ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 236-242).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by María de Jesús Díaz-Pérez.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
761013717 ( OCLC )
ocn761013717
Classification:
LD1193.L566 2011d D59 ( lcc )

Full Text
HOW ARE MIGRANTS SELECTED ON HEALTH?
EVIDENCE FROM A MEXICO COLORADO LABOR STREAM
by
Maria de Jesus Diaz Perez
B.A., Institute Tecnologico y de Estudios Superiores de Occidente, 1993
M.A., Universidad Nacional Autonoma de Mexico, 2003
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences
2011


2011 by Maria de jesus Diaz-Perez
All rights reserved.


This thesis for the Doctor of Philosophy
degree by
Maria de Jesus Diaz-Perez
has been approved
by
Richard Miech
1(V2ot)
Date


Diaz-Perez, Man'a de Jesus (Ph.D., Health and Behavioral Science)
How Are Migrants Selected on Health? Evidence from a Mexico-Colorado Labor Stream
Thesis directed by Assistant Professor Paula Fomby
ABSTRACT
The purposes of this project were first to determine if selection by positive
health was present in a bounded male-only labor migration stream from Mexico to
Colorado; and second, to test two plausible mechanisms for migrant selection: network-
selection or self-selection. This was a mixed-methods study. The quantitative
component was a cross-sectional survey of 384 adult males in Mexico and Colorado to
evaluate health status and behaviors; characteristics of the migrant network; and
migration decision-making factors. The qualitative component used 22 semi-structured
interviews of men with migratory experience in Mexico and Colorado and the recruiter
of the labor program to understand the role of health status and access to migrant
networks on the migration and recruitment decision-making processes.
Findings support partially the existence of selection by positive health in this
migration stream: Having lower depressive symptomatology and better relative self-
rated health increased the probability of being a migrant. Having normal blood pressure
or body mass index, and healthy diet behaviors decreased the likelihood of being a
migrant. The results of quantitative models did not provide evidence suggesting that
the network had a role in selecting men with positive health status. However, the
presence of a network predicted significantly being part of the migrant group. Overall,
quantitative findings emphasized self-selection by subjective health, but qualitative
findings indicated that social networks also function to select out potential migrants
based on a specific construction of being fit for the job that includes indicators of well-


being not captured in the quantitative component of the study. In particular, in
qualitative interviews men acknowledged the importance of personality and mental
health characteristics necessary for coping with the demands of life as a migrant
laborer. Qualitative information also provides important insight as to how and when
the selection process takes place.
These findings suggest that the positive migrant selection for health on some
indicators that has been observed in the literature might be the result of using
inadequate comparison groups. Qualitative findings provide evidence of the need for
careful selection of health indicators.
This abstract accurately represents the contents of the candidate's thesis. I recommend
its publication.


DEDICATION
I dedicate this dissertation with deepest gratitude and love-
To my parents, Victory Carmen, for their wisdom and unconditional love
To Dan, for providing a needed peace haven along this journey


ACKNOWLEDGEMENTS
I would like to express my infinite gratitude to Dr. Paula Fomby, my thesis
advisor, for her guidance, patience, and constant support that lead to the completion of
this project. I am grateful to all Committee members: Dr. Sheana Bull for supporting this
research and inviting me to apply for a LUCHAR fellowship; Dr. Richard Miech for his
encouragement to adopt a binational project, it has been rewarding; Dr. Jean Scandlyn
for her guidance on the qualitative piece of this project; and to all members for their
substantive contributions to this work.
This project would not have been possible without the help from many people in
Mexico. From the Ministry of Health in Guanajuato, I would like to thank Dr. Armando
Perez, Dra. Silvia Quintana, Dr. Samuel Rodriguez, Dra. Araceli Cortez, and Dra. Maria de
Jesus Gallardo. I would like to recognize the work of my team of interviewers: Cristina
Salmeron, Monica Cruz, Ana Isabel Garcia, and Juan Carlos Rodriguez. A very sincere
thanks to the family Castorena Martinez, my godmother Rosita, her husband
Constantino, Tino, Haydee and Luis Fernando for their hospitality during all my trips to
Mexico and their friendship. Muchas gracias a todas y todos.
To my cohort, thanks for your friendship and encouragement; you have been
essential to my education. Specially, I would like to thank Sharon Devine for her expert
guidance on a variety of issues, including Human Subjects.
I thank my team of interviewers in Colorado: Miriam Ceja, Elva Medina, Tanya
Navarro, Jessica Perez del Olmo, and Charlene Shelton for their consistent work and
commitment even in difficult work conditions.
I want to recognize my employer Salud Family Health Centers, whose solid
reputation in the local community made this project possible. A special thanks to the
Migrant Health Program team, lead by Clara Cabanis for their instrumental help. My
deepest gratitude to Tillman Farley, Saluds Medical Services Director: for believing in
my contribution to Salud, for his support through the years and for inspiring my work
with his unfaltering effort to improve access and the quality of health care for Mexican
immigrants.
This project was possible with financial support from a National Science
Foundation Doctoral Dissertation Improvement Grant (SES-0902484J; a University of
Colorado Scientific Team Grant awarded to my Committee members under the direction
of Dr. Richard Miech; a fellowship stipend from the LUCHAR project (Latinos Using
Cardio Health Action to Reduce Risk, NHLBI grant HL079208J; and support from the
Department of Health and Behavioral Sciences.


TABLE OF CONTENTS
FIGURES.....................;................................................XI
TABLES.......................................................................XIII
CHAPTER
1. INTRODUCTION................................................................1
Latino health paradox and potential explanations............................1
Healthy migrant effect......................................................4
2. HEALTH SELECTION OF IMMIGRANTS:
EVIDENCE AND POTENTIAL EXPLANATIONS........................................11
The healthy migrant effect: brief review of the evidence...................11
Health status of Mexican population: major characteristics.................20
Understanding health selection of immigrants: contributions from migration
research...................................................................25
Using rational-choice theory to explain migrant selection on health......27
Using social capital theory to explain migrant selection on health.......29
A framework to study migrant health selection..............................33
3. RESEARCH DESIGN AND METHODS................................................39
Quantitative component.....................................................42
Sample and participants..................................................43
Procedure................................................................45
Definition of variables..................................................47
Instruments..............................................................49
Data analyses............................................................58
Qualitative component......................................................64
Sample and participants..................................................65
Instruments..............................................................66
Procedure................................................................66
Data analyses............................................................68
viii


4. QUANTITATIVE RESULTS
69
Descriptive results.......................................................69
Sociodemographic characteristics........................................69
Health status indicators................................................70
Migratory history and motivations to migrate............................79
Migrant and social networks.............................................82
Testing the presence of a healthy migrant effect..........................85
Dependent variable: migrants vs. non-migrants...........................86
Dependent variable: migrants vs. return-migrants........................95
Testing self-selection and network-selection hypotheses..................106
Dependent variable: migrants vs. non-migrants..........................108
Dependent variable: migrants vs. return-migrants.......................118
5. QUALITATIVE RESULTS......................................................133
Context of decision-making: motivations to migrate.......................135
Role of health on migration decision-making..............................137
Connection to the program................................................140
Who to sponsor? Characteristics of men who are good for this program.....142
The contractor...........................................................143
Summary..................................................................146
6. DISCUSSION...............................................................151
Migrant men have worse objective health than men who did not come to the US....152
Migrant men have better subjective health than men who did not come to the US..154
How are migrants selected on health? a review of the evidence............157
Limitations..............................................................162
Contributions............................................................165
APPENDIX
A. SELECTED STUDIES PROVIDING EVIDENCE OF A HEALTHY MIGRANT EFFECT
FOR LATINO OR MEXICAN IMMIGRANTS.........................................167
B. HUMAN SUBJECTS APPROVALS AND INFORMED CONSENT DOCUMENTS IN
SPANISH AND ENGLISH......................................................176
IX


C. SURVEY QUESTIONNAIRE USED IN MEXICO AND COLORADO IN SPANISH AND
ENGLISH TRANSLATION...........................................199
D. INTERVIEW GUIDES FOR MIGRANT/RETURN MIGRANTS AND KEY INFORMANTS
IN SPANISH AND ENGLISH TRANSLATIONS.......................232
BIBLIOGRAPHY
236
x


LIST OF FIGURES
FIGURES
2.1 MAP OF MEXICO WITH REGIONALIZATION USED IN EPIDEMIOLOGICAL
STUDIES..........................................................21
2.2 MIGRANT HEALTH SELECTION ACCORDING TO SELF-SELECTION HYPOTHESIS.... 35
2.3 MIGRANT HEALTH SELECTION MIGRANT HEALTH SELECTION ACCORDING TO THE
NETWORK-SELECTION HYPOTHESIS.....................................37
4.1 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. NON-MIGRANT) PREDICTED
BY SINGLE POSITIVE HEALTH STATUS INDICATORS......................92
4.2 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. NON-MIGRANT) BY SES AND
RED MEAT CONSUMPTION.............................................93
4.3 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. NON-MIGRANT) BY BODY
MASS INDEX AND AGE...............................................94
4.4 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT)
PREDICTED BY SINGLE OBJECTIVE HEALTH STATUS INDICATORS..........102
4.5 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT)
PREDICTED BY SINGLE SUBJECTIVE HEALTH STATUS INDICATORS.........103
4.6 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT) BY BODY
MASS INDEX AND SOCIOECONOMIC STATUS.............................104
4.7 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. NON-MIGRANT) PREDICTED
BY POSITIVE SUBJECTIVE AND OBJECTIVE HEALTH STATUS..............115
4.8 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. NON-MIGRANT) BY NUMBER
OF FAMILY MEMBERS IN THE UNITED STATES AND SOCIOECONOMIC STATUS.116
4.9 STANDARDIZED COEFFICIENTS OF SUJBECTIVE AND OBJECTIVE HEALTH STATUS
INDICATORS IN LOGISTIC REGRESSION MODELS TO PREDCIT BEING A MIGRANT (VS.
NON-MIGRANT)....................................................117
4.10 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT)
PREDICTED BY POSITIVE SUBJECTIVE HEALTH STATUS AND OBJECTIVE HEALTH
STATUS..........................................................126
xi


4.11 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT) BY
POSITIVE OBJECTIVE HEALTH STATUS AND SOCIOECONOMIC STATUS....127
4.12 ADJUSTED PROBABILITY OF BEING A MIGRANT (VS. RETURN-MIGRANT) BY
POSITIVE SUBJECTIVE HEALTH STATUS AND NON-FAMILY RELATONSHPS WORKING
IN UNITED STATES.............................................128
4.13 STANDARDIZED COEFFICIENTS OF SUBJECTIVE AND OBJECTIVE HEALTH STATUS
INDICATORS IN LOGISTIC REGRESSION MODELS TO PREDICT BEING A MIGRANT (VS.
RETURN-MIGRANT)..............................................130
6.1 HOW ARE MIGRANT SELECTED ON HEALTH? A PROPOSED FRAMEWORK..158
Xll


LIST OF TABLES
TABLES
3.1 Overview of the study......................................................42
3.2 Municipalitys male population and sample distribution by age groups.......44
3.3 Definitions of subjective health status indicators.........................51
3.4 Definitions of objective health status indicators..........................54
3.5 Description of model building process for aim 2............................60
3.6 Description of model building process for aims 3 and 4.....................63
4.1 Sociodemographic characteristics...........................................71
4.2 Subjective health indicators...............................................72
4.3 Presence of chronic conditions.............................................73
4.4 Objective health indicators................................................75
4.5 Migratory history and motivations for labor migration......................80
4.6 Migrant and family networks................................................84
4.7 Demographics, subjective and objective health status indicators: migrants vs. non-
migrants...................................................................87
4.8 Logistic regression models predicting being a migrant vs. non-migrant......88
4.9 Demographics, subjective and objective health status indicators: migrants vs.
return-migrants...........................................................97
4.10 Logistic regression models predicting being a migrant vs. return-migrant..98
4.11 Logistic regression models with aggregated health indicators predicting being a
migrant vs. non-migrant..................................................Ill
xiii


4.12 Comparison of fit measures of models predicting being a migrant
vs. non-migrant.........................................................118
4.13 Logistic regression models with aggregated health indicators predicting being a
migrant vs. return-migrant..............................................121
4.14 Comparison of fit measures of models predicting being migrant
vs. return-migrant......................................................131
5.1 Description of interview participants...................................134
xiv


CHAPTER 1
INTRODUCTION
Latino Health Paradox and Potential Explanations
The Latino health paradox is a term used to describe the common observation
that by many measures Latino health is superior to what one might expect given their
socioeconomic status. In particular, Latino health levels are better than those of African
Americans, although they share similar economic positions, and are often above those of
non-Latino whites, whose economic resources are far superior. With some exceptions,
Latino age-adjusted death rates are actually lower for most diseases than those of non-
Latino whites [Vega, Rodriguez, Gruskin, 2009; Markides & Eschbach, 2005; Hummer,
Biegler, De Turk et al., 1999; Hummer, Rogers, Amir, Frisbie, 2000; Palloni & Morenoff,
2001; Abraido-Lanza, Dohrenwend, Ng-Mak, & Turner, 1999; jasso, Massey,
Rosenzweig, & Smith, 2004).
Understanding the mechanisms underlying the Latino health paradox has
motivated substantial research and debate. Some researchers have suggested that the
Latino paradox is the result of poor measurement or inconsistencies on assigning ethnic
categories [e.g. Hunt, Resendez, Williams et al., 2003; Patel, Eschbach, Ray & Markides,
2004; Smith & Bradshaw, 2006) or that is a result of the "salmon bias effect, which
proposes that many Latinos return to their place of origin to die or when they become ill
(Franzini, Ribble & Keddie, 2001; Palloni & Arias, 2004). Overall, two plausible
explanations that might be complementary have dominated the Latino health paradox
literature [Abraido-Lanza, Armbrister, Florez, & Aguirre, 2006; Markides & Coreil,
1986). One concerns the protective effects of culture and norms within Latino families
and communities. This explanation proposes that Latino populations have a cultural
buffering effect characterized by close social and family ties that reinforce norms
proscribing risky health behaviors and promoting good ones [Vega & Amaro, 1994;
1


Scribner, 1996; Abraido-Lanza et al., 2006). The second topic is related to the healthy
migrant effect, where Latino migrants are seen as inherently healthier. This proposition
is not explicit on whether the better health of immigrants "reflects their generally
superior health habits, behaviors, and conditions in the sending countries relative to the
US or whether it is principally due to the health selectivity among migrants compared to
those who stayed (Jasso et al, 2004, p. 238-239). Despite the debate about how well
these propositions can explain or not the Latino health paradox, many authors agree on
the existence of a health selection of immigrants.
The health selection of immigrants has been documented not only in the United
States but also in other important immigrant receiving countries, like Australia
(Australian Institute of Health and Welfare, 2000), Canada (e.g. Gushulak, 2007), and
Germany (e.g. Razum, Zeeb, & Rohrmann, 2000); and for immigrants from different
sending countries. In the United States, the healthy migrant effect has been documented
for a variety of health indicators, e.g. adult and infant mortality (e.g. Markides &
Eschbach, 2005; Hummer, Power, Pullum et al., 2007); adult morbidity (Singh &
Siahpush, 2002); health status (Cho, Frisbie & Rogers, 2004); birth outcomes
(Rosenberg, Raggio, & Chiasson, 2005); mental health disorders (Alegria et al., 2007,
2008); and specific health behaviors like smoking and drinking (Singh & Siahpush,
2002; Lopez-Gonzalez, Aravena, & Hummer, 2005). The breadth of evidence regarding
immigrants health selection makes this topic theoretically appealing since it defies a
hypothesis central to social science health research, that socioeconomic status predicts
health outcomes, given that immigrant populations tend to have lower socioeconomic
status than other groups in receiving societies.
Latinos are the largest minority group in the United States. Although, this
category includes population with origins from at least 15 different countries in Latin
America and the Caribbean region, Mexicans represent 65.5% of this group (Pew
Hispanic Center, 2009). From 1996 to 2006, the Mexican-origin population in this
2


country increased from 18.7 to 28.3 million, and over 40% of this increase was due to
immigration. Mexican population is one of the five largest immigrant groups in almost
every state in the United States. By 2006, fourteen states were identified with
populations of Mexican immigrants over 100,000. One of these states is Colorado with
an estimated number of 268,000 Mexican immigrants (SSA & UC, 2007; Pew Hispanic
Center, 2008). A slower growth in immigration has been observed in recent years
probably due to the economic recession in this country. However, in states like North
and South Dakota, New Hampshire, West Virginia, and Maine, the Mexican population
has increased over a 100 times from 2000 to 2007 (Leite, Angoa & Rodriguez, 2009).
Compared to other groups in the US, Mexican immigrants are poor. Over one-
quarter of recent Mexican immigrants adults [27.2%) live in families with annual
incomes below the federal poverty level compared to long-stay MX immigrants [20.7%),
US-born MX Americans [14.8%) and US-born non-Latino whites, who have the lowest
level of poverty (7.9%). Among Mexican immigrant children, the poverty rate reaches
43% (Pew Hispanic Center, 2009). They have lower education: 42% of Mexican
immigrants have less than nine years of education; and hold occupations with the
highest rates of injury and death (Pew Hispanic Center, 2009; SSA & UC, 2007). Despite
their low socioeconomic status, it has been documented that Mexican immigrants,
compared to US-born Mexicans, and even with other groups in the US, have more
favorable health outcomes. For example, data from the 2000 National Health Interview
Survey shows that only 6.8% of recent Mexican immigrant adults characterized their
health as fair or poor compared to 11.5% of long-term Mexican immigrants, 10.8% of
US-born Mexican Americans, and 9.2% of US-born non-Latino whites (CONAPO, 2005).
This unexpected piece of evidence is thought to be one of the components of what has
been widely described as the "Latino health paradox.
Identifying the determinants of the good health of those who immigrate to the
United States as well as the forces that shape their health following immigration is of
3


