Content uploaded by Dr. Carlos Siordia
Author content
All content in this area was uploaded by Dr. Carlos Siordia on Dec 17, 2014
Content may be subject to copyright.
MEXICAN-ORIGIN INTERREGIONAL MIGRATION FROM THE SOUTHWEST:
HUMAN, HOUSEHOLD, AND COMMUNITY CAPITAL HYPOTHESES
A Thesis
by
CARLOS SIORDIA
Submitted to the Office of Graduate Studies of
Texas A&M University
in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE
May 2009
Major Subject: Sociology
MEXICAN-ORIGIN INTERREGIONAL MIGRATION FROM THE SOUTHWEST:
HUMAN, HOUSEHOLD, AND COMMUNITY CAPITAL HYPOTHESES
A Thesis
by
CARLOS SIORDIA
Submitted to the Office of Graduate Studies of
Texas A&M University
in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE
Approved by:
Chair of Committee, Rogelio Saenz
Committee Members, Dudley L. Poston
Marco Portales
Head of Department, Mark Fossett
May 2009
Major Subject: Sociology
iii
ABSTRACT
Mexican-origin Interregional Migration from the Southwest:
Human, Household, and Community Capital Hypotheses. (May 2009)
Carlos Siordia, B. S., University of Texas-Pan American
Chair of Advisory Committee: Dr. Rogelio Saenz
This research addresses the question of what factors lead Mexican-origin
individuals living in the U.S. to seek a new residence outside their Southwestern state of
residence. The analysis examines three hypotheses: (1) the human capital hypothesis
that college graduates have higher odds of migrating out of the core region than those
with less than a high school education; (2) the household social capital hypothesis that
posits that the presence of a household member born outside the core increases the odds
of migration; and (3) the community social capital hypothesis which states that
householders residing in an area with community social capital will have higher odds of
leaving the core than those living in areas with no community social capital.
These hypotheses are investigated using three models: (1) a full model that
includes both native- and foreign-born Mexican-origin householders; (2) a native-born
model which includes only native-born Southwest householders; and (3) a foreign-born
models that includes only foreign-born Mexican-origin householders. By using the
Saenzian region-concepts of core, periphery, and frontier, I find: (1) limited support for
iv
the human capital hypothesis; (2) consistent support for the household social capital; and
(3) no support for the community social capital.
The analysis is important to sociological theory and demography because it
specifically endeavors to explain how the connections between three kinds of capital—
human, household, and community—shape the decision to leave the Southwest for other
regions of the country. By computing statistical and theoretical particulars, the thesis
ascertains that migration-selectivity theories regarding the general population are useful
in theorizing Mexican-origin interregional migration. Findings expand existing
sociological literature by theorizing how human, household, and community capital
operate under the Saenzian regions to shape the interregional migration of the growing
Mexican-origin population of the U.S.
v
TABLE OF CONTENTS
Page
ABSTRACT .............................................................................................................. iii
TABLE OF CONTENTS .......................................................................................... v
LIST OF FIGURES ................................................................................................... vii
LIST OF TABLES .................................................................................................... vii
CHAPTER
I INTRODUCTION ............................................................................... 1
II LITERATURE REVIEW ..................................................................... 7
Human Ecology, Migration, and Capitals ..................................... 8
Interregional Migration and Saenzian Concepts ............................ 18
Hypotheses ..................................................................................... 29
III METHODOLOGY .............................................................................. 31
Data Source .................................................................................... 31
Sample Selection and Saenzian Regions ........................................ 31
Dependent Variable ........................................................................ 38
Independent Variables .................................................................... 39
Control Variables ........................................................................... 52
Multinomial Logistic Regression ................................................... 54
IV ANALYSIS .......................................................................................... 59
Introduction ................................................................................... 59
Descriptive Statistics ...................................................................... 59
Cumulative vs. Multinomial Model ............................................... 63
Organization of Tables ................................................................... 67
Interpreting Coefficients ................................................................ 68
Multinomial Logistic Model for Full Sample ............................... 69
Multinomial Logistic Model for Native-born Sample ................... 75
Multinomial Logistic Model for Foreign-born Sample .................. 78
vi
IV SUMMARY ......................................................................................... 84
Conclusions .................................................................................... 84
Shortcomings .................................................................................. 88
Future Research .............................................................................. 89
Implications .................................................................................... 91
REFERENCES .......................................................................................................... 94
VITA ......................................................................................................................... 104
vii
LIST OF FIGURES
Page
Figure 1 Percent of Mexican in Continental U.S.
by State and Equal Intervals . ............................................................ #
Figure 2 Percent of Mexicans in Continental U.S. by State:
Used Saenzian Categories ................................................................. #
Figure 3 Community Capital in The Southwest:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
Figure 4 Southwest PUMAs with and without Community Capital:
By 2000 5% PUMA ......................................................................... #
Figure 5 Community Capital in Arizona:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
Figure 6 Community Capital in California:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
Figure 7 Community Capital in Colorado:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
Figure 8 Community Capital in New Mexico:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
Figure 9 Community Capital in Texas:
Percent of Non-core Mexican-origin In-migrants
by 2000 5% PUMA ........................................................................... #
viii
LIST OF TABLES
Page
Table 1 Dependent Variable ........................................................................... #
Table 2 Independent Variables ....................................................................... #
Table 3 Control Variables ............................................................................. #
Table 4 2000 Frontier States ......................................................................... #
Table 5 2000 Periphery States ....................................................................... #
Table 6 2000 Core States ............................................................................... #
Table 7 Descriptive Statistics:
All Mexican-origin Householders (73,824) .. ................................... #
Table 8 Multinomial Results for Full Model .................................................. #
Table 9 Descriptive Statistics:
Native-born Mexican-origin Householders (33,297) ....................... #
Table 10 Multinomial Results for Native-born Model ..................................... #
Table 11 Descriptive Statistics:
Foreign-born Mexican-origin Householders (40,530) ..................... #
Table 12 Multinomial Results for Foreign-born Model ................................... #
1
CHAPTER I
INTRODUCTION
Demographic analyses of the movement of humans extend back over a century
with the work of Ravenstein based on migration in Europe (1876, 1885, 1889). In the
United States, initial interest on migration centered on rural-to-urban migration and the
movement of blacks out of the South (Anonymous 1936; Hamilton 1959; Taeuber &
Taeuber 1938; Thompson & Bogue 1949; Wattenberg 1948). Research on the
movement of Mexican Americans1 began in the 1970's (Alexander 1979; Beale 1973;
Brunton 1973; Cardenas 1976, 1977; Boswell 1979; Estrada 1976; Grebler 1970;
Moore 1970a, 1970 b; Romo 1978; Weeks 1979) and continued in the 1980’s (Arreola
1985; Bean et al. 1988; Garcia 1981; McHugh 1989; Saenz 1989a, 1989b) and the
1990’s (Frey et al. 1998; Saenz et al. 1994, 1997, 1999).
Investigations of the migratory patterns of Mexican-origin people has increased
significantly in recent years (e.g., Borjas 2006; Cai 2007; Crowley et al. 2006; Durand et
al. 2006; Ellis et al. 2006; Fernandez et al. 2007; Foulkes et al. 2000; Frey et al. 2005;
Gilroy 2007; Gouveia et al. 2000; Gurak et al. 2000; Guzman et al. 2002; Hiller et al.
2007; Saenz et al. 2004, 2007; Yankow 2003). Even though studies concentrating on the
migratory behaviors of Mexican-origin individuals have only existed for about three
This thesis follows the style and format of the American Sociological Review
1The umbrella label of "Latino/a" cannot be used interchangeably with Mexican-origin because while all
Mexican Americans are Latinas/os not all Latinos/as are Mexican American. Thus, when the term
Hispanics is used, it refers to Mexican Americans and all other Latin-origin individuals who fall under the
ethnic category.
2
decades, the fact that Latinos have become America’s largest ethnic minority group
(U.S. Census 2006) has recently increased the importance of studying this population.
Despite this research, there is no published study that uses census data from other
years to explore how human, household, and community capital influence the likelihood
of interregional migration in the U.S. Mexican American population. By using network
theories by Fred E. Katz (1966), Mark S. Granovetter (1973), and Charles Tilly and C.
Harold Brown (1967), the thesis seeks to advance migration theory by evaluating how
individual level attributes and social contacts may influence the likelihood of out
migrating from the Southwest (Arizona, California, Colorado, New Mexico, and Texas)
into other U.S. regions.
Advancing sociological knowledge by studying how the different forms of social
capital relate to the migratory behavior of Mexican-origin individuals will help
sociologists develop an understanding of how they move away from what has typically
been characterized as their "homeland" (Anaya & Lomeli 1989) and into regions of the
United States that have traditionally had fewer Mexican-origin people. This
understanding will help evaluate theories of social networks and in particular how
different forms of social capital influence their geographical location through the U.S.
This research based on the 2000 5% Public Use Microdata Sample (PUMS)
focuses on Mexican-origin migration from the Southwest region occurring between 1995
and 2000. The project makes the Mexican-descent population in the Southwest the foci
of the study for various reasons. In general, the Southwest is an important region for
Mexican-descent people (Galarza 1981, Samora & Simon 1993; Nostrand 1975).
3
Mexican-origin people have populated Southwestern states for many decades (Watson &
Samora 1954) extending back prior to the settlement of European pilgrims in the first
colonies (Bushee 1923; Pachon & Moore 1981). The Southwest has for the most part
retained the heaviest population concentration of Mexican-descent people in part due to
the historical and continuing Mexican immigration (Bogardus 1930; Calavita 1992;
Gwin 1921; Samora 1971; Pantoja, Menjívar, & Magaña 2008).
The primary cause for studying the migratory behavior of Mexican-origin
individuals living in the Southwest is that this ethnic U.S. population has experienced
and continues to experience unique socio-cultural and economic circumstances in the
region. For example, the Southwest geographical area has its own regional identity
(Weber 1982), its own politics (Shockley 1972), and its own culture (Perrigo 1971) due
in large part to the presence of pre- and -post U.S. Mexicans in the region. Mexican-
origin individuals in the region have experienced a wide range of social environments
ranging from violent social oppression against Mexicans (Paredes 1958) to revolutionary
social movements (Gutierrez 1998). The Southwest is filled with both wonderful and
terrifying stories that inform modern day Mexican-origin discourse in North America
and how their residence in or pilgrimage from Aztlán (Hernandez 1975) influences their
lives.
Early works have given the Southwest region special attention by pointing out
how race and class operated very differently within it in comparison to other U.S.
regions (Barrera 1979). For example, law scholars have pointed out that in post-
conquest Southwest states Mexicans became invisible under the law in order to keep
4
them from entering the landholding or claimant class (Cameron 2000). Because the
Mexican-origin population in the Southwest has experienced a very unique socio-
political struggle (Acuña 1972, 2006), the key socio-political Chicano movement that
served in large part to awaken the national identity of Latinos/as in general during the
1970s and today through academia and various communication agencies took roots
within the southwestern states (Munoz 1989; Trujillo 2005).
Ever since Mexico was militarily forced to sign the Treaty of Guadalupe Hidalgo
in 1848 (Griswold 1990) the Southwest has continue to host the largest Mexican-descent
population in the U.S. Existing maps clearly show the pronounced presence of
Latinos/as in the Southwest states by county in 2006 (Census 2007a). The Southwest
continues to be a special place for Mexican-origin individuals.
The continual presence of Latinas/os in the southwestern states has given rise to
many special ethno-racial struggles and experiences that may have helped defined their
population identity within the U.S., allowing previous researches and the current
enterprises the means to justify that it be given special attention (Galarza, Gallegos, &
Samora 1970; Grebler, Moore & Guzman 1970; Griswold 1984; Moore 1970a; Trujillo
1974; Saenz, Cready, & Morales 2007; Saenz & Davila 1992). Recent data continues to
indicate that southwestern states have the highest percent of Latino/a concentrations in
the U.S2. For example, in 2006, New Mexico's total population was 45% Hispanic,
while California had 36%, Texas 36%, Arizona 29%, and Colorado 20% (Census
2 Earlier writers have pointed out that the Hispanic and Latino/a umbrella label hides the fact that Mexican
Americans have been the bulk of the Southwester Hispanic population for many decades (McWilliams
1949; Saenz, Cready, & Morales 2007).
5
2007b). Nevada was the only state not currently included in the Southwest to have a
higher Latino/a concentration (24%) than Colorado, the southwestern state with the
lowest Latino/a concentration. The major point in all this is that most Mexican
Americans have and continue to make the Southwest their home in the U.S.
Another reason why Mexican-origin individuals in the southwestern region are
the center of attention in this research is that they have been unfortunate recipients of
high poverty rates3 and low levels of socioeconomic status for many years. For
example, in 2006 about 21% of Latinas/os were in poverty and except in New Mexico
and all other southwestern states had poverty ratios higher than the national ratio
(Bishaw & Semega 2008). This makes poverty among Latinos/as in the Southwest a
special issue because interregional migration to non-southwestern states represents a
strategy for exiting poverty with geographic mobility associated with social mobility.