critical importance to understanding ethnic health differences in this country (Jasso et
al., 2004; Donato, Wakabayashi & Kanaiaupuni, 2006). Immigration is and will be one
of the driving forces in accounting for the growth of the American population. Recent
estimates indicate that the American population will increase by 120 million people
over the next 50 years and 66% of this increase will be the consequence of immigration.
Mexican immigration to the US will continue and the health status of Mexican
immigrants and their descendants will play an increasingly central role in shaping
health outcomes of the American people (Jasso et al., 2004).
Several studies have documented the concerning decline in health status as
immigrants spend more time in the United States (Singh & Siahpush, 2002; Cho et al.,
2004; Lopez-Gonzalez et al., 2005). By studying the forces responsible for the better
health of recent immigrants we might be able to inform interventions that promote
these same forces and extend their "health protective effects while immigrants
integrate in their new society.
Healthy Migrant Effect
Mexican immigrants who are new to the U.S. show remarkably high levels of
health, a phenomenon known as the "healthy migrant effect." Across a number of
outcomes their health is (a) better than African Americans, (b) more favorable than
their US born Mexican counterparts, who usually have a better socioeconomic position,
and (c) comparable to White populations (Lopez-Gonzalez et al., 2005; Cho et al., 2004;
Palloni & Morenoff, 2001). What makes their positive health status notable is that it
belies the expectation that they would have poor health (Lynch & Kaplan, 2000) based
on the fact that as a group immigrants have low education, earn a low income, and one
out of four lives below the poverty line (Mexican Secretariat of Health (SSA) &
University of California (UC), 2007; Pew Hispanic Center, 2009). Specification of the
underlying reasons for this immigrant health advantage is important not only for
4


immigrants but, more generally, to identify the resources people draw on to maintain
good health in the face of adversity.
Researchers have hypothesized that the healthy migrant effect is the result of a
selection process, through which those who are better off tend to move; or the result of
the more healthful behaviors that are the norm in the countries of origin of immigrants
(Markides & Coreil, 1986; Jasso et al., 2004; Gushulak, 2007). However, most of the
evidence available about the healthy migrant effect is descriptive: no elaboration has
been offered as to how this selection process takes place.
The proposition that immigrants are selected by health coincides with the
widely accepted observation in migration research that migration is not a random
event, and that not everyone in source countries has an equal chance to migrate (Liebig
& Sousa-Poza, 2004; Chiswich, 1999; Massey, Arango, Hugo et al., 1998). However, who
is selected for migration and how could be explained in different ways depending on the
theoretical perspective adopted. I propose that current explanations of how migrants
are selected on a variety of characteristics could be used to understand how migrants
are selected on health characteristics.
For example, the Rational-Choice Theory describes that potential immigrants
make a decision to migrate after a cost-benefit calculation leads them to expect a
positive return, usually monetary, from the migratory movement. The likelihood of
migration would be increased by individual human capital characteristics like health.
Following this model, a potential labor migrant would need a positive personal
evaluation of her/his own health status, including physical health, to be able to increase
potential earnings through uninterrupted work and emotional/psychological strength
to withstand separation from family members and the demands of adaptation. Because
a prospective migrant would weigh several elements to make the decision to migrate,
5


the health status of migrants, other things equal, is expected to be better than that of
non-migrants.
Another description of how migrants are selected comes from the Social Capital
Theory of migration. The leading prediction of this perspective is that people who are
related to migrant networks have access to social capital that significantly increases the
likelihood that they will migrate. It is possible that the health selection of migrants takes
place because the social process responsible for "selecting" new migrants takes care of
their "health selection". In the case of migrant male labor selection, new immigrants are
sponsored by the migrant network when they are believed to be strong, able to work,
and resilient. From the Social Capital perspective, a migration move might include an
evaluation of individual attributes and costs; however, in this case migration would only
take place in the presence of a migrant network. A positive self-evaluation of health
status might be present but it would be the networks validation of "good health" that
determines if an individual can be sponsored for international migration.
These two perspectives might be complementary. A migrant could have a
positive evaluation of his own health, and consider his health status a necessary asset to
migrate successfully; however, this would not be enough to carry out the migration
move unless he is part of a network that makes the trip possible. Whether migration
can be better predicted by a personal attribute, like health, or a persons access to a
migrant network, the migration decision process ends with the move of an individual.
The prospective migrant probably needs to evaluate several factors before making a
determination of looking for labor abroad or staying in the community of origin.
Therefore to study the role of health status and/or access to migrant networks on
predicting a migration move, we need to take in consideration the context in which the
migration decision making takes place and the different factors involved in this process
(De Jong, 2000; Massey, Arango, Hugo et al., 1998).
6


The purposes of this project were first to determine if the healthy migrant effect
took place in a bounded male-only labor migration stream from Mexico to Colorado,
whether measuring health status or health behaviors; and second to test two plausible
ways in which migrant selection by health might take place. Following the two
perspectives mentioned above, I proposed that the migrant selection by health could
operate through a "network-selection" or a "self-selection process.1
This was a nested mixed-methods study driven by a deductive theoretical
approach. The quantitative component was a cross-sectional survey in Mexico and
Colorado to evaluate health status and behaviors; characteristics of the migrant
network; and motivations to migrate (e.g. personal, economic, social, and cultural) of
men in Colorado and Mexico. The qualitative component used semi-structured
interviews with men with migratory experience in Mexico and Colorado and the
recruiter of the labor program to understand the role of health status and access to
migrant networks on the migration and recruitment decision-making processes. The
specific aims of this project were:
To describe selected health status indicators; migrant network characteristics; and
motivations to migrate for adult males who reside in Mexico (non-migrants; return-
migrants) and in Colorado (migrants).
To determine if being a migrant in Colorado is predicted bv a positive health status.
To determine the extent to which being a migrant in Colorado, sociodemographic
characteristics and other motivations to migrate being equal, is directly determined
by a positive health status.
1 These terms are used to make emphasis on the main proposition of the two theoretical
perspectives guiding the hypotheses of this research project.
7


To determine the extent to which being a migrant in Colorado, sociodemographic
characteristics and other motivations to migrate being equal, is indirectly
determined by a positive health status, via access to a migrant network.
To understand the role of health and migrant networks on migration participation
from the perspective of men with migration experience and those who hire and
recruit migrants.
It was expected that if health selection of migrants followed a self-selection
process, migration would be predicted by positive health status. If the health selection
of migrants followed a network-selection process, migration would be predicted mainly
by presence of a migrant network. These two mechanisms, however, were not expected
to be exclusive and migration behavior could be predicted to some extent in both ways.
The findings of this study support partially the existence of a healthy migrant
effect in this migrant stream from Mexico to Colorado. For the dependent variables
modeled: migrant vs. non migrant and migrant vs. return migrant, positive subjective
health indicators [low depressive symptomatology and better relative self-rated health)
increased the probability of being a migrant; while positive objective health indicators
(normal blood pressure and BM1, and healthier diet behaviors) decreased the
probability of being a migrant. The effect of these health indicators was strong and
remained almost intact when sociodemographic variables and time spent in the US for
the sample of migrants and return migrants were considered in the models.
The results of these models provide evidence favoring a self-selection process
by positive health only for subjective health indicators. Qualitative findings provided
information supporting the importance of subjective health status in agreement with
the quantitative component. Participants acknowledged the importance of personality
features and mental health characteristics as necessary to endure being away from their
8


families and familiar environment; as well as coping with the demands or conditions of
life as a migrant laborer.
Quantitative results do not provide evidence suggesting that the network has a
role selecting men with positive health status. However, presence of a network,
whether family or other relationships, has a significant role on predicting migration.
The essential role of the network making migration possible for men is confirmed by
qualitative findings. Overall, quantitative findings emphasized self-selection by
subjective health, but qualitative findings indicated that social networks also function to
select out potential migrants based on a specific construction of being fit for the job that
includes indicators of well-being not captured in the quantitative component of the
study. This construction includes characteristics like the ability to do the job and the fact
that a man does not drink excessively or that a man is aggressive. Qualitative
information also provides important insight as to how and when the selection process
takes place.
This chapter introduces the problem studied in this research project and
presents an overview of the organization of this document. Chapter 2 provides the
background to the research problem studied in this project. First, it presents a
summary of the evidence available regarding the healthy migrant effect for Latino and
Mexican populations. Second, it presents a brief review of the most important
characteristics of health status in Mexican population. Third, it briefly describes two
theories currently used to explain a migration move and how they can be used to
explain why selection of immigrants by health might take place. Last, it describes the
framework that guides this study of the health selection of migrants.
Chapter 3 describes the research design and methods used to conduct this
study. In this chapter the reader will find the description of the population and sample
targeted in this project. It includes a detailed description of the questionnaire and
9


variables included in the study; as well as the procedures used in Mexico and Colorado
to conduct this research project.
Chapters 4 and 5 describe the results for the quantitative and qualitative
components of this study respectively. Chapter 6 discusses the main findings, and
describes limitations and contributions of this research project.
10


CHAPTER 2
HEALTH SELECTION OF IMMIGRANTS:
EVIDENCE AND POTENTIAL EXPLANATIONS
The Healthy Migrant Effect: Brief Review of the Evidence
The healthy migrant effect (HME) states that individuals who migrate are
healthier than their counterparts who do not migrate (Markides & Coreil, 1986) and this
fact has been documented in several countries and for a variety of health indicators
(Jasso et al., 2004). Mainly motivated by discerning the Latino health paradox, the
number of studies documenting the healthy immigrant effect has grown considerably in
the last ten years. However, it has been only recently have more researchers included in
their studies, beyond ethnicity, variables like nativity, immigrants national-origin and
duration of residence in this country (e.g. Singh & Siahpush, 1999; Cho et al., 2004). The
absence of these variables has been acknowledged as a serious limitation mostly
consequence of the databases available for such studies, mostly collected through
government sponsored national health surveys in the country of reception (e.g. Lopez-
Gonzalez et al., 2005). Early literature providing evidence of a HME focused on
mortality, whether in adult or infant populations (e.g. Markides & Coreil, 1986;
Hummer, Biegler, De Turk et al., 1999). Recently, the study of the HME includes a
variety of health indicators and behaviors. This review will focus on studies that
provide evidence of the HME in Latino and Mexican-origin adult populations and
included health indicators other than mortality (see Appendix A for a description of
these studies).
There are three types of study designs that document the existence of a HME
(Gushulak, 2007). One type compares outcomes in a group of immigrants with those in
a demographically similar component of the host population. Using data from national
health surveys, researchers have compared health indicators among different
11


racial/ethnic subgroups, including variables like nativity and length of residence (e.g.
Singh & Siahpush, 2002; Cho et al., 2004; Lopez-Gonzalez et al., 2005; Alegria et al.,
2008) providing convincing evidence about the superior health of immigrants. This
type of study (see Appendix A) has been the most prevalent on describing the HME
among Latinos or Mexicans in the US. Evidence from these studies shows that
compared to US-born Latinos; foreign-born Latinos have lower rates of smoking,
drinking, chronic diseases and mental disorders. With the exception of a study
conducted by Redstone & Frank (2008) described later in this chapter, studies
considering national origin have found an even larger advantage for Mexican
immigrants compared to their US-born counterparts for outcomes like self-rated health,
activity limitations, days in bed due to illness (Cho et al., 2004), and mental disorders
particularly substance abuse (Alegria et al., 2008).
Crimmins, Kim, Alley et al (2007) conducted the only study that measures
biological risk profiles including blood pressure (systolic and diastolic pressure and
pulse), metabolic (cholesterol and hemoglobin), and inflammation risk factors (C-
reactive protein, fibrinogen, and albumin). In this study, the authors found that US-born
Mexicans showed a higher level of biological risk relative to foreign-born Mexicans, also
in agreement with the health selection hypothesis. In contrast, the few studies that
document physical activity show that Latino immigrants have lower rates of this
positive health behavior compared to US-born Latinos and for this health behavior,
acculturation to this country has a positive effect (Sanghavi, McCarthy, Phillips, & Wee
etal., 2004).
The main limitation of the type of studies mentioned above is that even when
they present a strong suggestion of a HME, as Jasso et al. (2004) point out, the
appropriate comparison group to evaluate the health selection of migrants is the health
of those who remain in the sending countries at the time of immigration. On the
positive side, US health indicators for the US as a whole are better than health indicators
12


in the immigrant's countries of origin, as it is the case of Mexican immigrants. The
assumption is that if immigrants have better health than the rates reported for their
countries of origin then they are not representative of the sending population but are
positively selected to migrate. I would argue, though, that comparisons of immigrants
with those who stay in Mexican communities of origin are still necessary since
international migration is not equally distributed in the Mexican population, nor are
BMI or other health indicators.
The second type of study is longitudinal, where a population of immigrants may
be followed to delineate changes in health outcomes or characteristics over time. There
is only one example of this type of study, the New Immigrant Survey, a panel survey of a
nationally representative sample of new legal immigrants. This survey includes follow-
ups at three, six, and 12 months from the date immigrants received their green card.
Using data from this National Survey, Redstone [2007) analyzed the dietary changes
among Latinos and found that most recent immigrants have healthier dietary habits.
Latinos tend to adopt negative habits like increased consumption of fat or junk food as
they spend more time in the US. Although there is no information available about
specific dietary changes from cross-sectional surveys, the increased rate of obesity
among immigrants, particularly among those groups with longer time of residence in
the US (Sanghavi et al., 2004) would be congruent with the finding reported by
Redstone.
In 2004, Jasso et al. compared the self-reported health as well as the presence of
chronic conditions of immigrants with the health of US-born population. The authors
found that immigrants were quite healthy; they reported lower rates of chronic
conditions even after controlling for age and having visited a doctor in the previous
year. However, they also found that the self-reported health of immigrants improved
over time. This finding does not agree with other studies that have observed a negative
effect of duration of stay in the US on the health of immigrants, which may be result of
13


the short time span considered (one year) or the use of self-reported health as health
status indicator. The authors suggest that the report of better health might be explained
by a perception of better quality of life of these new immigrants. It is likely that one
year after obtaining permanent residence in this country, a better job, healthcare, and
living opportunities arise, which might affect peoples evaluation of their health status.
The study conducted by Jasso et al. (2004, not described in Appendix A) did not analyze
immigrants by ethnic or national origin group therefore we do not know if health would
improve for immigrants of all origins. The finding about health improving with time
spent in the US could be related to the aggregation of immigrants with very dissimilar
origins and time of arrival. Ultimately, this type of study speaks even more indirectly
about the possible health selection of immigrants, since they only measure immigrants
health status compared to the population in the receiving society (in this case US
general population).
Trying to solve some of the problems just mentioned, Redstone & Frank (2008),
more recently conducted a study, using the first wave of the NIS survey, to determine if
health selection among immigrants varied across regions and the extent to which health
selectivity accounted for variation in immigrant's health outcomes. The analysis relied
on self-reported health compared to others in their home country, at time of the first
interview, and their self-rated health at the time the survey took place. Compared to
other immigrants, Mexicans were the least likely to experience positive health selection.
This weak selection could be partially explained by the mechanism that is most common
among Mexican immigrants to obtain permanent residency (family preference). The
authors conclude that Mexican migrants might have less positive health selection than
immigrants to the US from other countries since the costs of migration for people
obtaining their permanent residency through family relationships might be lower than
the costs for people who obtain permanent residency based on employment.
14