Mexican-origin people make up most of the Latino/a population in southwestern states.
The presence of high levels of poverty make "new destinations" (Zuniga & Hernandez-
Leon 2005) instrumental in potentially relinquishing the Mexican-descent population
from the continual grip of poverty in the Southwest.
The spotlight on the Southwest--in the context of interregional migration--is
important because it holds special social, historical, economic, and political meaning for
Mexican-origin individuals in the U.S. As Mexican-origin individuals alter their
population distribution across the U.S. (Bernstein 2007), migration out from the
Southwest may lend a hand in helping them experience economic and social "upward"
3 For a detailed discussion of how poverty is measured please see Dalaker 2001.
6
mobility, allowing them to continue to change the "face of America" in general
(Rodriguez et.al., 2008; Saenz et.al. 2003; Saenz 2004) and that of their own ethnic
population.
In examining Mexican-origin interregional migration from the Southwest region,
this study seeks to expand demographic theory and methods. By using multinomial
logistic modeling, this research endeavors to help social theory further open the door for
new insights (Gale 1973) on the complex migration patterns among the Mexican-origin
population in the United States. In particular, the research addresses the question of how
human, household, and community capital are associated with Mexican-origin
interregional migration from the Southwest into other U.S. regions.
The thesis consists of five chapters. This first chapter highlights the goals of the
research undertaken here. Chapter II examines the literature on interregional migration.
This section introduces the Saenzian concept related to the movement of Mexican-origin
people to specific areas of the county in more detail. Chapter III describes the data and
the methods used to carry out the analysis. Chapter IV discusses the findings. Finally,
Chapter V summarizes the findings and concludes with a discussion of the implications
of the results.
7
CHAPTER II
LITERATURE REVIEW
This chapter provides an overview of the literature on internal migration, with
specific attention to interregional migration. The chapter provides a discussion based on
the major theoretical components of the analysis, namely the Saenzian concepts and the
different forms of capital (human and social) that are related to movement of Mexican-
origin individuals from the Southwest region. The chapter describes how various
theories frame our understanding of migration.
By drawing on ecological demography theories, I first delineate why migration is
such an important dynamic in populations. A discussion of how social networks help
explain the processual migration perspective follows. This information helps capture
the economically-laden logic underlying ideas used to frame how human and household
and community social capital influence the decision of Mexican-origin individuals to
migrate out of the Southwest. The section delineates the three main human and social
capital factors used to establish the three main hypotheses under analysis.
In the second section, the interregional migration literature is explored to explain
why the Southwest region is the foci of this study. This section introduces the
cornerstone idea of the empirical project: the Saenzian concepts.
The literature review concludes with a discussion of how interregional migration
selectivity occurs. The theories and various factors introduced earlier are used in this
8
section to encapsulate the main idea of the thesis: that interregional migration is
influenced by various types of human and social capital mechanisms.
Human Ecology, Migration, and Capitals
Human ecology theory (Hawley 1950) concerns itself with how human
populations organize themselves in relation to their sustenance-producing activities
(Poston & Frisbie 2006). Ecological demography posits that populations attain
sustenance equilibrium by using the three main demographic processes: (1)
births/fertility; (2) deaths/mortality; and (3) moving/migration (Poston & Frisbie 2006).
Research on the "ecological complex" (Sly 1972) in migration has been around since the
sixties (Hawley 1968; Davis 1963).
Although some have pointed out how human ecological theories parallel conflict
theories (Hawley 1984), it may follow more of a quasi-functionalist perspective. The
macro-theory basically posits that populations maintain equilibrium between their size
and resources by redistributing themselves through the three main demographic
processes, with migration being the most efficient in affecting population change
(Poston & Frisbie 2006).
An oversimplified example of how equilibrium is attained among a group of
people (i.e., a population) that reside in a geographical location where food becomes
scarce, would be that these people would have to find a way to distribute their food in
such a way that all members are sustained—kept alive. In the event that food access
continually decreased, the population would have to find a way to limit the number of
9
"mouths" it has to feed. The population may then attain sustenance equilibrium by either
having fewer births, more deaths, or people migrating out of the group. No matter what,
they would have to change the size of their group relative to their resources. Having
fewer births will only benefit them at a much latter time. Increasing the amounts of
deaths in the group may create ethical dilemmas that may dissuade the group using such
solutions. In this hypothetical scenario, migration offers the quickest most acceptable
response to disturbances in the equilibrium.
Some researchers have pointed out the U.S. population is becoming less mobile
(Wolf & Longino 2005), while others have indicated that mobility has remained high in
the civilian population mobility over the last few decades (Franklin 2003; Pingle 2007).
Migration is a dynamic process that helps populations regulate themselves: it is an
exceptionally dynamic demographic process. So much so that Charles Tilly wrote that
throughout American history "...the highly variable knitting together of sending and
receiving networks, shaped the aspirations, opportunities, strategies, fortunes, and
accomplishments of most Americas" (1990: 93). This is why I am investigating human
migration.
The macro perspective human ecology offers can help us understand that
individuals migrate for more than just personal reasons: there are structural influences on
the decision to migrate. Human organisms are framed as rational economic agents in the
hypotheses of the research. Humans are seen as having a desire to maximize both their
monetary and psycho-emotional benefits. They are believed to have the ability to
perceive opportunities that may help them maximize their benefits and that their ability
10
to act on such perceptions is affected by different demographic factors. The mere
"possibility of migration" has been shown to foster human capital formation (Katz &
Rapoport 2005:273). The thesis seeks to discern how existing human and social capital
are associated with the decision to migrate.
The economic framing of the individual thus assumes that in general most
humans want to move up the social hierarchy. More specifically, Mexican-origin
individuals have the ability to perceive how their individual skills and social networks
can be used to gain the greatest returns on their various forms of human and social
capital and their ability to act on such perceptions varies by different socioeconomic and
demographic factors, such as marital status and education.
I will now discuss how the term social network is being used more in a
metaphorical sense than an empirical one.
The word network refers to the idea that people are interconnected in a social
system. That is, humans interact with others and their interactions with others vary in
strength. In most instances, family members would be seen as having a strong
connection while co-workers may be seen as having a weak connection (Granovetter
1973; Katz 1958). Human relationships are then imagined as connections that unite
various nodes (i.e., humans) in a social network, where migration flows of human
organisms through "trust networks" are important for individuals', families', and national
economies (Tilly 2007). The thesis is interested in assessing whether different forms of
social (and human, see below) capital affects migratory behavior as individuals navigate
through the various "social links" (Katz 1966).
11
While the ecological perspective is not used directly in modeling migration
decisions, recognizably populations tend toward achieving sustenance equilibrium and
that social networks vary in strength and usefulness. The thesis investigates how micro-
level attributes like level of education and the macro-level factor of having many in-
migrants in the community influence the likelihood of migrating. As such, it is only
concerned with how Mexican-origin individuals use their personal and public forms of
capitals in searching for upward social mobility. It is ultimately interested in
ascertaining if social factors play a role in the decision to migrate.
There are three types of human and social capital theories used in the analysis:
human capital and household and community social capital. These are treated as
"migration capitals." That is, they are being framed as resources that can influence
migration processes.
Capital theories have been utilized in migration studies (Massey et al. 1997;
Saenz et al. 2007) to understand how the different forms of human and social capital
influence migration behavior. By conceptualizing social network interactions, research
has been able to compare the value4 of different types of ties (Coleman 1988; Litwak
1960; Rapoport 1953).
In general, human capital theory refers to the individual’s characteristics and it is
most commonly measured by level of education (Davila et al. 1990; Foulkes et al. 2000;
Giordono 2000; Greenwood 1985; Kritz et al. 1994; Saenz 1991, 1989; Schwartz 1976).
4 It is beyond the scope of this study to discuss how social networks/ties/contacts unequally benefit
individuals (Menjivar 2000), how their equilibrium is altered by context (Jones 1948) and how they are
simultaneously embedded with advantages and disadvantages (Takahashi & Magalong 2008).
12
Human capital is viewed as a private entity because the individual “owns” it. In keeping
with tradition, this study will also use education as a proxy for human capital.
Educational attainment is important in predicting migration because many
migration studies have shown it to be a significant factor for estimating the probability
of migration (for example see Domina 2006). There are several reasons why migration
research should consider the individuals' human capital. For example, human capital has
been shown to influence an individual's position in the social hierarchy (Wall, Ferrazzi,
& Schryer 1998) and that labor market returns vary by the level of human capital
(Redstone 2006).
This socioeconomic mobility associated with education makes human capital an
important component of population redistribution. If education influences the ability to
influence one's socioeconomic status and it is highly associated with migratory behavior,
then it should be included in migration research. Including such an important factor is
contextualized by using the Saenzian concepts. Evaluating how the core, periphery, and
frontier regions play a role in understanding the various human and social capital in
relation to interregional migration shows how education is involved in migration
processes.
Education has the potential to provide the means by which to act on the pursuit of
a better quality of life. Of the three forms of human and social capitals being studied,
human capital is the most resourceful in navigating formal interactions. A
straightforward example of this can be given when we consider to employment
opportunities. Previous research has shown that interregional movers have "greater
13
propensities for upward mobility...when upward mobility implies increase in income"
(Ge Lin 2006). Researchers have also found that returns to educational investments
vary by state (Goetz & Rupasingha 2003) and that skilled workers respond to market
disequilibrium by moving into the more attractive regions, while the unskilled are
immobile and move to these attractive regions only after their more skilled counterparts
move in (Michel, Perrot, & Thisse 1996).
All individuals may benefit from network contacts in finding their way through
"multiregional occupational labor market systems" (Sweeney & Goldstein 2005:314) to
gain employment, but those with college degrees may rely on impersonal links more
easily than individuals with less than a high school education --specially when the job is
in a different region from where they are currently residing.
Having a college degree may also mean that extra avenues of information (like
internet job postings) are available to inform the person of more profitable formal
economic sectors. Because individuals with a college degree are hypothetically more
skilled in navigating through weak-formal networks, it is believed that the migration of
those with less than a high school education may rely more heavily on strong and
informal social networks than for those with a college degree.
Implicit in this logic is the notion that weak ties are more abundant than strong
ties and that individuals who can navigate more efficiently through weak ties have the
greatest opportunities for maximizing their skills. Weak ties may also be formal or
informal. I use the term "formal" to refer to quasi-professional associations and
"informal" to tentatively name more socially intimate and non-professionally related
14
associations. By way of example, a formal weak tie may be an individual who lives far
away and can only be contacted via their business email, phone, fax, or internet website.
An informal weak tie may a be an acquaintance known through a third cousin. I believe
those with a college education have more training in how to navigate formal-weak ties.
That is, those with a college education may on average have more skills in using
different forms of technology, like the internet, to successfully make contacts with semi-
strangers in geographically distant places.
As a consequence, those with less than a high school education are perceived as
having more limitations in navigating weak formal networks. This is a crucial point
because current race problems in America allow and create different forms of
discrimination for all potential Mexican-origin migrants. That is, they may be
navigating through potential hostile environments and facing more social obstacles in
their ability to migrate. As a consequence, those who are able to enter non-ethnic
networks (like formal weak-ties) may have a higher likelihood of leaving the Southwest.
Human capital is a privately owned individual attribute that can aid the
movement through networks imbedded in structural systems where job attainment is
achieved through a supposed competitive process. Those with more human capital are
more empowered to influence their position in the social hierarchy. As a consequence,
Mexican-origin individuals living in the Southwest that perceive migration as a tool with
the potential to help with upward social mobility may act on the desire to leave the core.
Those with less formal education are hypothetically moving out of necessity while those
with more education move out of the desire to improve the returns to their education.
15
A classical study of migration details why and how education plays an important
role in migration process (Ritchey 1976). As a result, a main hypothesis under
observation evaluates if education is important in predicting migration behaviors and if
different forms of social capital operating through social networks are important factors
in predicting migration after this human capital factor is statistically controlled. There
are three hypotheses under investigation. In the first, I hypothesize that householders
with a college degree will have a higher likelihood of migrating out of the core region
compared to those with less than a high school education.
Household social capital has been used to hypothesize how the household’s
network contacts are associated with interregional migration (Saenz 2007). Some have
recently reported that households are the social unit through which individuals
experience their neighborhood and make residential location decisions (Ellis, Wright, &
Parks 2006). The key difference between human and household social capital is that in
general terms human capital theory focuses on the individual’s characteristics while
household social capital theory concentrates on the household’s characteristics to
interpret likelihoods of migration. Human capital is imagined as existing in the
individual, while household capital moves within a group of significant others—usually
family (or more broadly, household) members. Both are viewed as private entities
because they belong to the individual or his/her household unit.