One of the limitations of the studies described here conducted with data from
the National Immigrant Survey (Redstone & Frank, 2008; Redstone, 2007; Jasso et al.,
2004) is that they only included immigrants who obtained permanent legal residence,
excluding many unauthorized immigrants. Also, the surveys first measurement does
not necessarily represent the time of first arrival to this country, since obtaining legal
residence might take from one year to two decades. Despite this limitation, the study by
Redstone & Frank (2008) uncovers the importance of socioeconomic profiles on
explaining not only positive health status but also their contribution to explaining
positive or negative changes on health since migration. When they consider
socioeconomic variables like education and occupation of immigrants from different
origins, they found that the weak selection of Mexicans was explained in part by their
lower socioeconomic profile compared to other immigrants who were more educated
and held better occupations. Wealthier immigrants were more likely to be healthier.
The third type of study that compares health status indicators of a population of
immigrants to indicators in their country of origin, although still not very common, has
increased in the last few years. In 1998 (not shown in Appendix A), Vega, Kolody &
Aguilar-Gaxiola et al. compared the life-time prevalence of several mental health
disorders of Mexican immigrants, US-born Mexicans and Mexicans residing in Mexico
City. They did not find evidence for a HME since Mexican immigrants had lifetime rates
of mental disorders similar to those of Mexico City residents. However, the study
provided strong evidence for a negative acculturation effect: immigrants who had lived
in the US for 13 years or more had a higher risk of having a mental health disorder
compared to recent immigrants. Unfortunately, the study conducted by Vega et al.
(1998) used prevalence rates from Mexico City, which might not be the best comparison
group for Mexican immigrants. Although, urban migration from Mexico to the US has
increased, still a great proportion of Mexican immigrants in the US have rural origin
(Riosmena & Massey, 2010). The negative acculturation effect on the mental health of
15


immigrants has been confirmed in a study conducted with a large national database
(Alegria et al., 2008). In this study, the authors found that recent Mexican immigrants
have much better mental health than immigrants who have resided for a longer time in
this country.
Recently, Barquera et al. [2009) compared the results of the Mexican National
Health Survey [2000, 2006) to the NHANES [1999-2000 and 2005-2006). The authors
describe that despite an increasing obesity trend in Mexico, rates for overweight and
obesity [based on BMI) are higher for Mexican Americans compared to Mexicans in
2000 and 2006. For the most part, data described in this study does not show evidence
of positive health selection. For example, in 2000 the rate of overweight for Mexican
males was 41.8%, while the rate of overweight for Mexican immigrant males in the US
was 45.2%. In the case of the obesity rate, there is some suggestion of selection; the
rate of obesity among Mexican males was 16.7% in 2000, while the rate of obesity
among their counterparts in the US was 14.7%. In 2006, the difference between
Mexican and Mexican immigrant rates of obesity decreased.
A recent study used data from NHANES 1999-2004 and the National Health
Survey 2000 to compare patterns of hypertension between Mexican adults in Mexico
and the US [Barquera, Durazo-Arvizu, Luke, Cao & Cooper, 2008). The rates described
by the authors suggest a positive health selection of Mexican immigrants. Sex and age-
adjusted prevalence rates of hypertension were higher in Mexico [33.3%) than among
Mexican immigrants (17.3%) and US-born Mexicans (22.4%).
These two studies provide valuable information; however they do not consider
individual migration history or regional migration rates. Accounting for these variables
might provide a different picture of the hypertension and obesity rates of Mexicans and
might provide more definite evidence regarding migrant health selection.
16


The one study that most precisely tests migrant health selection (Rubalcava,
Teruel, Thomas, & Goldman et al., 2008) used representative longitudinal data from the
Mexican Family Life Survey to determine whether Mexicans who migrated to the US
between 2002 and 2005 were healthier than other Mexicans. Mexicans who migrated in
this time period were interviewed in their places of residence in the US or in Mexico if
they had returned. Health status was measured in the first wave, before migration to
the US took place, so the possibility of measuring the impact of migration on health
status was eliminated. The authors found that even when health significantly predicted
subsequent migration, the associations were weak, and only a few health indicators
(normal BM1 and blood pressure) were statistically significant. In this study, the
authors included place of residence in Mexico as an indicator of migration intensity to
account for the fact that not all communities have equal participation on migration
flows to this country. The fact that this variable had a significant marginal effect might
suggest that what was not measured in these particular places with high migratory
intensity could be of great importance explaining the role of health on predicting
subsequent migration and that place of residence might be an important consideration
when studying the health of immigrants. For this study, only young Mexicans (15-29
years old) were included since they were more likely to migrate.
Despite its limitations, the evidence regarding the existence of a HME motivates
a deeper exploration with a more carefully designed assessment to determine the
veracity of this phenomenon. Comparisons made between Latino or Mexican
immigrants against US-born populations show that immigrants in general report a
health advantage in a range of health outcomes (See Appendix A). Some health
outcomes have been studied more consistently than others. For example, in contrast
with the number of studies addressing BMI, physical activity and diet have seldom been
studied. Physical activity (PA) has been found to be higher among US-born groups and
17


to increase as immigrants spend more time in this country, which is apparently at odds
with the health behaviors hypothesis to explain the HME.
The HME is believed to be a consequence of the characteristics of the regions of
origin of many immigrants, where environmental conditions promote behaviors like
physical activity, which has been associated to the prevention of chronic diseases. It is
also thought that only the youngest and healthiest people from places of origin migrate
since they are the ones who can cope successfully with the physical, psychological and
sociological demands imposed by the process of immigration (Gushulak, 2007). To test
any of these two hypotheses, only comparisons between sending and receiving
communities could be useful (Palloni & Arias, 2004). The only study that evaluates the
role of health status on predicting future migration including those who stay in their
places of origin as well as people who migrate to the US found only weak evidence for a
HME and did not include some variables that could be of great importance to elucidate
the health selection of immigrants (Rubalcava et al., 2008). For example, we dont know
if people who migrated did so for first time or they had years of cumulative experience
in the US, which might have contributed to weaken their health status, according to the
acculturation hypothesis.
Although studies about the HME have increased in late years, most research is
limited since it compares the health of recent immigrants with the health of US-born
people of the same origin or even with general US-born populations; there are still only
a few studies that consider national origin (See Appendix A). This variable is important
since it has been shown before than when the Latino ethnicity is divided by national
origin, different health patterns arise (e.g. Hummer et al., 1999). I would argue that is
also necessary to go beyond national origin and include variables that speak more about
the particular conditions of places of origin as it is suggested in the findings of
Rubalcava et al. (2008). Using self-reported health status, immigrants were not more
likely to have better health than non-immigrants, they would only report to be "as
18


healthy" as others who were their same age. This finding might suggest that the health
reference has been set against a group of peers, probably previous migrants and that in
a personal evaluation of own health status to migrate, an individual might determine
that he/she only needs to be as healthy as those who have departed before. Therefore
the health norm might be different from region A to region B or across different levels of
socioeconomic status.
Since the HME implies that there is a selectivity process on who migrates and
who does not from communities of origin, there is still a great need of more research
that establishes the right comparison group, those who do not migrate and stay in the
communities of origin (Jasso et al., 2004; Gushulak, 2007; Rubalcava et al., 2008). To
understand how migrants might be selected by health, we need to conduct studies
including variables that speak about the context in which migration decisions take
place, as well as the context in which the health status of those potential migrants is
shaped. Furthermore, to study the health selection of immigrants comprehensively, the
ideal design would include measures of health status of people living in Mexico, in the
same communities of origin of those immigrants who live in the US. Ideally the health
status of these two groups would be measured over time and studies would consider
age, life stage, socioeconomic position, legal status, as well as other important
contextual variables (Jasso etal., 2004).
19


Health Status of Mexican Population: Major Characteristics
Mexico is a country in epidemiologic transition. This transition is characterized
by a rapid increase in obesity and chronic diseases with a slow decrease of under
nutrition and infectious diseases; polarization across the country, in which the more
socioeconomically developed areas are experiencing a higher burden of chronic
diseases compared to less developed areas, where there is still under nutrition and
higher rates of infectious diseases; and a process of homogenization, where less
developed areas are getting closer to the chronic disease prevalence in the rest of the
country. For example, in the Southern region (a less socioeconomically developed area)
diabetes mellitus mortality increased 92.3% from 1980 to 2000 compared to 24.5% in
the Northern region (a more socioeconomically developed area) during the same period
(see Figure 2.1). Dramatic increases in mortality for diabetes, hypertension and acute
myocardial infarction have been documented. From 1980 to 1998, increases for these
mortality causes ranged from 53% to 62%. Heredity plays a role in all but the three
causes of death share common risk factors such as obesity, inadequate dietary intakes
and physical activity (Rivera et al., 2002, 2004; Barquera et al., 2009).
Mexico currently faces an obesity epidemic. The epidemic is growing at such a
pace that prevalence statistics become rapidly outdated. When the BMI estimated with
data from a national survey conducted in 2006 was compared with the 2000 Mexican
National Health Survey, the prevalence of excess weight gain (BMI > 25) among adults
in Mexico increased by about 12 percent (Barquera et al., 2009).
20


Regions in Mexico
Mexico City
Central region
~1 Southern region
i I North region
n
^ Guanajuato
Mexicos Regions used in Mexican National Health Survey 2000 and Mexican National Health and
Nutrition Survey 2006.
FIGURE 2.1 MAP OF MEXICO WITH REGIONALIZATION USED IN
EPIDEMIOLOGICAL STUDIES
Barquera et al. (2009) used results of the Mexican National Health Survey 2000
and ENSANUT (Mexican National Health and Nutrition Survey 2006) to describe the
obesity prevalence in Mexican population. BMI among adult males was distributed as
follows: 31.8% were normal, 43.2% were overweight, and 23.5% were obese. The
obesity rate was the highest for the age group 50 to 59, and the overweight rate was
equally high for age groups from 30 to 59. After Mexico City, the region with the highest
overweight rate was the Central region (which includes the state of Guanajuato);
however, it was the North region the one with the highest rate of obesity after Mexico
City. Despite the difference in prevalence of excess weight between most and less
developed regions in Mexico; on average people in all regions are overweight.
21


The average body mass index in Mexico is overweight regardless of age group,
region of the country, rural/urban origin, and SES level or education. Rates of
overweight and obese are significantly higher in urban regions compared to rural.
Regarding SES (measured by assets and household characteristics), the rate of
overweight is equally high across SES levels but the rate of obesity is significantly higher
for medium and high SES levels.
Physical activity has not been measured consistently in national surveys in
Mexico, therefore changes cannot be evaluated but its association with obesity has been
studied. Recently, Gomez, Hernandez-Prado, Morales & Shamah-Levy (2009) evaluated
the association between physical activity and BMI in Mexican population. In general,
the authors found that levels of physical activity tend to be high for the adult population
in general; particularly 69% of males were classified in the high physical activity
practice category. They also found a negative association between levels of physical
activity and prevalence of overweight and obesity and this association was more
consistent among men than women. Most men in the low physical activity category
were overweight or obese. Adult men in the high physical activity category were 27%
less likely to be overweight or obese compared to men in the low category.
According to data from the National Health survey (2000), the rates of
hypertension range from 21.3 in population 25 to 34 years old to 55.4 in population
between 55 and 64 years old. Barquera, Durazo-Arvizu, Luke, Cao & Cooper (2008)
found that among Mexicans the hypertension-education association exhibited a U-
shaped relationship, with a hypertension prevalence of 33.9, 28.1 and 33.1 for those
with less than high school, completed high school and more than high school
respectively. The non-linear association between SES and hypertension in developing
countries, elevated levels of hypertension for low as well as high SES individuals, has
been previously reported. Smoking, an associated risk factor to hypertension and other
22


chronic diseases shows a positive gradient with socio-economic status in Mexican
population.
According to Prentice (2006), the environmental drivers of the obesity epidemic
are: world's food supply and diets have been sweetened tremendously; edible oil intake
has grown very rapidly; energy density of diets seems to be growing rapidly; and intake
of animal-source foods is increasing rapidly in the low-income world. This trend along
with a sedentary life and a potential genetic susceptibility are all factors that have
impacted the growth of overweight and obesity prevalence in Mexican population
(Olaiz-Fernandez, Rivera-Dommarco, Shamah-Levy et al., 2006). In addition to these
environmental factors, in many developing countries there is obesity acceptance
especially among middle-aged women.
Based on data from a National Income and Expenditure Survey, Rivera et al.
(2002) documented increasing trends in the quantities of sugars and refined
carbohydrates purchased and more particularly the quantities purchased of soda, which
could be associated with the increased obesity rate and mortality from chronic diseases.
Fernald (2007) analyzed the relationship between BMI and SES, as well as the
contribution of behavioral characteristics like soda and alcohol consumption in a
sample of rural poor population in Mexico. She used data from a National Social Welfare
Survey conducted in 2003, which was designed to be representative of the poorest rural
communities in seven states. In this sample, the majority of men and women were
classified as obese. In multivariate analyses, education, occupation, housing conditions
and assets all contributed independently and significantly to the prediction of BMI. Age
adjusted BMI was also associated with increased consumption of carbonated sugary
beverages in both sexes; and SES was positively associated with the consumption of
carbonated beverages and alcohol. BMI was positively associated with SES, regardless
of how it was measured in this low-income population of adults in rural Mexico. These
findings suggest that increased economic resources may allow people to purchase and
23


consume larger number of high-calorie beverages that are further associated with
increased BMI among this rural poor population.
The only study that has truly tested the health selection effect from Mexico to
the US, found that migrants were selected by education but evidence of selection by
health was weak (Rubalcava et al., 2008). Buttenheim, Godman, Pebley et al. (2010)
hypothesize that the negative health status prevalent among people with higher SES in
Mexico might be the explanation for this weak selection by health. Using data from the
Mexican National Health Survey (2000), the authors describe rates of obesity and
smoking for adult population residing in high and low migration regions. They found
that the prevalence of obesity was 22% vs. 16% and for smoking was 40% vs. 31% for
men living on high and low migration regions respectively. They also found that men in
high migration regions were more educated than men in low migration regions (8.6 vs.
6.3 years of education). These differential rates might be the result of long-standing,
frequent seasonal and circular migration from Mexico to the US. In one side migration
brings to poor rural communities excess income that increases purchase power. On the
other side, circular migration facilitates the diffusion of consumption patterns and other
health determinants in sending communities in Mexico and receiving communities in
the US; and these influences are reciprocal. In the presence of these patterns, the study
of health selection does not seem possible if we do not consider individual and
community indicators of migration participation. These differences also suggest that
selection of immigrants would be different depending on the level to which migration
has been a feature of their communities of origin.
24


Understanding Health Selection of Immigrants:
Contributions from Migration Research
Health is a central element of well-being and an indispensable condition for the
development of a person's productive potential. Therefore, good health constitutes an
essential asset for the integral development of an immigrants capacities for performing
labor and for social participation (CONAPO, 2005). It is therefore expected to observe
that Mexican immigrants, particularly labor immigrants, are healthy enough to be able
to cross the border and find a job in the US. However, the HME hypothesis states that
individuals who migrate are not only healthy but healthier than their counterparts who
do not migrate, implying that there is a selectivity process (Markides & Coreil, 1986).
It has been well-established in migration research that migrants are not a
random sample of the population in the source countries and that they tend to be
favorably selected for labor success (Liebig & Sousa-Poza, 2004). It has been described
that labor migrants, on average, tend to be more able, ambitious, aggressive,
entrepreneurial or otherwise more favorably selected than similar individuals who
remain in their place of origin (Chiswich, 1999). Mexico-US migration research has
documented that it is not the poorest or the least educated the ones who migrate to the
US, a fact that supports the selectivity hypothesis (Massey, Arango, Hugo et al., 1998).
The proposition of a HME agrees with these findings since it assumes that the
health status of migrants is not a random sample of the health status of the population
in their places of origin, implying a selection process by health. For the most part, theory
has not been explicitly used to study the health selection of immigrants. As a result,
most studies concerned with health selection of immigrants seem to have the premise
that people in home countries have equal access to migration. The few studies that
provide an explanation of how migrants might be selected by health seem to embrace
only one theoretical perspective of migration, an economic theory, when they talk about
health selection of immigrants. For example, Rubalcava et al. (2008) introduces health
25


as a migration predictor along with education as a human capital indicator; Jasso et al.
(2004] elaborate a model where the selection by health of immigrants would be
necessary in order to obtain net positive returns from migration. The use of this
explanation can be the result of limitations of the data, since most studies about the
HME use data from surveys that were not designed with the purpose of measuring a
HME and even less to explain why migrant selection on health might takes place. In
addition, the selection of immigrants on health can only be described and explained
when the right comparison group is studied, the population of origin that was not
selected for migration.
There is no single theory to explain the emergence and perpetuation of
international migration that is widely accepted and the theories developed are not
necessarily contradictory since they conceive causal mechanisms of migration at
different levels of aggregation (Massey, Arango, Hugo, et al., 1998; Massey & Espinosa,
1997], Some theories provide hypotheses than can be tested at individual and group
levels. At the individual level, Rational-Choice Theory, the micro-theory version of
neoclassical economics, as well as Social Capital Theory might provide alternative
explanations to understand the role of health status on labor migration participation.
These theories that have explained successfully other aspects of migrant selection can
be extended to explain selection of migrants based on health status characteristics.
These two theories could be complementary to some extent. Rational-choice
theory assumes that everyone in the home country has common access to the migration
opportunity, and whether they participate or not is a function of their individual
characteristics and their own determination of their fitness. While, Social Capital theory
considers that a person's access to the migration opportunity is filtered through
networks and that this network influence whether that person actually does migrate. In
one proposition, Rational-choice theory, migration is the outcome of an individual
decision, while in Social Capital theory, migration is an outcome determined by social
26


forces. Immigrant selection on health could be explained as the outcome of an
individual's decision after considering costs and benefits of migration participation or
could be explained as the outcome of social forces at play, like access to a migrant
network. Understanding the mechanisms that might bring healthier immigrants to this
country in the first place, is the first step to elaborate better models that explain the
forces contributing to immigrant health at arrival and when they adapt to their new
country. In this section, I explore how these two perspectives of migration theory might
explain the selection by health of immigrants.
Using Rational-Choice Theory to Explain Migrant Selection on Health
The Rational-Choice theory proposes that potential immigrants make a decision
to migrate after a cost-benefit calculation leads them to expect a positive return, usually
monetary, from the migratory movement. Given the right macroeconomic conditions,
like differential in earnings and employment rates between sending and receiving
places, a potential migrant would go to wherever the expected net returns of migration
are greatest. The likelihood of migration would be increased by individual human
capital characteristics (like education, experience, training, language skills); or by
individual characteristics, social conditions, or technologies that lower migration costs.
Because of variation on human capital and individual characteristics, individuals living
in the same places can display very different propensities to emigrate (Massey et al.,
1998).
Health has been widely acknowledged as an important component of an
individuals human capital that can directly increase earning capacity. Healthier
individuals are generally more energetic and robust, characteristics that would allow
them to be more productive by using their skills. But before migrants can obtain the
higher wages associated with greater productivity, they need to consider some costs
which include travelling and maintenance while moving and looking for work, the effort
27


involved in learning a new language and culture, the difficulty experienced in adapting
to a new environment, and the psychological costs of cutting old ties and forging new
ones. All these investments would also require that immigrants are healthy, physically
and emotionally. Where the costs of migration are greater, migrants will self-select on
better health to a greater extent. Therefore, in the individuals evaluation process of
costs and gains of migration, health could be part of the equation in several ways (Jasso
etal.,2004).
If health is such an important element of this self-selection process then some
awareness and evaluation of personal health status would be needed on the migration
decision-making process. The decision-making process would take into account the
probability of employment at the destination; the earnings if employed at the place of
destination; the probability of employment in the community of origin; earnings if
employed at the community of origin; the probability of avoiding deportation from the
area of destination (when migration is undocumented]; and the costs of movement
(including psychological costs] (Massey et al., 1998]. Access to a network of prior
migrants would be also likely to influence whether an individual migrates. Networks
provide information about crossing the border and finding employment that lower an
individuals migration costs (Orrenius & Zadvodny, 2001). In addition to these
considerations, other individual factors should be taken into account like previous
experience, knowledge of the environment in the place of destination, and ability to
perform the job, which depending on the type of job that is being pursued, would
involve more or less physical fitness, strength and endurance, all of them components of
health status.
Following this model, a potential labor migrant would need a positive personal
evaluation of her/his own health status, including physical health, to be able to perform
their skills to increase potential earnings through uninterrupted work and
emotional/psychological strength to withstand separation from family members and
28


the demands of adaptation. The notion that a potential migrant has to make an
assessment of their own health as part of the evaluation of costs and benefits for
migration implies that awareness of good health needs to be present and health
selection might likely be based on health indicators that can be judged by the
prospective immigrant, like self-rated health or presence of psychosocial symptoms.
Because a prospective migrant would weigh several elements to make the
decision to migrate, the health status of migrants, other things equal, is expected to be
better than that of non-migrants. Furthermore, the health status of migrants would be
better among those who have fewer positive attributes to weigh on their decision, for
example lower previous migratory experience; smaller or not as close migrant network;
lack of documents to cross the border and work.
Using Social Capital Theory of Migration to Explain Migrant Selection on Health
The leading prediction of Social Capital theory of migration is that people who
are socially related to current or former migrants have access to social capital that
significantly increases the likelihood that they will migrate. A characteristic of social
capital is that it may be translated into other forms of capital, notably financial capital,
in this case foreign wages and the remittances they permit. People gain access to social
capital through membership in networks and social institutions and then convert it into
other forms of capital to improve or maintain their position in society (Palloni, Massey,
Ceballos et al., 2001).
In the case of international migration, people in sending communities gain
access to social capital through their migrant networks. "Migrant networks are sets of
interpersonal ties that connect migrants, former migrants, and non-migrants in origin
and destination areas through ties of kinship, friendship, and shared community origin."
(Massey et al., 1998, p. 42). The first migrants who leave for a new destination have no
social ties to draw upon, and for them migration is costly, particularly if it involves
29


entering another country without documents. After the first migrants have left,
however, the potential costs of migration are substantially lowered for friends and
relatives left behind. Because of the nature of kinship and friendship structures, each
new migrant creates a set of people with social ties to the destination area. When these
networks are well developed, a job at the place of destination seems at easy reach for
most community members, making emigration very attractive as a reliable and secure
source of income (Massey et al., 1998).
Once international migration begins, it tends to expand over time until network
connections have diffused so widely in a sending region that all people who wish to
migrate benefit from having ties to persons who have already migrated. As this
organization develops to support, sustain, and promote international movement, the
international flow of migrants becomes institutionalized and independent of the factors
that originally caused it. In this sense, international migration causation is cumulative
since each act of migration alters the social context within which subsequent migration
decisions are made, typically in ways that make additional movement more likely
(Massey et al., 1998).
In those communities where international migration has become
institutionalized, the easy reach of international migration along with reduction of costs
and risks associated with the growth of migrant networks appears to be in opposition to
the notion of a strict selection of migrants, whether by health or other traits. Following
propositions of social capital theory, it seems that the only selection that might take
place is having access to a resourceful network, and individual attributes, like health,
would be less important. However, we have reviewed the evidence regarding the health
status of Mexican immigrants compared to other subgroups of the population, which
shows that Mexican immigrants are healthier than US-born Mexicans, fact that suggests
that selection by health still might operate, in spite of how institutionalized migration to
the US from Mexico is.
30


It is possible that the health selection of migrants takes place because the social
process responsible for "selecting new migrants takes care of their "health selection.
Some authors have suggested that the resources provided by the migrant network,
whether financial, information, or emotional support, which facilitate the migration
movement could also be related to better health outcomes of migrants (Palloni &
Morenoff, 2001}. This notion is supported by evidence showing an association between
access to social networks and health outcomes (Berkman & Glass, 2000). Positive
health selection would be possible since in these sending communities of origin,
migrant networks might be able to provide key resources for better health differentially
for those who might join the international migration pathway (Donato, Kanaiaupuni, &
Stainback, 2003; Kanai'aupuni, Donato, Thompson-Colon, & Stainback, 2005}.
If new migrants sponsored by a migrant network were not selected by health,
their ability to help others would end very soon, their access to work in the US would be
limited, and they could not invite/recommend others to participate of the benefits of the
relationships created with US employers. In the case of migrant male labor selection, I
propose that there might be a network-selection by health, where new immigrants are
sponsored by the migrant network when they are believed to be strong, able to work,
and resilient. The selection by health of migrants through their access to social capital
would follow a different mechanism to what is proposed by rational-choice theory. In
the case of rational-choice theory, a migration movement will result from an evaluation
of positive net returns considering individual attributes and costs; where having access
to a migrant network will reduce the cost of migration, and a positive evaluation of
personal health status is a crucial individual attribute. From the social capital
perspective, a migration movement will result also from an evaluation of individual
attributes and costs; however in this case the movement would only take place in the
presence of a migrant network. A positive self-evaluation of health status might be
present but it would be the network's validation of good health that determines if an
31


individual can be sponsored for international migration. If the migrant network
validates good health of a prospective immigrant, selection on health will likely happen
based on health status characteristics that are more evident to others and not
exclusively based on perception (like health behaviors or fitness).
These two perspectives, even when contrasting, might be complementary. A
migrant could have a positive evaluation of health, and consider his health status a
necessary asset to migrate successfully, however this wont be enough to carry out the
migration movement unless he is part of a network that will make the trip possible. One
of the entry requirements to the migrant network might be a good health status, as
evaluated by the migrant network.
My discussion of how these two perspectives could help us explain the selection
by health of migrants is centered on male labor migration. There is extensive literature
that documents different migration dynamics for women and men. Therefore even if we
use the same theoretical perspectives to understand female migration, we might need a
separate set of hypotheses to explain womens selection by health. This difference on
migration dynamics between men and women has been widely acknowledged and
referred to as gendered process of migration (e.g. Donato & Patterson, 2004; Mahler &
Pessar, 2006; Donato, Gabaccia, Holdaway et al., 2006).
Whether a migration movement can be better predicted by a personal attribute,
like health, or a person's access to a migrant network, the decision to migrate is still
made by an individual. This prospective migrant probably needs to evaluate several
factors before making a determination of looking for labor abroad or staying in the
community of origin. Therefore, to study the role of health status and/or access to
migrant networks on predicting a migration movement, we need to take into
consideration the context in which the migration decision-making takes place and the
32


different factors involved in this process (De Jong, 2000; Massey, Arango, Hugo et al.,
1998).
To understand the process of health selection, we need to study the role that a
personal evaluation of health status, along with other individual and social factors play
in the final decision of moving across the border to work. In this case, health status
would be considered a precious individual resource that will increase skill performance
and productivity by allowing more work hours once in the US; being healthy would
increase the likelihood of being successful in the case of taking undocumented crossing;
being healthy would allow immigrants to be resilient to the effects of separation from
the nuclear family, sometimes isolation, and dealing with the demands of adapting to a
new environment. From an economic perspective, a positive evaluation of health status
would be a necessary salient feature for those who migrate; while from a social capital
perspective, access to a migrant network would make migration possible. Furthermore,
from a social capital perspective, a migrant networks validation of an individual's good
health might be needed; men who are healthier would be "more likely helped by the
network to migrate. If selected, the migrant network would give a potential migrant
access to needed resources, which could be translated as an invitation to come to the US,
the contact of a contractor, specific job offer, and/or financial resources.
A Framework to Study Migrant Health Selection
Researchers have hypothesized that the healthy migrant effect is the result of
either a selection process, through which those who are better off tend to move; or the
result of the more healthful behaviors that are the norm in the countries of origin of
immigrants (Markides & Coreil, 1986; Jasso et al., 2004; Gushulak, 2007). Testing these
two hypotheses requires comparisons between sending and receiving communities
(Palloni & Arias, 2004).
33


As stated above, the proposition that immigrants are selected by health
coincides with the widely accepted observation in migration research that migration is
not a random event, and that not everyone in source countries has an equal chance to
migrate (Liebig & Sousa-Poza, 2004; Chiswich, 1999; Massey, Arango, Hugo etal., 1998).
However, who migrates is explained in different ways depending on the theoretical
perspective adopted. As 1 described previously, Rational-choice theory and Social
Capital theory applied to migration provide the current framework to test potential
mechanisms of health selection of immigrants. Here, I will describe more specifically
how these two theories might explain migrant health selection through two different
mechanisms: self-selection and network-selection.
From a Rational-choice theory perspective, people migrate because after an
evaluation of costs and benefits, they determine that they will obtain a net positive
"economic" return from migration. In the individuals evaluation of costs and gains of
migration, health would be part of the equation. A potential labor migrant would need a
positive personal evaluation of his own health status, including physical health to be
able to perform their skills and increase potential earnings through uninterrupted
work; and emotional/psychological strength to withstand separation from family
members and the demands of adaptation. Because a prospective migrant would weigh
several elements to make the decision to migrate, the health status of migrants, other
things equal, is expected to be better than that of non-migrants (Jasso et al., 2004).
Because this mechanism indicates that the health selection is the outcome of an
individual's decision, it is described as a self-selection hypothesis (see Figure 2.2).
To test the self-selection hypothesis, it is necessary to take into consideration a
variety of factors that might motivate an individual's participation on migration.
Factors that are relevant to make a decision to migrate are included in this framework.
These factors are defined as the motivations that prompt the migratory movement and
include specific economic situations (e.g. having to pay a debt or wanting to save
34


money), personal characteristics (e.g. having previous migratory experience or
considering themselves hard workers); social reasons (e.g. being encouraged by friends
and family); or normative motivations (e.g. all young males in town go to the US). It is
necessary to test the self-selection under equal motivations since people who are more
motivated to migrate might be willing to assume more risks.
The self-selection mechanism is illustrated in Figure 2.2. The graphic describes
that in this male-only agricultural labor migration stream, where participants share the
same context of origin and equal motivations to migrate, men with positive health status
will be more likely to migrate than men with poorer health.
CONTEXT OF MIGRATION
Positive Objective Indicators of Health Positive Subjective Indicators of Health Migration

Equal motivations to migrate: economic, personal, social, normative
FIGURE 2.2 MIGRANT HEALTH SELECTION ACCORDING TO
SELF-SELECTION HYPOTHESIS
The health status indicators included in this study are classified as objective or
subjective. Subjective health status indicators are defined as those based on self-
perception (like self-rated health and depressive symptomatology) or private behaviors
and attributes that are not immediately apparent to others (like presence of chronic
diseases or health care use). Objective health status indicators are those based on
measurements (like weight and height to estimate BMI, blood pressure, fitness level,
physical activity) or those that can be observed by others (like smoking, drinking and
diet behaviors).
35


The notion that a potential migrant has to make an assessment of their own
health as part of the evaluation of costs and benefits for migration implies that
awareness of good health needs to be present and that the subjective health indicators
will be more important. Objective health status is still important since it is expected to
be associated with subjective health status but not as much as perceived health. The
arrow leading directly from Positive Subjective Health Status to migration illustrates
this expected relationship in Figure 2.2.
According to the propositions of Social Capital Theory of migration, men with
access to migrant networks have a higher likelihood to migrate. A migrant network is
defined by the number and nature of relationships a person has with other individuals
who have US experience. In cross-sectional surveys, migrant networks are measured at
the individual level by the presence of family members who are current or past
migrants, and/or the number of friends or acquaintances with US experience; and at the
community level, as the fraction of a communitys inhabitants with prior migrant
experience (Espinosa & Massey, 1998). In this study, the migrant network is measured
as the presence of a set of non-family relationships and the number of family members
with current US experience. Based on findings of migration research, it is expected that
the presence of this network will increase the likelihood of migration.
The network-selection mechanism is illustrated in Figure 2.3. The graphic
describes that in this male-only agricultural labor migration stream, where participants
share the same context of origin and equal motivations to migrate, men with positive
health status will be more likely to migrate via access to a migrant network. For the
network-selection mechanism, objective health indicators are expected to be more
important since they include behaviors that are observable. This hypothesis implies
that it is the network's validation of good health which determines if an individual can
be sponsored for migration. That is, if migrant networks are selecting members based
on health status, access to the network will mediate the association between objective
36


health status and the probability of migration. Under the migrant network hypothesis,
subjective health indicators are still important since they are likely associated to
objective health status but are not expected to be as important as the objective
indicators.
FIGURE 2.3 MIGRANT HEALTH SELECTION ACCORDING TO
THE NETWORK-SELECTION
Figure 2.3 shows the migrant network in an upper level to make reference to the
meso-level of social networks. As stated before, the migrant network in the quantitative
part of this study is an individual level measure, however the qualitative component of
the project will provide important additional information about the role of the migrant
network on making migration possible as well as information on the health
requirements that a group of migrants would have to "select" other men for migration.
This information will help us understand how migrant networks might work on
selecting potential participants by health.
Under the network-selection hypothesis, motivations to migrate are still
considered because regardless of the access provided by the network, there is still an
37


individual evaluation of the incentives to participate in migration. The two mechanisms
described here might be complementary. A migrant could have a positive evaluation of
health, and consider his health status a necessary asset to migrate successfully, however
this wont be enough to carry out the migration movement unless he is part of a
network that will make the trip possible. The qualitative component provides
additional information regarding how health is considered at all in the decision to
migrate, how a potential migrant determines that he is healthy or fit for a labor
opportunity. Qualitative information might also provide insight on the relationships
between subjective and objective health status indicators.
This framework describes two mechanisms of migrant selection. The network-
selection hypothesis to an extent builds on the self-selection hypothesis. It is true that
an individual makes a determination of their own fitness to migrate but only the
migrant network makes migration possible. The outcome would be that healthier
people migrate because being "selected" by the network would take care of the health
selection. The notion that selection by a network and selection by good health might be
concurrent is based on an abundant body of literature regarding the relationship of
social networks and good health outcomes (Kawachi, Subramanian, Kim, 2008;
Berkman & Glass, 2000). However, it is not the intention of this study to provide
explanations of the multiple ways in which networks and good health might be
associated. This is the first step to test if the health selection of immigrants is the
resultant of a social process instead of an individual determination.
38