Household social capital theory also helps explain how household network
connections may influence the likelihood of migration. The logic here is that individuals
with a household member born outside the core may have a higher likelihood to migrate
16
out of the homeland because they have verifiable contacts outside the core. This thesis
uses the presence of a person born outside the core in the household to establish if the
householder has any periphery or frontier contacts. If there is a non-core person who
was born outside the core (more details below) in the household, then the householder is
said to have household capital. Because a Mexican-origin individual with household
capital has verifiable non-core network contacts, it is more likely he/she will have a
higher probability of leaving the Southwest.
By statistically modeling the three regional categories, the thesis examines if the
presence of a non-core place of birth in the household is actually influencing the
householder's likelihood of interregional migration. This leads us to the second of the
three main hypotheses under investigation. I hypothesize that householders with a
household member born outside the Southwest region have a higher likelihood of
migrating out of the core region compared to householders who do not have a household
member born outside of the homeland. More specifically, those with periphery
household social capital will be more likely to out-migrate to a periphery state and those
with frontier household social capital will be more likely to leave the core for a frontier
state.
Research on community social capital has also been employed in migration
studies to hypothesize how individuals benefit from social connections in their
communities (Ayala-Garcia 2003; Massey 1990; Saenz 2007). By concentrating
attention on community characteristics, researchers have theorized how communities’
17
network-contacts shape migratory behavior. The loci of resources here are perceived to
rest within the communal geographic space of the individual.
The major difference among these three capital-based concepts is that
community social capital is a context factor, while human capital and household social
capital are both individual- level factors. Community social capital is a "public good"
while the other two are "privately owned". In this paper, community social capital refers
to the relative presence of persons in the local area [Public Use Microdata Area
(PUMA)] (where householders were living in 1995) who migrated from outside of the
Southwest between 1995 and 2000.
Accounting for the context variable of community social capital may aid
migration studies by observing how communal resources provide potential network
contacts. I will give a straightforward example. A person living in a core community
(i.e., PUMA) where there are lots of Mexican-origin persons who migrated recently from
outside of the core is potentially exposed to many frontier and periphery network links in
comparison to a householder living in community with very little of such migrant. The
logic is that the degree to which there are members present from outside the core in the
community affords different potentialities for connecting with periphery or frontier
networks. As a consequence of these assumptions and ideas, the individual living in an
area with community capital is seen as being more likely to migrate out of the core
region. It is therefore hypothesized that Mexican-origin individuals residing in 1995
communities that have relatively high levels of Mexican-origin periphery and frontier in-
18
migrants will have a higher likelihood of migration than those who live in areas with low
levels of in-migrant concentrations.
Interregional Migration and Saenzian Concepts
Interregional migration in the United States has been studied for many decades
(Ross & Truxal 1931). Studies examining interregional migration by race have also
been around for many years (Cebula 1974; Crowder & South 2005; Fulton 2007;
Greenwood & Gormely 1971; Saenz, Cready, and Morales 2007). Research focusing in
on Mexican-origin individuals living in the Southwest (Saenz 1989a) has suggested that
general migration selectivity theories need to be examined to see if they can be
generalized to varying ethnic groups (Saenz 1989b). Recent work has shown that
“minority status” individuals have different “onward migration” behaviors than whites
(Wilson, Berry, & Toney 2008) and that Latinos are more likely to “return migrate” than
whites (Wilson et.al. 2008). Because Mexican-origin interregional migration may be
different than non-Hispanic White or Black movements, this project seeks to provide
information that can help better understand the interregional migration experience of
Mexican-origin individuals living in the Southwest.
Much is known about interregional migration, but this complex phenomenon has
many worthwhile questions that remain to be explored in the Mexican-origin population.
Interregional migration is defined here as an event that occurs when an individual moves
from one region to another in the continental U.S. The review of the existing literature
19
on interregional migration is narrowed to focus on how various demographic
characteristics influence migration behavior.
The states in the continental U.S. are used to conceptualize the three Saenzian
regions. The states are categorized into regions by measuring the Mexican-origin
population presence (discussed in more detail below). Accounting for where potential
migrants move is important because the place of destination plays a role in the decision
to migrate. For example, a recent study found that communities vary in level of
"attraction" to prospective in-migrants (Lekwa, Rice, & Hibbing 2007). Thus creating
regions based on an ethno-racial measure helps this research account for how place of
destination plays a role in either "pulling in" or detracting potential Mexican-origin
interregional migrants.
The thesis uses the Saenzian concepts of core, periphery, and frontier to
categorize the three ethno-racial regions. Because the U.S. Mexican-origin population is
most heavily concentrated in the core region, it hypothetically exerts the most powerful
pull for Mexican-origin individuals. The periphery follows in Mexican-origin
concentration and thus in attraction and this region is followed by the frontier where
very few Mexican-origin people can be found relatively speaking.
The interconnectivity of migration flows is complex and different for each state,
but in general within-county migrations have been typically associated with life-cycle
states (Sjasstad 1962) and inter-county migrations and beyond have been seen more
driven by economic reasons. Classical work on this topic clearly states that "economic
repulsions and attractions have long been emphasized as the primary forces motivating
20
migration" (Westefeld 1940). This investigation is concerned with evaluating
interregional migration and consequently assumes that Mexican-origin south-westerners
are in general behaving as logical actors motivated by both economic and non-economic
reasons.
A key element addressed is how the potential desire to be among co-ethnics plays
a role in moderating economic motives as they influence migratory behavior. Some
have recently shown that the individuals own assessment the potential climate of
reception is associated with migration (Valdivia et.al. 2008). For these reasons the
multinomial statistical models being used with 2000 Census data are believed to show
how the different forms of human and social capital differ in significance and influence
when place of destination is considered.
Interregional migration is a multi-dimensional phenomenon. It can separated
down into two general processes: deciding to migrate and where to migrate. The
previous section delineated some of the fundamental dynamics involved with the
decision to migrate. I will now discuss how the "where" factor interacts with the
individual's characteristics and in particular will outline how the Saenzian concepts help
crystallize the way various types of capital operate to influence interregional migration
decisions.
In 1991 Rogelio Saenz published a study using the core, periphery, and frontier
terms to quantitatively model Mexican-origin interregional migration. By introducing
these demographic concepts, Saenz expanded ethnic-migration theories. He
operationalized spatial-population constructs that helped articulate how Mexican-origin
21
selectivity factors operate in interregional migration from: highly concentrated ethnic
areas (the core region) to the two less ethnically-populated regions of the U.S. (the
periphery and frontier). Accounting for ethnic context is necessary because research
has found that it is an important determinant of internal migration (Kritz & Nogle 1994).
This approach is methodologically useful to the sociology of migration because it
addresses the important element of physical and ethnic distance (see Bogue et al. 1949;
Bright 1941; Stouffer 1940, 1960; Westefeld 1940) and place of destination (see
Fotheringham et. al. 2000; Wolpert 1967) in studying interregional migration. The core,
periphery, and frontier designations—hereafter referred to as the Saenzian concepts—are
very useful because they directly investigate factors involved in Mexican-origin
interregional migration research.
The core region is also referred to as the homeland because it has been
historically populated by Mexican-origin people and because it currently holds the
largest Mexican-origin population concentration in the U. S. The continual presence of
Mexican-origin people in this sizeable geographic area gave birth to the "core"
categorization.
The theoretical framework that informs this thesis research draws on several
different traditions. At the center of this study is the time-honored theoretical
assumption that migration evolves through stages (Bohning 1972; Marks 1983; Massey
1987, 1990; Saenz 1991). According to this view of the process of interregional
migration, the cumulative causation of migration begins when individuals that migrate
become an information resource to those “back home” who may be interested in
22
migrating. This “feedback” (Reichert 1981) process is then assumed to establish a
mechanism whereby migration becomes processual: past migrants influence those with
whom they are linked by providing them with a migration network contact. This is how
chains of migrants are formed linking distant places (Tilly 1990).
A general example using Saenzian concepts would be that feedback mechanisms
are established when members of one’s ethnic group migrate out from the core into the
periphery or to the frontier area. When the core out-migrants break ties with the
Southwest, they serve as a source of information to their connections "back home."
Because previous studies have found that knowledge networks can help the migration
process (Lee 1966), it is assumed that knowledge of what the periphery or frontier
regions are like and how to migrate to them with affordability can increase the likelihood
of out migration from the core. Ultimately, knowledge-based movements trigger the
internal momentum necessary for the creation of the migratory-contact networks (see
Gurak 1992; Katz 1958) that form migratory feedback mechanisms. This processual
formulation of interregional migration directs the overall theoretical approach in this
thesis.
In the example above, the periphery and frontier migrants consequently set up the
first stage of the migration process by establishing a network contact from a particular
region back to the core. Social networks operate through a processual migration
sequence that pulls and pushes individuals in such a way that sustenance equilibrium is
attained through population redistribution.
23
The structural and individual influences on migration behaviors are also a
function of various pull5 factors that shape the level of attraction a region may have to
potential migrants. Interregional migration research has historically sought to
investigate how certain factors—such as a robust economy—function to attract potential
migrants and how other elements—such as high local unemployment rates—influence
the decision to leave the area. In general, context is seen as influencing migration
patterns among Mexican-origin householders because the racial-ethnic composition of
the place of destination is believed to play a role in promoting or dissuading the decision
to move. "Social links" (Katz 1966) then are central in the most dynamic demographic
process known to humans.
The underlying logic of pull/push thoughts is that place of residence and
potential place of migration interact with the individuals’ demographic characteristics to
either pull or push them from one geographic areas to another. For example, the core’s
grip can be seen as loosening its grip on the Mexican-origin population at the beginning
of the 20th century when agricultural, manufacturing, and railroad industries produced
several pull forces that weakened the grip—resulting in increased out-migration of
Mexican-origin individuals from the Southwest region of the country (Saenz 1991).
Accounting for where the individual lives with a community social capital variable is
very important. The Saenzian concepts are necessary for migration research of minority
populations because they help enhance statistical models by taking into consideration the
racial-ethnic context of where a potential migrant may move to.
5 For a discussion of the conceptualization of the “pull” factor, see Bogue 1949 and Herbele 1938. Lee
(1966) uses “plus” to discuss similar migration factors
24
As stated previously, the periphery and frontier regions are so labeled because
they have, relatively speaking, smaller Mexican-origin populations than the core. The
regions have exerted different pull and push6 factors on Mexican-American migration
over time. The key difference between the Saenzian ethno-racial regions is on how each
of them provides different social network contacts.
The core is densely populated by Mexican-origin individuals. It is assumed that
this heavy co-ethnic concentration provides Mexican-origin people in the Southwest
with many networks that may potentially increase their chances of maximizing profits
with their skills. In states where co-ethnic concentrations are low, as in the frontier or
periphery states, it is assumed that social networks will be more limited, making the
ability to maximize skill more problematic. Migrating to a non-core state then presents
more social and economic obstacles for the potential migrant. The limited amount of
migration chains entering the Southwest from the periphery and frontier region then
make the decision to leave the core a more challenging alternative. As a result of all
these propositions, I argue that economic pull and push factors are important along with
noneconomic pull and push factors like racial-ethnic concentrations and the presence of
in-migrants from non-core regions.
A simple example of how the Saenzian regions may exert pull or push factor can
given to help illuminate the reader. A New Mexico resident migrating to North
Carolina may face different issues than if he/she were to migrate to Texas. Why? There
are several reasons. The first is the amount of geographical distance required to relocate.
6 For a discussion of the conceptualization of “push” factor, see Bogue 1949 and Herbele 1938. Lee
“minus” to discus similar migration factors (1966).
25
Moving to North Carolina from New Mexico is further than moving to Texas. This
geographical distance then interacts with perceived social distances. In this project, the
social distance is being measured by like ethnic concentration. Thus, moving to North
Carolina means potentially living in an area with less people that look like me while
moving to Texas will probably mean the potential migrant has more chances of living in
an area where there are more co-ethnics.
It follows then that the core will exert a greater pull/retention power than the
other two regions--although the periphery and frontier may offer more and different
types of avenues for upward social mobility. For example, many of the core
communities where Mexican-origin people reside are burden with chronic poverty. So
even though they reap the social benefits from living among co-ethnics, they pay a
penalty for residing in an area with limited labor market opportunities. For example, the
Southwest has heavy concentrations of immigrants and resent research has found
evidence that immigrants adversely impact the employment opportunities of native-born
workers (Camarota 2007). These generalizations vary by the individuals level of
education, but even those with a college degree may have lower returns on their
education than if they lived in a non-poor community. In short, residing in a periphery
or frontier state may mean that the Mexican-origin person is able to tap into a more
resource laden area where their minority status may even help them be more competitive
in maximizing their skills (like being bilinguals).
As discussed in the introduction, the Mexican-origin individuals living in the
Southwest are very frequently exposed to chronic poverty. Thus, in spite of how the
26
networks and the desire to be among co-ethnics may function to retain the potential
migrant within the Southwest there are other pull factors exerting an effect from outside
the core. In this case, the frontier and periphery may offer different levels of attraction
that may vary by economic opportunities and perceptions of how the host community
may ease the integration of the perspective migrant. For example, research has found
that trust towards Whites varies in the Latino population (Chaves, Wampler, Burkhart
2006). Such pull factors from the different Saenzian regions exert dissimilar influence
in pulling and pushing people. This variation is belief to have a significant association
with how individuals vary in their region-specific contact availability. That is, what
kinds of community and household capital they have. This is why the Saenzian concepts
are so instrumental when accounting for community capital.
A recent study on Mexican American internal migration patterns between 1985
and 1990 was conducted using the Saenzian concepts (Saenz et al. 2007). This research
and others have used census data from earlier decades, but no study similar to this one
has been undertaken using data from the 2000 census. This thesis examines 2000 census
data and uses statistical models that better integrate individual and contextual factors
compared to extant research based on the Mexican-origin population.
Migration selectivity occurs at both macro and micro levels. At the macro level
there are context phenomena exerting pull/push forces and at the micro level there are
individual level characteristics that both invoke or limit the ability to move.
Selectivity is an overall important factor because research has shown that
migration is a highly individualistic decision (Bogue 1959, 1992b; Feliciano 2005; Shaw
27
1975; Thomas 1938) in the Latino/a population (Saenz 1989, 1992, and 2007). It has
been established that economic motives influence migration (Weeks 1996) and that these
motives are also associated with age, race, education, and sex as the significant factors
influencing individual selectivity. Even bilingualism has been found to have a
significant association with migration (Golash-Boza 2005). These are introduced in the
model to statically control for their potential influence on the probabilities of migration.
Since they are not the central focus of the research, they are only briefly discussed
before moving on to a discussion on how social networks are also influence by the
individuals' demographic characteristics.
Selectivity fundamentally focuses attention on how certain characteristics are
highly probabilistic and sociologically significant to interregional migration. In other
words, the likelihood of migration is said to either increase or decrease when certain
statuses—like being a college graduate versus not —are “selected” (active) or “de-
selected” (inactive). For example, the possession (active state) of high formal education
increases the likelihood (selection) of interregional migration when the formally
educated individual is compare to those with a high school diploma or some college.
Understanding selectivity is important because migration research is
contextualized in associations between demographic and environmental characteristics
in migratory behavior. Research has shown that selectivity for “frontier” migrants is
more evident than for “periphery” migrants (Saenz 1991), meaning that demographic
characteristics make moving out to the frontier a more selective process. Hence, the
heightened selectivity for the frontier migrant becomes more evident than the periphery
28
migrant because of the social and economic complexes that influence mobility (Blanco
1963; Lee 1966). The likelihood of migration is influenced by various factors that offer
different selectivity options to potential migrants.
Age has been a part of the migration research lexicon for many years because it
has been found to have an association with migration patterns (Johnson et. al. 2005;
Smith-Buani 2001). Generally, people move most in their early adult years between age
18 and 30. Responsiveness to perceived income opportunities have been shown to vary
by race (Greenwood 1975) and males have consistently been shown to migrate more
than females (Curran et.al., 2006; Cebula 1974; Donato et. al. 2006).
It is important to recognize that social networks are not neutral links: access to
them is affected by the individual's characteristics and statuses. Researchers have shown
that linking up to the many hypothetical social networks varies by different factors
(Entwisle et.al. 2007). Classical work has shown that attributes like sex and age may
play a role in how selective factors operate in the forces that govern migration process
(Hobbs 1942).
Race is one such ascribed status that still matters in the American experiment.
Some have even said that classifying people is "necessary to organize and maintain an
empire" (Hacking 2005: 116). This is when moving to a region where there are more
people "like me" may influence the potential migrant to evaluate how socioeconomic
returns may vary by region (Aguilera & Massey 2003). The Saenzian concepts capture
the racial-ethnic composition of the place of destination and allows the statistical
inclusion of such an important ethno-racial geo-spatial factor. When potential migrants
29
contemplate their potential destinations, they are--hypothetically speaking--likely to
consider how their access to local social networks in the community of reception will
influence their ability to attain upward social mobility. This is capture by using the
Saenzian regions.
In summary, the literature clearly shows that interregional migration is complex
and that for the most part individuals are treated as economic agents who seek to
improve their quality of life by living in areas that allow them to maximize their profits.
The thesis draws from all these different social thoughts to examine how human capital
and, household and community social capital are associated with interregional migration
among Mexican-origin individuals living in the U.S.
Hypotheses
By using the various schools of thoughts, the present study will test three
different hypotheses to evaluate if interregional migration theories of the general
population can be used to understand the migration of the Mexican-origin population
living in the Southwest. To reiterate below are the three hypotheses:
Human Capital Hypothesis:
I hypothesize that householders with a college degree will have a
higher likelihood of migrating out of the core compared to those with
less than a high school education.
Household Social Capital Hypothesis:
30
I hypothesize that householders with a household member born outside
the Southwest region have a higher likelihood of migrating out of the
core region compared to householders who do not have a household
member born outside of the homeland. More specifically, those with
periphery household social capital will be more likely to out-migrate to
a periphery state and those with frontier household social capital will
be more likely to leave the core for a frontier state.
Community Social Capital Hypothesis:
I hypothesize that Mexican-origin individuals living in a PUMA with
community social capital will have a higher likelihood of migration
than those without community social capital.
31
CHAPTER III
METHODOLOGY
Data Source
The hypotheses introduced above will be examined using data from the U.S.
Bureau of the Census' 2000 5% Public Use Microdata Sample (PUMS). This dataset
represents one of the best sources of information to research the demographic and
socioeconomic patterns of Mexican-origin householders and other racial and ethnic
groups.
Sample Selection and Saenzian Regions
The unit of analysis in the study is the householder (head of household).
Householders are only included in the analysis if they lived in a southwestern (Arizona,
California, Colorado, New Mexico, and Texas) state in 1995 and in one of the
continental U.S. states (including the District of Columbia) in 2000. The householders
in the analysis are further selected by only including those who report that they are of
Mexican origin and who were between the ages of 30 and 64 in 2000. The age-specified
selection allows the analysis to exclude persons whose migration may be potentially due
to non-labor market reasons (e.g., college students and military personnel).
As a consequence of these stipulations, the analysis is composed of householders
between the ages of 30 and 64 who are of Mexican-origin and who lived in a
southwestern state in 1995 and in one of the continental U.S. states in 2000. This is the
group of people who make up my sample used in the analysis.
32
The householders in the sample are divided up into three different categories
based on their region of residence in 2000 based on the Saenzian concepts. These
designation are established through the following procedure7.
The percentage of Mexican-origin population in the overall state population is
computed. The percentage of Mexicans in a state is calculated by dividing the number
of Mexican-origin individuals living in state i in 2000 by the total population living in
that state i in 2000 with this product multiplied by 100. For example, in 2000, Alabama
had a total population of 4,447,100 and a Mexican-origin population of 44,522--thus, the
following computation is carried out: (44,522/4,447)*100=1.0%. The percentage of the
Mexican-origin population relative group size in all continental states is obtained in this
fashion. These percentages are then used in accordance with the Saenzian concepts to
classify the continental states into the core, periphery, and frontier.
This approach used to create the Saenzian regions replicates that used in previous
research (Saenz 1990; Saenz et. al. 2007). The states that have been typically
categorized as southwestern (core) are treated the same here. Hence, Arizona,
California, Colorado, New Mexico, and Texas make up the core region.
After the core states are taken out, then the average absolute size and relative
percentage of the Mexican-origin population across the 44 mainland non-southwestern
states in the year 2000 is calculated. The relative percent average for 2000 is 2.4%.
That is, the Mexican-origin population accounted on average for 2.4% of the populations
of non-core states in 2000.
7 Other researchers have recently used the same terms in studying a different type of migration in the
western part of the United States (Henrie & Plane 2008).
33
The relative size (percentage) average is then used to categorize the non-
southwestern states into periphery and frontier regions. Non-Southwest states above the
relative average of 2.4% Mexican-origin concentration are assigned to the periphery
region and the remaining non-southwestern states that are below the relative average are
assigned to the frontier region. All of these procedures then lead to the regions being
comprised as follows: the core is made up of 5 states, the periphery of 14 states, and the
frontier of 30 states. This is a total of 49 states (including the District of Columbia) used
in the study. Tables 1 through 3 display each of the states by region and their
corresponding Mexican-origin population percents.
Table 4
2000 Frontier States
State Percent
Mexican-Origin
Arkansas 2.29%
Florida 2.28%
Michigan 2.22%
Iowa 2.09%
Minnesota 1.94%
Delaware 1.66%
Missouri 1.39%
New York 1.37%
Tennessee 1.36%
South Carolina 1.32%
Montana 1.30%
New Jersey 1.22%
Virginia 1.05%
Alabama 1.00%
D. C. 0.89%
34
South Dakota 0.84%
Ohio 0.80%
Kentucky 0.78%
Mississippi 0.76%
Maryland 0.75%
Louisiana 0.72%
Connecticut 0.69%
North Dakota 0.67%
Rhode Island 0.56%
Pennsylvania 0.45%
New Hampshire 0.37%
Massachusetts 0.35%
West Virginia 0.24%
Maine 0.22%
Vermont 0.19%
Source: U.S. Census 2000 SF-1
Universe: State’s total population
Table 4 above reports that Vermont, at 0.19%, is the frontier state with the lowest
percentage of Mexican-origin individuals and that in the same region Arkansas is the
state with the highest Mexican-origin concentration (2.3%).
Table 5
2000 Periphery States
State Percent
Mexican-origin
Nevada 14.30%
Illinois 9.21%
Oregon 6.27%
Idaho 6.13%
Utah 6.11%
Washington 5.60%
Kansas 5.52%
Nebraska 4.15%
35
Wyoming 4.04%
Oklahoma 3.85%
Georgia 3.36%
North Carolina 3.06%
Indiana 2.52%
Wisconsin 2.36%
Source: U.S. Census 2000 SF-1
Universe: State’s total population
Table 5 shows that in the periphery region, Wisconsin (2.4%) is the state with the
lowest Mexican-origin percentage and that the periphery state of Nevada has the highest
percentage (14.3%). Note that Nevada has a larger concentration of Mexican-origin
people than Colorado, a core state. In order to make this project as comparable as
possible to previous ones, Nevada is treated as a periphery state and Colorado as a core
state8.
Table 6
2000 Core States
State Percent
Mexican-Origin
California 24.96%
Texas 24.32%
Arizona 20.77%
New Mexico 18.14%
Colorado 10.48%
Source: U.S. Census 2000 SF-1
Universe: State’s total population
8 A co-authored presentation including the author of this thesis shows that Nevada is still significantly
different from the other five Southwest states (Saenz & Siordia 2008). Perhaps in the future as Nevada
becomes more similar to the Southwest states, it will become part of the core.
36
Of the core states listed in Table 6, Colorado is the state with the lowest
Mexican-origin percentage at 10.5% and California has the highest percentage (25.0%).
As a whole, the frontier has an average Mexican concentration of 1.1%, the periphery
has 5.5%, and the core has an average 19.7%. Figures 1 and 2 visually show how the
states fit within the scheme of the Saenzian concepts.
Figure 1
37
Figure 2
Through these designations I classify Mexican-origin householders living in the
Southwest in 1995 into one of three possible migration categories: (1) regional non-
migrants—lived in the Southwest in 2000; (2) periphery migrants—lived in a periphery
state in 2000; or (3) frontier migrants—lived in a frontier state in 2000. These three
migration categories constitute the trichotomous dependent variable used in the analysis.
I will now discus in more detail how the three migration categories that make up
the dependent variable are created.
38
Dependent Variable
The migration category is assessed by using the “five-year” migration census
question. The question was first introduced in the 1940 U.S. census and has been used
ever since to study migration (Bogue 1949; Lee et al. 1960; Saenz, Cready, Morales
2007). In 2000 the question asked long form respondents to report their place of
residence in 1995 and 2000. By using this five-year question Mexican-origin
householders are categorized according to their response: (1) regional non-migrants are
Mexicans who lived in the Southwest in both 1995 and 2000: (2) periphery migrants
are Mexicans who lived in the Southwest in 1995 and in the periphery region in 2000:
and (3) frontier migrants are Mexicans who lived in the Southwest in 1995 and in the
frontier region in 2000 (more details given below).
The dependent variable is made up of three categories: regional non-migrants,
periphery-migrants, and frontier-migrants. These categories are unordered. That is, I
cannot say they differ in respect to some absolute value other than the arbitrary
requirements I have outlined in their creation. I will test a cumulative model to make
sure that they are unordered. The statistical test, discussed in more detail below, will
validate the use of a multinomial model.
In a multinomial model, one of the categories has to be the reference category.