CHAPTER 3
RESEARCH DESIGN AND METHODS
The purposes of this project were first to determine if the healthy migrant effect
took place in a bounded male-only migration stream from Mexico to Colorado,
measuring a variety of health status indicators; and second to test two plausible ways in
which migrant selection by health might take place: a "network-selection or a "self-
selection" process.
The study was conducted with a male-only agricultural labor stream from
Mexico to Colorado. All men were recruited in the same region of Mexico and hired to
perform agricultural work. These men arrived in Colorado in early April depending on
weather and were expected to work for a period of approximately six months. Some of
them had participated in the program for more than one year. Studying the migrant
selection with this particular group of temporary farmworkers defined who was
considered migrant, non-migrant and return-migrant in a very specific manner (see
Definition of variables section).
This was a nested mixed-methods study. A deductive theoretical approach
guided the study but a qualitative component was included complementing, supporting
and strengthening quantitative findings. The quantitative aspect of the study provided a
test of the way in which specific health status indicators and the presence of a migrant
network might explain who migrates and who does not. Measures of the network were
egocentric, as the migrant/non-migrant individual reported personal resources that
provided them with different types of support. The effect that the presence of the
network had on predicting migration whether directly or mediating the effect of health
status indicators was measured. The assessment of health status included different
indicators, some of them based on a personal evaluation of health (like self-reported
39


general or relative health), on measurements [like BMI estimated with measured weight
and height) or on reported observable behaviors (like smoking, drinking). This
evaluation of health was crucial to the test proposed by this study because it provided
the opportunity of exploring if there was a differential role of health status in the
migration behavior when health was measured subjectively compared to objectively.
The qualitative component of this study provides complementary information about the
ways in which networks might "select" new migrants. The qualitative component
provides information about the context in which a decision to migrate was made and
explores the different social and family influences on making this decision, as well as
personal considerations like health.
The advantages of using a bounded migration stream to conduct this study are:
(1) all migrants are recruited to perform the same type of job, which implies that they
would require the "same individual attributes" to be successful and would homogenize
the assumed health screening criteria; (2) all migrants are men. Extensive evidence is
available about the differential gender dynamics of migration (e.g. Donato, Gabaccia,
Holdaway et al., 2006); (3) all men come from the same region in Mexico, which sets
clear geographic boundaries to define a comparison group selected from those who did
not participate in this specific job opportunity. Also, since they all come from the same
communities of origin, we can assume they have been socialized to similar behavioral
norms and environment that might contribute to their health status.
The disadvantages are: (1) this migration is meant to be temporary, which sets
different expectations for participants (and maybe different selection criteria); and (2)
this is a group of legal temporary workers, which is not a situation highly representative
of a large proportion of immigrants. The fact that this is documented and temporary
migration might imply that men are not as "strictly" selected for health as if migration
were undocumented and meant to last indefinitely or for a long time. Most of the time,
undocumented border crossing is physically strenuous (e.g. might require walking long
40


distances, crossing a water stream, or trespassing fences], which might prevent many
men and women from trying, if they do not consider themselves fit enough to endure
such challenge. Long-term immigration might also prevent many from trying, since they
might determine they do not have the emotional strength to withstand indefinite
separation from their families and known environment. (3) Ideally, the health of
migrants would be measured before they make their first trip to Colorado, which was
not feasible in this study. This is considered the most important limitation to test the
healthy migrant effect and I tried to overcome it by including alternative analyses that
show the impact that time already spent on the US might have on health status
indicators.
Despite its limitations, this study might set the basis for a larger, more
comprehensive study that includes different types of immigrants, working in different
occupations. The hypotheses studied in this project might be equally applicable to male
populations of undocumented immigrants working in different environments.
Assuming that legal migration involves less personal burden and/or risk, we can only
expect that the selection by health of this group is not as strict as it could be in the case
of undocumented migration, limiting the generalizability of these findings.
The study took place in both ends of an agricultural migration flow. This
migration stream was that of men recruited to work temporarily in agriculture in a rural
area located northeast of Denver in the state of Colorado. All men hired by a local
grower (approximately 200] came every year from different villages from one
municipality in the state of Guanajuato, Mexico. An overview of the components of this
study is presented in Table 3.1.
41


Table 3.1. Overview of the Study
Mexico Colorado
Quantitative
Purpose Considering other known migration decision-making factors, to determine the extent to which being a migrant is predicted directly by a positive health status or indirectly by positive health status via access to a migrant network
Method Survey collected with a group of 220 18- 49 yo males in five randomly selected towns of Valle, Guanajuato, Mexico Survey collected with a group of 164 adult male farmworkers in northeast Colorado
Qualitative
Purpose Understanding the role of health and migrant networks on the decision- making process for those who migrate and those who recruit, connect, and hire migrants
Method 6 interviews with return-migrants recruited following a snowball technique. 2 Interviews with key informants [contractor, majordomo) recruited with snowball technique 14 interviews with migrant males recruited following a maximum variation strategy
Quantitative Component
The quantitative component of the study was a cross-sectional survey of adult
males participating in a Mexico Colorado labor migration stream; as well as males
from the same places of origin in Mexico residing in their hometowns at the time the
survey took place.
The specific aims for the quantitative phase of this project were:
1. To describe selected health status indicators, the migrant network and motivations
to migrate for adult male non-migrants or return-migrants who reside in Mexico,
and migrants working in Colorado.
2. To determine if being a migrant in Colorado is predicted by a positive health status.
42


3. To determine the extent to which being a migrant in Colorado, other motivations to
migrate and sociodemographic characteristics equal, is directly determined by a
positive health status.
4. To determine the extent to which being a migrant in Colorado, other motivations to
migrate and sociodemographic characteristics equal, is indirectly determined by a
positive health status, via access to a migrant network.
It was hypothesized that if the health selection of migrants followed a self-
selection process, migration would be predicted by a positive health status, particularly
subjective health status. If the health selection of migrants followed a network-selection
process, migration would be predicted by access to the migrant network. More
precisely, access to a network would partially explain the association between health
status and migration. These hypotheses were not necessarily exclusive. To some
extent, migration behavior might be predicted by a positive health status and having
access to a migrant network. Furthermore, it was expected that the effect of the
network predicting migration varied depending on how healthy a prospective migrant
was, which was tested as an interaction effect described later in this chapter.
Sample and Participants
Adult males (18-49 years old) residing in fourteen villages in one municipality in
the state of Guanajuato were the population for this study.2 Five out of those 14 towns
were selected randomly. Interviewers visited all households in selected towns and
obtained information about 18-49 year old men, if they had been in the United States in
the previous year, and if they were in town at the moment. All 18-49 year old males
living in those five communities who were in town at the moment of the survey were
considered eligible to participate in the study. Five hundred and eighteen eligible men
2 The researcher collected a list with towns of origin of men working in this labor program,
previous to the conduction of the study.
43


were identified. Of those men, 149 were working in the US and 369 were residing in
Mexico at the moment the survey took place. One hundred and thirty six men were
never reached by the interviewers team, who visited their residences up to three times.
Thirteen men refused to participate in the survey. The response rate in Mexican
communities was 94.4% with a total sample of 220 men.
In Colorado, the group of Mexican men participating in the agricultural
temporary work program during 2009 was 200. Men working in this program who
were recruited from the selected municipality in Guanajuato were eligible to participate
in the study (n=174). The response rate in Colorado was estimated at 94% with a total
sample of 164 men. Ninety-eight percent of the men interviewed in Colorado came from
the state of Guanajuato in Mexico, out of this group 95% came from the one municipality
in Guanajuato. To estimate the total of eligible men in Colorado we relied on reports
from participants as it is explained later in the procedure section. The age distribution
of the total sample (including Mexico and Colorado) compared to the populations age
distribution in the municipality in Guanajuato is described in Table 3.2.
Table 3.2. Guanajuato municipality's male population and sample distribution of
by age groups
18-29 30-39 40 and over Total
Population* 10,400 (43.5%) 7,419 (31.0%) 6,094 (25.5%) 23,913
Sample 172 (44.8%) 113 (29.8%) 99 (25.8%) 384
* Source: 2005 Population and Household Count, Mexico: Instituto Nacional de
Estadistica, Geografia e Informatica (INEGI)
44


Procedure
Questionnaires were conducted face-to-face, in Spanish, by trained interviewers
in Mexico and Colorado. The questionnaire administration took 30 minutes including
measurements. In Mexico, four interviewers received a 36-hour training including role
playing. The topics of the training were: objectives of the study, procedures for
participant recruitment, informed consent process, interview skills, measurements and
use of the questionnaire. Interviewers and researcher were trained in Mexico by a
person with specialization in anthropometries on the proper way to take health
measures, particularly waist and hip circumference.
In Mexico, the interviewers went to five randomly selected villages. In those
villages, they visited all households in the communities and conducted a census of 18 to
49 year old males. For each of those males, interviewers asked if they had been in the
United States within the previous year, and if they were in town at that moment. All 18
to 49 year old men in town were eligible to participate in the interview. If there was a
man within this age range in the house, interviewers would try to interview him at that
moment or ask for a better time to reach him. Interviewers tried to reach eligible men
up to three times.
Fieldwork in Mexico took place during the months of April and May, instead of
December and January as it was originally planned. As a consequence, many men were
not reachable because they had already migrated to this country for the year or were
working out of town almost every day for most of the day.
The project was introduced to participants as a study with the objective to
assess the relationship between the health status and behaviors of men and their
participation in migration to the US. Potential participants were given a brief
description of the type of questions and measures included in the questionnaire.
Interviewers explained to potential participants that the study was conducted by the
45


University of Colorado in Denver in collaboration with the states Ministry of Health.
The interviewers explained to participants that their towns were selected through a
random procedure, that their participation was voluntary, and the information they
provide would be analyzed in a group and never individually; and that any information
that could identify them would not be recorded in the questionnaire. All procedures
were approved by the states Ministry of Health in Mexico and COMIRB (see Appendix
C). The researcher established collaboration with the Public Health Research Institute
at the Ministry of Health in Guanajuato. This project received full support from them.
They recruited interviewers, provided physical space for training and storage of surveys
and equipment, loaned back packs and stadiometers for the studys fieldwork in Mexico.
In the United States, five interviewers received similar training to the one
described above, with the exception of the recruitment procedure. The researcher
trained the interviewers in the US on the proper way to take health measurements as
she was trained in Mexico. In Colorado, all migrant farmworkers recruited from
Guanajuato and working for a specific employer were eligible to participate in the study.
The researcher contacted the employer to explain the study and ask for his help inviting
farmworkers to participate in the study. With the employer's and contractor's
agreement, farmworkers received a flyer from the crews supervisor with an invitation
to participate in the study. The employer and contractor suggested the researchers to
visit farmworkers housing at the end of the work day. The research team set up stands
outside of farmworkers residences so when workers came back from work, researchers
would be available if they were interested on being part of the study. To determine the
number of eligible men, we relied on reports of participants who approached the
research team. Participants told us how many men were living in each unit and how
many of them were from Guanajuato. We explained the purpose of the study to all
participants in the same way that it was explained to men in Mexico and then invited
them to participate at their convenience. In Colorado, participants were given an
46


incentive of ten dollars as a compensation for their time. Incentives for participants in
Mexico were not approved by the Ministry of Healths office in Mexico, so they were not
distributed.
In Mexico and Colorado, all participants were informed of the purpose, risks and
benefits of the study and signed an informed consent form before the interview took
place. Approval of these procedures was obtained from the Colorado Multiple
Institutional Review Board (COMIRB) at the University of Colorado at Denver and
Health Sciences Center and the IRB from the Public Health Research Institute from the
states Ministry of Health in Mexico (see Appendix B).
Definition of Variables
Dependent variable
Migration. Study participants were part of any of the following categories based on their
migratory behavior.
Migrants. All males interviewed in Colorado who participate in the agricultural labor
program.
Non-migrants. Males interviewed in Mexico who have not been to the US for labor
reasons ever in their lives.
Return-migrants. Males interviewed in Mexico who have worked in the US
sometime in their lives.
For descriptive analyses, comparisons were made among these three groups.
For logistic regression analyses, two sets of models were conducted for aims 2 to 4 of
this project:
Migration 1: being a Migrant vs. Non-migrant.
47


Migration 2: being a Migrant vs. Return migrant.

Separate logistic regression models (rather than a multinomial logistic
regression model) were used because they include different, theoretically important
variables (i.e. cumulative time in US).
This survey excluded other migrant men who might have been working in other
states in this country or in other job opportunities in Colorado at the time the survey
was conducted in Mexican communities.
Independent Variables
Health status indicators. This study includes a variety of self-reported and measured
health status indicators as well as health behaviors. Some health indicators were
defined as objective, whether because they were based on actual measurements
conducted by the interviewers or because they were the report of observable health
behaviors. Other indicators were subjective, whether because they were based on self-
perception; they could not be observed by others or were considered a private behavior,
like health care utilization.
Objective health indicators: Work days lost due to illness; smoking, drinking,
physical activity, diet, BMI (based on height and weight measured with a portable
calibrated scale), waist and hip circumference (based on measures conducted by
interviewer), blood pressure and heart rate (measured with an electronic cuff),
fitness level (based on a 2-minute walking test conducted by interviewer).
Subjective health indicators: Self-reported health status, relative self-reported
health status, presence of chronic diseases, health care use.
Migrant network. The migrant network was defined in reference to the Mexico-Colorado
migration stream as an indicator that included the number of members of the social
network and the nature of their relationship (family members or kinship, friendship,
48


shared community of origin) who had agricultural labor experience in Northeast
Colorado. It was also explored the size of social network formed by important
relationships that were not family, and by family; as well as the US experience of these
networks.
Motivations to migrate. We measured factors relevant on the migration decision-making
process. These factors are defined as the basic motivations that prompt the migratory
movement and include specific economic situations (e.g. having to pay a debt or wanting
to save money), personal characteristics (e.g. having previous migratory experience or
considering themselves hard workers); social (e.g. being encouraged by friends and
family); opportunity (e.g. being invited to a labor program); or normative motivations
(e.g. all young males in town go to the US).
Instruments
The binational cross-sectional survey used a face-to-face interviewer
administered questionnaire. This questionnaire was specifically designed for the
project but includes sections adapted from other well-known surveys conducted with
similar populations or with similar goals. For example, demographic, migratory history
and social/migrant network questions were adapted from the Health Migration Survey
(see Donato & Kanaiaupuni, 2006) and the Mexican Migration Project (survey
questionnaire is available at the MMP website: mmp.opr.princeton.edu). The sections
evaluating health outcomes and behaviors were adapted from the National Health
Survey and the Mexican National Nutrition and Health Survey 2006 (Olaiz-Fernandez et
al., 2006). The section about motivations to migrate was created for this study. The
Colorado and Mexico versions of the questionnaire were created in Spanish and vary
slightly in the wording of some questions (see Appendix C with Spanish and English
translation of questionnaire). The Mexico and Colorado versions of the questionnaire
included the same sections described as follows:
49


1. Demographic and socioeconomic characteristics. This section assessed age; years
of schooling; occupational status and type of occupation; size and type of place of
residence (small, medium or large village, town, or city); and indicators of
socioeconomic status. The indicators of socioeconomic status include: an objective
indicator (i.e. a list of assets) and evaluation of subjective socioeconomic position (SES
ladder and a question inquiring how successful their families are compared to other
families in their town). The indicator Objective SES was built as a simple count of assets
that men reported owning in their households in Mexico. Men were requested to report
if they owned specific items or property, including large possessions like a home,
vehicles, or land; appliances like a blender; or electronics like a computer or DVD.
Participants interviewed in Colorado were asked to answer household related questions
based on their place of residence in Mexico.
2. Health status indicators. As it has been described previously, the health status
indicators included in this study were categorized in two groups as follows:
Subjective indicators: Self-reported general health and relative physical health
status; depressive symptomatology measured with the CESD; presence of chronic
diseases; and formal use of health care in previous six months. The CESD measure was
considered a positive indicator of subjective health if the score was 11 or lower, which
was the median of the scores distribution in this sample. The CESD scores in this
sample ranged from 0 to 40. According to scoring standards (Raddlof, 1977), a cut-off
point of 16 is considered a positive screening for depression. Other authors have
proposed the need to use different cut-off points depending on the goal of the study.
Since the goal of this measure is to characterize positive health status, it seems
appropriate to use a lower cut-off point. These indicators are described in Table 3.3.
50


Table 3.3. Definition of Subjective Health Status Indicators
Variable Definition Values Positive Health Status Indicator
Self-Rated General Health1 Answer to the question: In general, would you say your health is? Categorical values: 1] Excellent, 2) Very good, 3] Good, 4] Fair, 5) Poor 0 = Fair or Poor 1 = Excellent, Very good, Good
Self-Rated Relative Health2 Answer to the question: Compared to others your age, would you say your physical health is...? Better than others, about the same, worse than others. Categorical values: 1] Better than others; 2] About the same; 3] Worse than others 0 = about the same or worse than others your age 1 = Better than others your age
Presence of Chronic Conditions3 We explored the presence of 9 chronic conditions: hypertension, diabetes, heart problems, asthma, arthritis, ulcers or colitis. Positive answers were added. Values indicate numbers of problems and might range from 0 to 9. 0 = presence of at least one chronic disease 1 = absence of chronic disease
Depressive Symptomatology4 We used the CESD (Center of Epidemiologic Studies for Depression Scale]. This scale has 20 items that explore the presence of depressive symptomatology during the week previous to the interview. Frequency of occurrence is evaluated for each item where 0 means the symptom was not present at all and 3 means symptom was present 4 to 7 days. Values range from 0 to 60. Higher values indicate more symptomatology. The median value (11) of the distribution was used to determine health status. (Mow depressive symptomatology (below 11], l=high depressive symptomatology (11 and over]
Number of times saw DR Answer to questions how many times did you see the doctor in the last 6 months? Values indicate number of days. 0 = saw DR at least once 1= did not see DR
1 e.g. Acevedo-Garcia, Bates, Osypuk, & McArdle, 2010; 2 Rubalcava, Teruel, Thomas et al, 2008; 3 Singh & Siahpush, 2002; 4
Raddloff, 1977.