The reasons for this are outlined in detail below. The regional non-migrant category is
the reference group. Regional non-migrants represent the comparison group. Table 1
below offers a concise summary of the categories in the dependent variable.
39
Table 1
Dependent Variable
Variable Name Variable Description Variable Operationalization
Nonmigrant Householder did
NOT migrate Regional non-migrants are those
householders who reported living in the
Southwest in 1995 and in 2000
Periphery migrant Householder migrate
to Periphery region The Periphery migrants are those
individuals who lived in the Southwest in
1995 and report living in a Periphery state
in 2000.
Frontier migrant Householder
migrated to Frontier
region
The Frontier migrants are those individuals
who lived in the Southwest in 1995 and
report living in a Frontier state in 2000.
Independent Variables
The thesis includes three primary independent variables: (1) educational
attainment; (2) presence of a household member born outside the core region; and (3) if
their 1995 PUMA of residence is considered to have community social capital. Each of
these is used to test one of the three main hypotheses. Table 2 below summarizes the
three main independent variables under analysis.
Table 2
Independent Variables
Variable
Name Variable Description Variable Operationalization
Education Householders’
educational
attainment level
(human capital)
Level of education is measured with four dummy
variables:
(1) some high school (9-11)
(2) high school graduate
(3) some college
(4) college graduate.
*The reference category will be made up by
persons with eight or fewer years of education.
40
Member Presence of
household members
born outside the Core
(household social
capital)
Member is measured by two dummy variables:
(1) households were at least one (>5 year old)
member was born in a Periphery state
(2) households were at least one (>5 year old)
member was born in a frontier state.
*The reference category will be made up by
households were none of the people
living in it was born outside the Southwest.
Community Level of in-migrant
concentration in
community
(community social
capital)
The community social capital independent
variable measures the percent (%) of Mexican-
origin concentration in the community that moved
in from a Periphery or Frontier state between 1995
and 2000.
The first independent variable measures educational attainment and it is used as a
proxy for the individuals’ human capital. According to the hypothesis stated in chapter
2, I expect householders with a college education to have a higher likelihoods of
migrating out of the core than those with less than a high school education.
Educational attainment is broken up into five categories: (1) less than high school
education (0-8 years of education); (2) some high school--the responded has completed
between 9 and 12 years of schooling with no high school diploma; (3) high school
graduate; (4) some college; and (5) college graduate and above. Earlier studies indicate
that persons with eight or fewer years of education have lower likelihoods of migration
(Saenz et. al. 2007). Thus, the dummy variables being created for these categories will
be compared to those with less than high school, the reference category.
The second of the three main independent variables measures whether a
householder is connected to a household where a household member was born outside
41
the core region and over the age of five is present (the reason why age five is used will
be discussed below). If there is a household member over the age of five present in the
household, the head of household will be labeled as having household social capital.
Because a birth outside the core is considered to signify increased potential contacts with
the periphery and frontier region, this measure is used to ascertain if the householder has
a verifiable social network connection with a non-core region.
Recognizing the increased potential of “non-core social network contacts” in a
household unit with household capital will allow us to evaluate its influence on the
likelihood to migrate. A subtle complication with the measure is that a person born
outside the core need not be a “blood” relative. Thus, the strength of the link between
the householder and the person born outside the core may—hypothetically—vary.
By way of making the logic clear, I will provide a straightforward example. Say
participant Xi reports living in a household where someone was born in a frontier state.
The person associated with participant Xi may be a daughter, niece, or friend. Not only
can the relationships vary, but the amount of time that has passed since the birth may
also vary. Thus, participant Xi may have a daughter born outside the core 15 years ago,
or a niece born outside the core 10 years ago, or a friend in the household born outside
the core only 5 years ago. Which of these would provide the strongest link to non-core
networks? The potential for variation creates limitations that are more clearly addressed
at closing.
Not all is lost though. The basic premise is that the Mexican-origin householder
has at least one verifiable connection with a person born out of the core region. The
42
presence of such an association is expected to increase the householder's potential for
periphery and frontier networks—and thus increase the likelihood that they migrate out
of the core compared to householders who do not have a household member born
outside of this region.
Why are only household members five years of age and older the only persons
considered in the construction of this variable? Since the PUMS data does not allow the
reconstruction of households back to 1995 (the beginning of the migration period under
observation), a proxy must be used. In order to insure that children born outside the core
region after the beginning of the migration period (1995) are not included, the household
capital measure will only take into account household members who are five years or
older in 2000. This means that if a person associated with the householder was born in
1996, 1997, 1998, 1999, or 2000 outside the core, they will not help the householder
gain household capital. Only those born on or before 1995 have the power to move the
head of household from having no household capital status into having it. More details
on how the household capital variable was created, by considering the presence of a birth
outside the core and in what region the birth occurred, is discussed below.
The analysis dichotomizes household capital into two different variables in order
to investigate if the non-core place of birth is actually exerting an influence on the
likelihood of migration. Two things are taken into account in creating this variable:
1. Is there a person in the household over the age of five born outside the core?
2. If so, from what region was the person born?
43
The two pieces of information are then used to create two dummy variables. A
householder does not have household social capital if no member of their household was
born outside of the core region. If there is a non-core birth over the age of five present
among household members, the event is further broken down into whether the birth
occurred in the periphery or frontier region. An individual is thus said to have periphery
household social capital if there is a household member whose birth occurred in the
periphery. In the same logic, a householder is said to have frontier household social
capital if a household member was born in the frontier. The variable has three dummy
components as a result of these procedures: (1) households where at least one member
was born in a periphery state; and (2) households where at least one member was born in
a frontier state, and (3) the reference category is households where none of the people
living in it were born outside the Southwest.
The third and final independent variable measures the Mexican-origin
concentration in the community. This variable is relatively different than the first two
main independent variables because it is a context measure, whereas the other two are
individual-level measures. Community social capital is measured by counting how
many Mexican-origin individuals moved in from a periphery or frontier state between
1995 and 2000 into a core PUMA. The variable is used in the multinomial models to
investigate if residing in a community heavily populated with non-core in-migrants has
an association with the outmigration of householders who are the focus of the analysis.
This measure is used because it is believed to be a good way of assessing the presence of
periphery and frontier networks in the PUMA. In simple terms, a person is said to have
44
community social capital if they reside in an area where there are lots of non-core in-
migrants. The logic here is that residing in a community with many non-core in-
migrants increases the number of potential non-core migration connections which
enhances the possibility of migration out of the core region.
I will now discuss how community social capital is quantified. The percent of
Mexican-origin non-core in-migrants is first computed for each of the 475 southwestern
PUMAs. For a given PUMA in the core region, the number of Mexican-origin non-core
in-migrants (migrated to the core PUMA between 1995 and 2000 from outside of the
core region) is the numerator and the total Mexican-origin population in the PUMA is
the denominator. The product is multiplied by 100 to converted it to percentage form9.
After calculating the percentages, an average percent of non-core in-migrants across all
the PUMAs is computed. The results indicate that between 1995 and 2000 the average
core PUMA had 1.71% of its Mexican-descent population made up of non-core in-
migrants. About 75% of the southwestern PUMAs fell below a 2.05% concentration.
The PUMAs in the top quartile (25%) are then seen as affording their residence with
community social capital. This top quartile cut off was arbitrarily chosen by me.
The communities’ in-migrant percents are then dichotomized into those with
community social capital and those without it. Individuals residing in a PUMA with
2.05% concentration of in-migrants or more are considered to have community social
capital and those below are said to not have community social capital. PUMAs with
community social capital hypothetically increase potential non-core migrant network
9 The numbers were computed using the following: PUMAanOriginInTotalMexic
UMAigrantsInPNonCoreInM20001995 →.
45
contacts and thus they may increase the likelihoods of out migration from the core
region.
Figure 3 clearly shows that PUMAs bordering Mexico have very low community
capital and that each state has only a few PUMAs endowed with high levels of
community social capital. The potential of spatially analyzing clusters with GIS
software with this community social capital measure is discussed at closing.
Figure 3
46
In Figure 4, the reader can see how the PUMAs are dichotomized. This map makes it
visually apparent that except for a few PUMAs in California, all those bordering Mexico
are not considered to have community social capital. There is a slight pattern in the
PUMAs with community social capital to be next to a non-core state or a PUMA with
community social capital. These are interesting patterns mentioned more at length in the
last chapter.
Figure 4
47
Because PUMAs are relative small in scale when compared to states—especially
in metro areas—maps displaying the percentage level associated with community social
capital is given for all five southwestern states individually. There are also noticeable
differences in the ranges of concentration and patterns of aggregations by state. The
figures below (Figures 5 to 9) are given in order to demonstrate the many variations of
community social capital concentrations throughout the Southwest.
Figure 5
48
Figure 6
49
Figure 7
50
Figure 8
51
Figure 9
These figures show that New Mexico has the lowest percentages of PUMAs with
community social capital, while California has heavy concentrations of PUMAs with
community social capital in the north and south parts of the state. Of all the Southwest
state, Arizona has the most PUMAs with community social capital--located at the center
of the state. Texas has a broad variation of community social capital with the PUMAs
near Austin having more than 11% of non-core Mexican-origin in-migrants.
52
Control Variables
The analysis includes several control variables. The table below labeled Table 3
contains a concise summary of all the control variables. Age is controlled for by
measuring it with three categorical dummy variables: (1) 30-39; (2) 40-49; (3) 50-59.
Householders 60 to 69 years of age have been shown to be the less likely to migrate
(Saenz et. al. 2007) and thus they are the reference group.
Table 3
Control Variables
Variable
Name Variable
Description Variable Operationalization
Age Age of householder Age will be measured with three categorical
dummy variables:
(1) 30-34
(2) 35-44
(3) 45-54
(4) 55-64 reference group
Sex Sex of householder Gender is measured by a single dummy variable:
(1) male
(2) females reference group
Marital Marital status of
householder Marital status is also measured by a single dummy
variable:
(1) married
(2) not-married reference group
Nativity Refers to
householders’
nativity/immigration
period status
Nativity/immigration period status is established by
four dummy variables indicating immigrants’
period of entry into the United States:
(1) entered between 1985 and 1994
(2) entered between 1975 and 1984
(3) entered between 1965 and 1974
(4) entered prior to 1965
(5) native born reference category
53
Language The language that
the householder
most commonly
uses
Language is measured with two dummy variables
are used:
(1) bilingual-means that the householder can
speak both English and Spanish fluently
(2) non-English-means that the householder
mainly speaks English.
(3) mono-Spanish means that householder
mainly speaks Spanish at home and this is
the reference category
Children Presence of own
children under the
age of 18 in
household
Presence of children is measure with two dummy
variables:
(1) householder with the presence of an own child
below age 18
(2) householder with no own child below age 18
present is the reference category
Because gender matters in migration research (Mahler & Pessar 2006), it is
measured by sex as a single dummy variable: (1) male. Previous findings indicate that
females are less likely to be interstate migratnts (Saenz et. al. 2007) and thus, they
represent the reference category.
Marital status has also been to shown to have an association with migration
(White et. al. 2005). Marital status is measured by a single dummy variable: (1) married.
Persons who are not currently married represent the reference group. The not married
category was chosen as the reference group because research shows that married people
are less likely to migrate (Hobbs 1942; Long 1992).
In order to take into account any variation in the probability of migrating out of
the Southwest due to length of U.S. residence among foreign-born householders, the
study includes a series of “nativity/immigration period status” dummy variables in the
54
model. These are similar to those used previously in research (Saenz et. al. 2007). The
nativity/immigration period status is established in the model with both native-born and
foreign-born Mexicans by four dummy variables indicating immigrants’ period of entry
into the United States: (1) entered between 1985 and 1994; (2) entered between 1975
and 1984; (3) entered between 1965 and 1974; (4) entered in before 1965. Native-born
Mexican Americans are the reference category.
Models are also conducted to evaluate the native-born population and foreign-
born population separately. This allows the research to investigate the degree to which
the results of the hypotheses are consistent across native- and foreign-born individuals.
In the native born population there are no nativity/immigration dummy variables present,
while the foreign-born model contains the same dummies as above but uses the fourth
category (those entering the U.S. before 1965) as the reference group.
Language is an important factor in the Southwest and in migration research
(Martinez-Brawley et. al. 2006) and consequently is included as a control variable. Two
dummy variables are used: (1) bilingual (the householder speaks a language other than
English---presumably Spanish—at home and English “well” or “very well” and (2)
monolingual English speakers (the householder speaks English at home). The reference
category is monolingual Spanish speakers (the householder speaks a language other than
English—presumably Spanish—at home and speaks English “not well” or “not at all”).
55
Multinomial Logistic Regression
The unordered and categorical dependent variable (migration status in 2000) is
polychotomous10. The categories in the dependent variable were determined to be
unordered after a cumulative logit model was regressed using SAS. The results indicate
that the log odds assumption in cumulative models is being violated. Thus, a
multinomial logistic regression model is used in STATA to estimate the likelihood of
interstate migration.
Unordered categorical dependent variable models are nonlinear. This makes
ordinary least squares methods biased and inefficient. The trichotomous dependent
variable in this research is considered to be a nominal response and will be analyzed
using multinomial logit regressions. In this technique, logits are formed from contrast of
non-redundant category pairs in the dependent variable. Since the dependent variable
has three categories, it creates two equations. The multinomial logistic model (MNLM)
will estimate a logit coefficient for each of the three outcomes. The logits for regional
non-migrant are set to zero in order to represent the change relative to the Y=1 category
and when set to zero e0=1. This creates the following probability equations for our
model:
10 Polychotomous simply means that the dependent variable is made up of more than two categories.
56
Logistic models use maximum likelihood estimation methods and a logistic
probability distribution. As such, they make many assumptions. For example, it is
assumed that independent variables are mutually exclusive and errors are assumed to
have a standard logistic distribution. Diagnostics were examined and it was determined
that no assumptions are being violated.
Hypothesis testing can lead research to reject the null hypothesis when in fact it
is true. When a researcher reports that there is statistical significance among the
variables under investigation and there really is none, he/she is committing a type-1
error. The reverse—the investigation fails to reject the null hypothesis when it is false—
is called a type-2 error. Relatively speaking, making a type-1 error is the least desirable
outcome because it means that associations between variables are being reported as
significant when in reality there is no significance in the statistical associations. This
investigation has taken precautions, using diagnostic procedures in STATA, to insure
that a type-II error has not been made.
The thesis uses different variables and subdivides the sample to investigate
Mexican-origin interstate migration. First, it models the variables using all Mexican-
57
origin householders. Because the native11 and foreign12 born populations are shown to
differ in the first model, the first specified sample is then split into two groups. The first
sub-sample and second model is made up only of native-born, and the second sub-
sample and third model of foreign-born. By using the same variables as in the first
model, the information from the two sub-groups is used to run their multinomial models.
In sum, three multinomial models are conducted. Three tables showing their coefficients
are discussed below. Each of the models has a table giving the descriptive statistics for
each group.
All these variables and methods are then being used to evaluate the human
capital and household and community social capital hypotheses. Before moving on to
the findings, the hypotheses guiding the investigation are reiterated below.
• Human Capital:
I hypothesize that householders with a college degree will have a
higher likelihood of migrating out of the core compared to
householders with less than a high school education.
• Household Social Capital:
11 Native born individuals are those who were more in one of the 48 Continental U.S. states including
District of Columbia.
12 These include all those NOT born in Continental U.S. and D.C. These may or may not be legal citizens
of the U.S. The idea behind the formation of this group is that they have geographically come from
outside the U.S. mainland.
58
I hypothesize that householders with household social capital will have
higher likelihoods of migrating out of the core region compared to
those with no household social capital.
ÆMore specifically, I hypothesize that specific form of household
social capital will increase the odds of migration. That is, those with
periphery household social capital will be more likely to out-migrate to
the periphery region and those with frontier household social capital
will be more likely to leave the core for the frontier region.
• Community Social Capital:
I hypothesize that Mexican-origin individuals living in a PUMA with
community social capital will have a higher likelihood of migration
than those without community social capital.
59
CHAPTER IV
ANALYSIS
Introduction
The Saenzian concepts require that a trichotomous dependent variable be
modeled using a multinomial logistic regression. This chapter discusses descriptive
statistics, the reason why an unordered logit model is used and finally the very important
findings of the research. The chapter will then explain why three different multinomial
models are conducted to evaluate interregional migration.
The three multinomial logistic models only differ in the sample on which they
are based: 1) a sample of all Mexican-origin householders; 2) a sub-sample which
includes only native-born Mexican-origin householders; and 3) a sub-sample which
includes foreign-born Mexican-origin householders. Findings for each of the models
will be discussed and a brief discussion of how the control variables are operating will
be given.
Descriptive Statistics
A brief discussion on the descriptive statistics of the trichotomous dependent
variable will help show how householders are distributed across the three migration
60
categories in the main sample and two sub-samples. General descriptive statistics can be
found in Tables 4a, 5a, and 6a.
In the complete sample (n=73,824), 95.7% of the householders are regional non-
migrants, 2.8% are periphery migrants, and 1.5% are frontier migrants. The native-born
only group (n=33,297) has 96.3% regional non-migrants, 2.1% periphery migrants, and
1.6% frontier migrants. In the foreign born group (n=40,530), 95.1% are regional non-
migrants, 3.4 are periphery migrants, and 1.4% are frontier migrants. About 55% of the
householders in the sample are foreign born. In all cases at least 95% of the group is in
the regional non-migrant category. Most householder in my sample are not interregional
migrants and in all cases the smallest group is the frontier migrants.
The first of three independent variables of main interest is educational
attainment. I will briefly outline how the various educational attainment categories vary
in their migration-type percentages. In the complete sample, 94%13 of college graduate
householders (n=5,731) are regional non-migrants, 2.9% are periphery migrants, and
3.1% are frontier migrants. In the same sample, 96.1% of householders with some
college education (n=14,463) are regional non-migrant, 2.5% are periphery migrants,
and 1.4% are frontier migrants. Of those with a high school diploma or GED
(n=14,667), 96.3% are regional non-migrants, 2.5% are periphery migrants, and 1.2%
are frontier migrants. Householder with some high school education (n=13,910) have
95.3% regional non-migrants, 3.3% periphery migrants, and 1.4% frontier migrants. The
reference category of less than some high school (n=25,053) is 96% regional non-
13 Some of the percentages have been rounded for convenience.
61
migrant, 3% periphery migrant, and 1% frontier migrant. The college education
category is the only one where the regional non-migrant status claims less than 95% of
the group and where there are more frontier migrants than periphery migrants.
When the same variable of education is observed in the native-born population
there some similarities and differences. College graduates (n=4,145) are 94% regional
non-migrants, 2.8% periphery migrants, and 3.1% frontier migrants. Householders with
some college (n=10,061) are distributed as follows: 96.4% regional non-migrant, 2.1%
periphery migrant, and 1.5% frontier migrant.
In the native-born group high school graduates (n=9,377) are 97% regional non-
migrant, 1.8% periphery migrant, and 1.1% frontier migrant. Those with some high
school (n=5,892) have 97% regional non-migrant, 2% periphery migrant, and 1%
frontier migrant. The reference group of less than some high school education (n=3,819)
has 96% regional non-migrant status, 2% periphery migrant, and 2% frontier migrant.
Similar to the complete sample, the college education category in the native-born sample
is also the only one where the regional non-migrant category has less than 95% of the
group and where there are more frontier migrants than periphery migrants.
The human capital variable measure with educational attainment is now observed
in the foreign-born sample. College graduate Mexican-origin foreign-born householders
(n=1,586) are 94% regional non-migrant, 3% periphery migrant, and 1% frontier
migrant. Those with some college education (n=4,402) are 96% regional non-migrant,
3% periphery migrant, and 1% frontier migrant. High school graduate householders
(5,290) have 95% regional non-migrant, 3.7% periphery migrant, and 1.3% frontier
62
migrant. Householders with some high school education (n=8,018) are 94% no-migrant,
4% periphery migrant, and 2% frontier migrant. The reference group of those with less
than some high school education (21,234) are 95.5% regional non-migrant, 3%
periphery migrant, and 1.5% frontier migrant. There are some differences in the
foreign-born group. They have lower levels of educational attainment and their
distribution into the different migration statuses does not differ by much across
educational categories. This group is different in the human capital variable of
education in comparison to the vary similar complete and native-born samples.
Household social capital is measured with the second independent variable of
interest. The variable accounts for the presence of a household member who may have
been born outside the core prior the beginning of the migration period: 1995. In the
complete sample, those with household social capital (6,192) are 82.4% regional non-
migrant, 11.5% periphery migrant, and 6.1% frontier migrant. In the native-born group
those with household social capital (n=4,223), 88% are regional non-migrants, 7% are
periphery migrants, and 5% are frontier migrants. From the native-born group with
household capital (n=1,969) there are 71% regional non-migrants, 21% periphery
migrants, and 8% frontier migrants.
The last hypothesis under review is measure with community social capital. In
the complete sample, those with community capital (n=5,168) are 97.5% regional non-
migrant, 1.7% periphery migrant, and 0.8% frontier migrant. In the native-born group of
those with community social capital (n=2,860) 97.7% are regional non-migrant, 1.5%
are periphery migrant, and 0.8% are frontier migrant. The variable is distributed
63
differently in the foreign-born sample where those with community social capital
(n=2,308) are 97% regional non-migrant, 1.9% are periphery migrant, and only 0.1% are
frontier migrants.
Cumulative vs. Multinomial Model
Before moving on to discuss what the multinomial model is and how it is used in
the thesis, it is important to note that a valid argument may be made that the three
categories in the dependent variable are ordered. For example, I could hypothesize that
not migrating (i.e., staying in the core) is the first category because most people are more
likely not to migrate. Of those that do leave the core, I could hypothesize that most will
only move to the periphery because it is spatially closer and contains more co-ethnics.
In light of these thoughts I could then expect that my first category (regional non-
migrant) is followed by my second category (periphery migrant) which is in turn
followed by the third category (frontier migrant). Thus, I could argue that my
trichotomous dependent variable is ordered. I do not make such an argument in my
thesis. In order to make sure that I am not overlooking the validity of such an argument
I will test for it.
If the categories in the dependent variable are ordered, a cumulative logistic
regression would be required. If the categories are unordered (as I suspect they are), a
multinomial logit model would be more appropriate. I will test the fit of a cumulative
logit model using SAS in order to dismiss the possibility that the dependent variable is
ordered.
64
A cumulative logit regression requires that the proportional odds assumption,
also referred to as the parallel regression assumption, not be violated. This assumption
basically expects that the relationship between all comparisons be the same and this
allows the ordinal logistic regression to use only one equation. Testing for the
assumption is an important first step when evaluating if a cumulative model is
appropriate to model a dependent variable with more than two categories. Testing for
the proportional odds assumption will indicate if the categories in the dependent variable
are in fact ordered or unordered.
After running a cumulative logit model using the trichotomous dependent
variable of interregional migration, I can confirm that the assumption of proportional
odds is not tenable because there is a significant difference between the coefficients for
the dependent variable. The SAS cumulative logistic regression output indicates that
there is a violation of the proportional odds assumption. In technical terms, the slopes
for the cumulative logits are not common. This is a clear violation of the assumption
and in lieu of this significant finding, I conclude that modeling the data as ordered is
inappropriate. In simple language, the findings indicate that using a multinomial logistic
regression may be more appropriate for modeling my trichotomous migration-dependent
variable.
The main sample under statistical evaluation is made up of both native-born and
foreign-born Mexican-origin householders. The first model uses full sample to regress
the various independent factors on the migration dependent variable. The findings for
the first model are given in Table 8. This model contains a variable that controls for the
65
amount of time the Mexican-origin householder has lived in the U.S. In general the
variable compares foreign-born to native-born people. The coefficients are discussed at
the end of this chapter. The important thing being signaled here is that there is some
statistical evidence that foreign-born Mexican-origin individuals do have higher
likelihoods of outmigration when compared to native-born Mexican origin householders.
Length of residence is an important topic for migration research focusing in on
the Mexican-origin population. Issues of assimilation--implied in this variable--are
beyond the scope of this study. But it is possible that the native and foreign-born groups
are significantly different in many respects. For example, the native-born group may
have higher levels of educational attainment than the foreign-born group.
The graphs clearly show that the native-born group has higher levels of
educational attainment than the foreign-born group. For this and many other potential
reasons, the two groups are subjected to separate multinomial models. This will help us
understand if and how the three main hypotheses under investigation and all the other
control variables are operating for each of the two groups.
In all of the three models the migration dependent variable consists of three
potential category outcomes: regional non-migrant, periphery migrant, and frontier
migrant. Because there are three categories two comparisons are being made in each of
the models. The nonmigrant group is the base14 in all instances.
The primary questions of interest are on how the different forms of capital
influence interregional migration. Some research indicates that interregional migration
14 The comparison group is referred to as the “base” in STATA.
66
patterns between the native- and foreign-born are different (Gurak & Kritz 2000). The
findings of the first model reported in Table 8 also show that there are differences in
interregional migration between native- and foreign-born householders and that these
differences are more significant in the periphery vs. nonmigrant contrast than in the
frontier vs. nonmigrant equation. For example, foreign-born Mexicans who have only
been in the U.S. for 10 or less years have odds of being a periphery migrant versus
regional non-migrant that are 196% greater than the odds for native-born Mexicans.
Since there are differences between native- and foreign-born householders, it is
important to ascertain how the different forms of capitals are operating in each of these
two sub-groups. To understand the different dynamics between native- and foreign-
born householders I conduct separate multinomial logistic regressions for each group.