Objective indicators: Days of work lost due to illness, physical activity; smoking,
drinking, diet indicators [consumption of sugary drinks; red meat), height [measured
with a stadiometer) and weight [measured with a portable calibrated scale) to estimate
BMI, fitness measured with a walking test (distance walked in two minutes using a
pedometer and a stopwatch), waist and hip circumference (measured by interviewers
with a tape measure), and heart rate and blood pressure measured with an electronic
arm cuff (see Table 3.4 for descriptions of these indicators).
Height was measured in centimeters with a stadiometer, values ranging
between 130 and 200 centimeters were considered valid data for analysis. Weight was
measured using a calibrated electronic scale in kilograms. Height and weight were used
to estimate Body Mass Index (BMI) with the formula: weight (kgs)/[height (cms)].
Values between 10 and 58 were considered valid values for BMI. For descriptive
purposes, BMI was categorized as normal when values were below 25; overweight if
values ranged from 25 to 29.9; and obese if values were 30 and over (Panamerican
Health Organization, 2006). Being below 25 was defined as a positive health indicator.
Waist and hip circumference were measured by interviewers using a tape
measure. Valid values for hip and waist circumference were between 50 and 180
centimeters. A ratio waist to hip was estimated. Waist-to-hip circumference ratio has
been considered a good indicator of risk for metabolic syndrome, and national
guidelines indicate that a ratio above .95 is considered as an indicator of risk. This is the
value used as a cut point to determine healthy waist-to-hip-ratio (Olaiz-Fernandez,
Rivera-Dommarco, Shamah-Levy et al., 2006).
A two-minute walking test was used as an indicator of fitness. Participants were
asked to walk trying to cover as much distance as possible but without running, for two
minutes. A pedometer was placed on participants' waist, and two minutes were
measured with a stop watch. Walking longer distances is considered as an indicator of
52


better fitness. It was observed during fieldwork that in this sample, the fitness measure
might be more a reflection of how well men followed our instruction "walk as fast as
you can" than an indicator of their fitness level. In general, older men seemed to follow
the instruction better. Because men seemed to follow instructions differently, the
outcome of this test was not used in multivariate analyses.
Smoking, drinking, physical activity, and diet were also measured through self-
report. Physical activity was assessed with a series of 11 questions that asked the
participant first to respond how many days they were involved in vigorous or moderate
physical activity, walking or seating, and in those days, about how much time they spent
doing that (Olaiz-Fernandez, Rivera-Dommarco, Shamah-Levy et al., 2006). A measure
of time per week spent on each activity was obtained. An aggregated Physical Activity
indicator was built by adding time spent on each type of physical activity and
subtracting time spend seating. A cut point on these measures was determined using
the median values of the distributions. The aggregated indicator of Physical activity had
a median value of 36 hours per week. Because of the way this measure was obtained,
this result might be more a reflection of the immediate circumstances [longer work
hours) of migrant men and not necessarily a health behavior characteristic of this
group. Therefore, this measure was not included in the logistic regression models.
To determine positive health indicators of diet [weekly frequency of red meat
consumption, daily frequency of soda and fruit consumption); median values of the
distribution for these indicators were used also. Only red meat and soda consumption
indicators were used in final analyses (See Table 3.4).
53


Table 3.4. Definition of Objective Health Status Indicators
Variable Definition Values Positive Health Status Indicator
Blood Pressure [BP]1 Blood pressure was measured using an electronic cuff. Systolic and diastolic measures are reported separately. BP values over 120/80 are considered a health risk. Normal blood pressure is defined as values 120/80 or below; values of 140/90 and over are considered hypertension. Systolic BP < 120 Diastolic BP < 80
Waist to hip ratio [WHR]1 Waist and hip were measured in centimeters. A ratio between the two values was obtained. Values of 0.95 or below are considered low risk; .96 to 1 moderate risk; over 1 high health risk. Ratio < 0.95
Body Mass Index [BMI]1 BMI was calculated with the measure of weight in kilograms divided by the square of measured height in meters. Values of 25 and over are considered overweight and 30 and over considered obese. BMI <25
Heart Rate (HR]2 Measured with an electronic BP cuff HR would be categorized as above average as follows: for ages 18 to 25, if below 70. 26-35 below 71, 36-45 below 71, 46-55 below 72, over 55 below 72. HR <70 if 18 to 25 HR <71 if 26 to 55
Two-minute walking test3 Participants were asked to walk a fast pace for two minutes. Participants were given a pedometer and the time was measured with a time watch. Distance was recorded in meters.
Smoking1 Participants who have smoked at least 100 cigarettes in their lives and currently smoke. Values represent Yes/No 0 = non-smoker 1= smoker
Drinking five or more1 Participants were asked how many times they had five drinks in the previous month. Values indicate number of times they had five drinks in the month 0 = did not drink 5 prev month
previous to the survey. 1 = had 5 or more prev month at least once
1 These measures were defined following the Mexican National Health Survey of Health and Nutrition (Olaiz-Fernandez,
Rivera-Dommarco, Shamah-Levy et al., 2006];2 Perret-Guillaume, Joly, Benetos et al., 2009 3 Butland, Pang, Gross et al., 1982.


Table 3.4. (Cont.). Definition of Objective Health Status Indicators
cn
ui
Variable Definition Values Positive Health Status Indicator
Physical Activity1 11 questions that asked how many days of the week and how much time every day participants were involved in vigorous or moderate physical activity, walking continuously for at least ten minutes, and seating. For each of these activities an indicator of hours per week was estimated. An aggregated physical activity measure was also estimated as the result of the addition of time spent on different types of physical activity subtracting time spent seating. Values indicate number of hours spent on physical activity per week. The median value in the distribution is 36 hours. 0 = below 36 hours of physical activity 1= 36 hours of physical activity or above
Diet Questionnaire included two questions that asked about frequency of consumption of sugary drinks, and red meat. Values indicate respectively number of sodas, and times per week they eat red meat. Median values were used to determine indicators of positive health diet behavior. Number of sodas per week 2 or below Times per week had red meat 3 or below
Number of workdays lost Participants were asked how many days they lost work during the 6 months previous to ;he survey. Values indicate number of days. 0 = at least one workday lost 1 = no workdays lost
1 These measures were defined following the Mexican National Health Survey of Health and Nutrition (Olaiz-Fernandez,
Rivera-Dommarco, Shamah-Levy et al., 2006).


3. Migratory history. These questions explored previous labor migratory
experience in the US, general and specific to Colorado. Total number of trips to the US
and cumulative number of months they had spent in the United States, whether in one
or several trips. First and last occupations in the US, and states where they have worked
in the US.
4. Motivations to migrate. This section evaluated the relevance of different
economic, individual, social, normative and opportunity factors on deciding to migrate
or not. This portion of the questionnaire includes a list of situations that motivated or
would motivate participants migration to the US. The questions were designed
specifically for the study and included 24 statements that included economic, social,
personal, and normative and opportunity related motivations to migrate. The general
instruction for this set of questions was slightly different for men in Colorado and
Mexico. For men in Colorado, the general question was: When you were deciding to
come to work in Colorado how important was it to you that (particular statement).
While in Mexico, the general question was phrased: If you were deciding to come work
in the US, how important would be to you that (particular statement), to make a
decision to migrate. The respondent would answer how important was/would be each
of the 24 items in their decision to migrate for work to the US, using 0 (if the statement
was not important at all) to 10 (if the statement was extremely important).
A principal-component factor analysis with Varimax rotation was used to assess
the extent to which these items reflected the conceptual dimensions of each set of
measures as they were intended. We included items to evaluate economic, personal,
social, normative, and opportunity motivations. The two opportunity items were not
included as motivations to migrate, since most men who participated in the program
gave consistently the highest scores to these items.
56


Four factors that explained 78% of the variance were found: 1) Personal: 11
items and explained 30% of the variance; 2) Social, included 9 items and explained 23%
of the variance; 3) Economic, included 8 economic items, explained 17% of the variance;
4) Norm, included 2 items and explained 8% of the variance. Responses to these items
were summed by sub-scale as confirmed by the factor analysis (economic, personal,
social, and norm). A Cronbach's alpha reliability test to assess internal consistency was
conducted for each of the four subscales. Reliability coefficients ranged from .88 for the
Norm subscale with two items to .98 for the Personal subscale with 11 items.
5. Migrant and Social Networks. There was one specific question in the
survey asking from men in Mexico how many people working in agriculture in Colorado
they knew, and from men in Colorado how many people working in agriculture they
knew before coming to Colorado. The follow-up question was how many of those
people working in agriculture were family members. These questions were intended to
provide one of the measures of the migrant network specific to Colorado and
agricultural labor. For many men it was difficult to estimate how many people in
Colorado they knew. When they knew many people, their answers would vary from 20
or 30 to 200 or 300. It is also possible that numbers provided by migrant men
(interviewed in Colorado) were not the answer to how many people they knew before
they came to Colorado since that could be hard to recall. For all these reasons, this
network indicator was not used in the logistic regression models.
The questionnaire also included measures of egocentric networks describing
size, and types of support. It included a series of questions that assessed the number of
important non-family relationships; the resources each of these individuals had (e.g.
own a house, have stable employment, own a car, own land); if they worked in the US;
and if they had provided help getting a job and/or economic help. The questionnaire
also included a matrix that assessed the size of their family, their place of residence
(same town or other place in Mexico or in the US), frequency of contact with them, if the
57


family member was a regular source of advice, if family members had provided money
to the interviewee, or if the interviewee had provided any of them with money.
Answers to these sections were used to represent elements of migrant networks
included in the logistic regression models. Two indicators were used: having important
non-family relationships working in the US and number of family members residing in
the US. The first indicator is a dichotomy, where 0 means they did not report any
relationship working in the US and 1 means they had at least one relationship (the
maximum number was 2)3. The family indicator is a count of family members who were
living in the US at the time the survey took place and it ranged from 0 to 9.
Data analyses
Aim 1. To describe selected health status indicators, the migrant network and
motivations to migrate for adult male non-migrants or return-migrants who reside in
Mexico, and migrants working in Colorado.
Frequencies, central tendency and dispersion statistics were used as
appropriate to describe health status indicators, motivations to migrate, migratory
history and networks for migrants, return migrants and non-migrants. Pearson Chi
square and ANOVAs were used to compare health status indicators, migration decision-
making factors, migratory history and networks among those three groups.
3 It is important to remember that this is a count of non-family important relationships, meaning
people they would talk about very important matters, financial or child rearing issues. This low
number contrasts greatly with what was reported before on the number of people participants
known working in Colorado.
58


Aim 2. To determine if being a migrant in Colorado is predicted by a positive
health status.
The healthy migrant effect hypotheses are expressed as follows:
Hmeo There is no association between positive health status and being part of the
migrant group.
Hmei Positive health status is associated with being part of the migrant group.
Logistic regression analyses to predict being a migrant (vs. non-migrant) and
being a migrant (vs. return-migrant) were used to determine the extent to which being a
migrant was predicted by a positive health status. Due to sample size restrictions, the
number of variables included in the logistic regression models had to be reduced.4 The
final selection of predictors included in the models was based on the following criteria:
1. Migrant group differences in a given health indicator were not found statistically
significant in descriptive analyses. Pearson Chi Square values for Crosstabs for
dichotomous health indicators (0=negative, l=positive) by the dependent variable
(migrant vs. non migrant or migrant vs. return migrant) were not significant (p >
.05). We did not include presence of chronic diseases, smoking, and number of lost
days of work based on statistical significance. In adjusted models, some variables
that were considered theoretically important were included regardless of their
statistical significance like age and education.
2. Health indicator was considered redundant. We did not include Waist to hip ratio
because it seemed to be somewhat redundant with the Body Mass Index (BMI)
measure. The decision of including BMI was based on likelihood of comparability to
other studies.
4 Although there is no clear rule regarding sample size to conduct logistic regression analyses,
recommendations vary from 5 to 20 cases in the sample per variable included in the model
(Menard, 2010).
59


3. As it was mentioned before, physical activity and fitness indicators were not
included in logistic regression analyses.
For each dichotomous dependent variable (migrant vs. non migrant; migrant vs.
return migrant), we built nested models that included successively a set of predictors as
described in Table 3.5:
Table 3.5. Description of model building for Aim 2.
Models Depen dei Migrant vs. Non Migrant it variables Current Migrant vs. Return Migrant
1 Depressive symptomatology Self-rated general health Self-rated relative physical health Having seen DR in previous 6 months Depressive symptomatology Self-rated general health Self-rated relative physical health Having seen DR in previous 6 months
2 Blood pressure Body Mass Index Red meat consumption Soda consumption Having more than five drinks in previous month Missing work due to illness in previous six months Blood pressure Body Mass Index Red meat consumption Soda consumption Having more than five drinks in previous month Missing work due to illness in previous six months
3 Age Education Marital Status SES assets indicators Perceived SES Cumulative Time in US
4 Age Education Marital Status SES assets indicators Perceived SES
Models with interactions Interactions were tested between health indicators and sociodemographic characteristics. Interactions were tested between health indicators and sociodemographic characteristics and time in the US.
60


For each dependent variable, Model 2 provides the test for the healthy migrant
hypotheses stated for Aim 2, and Models 3 and 4 provide further adjustment adding
sociodemographic variables and cumulative time in the US for the sample of men with
migration experience.
We tested the presence of expected interactions between sociodemographic
indicators and subjective and objective single health status indicators. An interaction
term was considered significant and kept in the Final model if the difference between
the -2LL for the model without the interaction and the model with the interaction was
statistically significant. It was also considered if the individual effect of the interaction
term on the dependent variable was significant.
Individual effects of predictors as well as overall model fit measures were used
for interpretation of results. Continuous variables that were part of a significant
interaction were centered to facilitate interpretation. For each model, -2 Log
Likelihood, Chi square, Hosmer and Lemeshow goodness of fit test, a pseudo R square
measure [Nagelkerke), and overall percentage of correct classification of dependent
variable are described. For each predictor, unstandardized coefficients, standard
errors, and odds ratios with confidence intervals are described.
Aim 3. To determine the extent to which positive health status predicts being in
the migrant group [vs. non-migrant) and [vs. return-migrant), after adjusting for
sociodemographic characteristics and motivations to migrate.
The self-selection hypotheses are expressed as follows:
Hsso After adjusting for sociodemographic characteristics and motivations to migrate,
there is no association between positive health status and being part of the
migrant group.
61


Hssa After adjusting for sociodemographic characteristics and motivations to migrate,
positive health status is associated with being part of the migrant group.
Aim 4. To determine the extent to which being a migrant in Colorado,
sociodemographic characteristics other motivations to migrate being equal, is predicted
by a positive health status indirectly, via access to a migrant network.
The network-selection hypotheses are expressed as follows:
Hnso After adjusting for sociodemographic factors and motivations to migrate,
presence of a migrant network does not mediate the association between
positive health status and being part of the migrant group.
Hnsa After adjusting for sociodemographic factors and motivations to migrate,
presence of a migrant network mediates the association between positive health
status and being part of the migrant group.
To test Aims 3 and 4, two new series of nested logistic regression models were
built for each dependent variable: 1) migrant vs. non migrant; 2) migrant vs. return
migrant. For these models, two aggregated indicators of health status were created. 1
1) Subjective Health Status indicator that included: Self-rated relative health,
depressive symptomatology, and having seen a doctor in previous six months.
These variables were selected by adding one by one to a logistic regression model to
predict the outcome of interest. Those variables that did not contribute significantly
to the model [determined with the -2LL test explained before) were not included in
the aggregated indicator. The indicator is a count of positive single health indicators
selected. Therefore, it had a range from 0 to 3.
2) Objective Health Status indicator that included: Blood pressure, Body Mass Index,
Diet indicators (red meat and soda consumption). The method to select these
62


variables was the same as described above. The Objective Health Status indicator
was a count of positive single indicators and had a range from 0 to 4. For each
dichotomous dependent variable, a set of predictors was added successively to the
model [see Table 3.6):
Table 3.6. Description of model building for Aims 3 and 4.
Models Dependet Migrant vs. Non Migrant it variables Current Migrant vs. Return Migrant
1 Subjective Health Status Subjective Health Status
2 Objective Health Status Objective Health Status
3 Cumulative Time in US
3 and 4 Age, education, marital status, perceived and objective SES Age, objective SES.
4 and 5 Economic, personal, social, normative motivations to migrate Economic, personal, social, normative motivations to migrate
5 and 6 Presence of non-family relationships in US, number of family members in US Presence of non-family relationships in US
Models with interactions Interactions were tested between health indicators and sociodemographic characteristics and migrant network indicators. Interactions were also tested between sociodemographic characteristics and migrant network indicators. Interactions were tested between health indicators and: sociodemographic characteristics, time in the US, and presence of relationships in the US. Interactions were also tested between sociodemographic characteristics, time in the US, and presence of relationships in the US.
For each dependent variable, the full model against model 5 or 6 (with
motivations to migrate) was compared to test the hypotheses guiding this study. Model
fit measures between models with motivations to migrate and models with network
variables were compared. The size of the effect that health status indicators had in the
dependent variables from model 5 or 6 was compared, accordingly. Wald tests were
used to determine if the coefficients of main effects were significantly different among
these models.
63


Individual effects of predictors as well as overall model fit measures were used
for interpretation of results. For each model -2LL, Chi square, Hosmer and Lemeshow
goodness of fit test, a pseudo R square measure (Nagelkerke), and overall percentage of
correct classification of dependent variable are described. For each predictor
unstandardized coefficients, standard errors, and odds ratios with confidence intervals
are shown.
When variables were continuous with different ranges, coefficients were
standardized to allow for their interpretation. Continuous variables that were part of a
significant interaction were centered to facilitate interpretation of interaction terms.
Most statistical analyses were conducted with PASW Statistics 18. Wald tests were
conducted using STATA 11.
Qualitative Component
The qualitative component of the study sought to contribute to our
understanding of the context in which labor migration takes place by describing the role
of health and migrant networks on the decision-making process for those who stay in
MX, those who migrate and for those who recruit, connect, and hire migrants. This
information complements the quantitative findings that would test two plausible ways
in which selection by health of immigrants might take place. The aim of the qualitative
component of the study was:
To understand the role of health and migrant networks on making labor migration
possible from the perspective of those who participate or have participated in labor
migration to the United States; as well as from the perspective of recruiters,
contractors, and experienced migrants who might have helped friends and family to
participate in migration.
64


Sample and Participants
This was a qualitative explanatory study since it looked to identify specific
forces contributing to the occurrence of a phenomenon. The sampling was purposeful
because information-rich cases were strategically selected based on the study purpose
(Patton, 2002). More specifically, a maximum variation strategy was followed to recruit
the sample of migrants. This strategy allows capturing central themes and common
patterns across great variation of cases, which might vary on age, life stage, household
characteristics, migration history, and family migration history. A varied sample of
migrant men, based on different background characteristics in Colorado was obtained.
In Mexico, a chain referral technique was used to find participants.
The snowball sampling technique was also used to recruit key informants. At
the end of each of the interviews conducted with migrant men, whether in Mexico or
Colorado, they were asked to recommend people who were involved on the recruiting
or hiring for this labor opportunity to be reached later by the researcher. All references
lead to one person: the contractor of the program. Therefore, only two interviews were
conducted with key informants (the contractor for the program and a long-time
employee who is in charge of taking care of the facilities where workers reside in
Colorado).
A total of 22 Mexican men, 6 in Mexico and 14 in Colorado, and two key
informants were interviewed. Saturation was reached for the interviews conducted in
Colorado, fieldwork in Mexico could have continued, however the researcher decided to
stop it. The intention was to achieve maximum variation for the sample but the
fieldwork was stopped before that could be achieved (see Table 5.1, Chapter 5). At the
time fieldwork was conducted, that geographic area was suffering from a wave of
violence, this would make potential participants highly suspicious of talking to people
65


they did not know. Further, the researcher determined it was not a safe environment to
continue fieldwork.
Instruments
Two versions of a semi-structured interview guide were designed for the study:
one version for migrants in Colorado and return-migrants in Mexico; and a second one
for key informants (see Appendix D). The two versions collected basic background
information that included: age, birth place, household composition, individual and
family migratory history.
By asking participants to tell their migration stories, the interviewer explored
the elements that played a role in the decision-making process before the migration
move; and the role of health (strength, endurance) that participants perceived for
themselves and for others in their migration decision. Key informants were asked about
the migration sponsorship process in the community. Specifically how the recruitment,
screening, and selection process take place, why and how they decided to recruit in a
particular community and if health (based on appearance or formal screenings) was
considered when making a decision to recruit/hire an individual to migrate. All
migrants or return migrants interviewed were part of a migrant network. Therefore, all
of them, particularly experienced migrants, were in way key informants as to how the
sponsorship process of migration works. The qualitative approach allowed exploring
possible selection criteria used by the migrant network when sponsoring new labor
migrants.
Procedure
All potential participants were informed that the study had the purpose of
understanding how men make a decision to migrate for work to the US. Potential
participants were asked for an interview at their convenience. All of the interviews in
66


Mexico and most of the interviews in Colorado were conducted by the researcher in
Spanish. Only one of the interviewers in Colorado conducted some semi-structured
interviews. It was explained to potential respondents that their participation in this
study was absolutely voluntary, and the content of the interview was described as well
as how the information collected was going to be treated. They were told that the
interview would take from 30 to 60 minutes. The duration of the interviews ranged
from 15 to 30 minutes. The interviews with the contractor and the person in charge of
taking care of migrant housing took 30 minutes each. Two interviews with migrant
workers took also 30 minutes. All participants were asked for permission to audio
record the interviews. Recorded interviews were transcribed. One interview in Mexico
was not recorded because the participant did not grant permission. The researcher
made notes as the interview progressed and completed notes right after the interview
was finished.
Because of the recruiting technique used with key informants, the researcher
could not assure full anonymity but she guaranteed strict confidentiality. Recordings
and transcriptions of interviews were absolutely de-identified. Informed consent was
obtained from all participants. Approval of these procedures was obtained from
COMIRB and the states Ministry of Health in Mexico (see Appendix B).
Migrants in Colorado were contacted in person during evenings at their places
of residence while the survey phase of the study was also taking place. Return-migrants
residing in Mexican communities were recruited using chain referral. Once an interview
was conducted, the researcher will ask for a recommendation of another man who had
experience working in the US. The researcher visited four towns in the municipality of
origin of most migrant men included in the survey.
67


Data analyses
Content analyses with general categories established a priori (Hsieh & Shannon,
2005) was used. The first general category was decision-making process or what
motivate men to migrate. The second category was the role of health on men's decision
to migrate or in general as a requirement for migration. The third category was how
they were connected to this labor program; this category covers the role of the migrant
network on their migration. Finally, it was explored what participants perceived as
requirements to be sponsored for migration or what characteristics they would
consider when sponsoring friends or family. Information provided by the contractor is
described separately. Information provided by the second key informant (man in
charge of housing administration) was described along with other migrants.
The objective of the content analyses was to identify the categories mentioned
above in the interviews transcripts. First, all the transcripts were read to identify all
potential occurrences of the categories described. On a second reading of transcripts,
coding of highlighted text took place. Once categories were identified; information
regarding each of the categories was summarized and examples are provided.
Qualitative findings support, complement and also challenged both, theory and
quantitative findings and also provided insight about when migrant selection might take
place. Triangulation was reached through the use of different methods
(quantitative/qualitative); and theories (contrasting two theoretical approaches to
explain migration) (Berg, 2001; Patton, 2002). Qualitative results are described in
Chapter 5 and discussed along with quantitative findings in Chapter 6.
68


CHAPTER 4
QUANTITATIVE RESULTS
Descriptive Results
Descriptive results are presented for non-migrants (men interviewed in Mexico
with no previous US migratory experience); return-migrants (men interviewed in
Mexico who had US migratory experience); and migrants (men interviewed in Colorado
participating in an agricultural labor program).
Sociodemographic Characteristics of Ail Participants in Quantitative Study
Men in the sample were young (X=32.03, SD=9.7), the youngest men were those
with no migration experience (X=28.5). They had 6.35(SD=3.03) average years of
education, slightly above completed elementary school (six years) (see Table 4.1). Most
participants were married (77.9%) at the time of the survey and had children (73.4%).
The average number of children was 2.6(SD=1.64) and there was no difference between
men interviewed in Mexico or Colorado. About 25% of men in Mexico were
unemployed at the time of the interview. The majority of employed men in Mexico had
a job in agriculture. Most of them were living in the same communities where they were
born (95%, data not shown in Table) and 98% of these communities were villages
(87%) or small towns (11%).
Migrant men reported a significantly (p=.000) higher number of assets (X=10.7,
SD, 3.15) than non-migrant (X=8.91, SD=3.48) or return-migrant men (X=8.94, SD=2.29)
(shown in Table 4.1). A socioeconomic ladder was used to evaluate participants
perception of the living conditions of their families when compared to other families in
their hometown. Migrant men in Colorado were asked to make this comparison against
other families in their home in Mexico. Most men, whether in Mexico or Colorado,
placed their families in the middle of the ladder. Even when the differences among the
69


groups are not significant, migrant men tend to perceive their familys conditions lower
than non-migrant and return-migrant men. It was also asked how successful they
thought their family was compared to others: although differences are not significant,
more non-migrant men (11.76%) reported their families were more successful than
other families compared to return-migrant (6.62%) or migrant men (10%).
Health Status Indicators of All Participants in Quantitative Study
Subjective Indicators
The inclusion of subjective indicators was considered crucial to test the health
selection hypotheses of this study. These health indicators reflect private behaviors or
perceptions, which most likely are only known by the individual or his close
relationships. The health indicators considered subjective were depressive
symptomatology, self-rated general health, relative physical health, presence of chronic
diseases, and having been to the doctor in the previous 6 months (see Table 4.2). A
detailed definition of these variables can be found in Table 3.3 described in Chapter 3.
70


Table 4.1. Sociodemographic Characteristics of All Participants in Quantitative Study
(N=384)
Non Migrant (n=68) X(SD) Return Migrant (n=152) XfSD) Migrant (n=164) XfSD) Total (n=384) XfSD) P1
Age 28.53(10.04) 35.04(9.00) 30.68(9.44) 32.03(9.7) 0.0003
18-29 61.76% 31.58% 50.00% 44.79%
30-39 17.65% 32.24% 31.71% 29.43%
40 and over 20.59% 36.18% 18.29% 25.78% 0.000
Years of education 6.59(3.39) 5.77(2.76) 6.79(3.06) 6.35(3.03) 0.010b
Married 50.00% 85.53% 82.31% 77.86% 0.000c
Years of marriage 10.81(8.8) 13.98(8.5) 10.10(8.07) 12.14(8.57) 0.003
Children 47.06% 83.55% 74.85% 73.37% 0.000c
Number of children 2.61(1.91) 2.64(1.63) 2.5(1.59) 2.58(1.64) 0.789
Employed 75.00% 73.68% 100% 86.50% 0.000d
Objective SES2 8.91(3.48) 8.94(2.29) 10.72(3.15) 9.69(3.37) 0.000d
Perceived SES3 4.38(1.94) 3.97(1.50) 3.84(1.76) 3.99(1.7) 0.092
Compared to other 11.76% 6.62% 10.00% 8.97% 0.391
families in town, their
family in better is more
successful____________________________________________________________________________________
1 Statistical significance of ANOVAs for mean comparisons among three groups and Pearson Chi
Square tests for categorical variables;2 Number of assets reported by participants (range 0 to
24);3 Subjective comparison with other families in their hometown, position in ladder ranges
from 1 to 10. Significant differences between:a return-migrant vs. non-migrant and migrant;b
return-migrant vs. migrant;c return-migrant vs. non-migrant and migrant; d migrant vs. non-
migrant and return-migrant.
Participants were asked to rate their health with the question: In general how
would you rate your health? To this question, they could answer excellent, very good,
good, fair or poor. About 60% of men in the sample considered their general health
good, very good or excellent (see Table 4.2). When men were asked how they would
rate their physical health relative to other men their age, results were different:
compared to non-migrants (29.85%) and return-migrants (29.33%), significantly more
71


migrant men reported that their physical health was better than other men their age
(48.15%).
Table 4.2. Subjective Health Indicators (N=384)
Non migrant (n=68) XfSD) Return migrant (n=152) XfSD) Migrant (n=164) XfSD) Total (n=384) XfSD) p1
At least good general health (GH) 66.18% 59.21% 59.15% 60.42% 0.564
Better physical health compared to others their age 29.85% 29.33% 48.15% 37.47% 0.001a
Absence of chronic disease 70.59% 66.45% 66.46% 67.19% 0.805
Depressive symptomatology score 8.50(6.37) 8.47(5.92) 7.07(7.2) 7.87(6.6) 0.118
Score 11 or below 67.16% 69.33% 81.71% 74.28% 0.015*
Visited doctor at least once in previous 6 months 55.88% 54.61% 37.80% 47.66% 0.004*
1 Statistical significance of ANOVAs for mean comparisons among three groups and Pearson Chi
Square tests for categorical variables.* Significant differences found for migrant vs. non-migrant
and vs. return-migrants.
Men were asked if they had been diagnosed with a chronic disease by a health
professional. About 33% of men in the sample reported at least one chronic disease
diagnosis from a health provider. A higher proportion of non-migrant men did not
report the presence of chronic conditions (70.59%) compared to either Return migrants
(66.45%) or Migrants (66.46%); however this difference is not significant (see Table
4.2). Table 4.3 shows proportions of the presence of several chronic diseases by
migrant status. Hypertension, back problems and diabetes were the most frequent
chronic conditions in the sample. Hypertension was more common among migrant
men, although this difference was not significant (see Table 4.3).
72


Table 4.3. Presence of Chronic Conditions
Chronic Conditions1 Non Return Migrant Total P2
migrant fn=68] migrant fn=152] (n=164) fn=3841
Hypertension 8.82% 10.53% 17.79% 13.32% 0.080a
Diabetes 4.41% 12.50% 3.05% 7.03% 0.003b
Heart Attack 2.94% 2.67% 0.61% 1.83% 0.300
Arthritis 4.41% 4.61% 3.05% 3.91% 0.754
Back problems 10.29% 8.61% 10.43% 9.69% 0.847
Asthma, chronic bronchitis or 1.47% 3.29% 1.22% 2.08% 0.405
emphysema Ulcers, colitis, enteritis 2.94% 4.61% 9.20% 6.27% 0.112
1 Participants were told by a health professional that they had the condition,2 Statistical
significance of Pearson Chi Square tests. Significant differences found between:a migrants vs.
non-migrants and vs. return-migrants; b return-migrants vs. non-migrants and vs. migrants.
Migrant men reported significantly less depressive symptoms (X=7.07, SD=7.2),
than men in Mexico (See Table 4.2). The CESD scores for this sample were low in
general. The median distribution of the sample as a whole (11) was used as a cut point
to divide the sample in two groups: low and high depressive symptomatology. A
significantly smaller proportion of migrant men had scores above the median value as
shown in Table 4.2. About half of men in the sample reported they saw a doctor in the
six months previous to the study. This behavior was significantly less frequent among
migrant men than among non-migrant or return-migrant men (see Table 4.2).
For further analyses, a selection of the health indicators described above were
aggregated into a single indicator named Subjective Health Status (SHS) as a count of
positive individual health variables. This global indicator is used in analyses described
further in this chapter.
Objective Indicators
Some health indicators were defined as objective whether because they were
based on measurements conducted by interviewers (height, weight, waist and hip
73


circumference, blood pressure, heart rate, and a fitness indicator); or because they were
the report of observable behavior like smoking, drinking, physical activity, diet, and
missed days of work. Table 3.4 shows detailed definitions of these variables in Chapter
3.
Table 4.4 shows descriptive information of the objective health indicators by
migratory status. Measures of blood pressure were high for all men in the sample. Only
39% of non-migrant men had blood pressure measures of 120/80 or below; the
proportion of normal blood pressure readings is even lower for return-migrants
(17.36%) or migrant-men (11.18%).
Body Mass Index (BMI) was estimated from height and weight measures (see
Table 3.4 for definition, Chapter 3). The average BMI for all men in the sample was
27.27(SD=4.7). This indicates that on average this group of men is overweight, with the
highest BMI found among migrant men and the lowest among non-migrant men. All
groups had an average BMI over 25 (overweight category is defined by BMIs from 25 to
30) and 35.4% of migrant-men had a BMI of 30 and over (indicator of obesity).
Waist and hip circumference were measured to estimate Waist to Hip ratio.
Although the differences were not as striking as with the BMI indicator, a significantly
smaller group of Migrant men (27.89%) had non-risk Waist to Hip ratio (.95 or below)
compared to non-migrants (47.37%) or return-migrants (33.58%).
74


Table 4.4. Objective Health Indicators (N=384)
Non migrant (n=68) X(SD) Return migrant (n=152) X(SD) Migrant (n=164) X(SD) Total (n=384) X(SD) P1
Systolic 125.61(12.4) 129.98(13.9) 135.81(15.0) 131.76(14.6) 0.000=
Diastolic 76.58(9.5) 78.66(10.6) 82.19(13.18) 79.84(11.8) 0.002=
Blood pressure 120/80 or below 39.06% 17.36% 11.18% 18.43% 0.000=
Blood pressure 140/90 and over 14.06% 22.22% 39.13% 28.18% 0.000b
Body Mass Index (BMI) 25.58(4.3) 27.35(4.6) 28.37(4.7) 27.47(4.7) 0.000'
Low BMI 4.48% 0.00% 0.00% 0.80%
Normal 37.31% 33.56% 22.98% 29.71%
Overweight 50.75% 36.24% 41.61% 41.11%
Obese 7.46% 30.20% 35.40% 28.38% 0.000b
Waist to Hip Ratio ,96(.06) ,98(.07) ,97(.05) ,97(.06) 0.212
W to H Ratio .95 or below 47.37% 33.58% 27.89% 33.43% 0.030b
Heart Rate 78.16(10.9) 77.17(12.8) 72.80(10.7) 75.47(11.8) 0.001=
Average or better for age 29.70% 40.00% 48.40% 41.80% 0.032b
Distance (kms.) .133(.04) ,127(.04) .190(.07) ,157(.06) 0.000=
1 Statistical significance of ANOVAs for mean comparisons among three groups and Pearson Chi Square tests for categorical variables. Significant differences between:a migrant vs. non-migrant and vs. return-migrant;b non-migrant vs. migrant;c non-
migrant vs. return-migrant and vs. migrant.
75