The models will help better understand how the different forms of capital are associated
with interregional migration.
When the logits for native-born regional non-migrant are set to zero, for the
native-born Mexican MNLM we have:
67
and when the logits for foreign-born regional non-migrant are set to zero for the foreign-
born Mexican MNLM we have:
Organization of Tables
Several tables are used in the discussion of the findings. The three main tables
given below are Table 8, 5b, and 6b. I will briefly delineate the layout of the tables in
order to make more efficient use of them. Uniformity is maintained across all the three
main tables in order to facilitate the discussion and allow the reader a more succinct
reference tool.
Each table has two columns. The columns represent a comparison. The first
column shows the results when the contrast is made between periphery migrants and
regional non-migrants. The second column represents the frontier migrant and regional
non-migrant comparison.
68
Within each of the columns--in all the three main tables--there are three numbers
given: 1) the logit coefficient; 2) the exponentiated logit coefficient; and 3) the relative
risk ratio (referred to simply as the percent). The percent is obtained by simply
subtracting one from the exponentiated value and then multiplying the result by 100.
The numbers all represent the same result but in different ways. The percents or relative
risk ratios--sometimes referred to as percent change in odds ratio--are the numbers being
discussed in the thesis. The reason why these numbers are highlighted--rather than
others--is because percentages lend themselves to straightforward interpretations.
It is important that the reader keep in mind that all the main tables display all
coefficients signaling those with a significance level of .0515 or better with a single
asterisk. The three main independent variables of interest (human capital, household
social capital, community social capital) are given at the top of the list followed by the
various control variables. Each of the tables gives the models' likelihood ratio chi-
square (along with its probability), pseudo r-square, and the number of observations.
Interpreting Coefficients
The coefficients in all the models have similar interpretations. Since all my
factors are coded as dummy variables, the odds ratio outputs are interpreted by
comparing the variable of interest to its reference category16. The ratio is then
interpreted by comparing the category of interest to its reference group and then framing
15 As a matter of fact more than 85% of the coefficients are significant at the 0.0001 alpha level.
16 The reference category is given in the table for each variable and can be identified because it only
contains a "ref" statement whereas all the others contain numbers.
69
the comparison as it relates to the contrast under review. In technical terms, t
exponentiated number is the relative risk of being in the non-base category in relation to
the reference group. An example follows in order to ground these statistical
abstractions.
he
The model including the complete sample controls for the householder's sex.
The periphery- versus regional non-migrant contrast (found in the first column of Table
8) indicates that the MALE variable has a raw coefficient of .44 and when it is
exponentiated this number equals 1.55. As indicated by the "ref" input in the table, this
variable uses females as the reference category.
Using the interpretation discussion given above we would interpret the MALE
ratio in the first contrast of Table 8 by saying something like: "Males are 55% more
likely to be periphery migrants when compared to females."
Multinomial Logistic Model for Full Sample
This section discusses the results for the model using the full sample that
contains both foreign- and native-born Mexican-origin householders. Table 7 below
gives the descriptive statistics for the complete sample and the STATA multinomial
logistic model (from here on only referred to as MNLM) outputs for all Mexican-origin
householders are listed in Table 8 below.
Table 7
Descriptive Statistics: All Mexican-origin Householders (n=73,824)
Variable Mean SD
70
Community Capital
.07 .26
Household Capital
HH with Periphery Capital
HH with Frontier Capital
.04
.04 .21
.20
Educational Attainment
Some High School
High School Graduate
Some college
College Graduate & Above
.19
.20
.20
.08
.39
.40
.40
.27
Age
30-39
40-49
50-59
.32
.35
.24
.47
.48
.43
Male
.76 .43
Married
.71 .46
Year Entered U.S.
85-94
75-84
65-74
1964
.13
.21
.15
.06
.34
.41
.36
.24
Presence of Own Child <18
.16 .37
Language
MonoEnglish
Bilingual .16
.58 .49
.26
Table 8
Multinomial Results for Full Model
Periphery
vs.
Regional non-migrant
Frontier
vs.
Regional non-migrant
71
b Exp (b) % b Exp
(b) %
Capital Variables
Human Capital
8th grade or below
Some High School
High School Graduate
Some college
College Graduate & Above
Ref
.08
-.06
-.01
.09
Ref
1.08
.95
.99
1.10
Ref
8
-5
-1
10
ref
-.21
-.32
-.25
.49
ref
.81
.73
.78
1.63
ref
-19*
-27*
-22*
63*
Household Social Capital
Householder with Periphery Capital
Householder with Frontier Capital
Householder with no H-capital
2.66
.68
ref
14.28
1.98
ref
1228*
98*
Ref
.88
2.41
ref
2.41
11.18
ref
141*
1018*
ref
Community Social Capital
Has community capital
No community capital
-.66
ref
.51
ref
-49*
Ref
-.78
ref
.46
ref -54*
ref
Control Variables
Age
30-39
40-49
50-59
60-69
1.61
.81
.43
ref
4.99
2.25
1.54
ref
399*
125*
54*
Ref
1.83
.98
.26
ref
6.26
2.67
1.29
ref
526*
167*
29
ref
Male
Female .44
ref 1.55
ref 55*
Ref .26
ref 1.30
ref 30*
ref
Married
Not married
-.46
ref .63
ref -37*
Ref -.46
ref .63
ref -37*
ref
Year Entered U.S.
Native-born
85-94
75-84
65-74
1964 or before
ref
1.08
.53
.33
-.07
ref
2.96
1.69
1.39
.93
ref
196*
69*
39*
-7
ref
.40
-.18
-.24
-.39
ref
1.50
.84
.79
.68
ref
50*
-16
-21
-32
Presence of Own Child <18
No own children <18 present
-.13
ref .87
ref -13*
ref -.25
ref .78
ref -22*
ref
Language
MonoEnglish
Bilingual
MonoSpanish
-.25
-.39
ref
.78
.67
ref
-22*
-33*
ref
-.45
-.78
ref
.61
.64
ref
-39*
-36*
Ref
LR chi(2)
Probability > chi2
Pseudo R2
Number of observations
4273.57
0.0000
0.1404
73,824
72
*Significant at the p < 0.05 level or lower
(b)=Raw coefficient
Exb (b)=Factor change in odds for unit increase in X
%=Percent change in odds for unit increase in X
When comparing periphery migrants to regional non-migrants in this model, the
human capital variables are not significant. Human capital is, however, significant in the
frontier versus regional non-migrant contrast. The reason why may have to do with
migrating out to the frontier region poses more challenges and thus education becomes a
more significant predictor in migration. By way of foreshadowing, this association
remains present even when the model is native- and foreign-born specific.
As stated earlier, education is significant in the frontier and regional non-
migrants comparison. I will now outline the significant associations in more detail.
Mexican-origin individuals with some high school (-19%), high school graduates (-
27%), and those with some college (-22%) have a lower likelihood of being frontier
migrants (than regional non-migrants) when compared to individuals with less than a
high school education. On the other hand and as hypothesized, those with a college
degree are 63% more likely to be frontier migrants (than regional non-migrants)
compared to Mexican-origin individuals with less than some high school education.
The human capital hypothesis finds support in the frontier and regional non-migrant
contrast.
Strong support is found for the household social capital hypothesis across both
equations. Under the periphery and regional non-migrant comparison, householders
with any type of household social capital have higher likelihoods of being periphery
73
migrants (than regional non-migrants) when compared to individuals with no household
member over the age of five who was born outside the core.
As hypothesized more specifically, the periphery versus regional non-migrant
contrast shows that those with periphery household social capital have odds of being
periphery migrants that are 13 times greater than the odds for those with no household
social capital and those with frontier household social capital have 68% greater odds of
being periphery migrants than those without such household social capital. The frontier
vs. nonmigrant equations indicate that those with periphery household social capital
have odds of being frontier migrants (versus regional non-migrant) that are 141% greater
than the odds for those with no periphery household social capital, while they have
1018% greater odds of being a frontier migrants (than a regional non-migrant) compared
to those with no household social capital.
A clear pattern shows up here: those with periphery household social capital have
greater odds of being periphery migrants while those with frontier household social
capital have greater odds of being a frontier migrant when compared to Mexican-origin
householders with no form of household capital. The household capital hypothesis finds
supported in the full sample MNLM.
The community social capital hypothesis is not supported17, however the results
are significant but opposite of what was expected. Individuals living in PUMAs with
community social capital have lower odds of being migrants than those with no
community social capital. Table 8 shows that those with community social capital have
17 Background analysis has shown that the variable is working as hypothesized.
74
odds of being a periphery migrant (than a regional non-migrant) that are 49% lower than
the odds for those with no community capital. The odds of being a frontier migrant
(versus a regional non-migrant) are 54% lower than the odds for those with no
community social capital. A possible explanation here may be that in-migrants are
going to communities where there is a robust economy. If this is so, then residents of
those PUMAs may have less of an incentive to move out of the region because economic
conditions are favorable for them to stay.
In short, Mexican-origin individuals living in PUMAs with community social
capital have lower likelihoods of being migrants than regional non-migrants. This
finding fails to support the community social capital hypothesis that speculated that
Mexican-origin individuals living in areas with community capital would have higher
likelihoods of being migrants than those living in PUMAs with no community capital.
Before moving on to the following two models subdividing the full sample by
nativity status, I briefly discuss the findings for the control variables in the model.
The relationship between the control variables and migration are as expected.
They relationships are fairly consistent across all the models. Householders below 60
years of age have a greater likelihood of being migrants than those who are 60 or older.
For example, in the periphery vs. nonmigrant comparison those between the ages of 30
and 39 have odds of being periphery migrants vs. nonmigrant that are 399% greater than
the odds for those between the ages of 60 and 69.
75
In addition, males have greater odds of being migrants than females. Married
householders have odds of being frontier or periphery migrants versus regional non-
migrant that are 37% lower than the odds for currently unmarried individuals.
Year entering the U.S. is significant in most categories in the periphery and
regional non-migrant contrast--and only in one instance in the frontier and regional non-
migrant equation. The pattern is as expected: those who have less time living in the
U.S. have greater odds of being migrants (than regional non-migrants) when compared
to native-born individuals.
Finally, having an own child below the age of 18 is associated with lower odds of
leaving the core and being a monolingual-Spanish speaker is found to be associated with
higher odds of leaving the core. For instance, the presence of an own child is associated
with having odds of being a frontier migrant (vs. nonmigrant) that are 22% lower than
the odds for those householders without a child. In the case of language in the periphery
versus regional non-migrant contrast, both monolingual-English speakers (-22%) and
bilinguals (-33) have lower odds of being migrants compared to monolingual-Spanish
speakers and in the frontier and regional non-migrant equation the same pattern appears
and is significant.
Multinomial Logistic Model for Native-born Sample
This section discusses the results for the model using only the native-born
individuals in the sample. Table 9 gives the descriptive statistics for this group and the
STATA regression outputs are listed in Table 10 below.
76
Table 9
Descriptive Statistics: Native-born Mexican-origin Householders (n=33,294)
Variable Mean SD
Community Capital
.09
.28
Household Capital
HH with Periphery Capital
HH with Frontier Capital
.06
.07 .24
.25
Educational Attainment
Some High School
High School Graduate
Some college
College Graduate & Above
.18
.28
.30
.12
.38
.45
.46
.33
Age
30-39
40-49
50-59
.29
.34
.27
.45
.48
.44
Male
.71 .45
Married
.64 .48
Presence of Own Child <18
.11 .31
Language
MonoEnglish
Bilingual .29
.67 .45
.47
Table 10
Multinomial Results for Native-born Model
Periphery
vs.
Regional non-migrant
Frontier
vs.
Regional non-migrant
Variable b Exp (b) % b Exp
(b) %
Human Capital
77
8th grade or below
Some High School
High School Graduate
Some college
College Graduate & Above
ref
-.14
-.25
-.14
.08
ref
.87
.78
.87
1.08
ref
-13
-22
-13
8
ref
-.77
-.87
-.61
.06
ref
.46
.42
.54
1.07
ref
-33*
-13*
-39*
7
Household Social Capital
Householder with Periphery Capital
Householder with Frontier Capital
Householder with no H-capital
2.16
.78
ref
8.65
2.17
ref
765*
117*
Ref
.85
2.01
ref
2.33
7.51
ref
133*
151*
Ref
Community Social Capital
Has community capital
No community capital
-.42
ref
.66
ref
-34*
Ref
-.79
ref
.46
ref -54*
Ref
Age
30-39
40-49
50-59
60-69
2.04
1.01
.60
ref
7.66
2.73
1.83
ref
666*
173*
83*
Ref
1.96
.96
.18
ref
7.13
2.62
1.20
ref
663*
162*
20
ref
Male
Female
.42
ref 1.52
ref 52*
Ref .24
ref 1.27
ref 27*
Ref
Married
Not married
-.53
ref .59
ref -41*
Ref -.48
ref .62
ref -38*
Ref
Presence of Own Child <18
No own children <18 present
-.10
ref .90
ref -10
ref -.26
ref .77
ref -13
Ref
Language
MonoEnglish
Bilingual
MonoSpanish
-.89
-1.28
ref
.41
.28
ref
-59*
-72*
Ref
-.62
-.75
ref
.54
.47
ref
-46*
-53*
Ref
LR chi(2)
Probability > chi2
Pseudo R2
Number of observations
1638.15
0.0000
0.1352
33294
*Significant at the p < 0.05 level or lower
(b)=Raw coefficient
Exb (b)=Factor change in odds for unit increase in X
%=Percent change in odds for unit increase in X
Similar to findings with the full sample, human capital is not significant when
comparing periphery migrants to regional non-migrants among native-born
78
householders. Education is significant in the frontier versus regional non-migrant
comparison up to the some college category. The college degree category is no longer
significant as in the earlier analysis involving the full sample. The human capital
hypothesis finds no support when only the native-born Mexican-origin group is used
conduct the analysis.