Table 4.4. (Cont). Objective Health Indicators (N=384)
Non Return migrant Migrant
migrant (n=68) (n=152) (n=164) Total (n=384) P1
Vigorous activity (hrs per week) 19.54(18.0) 19.09(19.6) 17.80(23.24) 18.62(20.95) 0.798
Moderate activity (hrs per week) 8.76(9.2) 11.5(14.0) 23.23(23.4) 16.14(19.18) 0.000*
Walking (hrs per week) 6.86(10.5) 9.33(12.5) 26.06(23.92) 16.04(20.01) 0.000*
Seating (hrs per week) 12.82(9.9) 9.96(9.9) 21.47(16.6) 15.77(14.4) 0.000*
Physical activity (hrs per week)** 23.39(30.3) 32.09(34.3) 45.31(41.8) 36.80(38.3) 0.000*
Currently Smoking 29.41% 38.16% 30.67% 33.42% 0.276
Currently Drinking 61.76% 65.13% 72.05% 67.45% 0.232
Had 5 drinks at least once in last month 61.29% 65.28% 36.04% 49.53% 0.000*
Number of sugary drinks per day 1.84(1.4) 1.60(1.3) 2.53(2.3) 2.04(1.9) 0.000*
Pieces/portions of fruit per day 1.61(1.2) 1.77(1.2) 2.1(1.5) 1.88(1.3) 0.018b
Days p/week has red meat 2.33(1.7) 2.09(1.5) 4.35(1.8) 3.09(2) 0.000*
Missed work at least once previous 6 months 10.29% 13.82% 7.93% 10.68% 0.237
Missed work due to illness at least once previous 6 months 7.35% 11.84% 3.05% 7.29% 0.011c
1 Statistical significance of ANOVAs for mean comparisons among three groups and Pearson Chi Square tests for categorical
variables. Significant differences between:a migrant vs. non-migrant and vs. return-migrant;b non-migrant vs. migrant;c
migrant vs. return-migrant.
76


Heart Rate was measured using an electronic blood pressure cuff. More migrant
men (48.4%) had heart rates better than average for their age (see definition in Table
3.4, Chapter 3) compared to men interviewed in Mexico (non-migrant: 29.75% or
return-migrant: 40%). Men in the sample participated on a fitness test: They were
asked to walk as fast as they could for two minutes, and the distance covered was
measured. Migrant men walked significantly longer distances (.190 kms) than non-
migrant (.133 kms) or return-migrant men (.127 kms). These objective measures
suggest migrant men were fitter than men interviewed in Mexico. However, it was
observed during fieldwork that the fitness measure might be more a reflection of how
well men followed our instruction "walk as fast as you can" than an indicator of their
actual fitness level. In general, older men seemed to follow the instruction better.
Because men seemed to follow instructions differently, the outcome of this test is
described in this section but wont be used in further analyses.
The questionnaire included a series of questions to assess current physical
activity. Men reported how many hours per week and per day they would spend on
vigorous (activity that felt difficult and made them breathe hard) and moderate physical
activity; walking continuously for at least 10 minutes, or seating. A measure of hours
per week is described in Table 4.4. Migrant men reported more hours spent on
moderate physical activity and walking but also reported more hours spent seating. An
overall measure of physical activity that added time spent on activity and subtracted
seating shows that even after discounting time spent seating. Migrant men spent
significantly more hours per week doing any kind of physical activity (45 hours per
week) compared to either return-migrants (32 hours) or non-migrants (23 hours).
Participants reported all activities during their day, including work. Therefore, the
higher number of hours reported by migrant men is very likely a reflection of longer
work hours. Because of the way this measure was obtained, this result might be more a
reflection of the immediate circumstances (longer work hours) of migrant men and not
77


necessarily a health behavior characteristic of this group. Therefore, this measure wont
be included in future analyses.
Thirty-three percent of participants reported being current smokers, defined as
having smoked at least 100 cigarettes in their lives and were smoking at the time of the
survey. The higher proportion of smokers was found among return-migrant men and
the lowest among non-migrant men but the difference is not significant. About two
thirds of participants reported being current drinkers. The higher percentage of
drinkers was found among migrant men and the lowest among non-migrant men but
the difference is not significant. Drinking alcohol is a common behavior among Mexican
men, therefore it was assessed how frequently 'heavy drinking (5 drinks or more) took
place among this men. The prevalence of heavy drinking in the previous month to the
survey was the lowest among migrant men (36%), and the highest among return-
migrants (65%) and this difference was statistically significant.
Sugary drinks (i.e. juice, sodas, sport drinks), fruit and red meat consumption
were evaluated for participants. Migrant men reported an average of 2.5 sugary drinks
per day, amount significantly higher than either non-migrant (X=1.84) or return-
migrants (X=1.6). When asked about their fruit and red meat consumption, the same
pattern was found. Consumption of red meat among migrant men is remarkably higher:
Migrants reported they eat red meat about 4 days per week, compared to two days per
week reported by non-migrant and return-migrant men.
Men were asked to report number of missed days of work in the previous 6
months to the survey, and how many of those days were missed due to health reasons.
Most participants did not miss work very often. In general, about 11% of participants
reported they had missed work in the 6 months previous to the study, and even a
smaller proportion reported they missed work for health reasons. However, differences
by migrant status were apparent: a significantly higher rate of return-migrant men
78


reported they missed work due to health reasons (11.8%), compared to non-migrant
(7.3%) or migrant men (3.05%).
For further analyses, these health indicators were aggregated into a single
indicator named positive Objective Health Status (OHS), by counting the occurrence of
selected positive individual health status variables.
Migratory History and Motivations to Migrate of All Participants in Quantitative Study
Participants were asked about their previous history of migration from Mexico
to the US (See Table 4.5). Men in the sample had an average of 3.5 trips to this country.
Migrant men reported more trips (4.7) compared to return-migrants (3.7). However,
migrant men had spent less cumulative time in the US than return-migrants (28 vs. 40
months respectively).
Return-migrants are men who were interviewed in Mexico but reported
previous experience travelling to this country for work. It is not known if these men
were retired or they were still participating in labor migration but missed the year
when the survey was conducted. The higher number of trips reported by migrant men
in this sample reflects the temporary circular migration pattern that was possible by
their participation in a documented agricultural labor program.5
5 Due to increased costs and risks, undocumented migration results in longer stays in this
country (Leite, Angoa, Rodriguez, 2009).
79


Table 4.5. Migratory History and Motivations for Labor Migration (N=384)
Non migrant (n=68) XfSD) Return migrant (n=152) X(SD) Migrant (n=164) X(SD) Total (n=384) XfSD) P2
Number of trips NA 3.71(3.4) 4.7(3.1) 3.49(3.4) 0.000
Cumulative time in the US (months) NA 40.10(45.42) 28.33(26.22) 33.91(37.01) 0.005
Have spent 24 months or more in the US NA 50.68% 40.74% 45.45%
Likelihood of working in the US following year1 Motivations to migrate 1.23(2.5) 3.99(3.6) 5.37(2.3) 4.33(3.2) 0.0003
Economic 7.26(4.3) 8.68(2.9) 7.76(1.4) 8.04(2.8) 0.001b
Personal 7.00(4.3) 8.67(2.9) 8.63(1.7) 8.36(2.8) 0.000c
Social 6.72(4.2) 8.19(3.1) 7.21(2.4) 7.51(3.1) 0.002b
Norm 5.37(4.5) 6.63(3.8) 7.89(2.7) 6.95(3.6) 0.000a
Opportunity Participated in CO labor 7.38(4.3) 8.80(3.0) 9.58(1.3) 8.88(2.9) 0.000a
program Invited to work in the US 0.00% 7.46% 97.56% 55.92% 0.000d
by: (n=6) (n=36) (n=135) (n=177)
Contractor/Majordomo 0.00% 8.33% 14.81% 12.99%
Relatives/friends 66.67% 47.22% 70.37% 65.54%
Nuclear family 0.00% 11.11% 12.59% 11.86%
Self 33.33% 33.33% 2.22% 9.60% 0.000d
1 Answers were 0=not likely at all to 10=extremely likely;2 Statistical significance of ANOVAs for
mean comparisons among three groups and Pearson Chi Square tests for categorical variables.
Significant differences between:a migrant vs. non-migrant and vs. return-migrant, and non-
migrant vs. return-migrant; b return-migrant vs. non-migrant and vs. migrant;c non-migrant vs.
return-migrant and vs. migrant;d migrant vs. non-migrant and vs. return-migrant
Participants were asked to rate how likely they were to participate in labor
migration the following year. This question was asked in a scale of 0 to 10 where 0
represented not likely at all and 10 represented extremely likely. Migrant men
80


considered a future trip significantly more likely than return-migrant or non-migrant
men.
Motivations to migrate were evaluated using a scale developed specifically for
this study (see Chapter 3 for detailed description). The items were grouped in five
dimensions that describe different types of motivations to migrate: Economic, personal,
social, normative, and opportunity. Migrant men were asked how important different
factors were to determine their decision to migrate for work and non-migrant or return-
migrant, how important different factors would be to determine their decision to
migrate for work. The scores are presented in a scale from 0 to 10, where 0 represent
those motivations were not important to make the decision; and 10 represents they
were very important. All types of motivations were considered important, with high
scores for most men in the sample. However, significant differences among the groups
for each of those factors were also observed (see Table 4.5). For example, for migrant
men, the factor scored the highest was opportunity (i.e. they were invited to a program;
offered the possibility of documents). Because this motivation was considered
redundant to being invited to the program, this dimension was eliminated of the
motivations to migrate from further analyses. No other differences were found in the
scores of motivations to migrate when comparing by age, marital status, education, SES
or time spent in the US (data not shown).
Almost all migrant men were formal participants in the labor program from
Guanajuato to Colorado; a small group of return-migrants had participated in the
program before the survey was conducted. Participants were asked if they had been
invited to participate in the labor program and who extended that invitation. One
hundred and seventy-seven participants reported they received an invitation to be part
of the Colorado agricultural labor program. Most of them were 'invited' or 'made aware'
by relatives and friends (65%), the contractor directly (13%), nuclear family (12%), and
a few of them expressed they found out about the labor opportunity by themselves
81


(9.6%). Even a few non-migrant men (6) reported they had received an invitation to
sign up for the program.
Migrant and Social Networks of All Participants in Quantitative Study
The migrant network is defined as a "set of interpersonal ties that connect
migrants, former migrants, and non-migrants in origin and destination areas through
ties of kinship, friendship, and shared community origin" (Massey et al., 1998, p 42).
The presence of a migrant network defined in reference to the Mexico-Colorado
migration stream was evaluated. It was also assessed the size, resources and US
experience of the social network formed by non-family important relationships and by
family.
Participants reported the number of people they knew who had agricultural
labor experience in Colorado and how many of those were family members. Since labor
migration to Colorado from the state of Guanajuato is very common, all men in the
sample knew people working in agriculture in Colorado. Migrants reported in average a
higher number of people they knew working in agriculture in Colorado (42) compared
to non- migrants (8) or return- migrants (12). However, the percentage of these people
who were family members was the largest for non-migrants (40%) compared to either
migrants (27%) or return-migrants (23%) (See Table 4.6).
Participants were asked to list relationships that were not family who they
relied on to talk about important matters, financial issues, or child rearing issues. Most
men stated they would usually talk about these issues with their wives if they were
married or with their father or grandfather if they were single. Most men reported zero
relationships (82.9%); 15.9% reported one relationship; and 1.6% reported 2
relationships.
82


Data shown in Table 4.6 shows averages of a count of the relationships
mentioned by men in the sample. For each of the relationships listed, they were asked
to mention the resources each of these persons had (e.g. own a house, have stable
employment, own a car, own land); if they worked in the US; provided help getting a job;
and provided economic help. As described in Table 4.6, migrant men reported a
significantly larger and resourceful network. In average, migrant men had more
relationships working in the US, more relationships who helped them economically; and
more relationships with resources (i.e. they were employed, owned a car and/or land).
The size of participants family, their place of residence (same town or other
place in Mexico or in the US), frequency of contact with them, and how often family
members were a source of advice, if they have provided money to the interviewee, or if
the interviewee had provided any of them with money were also assessed. Again,
migrant men seemed to report in average significantly larger families. However, they
did not report a larger number of family members residing in the US. Migrant men
reported they had asked for advice from significantly more members of their family
network; and they had given money to more family members than other men in the
sample.
83


Table 4.6. Migrant and Family Networks (N=384]
Non Return Migrant Total
migrant migrant
(n=68) (n=152) (n=164] (n=384]
XfSD] X(SD] XfSD] XfSD] P4
Number of people known in CO 8.13(9.9] 12.6(20.2] 42.8(59.9] 31.4(51.2] 0.000a
% of family members 40.0(41.7] 23.43(36.5] 27.1(37.1] 26.96(37.3] 0.304
Number of important non- family relationships1 .26(.5] ,12(.3] .51 (.6] .31 (-5] 0.000a
Number of relationship working in US1 ,13(.38] ,07(.26] 32(.5] .19(.4] 0.000a
Number of relationships helped economically1 ,20(.5] .09(.3] .35(.6] ,22(.5] 0.000b
Number of relationships with resources12 ,29(.9] ,27(.8] .94(1.4] .56(1.2] 0.000a
Number of family members1 5.56(2.7] 6.73(2.9] 7.49(3.3] 6.80(3.1] 0.000c
Family members living in US1 .39(1.1] .70(1.2] .48(1.2] .56(1.2] 0.140
Has asked advice from a 2.01(2.0] 2.7(3.1] 3.22(2.9] 2.78(2.9] 0.016b
family member12'3 Has received money from a family member3 2.55(2.8] 2.21(3.0] 2.55(2.8] 2.41(2.9] 0.545
Has given money to a family member3 1.95(2.1] 2.4(3.22] 3.5(3.1] 2.76(3.1] 0.000a
1 Count of all relationships with those characteristics;2 Resources include house, car, stable
employment and/or land;3 Number of family members that participants received advice,
money or gave money; 4 Statistical significance of ANOVAs for mean comparisons among three
groups and Pearson Chi Square tests for categorical variables. Significant differences between:a
migrant vs. non-migrant and vs. return-migrant; b migrant vs. return-migrant; c non-migrant vs.
return-migrant and vs. migrant.
In summary, descriptive results indicate that migrant men have higher rates of
positive subjective health indicators. On the other hand, migrant men in general had
higher rates of negative objective health indicators compared to other men in the
sample. Exception to this pattern are health behaviors like smoking or drinking,
although differences were not significant and fitness measures, like the walking test.
Compared to non-migrants and return-migrants, migrant men seemed to have larger,
more resourceful family and social networks.
84


Testing the presence of a healthy migrant effect
Aim 2. To determine the extent to which being a migrant in Colorado is predicted by a
positive health status.
The hypotheses are stated as follows:
Hmeo There is no association between positive health status and being part of the
migrant group.
Hmei Positive health status is associated with being part of the migrant group.
To meet this objective two sets of logistic regression models were built. The
dependent variable for the first set of models was a dichotomy between being a migrant
(interviewed in Colorado) and not being a migrant (never been to the US). For the
second set of models, the dependent variable was a dichotomy between being a migrant
(interviewed in Colorado) and being a return-migrant (interviewed in Mexico but with
US work experience). The adjusted models for the second dependent variable included
different, theoretically important variables (i.e. cumulative time in US), which precluded
the use of multinomial logistic regression. A selection of health status indicators
described above was included in the logistic regression analyses.
To test the presence of a healthy migrant effect, the multivariate analyses
included indicators of subjective and objective health status. As I describe in greater
detail below, measures of subjective health operated in the direction predicted by the
healthy migrant hypothesis, migrants who perceived themselves in good health were
more likely to migrate than to never migrate. However, indicators of objective health
status worked in the opposite direction, suggesting that adults who have stayed out of
the migration stream had better directly observable physical health. These findings are
interpreted in Chapter 6.
85


Dependent variable: Migrants vs. Non-Migrants
Table 4.7 describes the independent variables and covariates included in the
logistic regression models to predict being a migrant vs. non-migrant. Variables
described with means and standard deviations were included in the model as
continuous and variables described with percentages were included as a dichotomy (0-
1). For these variables, the percentage shown in the table represents the group with
value 1 in the models.
Table 4.8 describes the set of logistic regression models to predict being a
migrant vs. non migrant. In Model 1 all subjective health indicators (SHS) were included
in the model as a block. Each of these indicators significantly predicted being a migrant
independently and as a whole they significantly contributed to the explanatory power of
the model as reflected by the model fit measures. Of notice, rating health as at least
good decreased the likelihood of being a migrant. The rest of SHS indicators
significantly increased the likelihood of being a migrant.
86