Support for the household social capital hypothesis remains present, however
individuals with either form of periphery or frontier household social capital have
greater odds of being migrants than those with no household social capital. Household
specific capital still has the same effect of increasing the odds a bit more for the region
under contrast with the base. For example, the native-born model under the periphery
and regional non-migrant equation indicates that Mexican-origin householders with
periphery household social capital have odds of being a periphery migrant (versus a
regional non-migrant) that are 8 times greater than the odds for those without household
social capital. While the same contrast shows that those with frontier household social
capital only have 117% greater odds of being periphery migrants than regional non-
migrants when compared to individuals with no household social capital.
Nonetheless, the community social capital hypothesis is not supported here
either. The findings are significant but with an association contradicting the one
expected through the stated hypothesis. The coefficients indicate that native-born
Mexican-origin householders living in PUMAs with community social capital have
lower odds of being migrants (than regional non-migrants) compared to those with no
community social capital. This is the opposite of what was expected. Table 10 shows
79
that those with community social capital have odds of being a periphery migrant (versus
being a nonmigrant) that are 34% lower than the odds for those with no community
social capital. In sum, those living in areas with community social capital have lower
likelihoods of being migrants when compared to householders living in PUMAs with no
community capital.
The control variables stay the same as in the first model except for one
difference: the presence of an own child under the age of 18 is no longer significant.
Multinomial Logistic Model for Foreign-born Sample
Results for the foreign-born sample are given in Tables 11 and 12.
Table 11
Descriptive Statistics: Foreign-born Mexican-origin Householders (n=40,530)
Variable Mean SD
Community Capital
.06
.23
Household Capital
HH with Periphery Capital
HH with Frontier Capital
.03
.02 .17
.14
Educational Attainment
Some High School
High School Graduate
Some college
College Graduate & Above
.20
.13
.11
.04
.40
.34
.31
.19
Age
30-39
40-49
50-59
.36
.36
.22
.48
.48
.41
80
Male
.79 .40
Married
.76 .43
Year Entered U.S.
85-94
75-84
65-74
.24
.38
.27
.43
.49
.44
Presence of Own Child <18
.21 .41
Language
MonoEnglish
Bilingual .05
.52 .26
.50
Table 12
Multinomial Results for Foreign-born Model
Periphery
vs.
Regional non-migrant
Frontier
vs.
Regional non-migrant
Variable b Exp (b) % b Exp
(b) %
Human Capital
8th grade or below
Some High School
High School Graduate
Some college
College Graduate & Above
Ref
.13
.03
.05
-.04
Ref
1.14
1.03
1.05
.96
Ref
14
3
5
-4
Ref
-.02
.02
-.28
.67
Ref
.98
1.02
0.76
1.95
ref
-2
2
-24
95*
Household Social Capital
Householder with Periphery Capital
Householder with Frontier Capital
Householder with no H-capital
3.08
.56
ref
21.78
1.75
ref
2078*
75*
Ref
.96
3.04
ref
2.60
21.00
ref
160*
2100*
Ref
Community Social Capital
Has community capital
No community capital
-.92
ref
.40
ref
-60*
Ref
-.82
ref
.44
ref -66*
Ref
81
Age
30-39
40-49
50-59
60-69
1.13
.52
.22
ref
3.10
1.69
1.25
ref
210*
69*
25
ref
1.61
.96
.33
ref
5.00
2.61
1.39
ref
400*
161*
39
ref
Male
Female
.45
ref 1.57
ref 57*
Ref .31
ref 1.36
ref 36*
Ref
Married
Not married
-.40
ref .67
ref -33*
ref -.41
ref .66
ref -34*
ref
Year Entered U.S.
85-94
75-84
65-74
1964
1.48
.84
.51
ref
4.39
2.32
1.67
ref
339*
132*
67*
ref
1.09
.45
.28
ref
2.97
1.57
1.32
ref
197*
57
32
ref
Presence of Own Child <18
No own children <18 present
-.16
ref .85
ref -15*
Ref -.26
ref
.77
ref -13*
Ref
Language
MonoEnglish
Bilingual
MonoSpanish
-.43
-.21
ref
.65
.81
ref
-35*
-19*
Ref
-.66
-.39
ref
.52
.68
ref
-48*
-32*
Ref
LR chi(2)
Probability > chi2
Pseudo R2
Number of observations
2753.74
0.0000
0.1513
40530
*Significant at the p < 0.05 level or lower
(b)=Raw coefficient
Exb (b)=Factor change in odds for unit increase in X
%=Percent change in odds for unit increase in X
The human capital variable of education is even less significant among foreign-
born Mexican-origin householders. In the first model (full sample) all the education
categories were significant under the frontier and regional non-migrant contrast. In the
second model (native-born sample) being a college graduate is a significant predictor for
migrating to the frontier. In this third model (foreign-born sample) the only significant
category is the college graduate: those with a college degree have odds of being a
82
frontier migrant (versus regional non-migrant) that are 95% lower than the odds for
those with less than some high school education. In lieu of these findings, the human
capital hypothesis finds support when the foreign-born population is used to conduct the
analysis.
The household social capital hypothesis, however, is supported here as in the
previous two models based on the full sample and the native-born sample). Foreign-
born Mexicans with household social capital have greater odds of being migrants (than
regional non-migrants) compared to their counterparts that lack household social capital.
The household capital specific association with out-migrating to the specific region
where the capital was attained is even more pronounced in this group.
For example, foreign-born Mexicans with periphery household social capital
have odds of being a periphery migrant (versus a regional non-migrant) that are 20 times
greater than the odds for those without periphery household social capital. Among the
foreign-born, those with frontier household social capital have 21 times greater odds of
moving to the frontier compared to those with no frontier household social capital.
Nonetheless, the community social capital hypothesis is not supported here
either: the opposite of the expected is still occurring as in the first two models. Foreign-
born Mexican-origin householders living in PUMAs with community social capital have
lower odds of being interregional migrants than those who reside in areas with no
community social capital. The regressions indicate that those with community capital
have odds of being a periphery migrant versus regional non-migrant that are 60% lower
than the odds for those with no community social capital.
83
Summary of Findings
In sum, the analysis examined three hypotheses involving the relationship
between different forms of human and social capital and migration. Overall, the
hypothesis suggesting that more educated individuals have a greater propensity to
migrate out of the core region than less educated individuals receives support in the
analysis involving the full sample and the foreign-born sample. In addition, the
hypothesis suggesting that individuals with household social capital have a greater
tendency to migrate out of the core region than those without such social capital receives
consistent support. Finally, the hypothesis indicating that individuals living in areas
endowed with community social capital have a greater likelihood of migrating out of the
core region compared to those who live in communities that lack this form of social
capital is not supported, with the results consistently in the opposite direction.
The next chapter discusses the implications of these findings.
84
CHAPTER V
SUMMARY
It is my believe that advancing sociological knowledge by studying how the
different forms of human and social capital relate to the interregional migratory behavior
of Mexican-origin individuals across the Saenzian regions will help sociologist better
understand the factors associated with their migration.
A premised in this research has been that migration is a dynamic process by
which populations can find a sustenance equilibrium through which communities can
insure their survival. Although it is beyond the scope of this study, an underlying
assumption within this premise is that migration is important as human beings adjust to
their environment. In this final chapter I explain how the various hypotheses under
investigation help us better understand how the Mexican-origin population engages in
the migration process to insure its survival.
85
I will first outline the general conclusions of the study, then I will delineate some
shortcomings in the project and how future researchers may contribute further research
to this topic. I will conclude the chapter by discussing the implications of the study.
Conclusions
In examining Mexican-origin interregional migration from the Southwest region,
the thesis hopes to have expanded demographic theory and methods. By using
multinomial logistic modeling the research has expanded academic knowledge on the
complex migration patterns of Mexican-origin individuals residing in the United States.
The investigation addresses the following general question: How are human capital and
household social capital and community social capital associated with Mexican-origin
interregional migration from the Southwest into the other U.S. regions?
By using the various sociological schools of thoughts, the study tested three
different hypotheses to evaluate if interregional migration theories of the general
population are useful in helping us understand the migration of the Mexican-origin
population living in the Southwest. The results indicate that they are useful in this
pursuit.
The human capital hypothesis that householders with a college degree have a
higher likelihood of migrating out of the core compared to those with less than a high
school education is only given support in both the full and foreign-born samples. The
educational attainment variable loses significance in the native-born sample, where it
plays a significant role only in the mid-range educational categories. One of the reasons
86
for this findings may be that the human capital measure is significant because in general
foreign-born (who as a group have lower levels of educational attainment than the
native-born group) have greater odds of migrating than the native-born and those with
high education have even greater odds of migrating.
Because foreign-born Mexican-origin householders may be moving about in
different labor sectors than their native-born counterparts and maybe even in different
social networks where human capital is not as instrumental in actualizing the desire to
migrate, it is possible that the level of education variable is not as useful in expanding
our understanding of how the native-born successfully engage in migration social
networks relative to their human capital. The “formal”18 nature of human capital may
only matter in the migration process when the potential migrant needs to move through
more “formal” migration links.
Using household social capital, I hypothesized that those with periphery
household social capital would have greater odds of out-migration to the periphery
region than those with no household social capital or frontier social capital. By
extension, I also hypothesized that those with frontier household social capital would
higher likelihoods of out-migration to the frontier region than householders with no
social capital or with periphery social capital. These two specific expectations under the
household social capital hypothesis are strongly supported. This variable is significantly
associated with outmigration when modeling Mexican-origin interregional migration
18 By formal I mean that attaining higher levels of human capital require the individual to engage for
longer periods of time with formal institutions where successful completion of requirements is
acknowledge through formal certifications.
87
across nativity-status group. The presence of a household member born outside the core
region has a more powerful association among foreign-born householders than native-
born. The “informal”19 nature of household social capital may be exerting a consistently
significant influence on the interregional migration process in the Southwest among the
Mexican-origin population because it is a useful measure that actually detects migration
network links.
The last of three hypotheses, the community social capital hypothesis, posited
that Mexican-origin householders residing in a PUMA with community social capital
have greater odds of migration than those with no community social capital finds no
support. The original hypothesis may have relied too heavily on the idea that the
increased presence of accessible migration social networks in PUMAs with community
social capital would “pull” potential migrants more easily into the periphery and frontier
regions.
This measure may be like a double-edged sword. On the one hand, a PUMA
with community social capital does afford its residence the potential for increased
migration social contacts, while on the other hand it may also mean that if others (i.e.,
non-core in-migrants) are migrating to it, the PUMA may be experiencing a healthy
economy which may mean better labor market conditions for its Mexican-origin
residents. Thus, it is necessary that future research find a way to control for the
19 In most cases household associations are developed through interpersonal relationships with others
through informal settings. Thus, by informal I mean that household members are a privately held and
informally attained capital.
88
PUMA’s economic situations while accounting for how community social capital may
be associated with interregional migration among in the Mexican-origin population.
By using Saenzian concepts, the research assessed the extent to which Mexican-
origin individuals migrated to the different regions of the country and the factors are
associated with their out-migration from the core into the rest of the U.S. By using
multinomial logistic models I have measured the association of human capital and
household social capital and community social capital with interregional migration
among Mexican-origin U.S. residents.
In general I find that migration-selectivity sociological theories derived from
studies of the general population and Saenzian region concepts are useful in explaining
Mexican-origin interregional migration. While the findings are unable to support the
community social capital hypothesis and find limited support for the human capital
hypothesis, they are able to lend strong support for the household social capital
hypothesis.
Shortcomings
One of the limitations of the study is that the Saenzian regions are created using
state level aggregations, it calculates the percent of Mexicans by taking quantities
representative of the whole state. For example, Florida is considered a frontier state in
2000 when its total population of 9,938,444 contains 220,769 persons of Mexican-origin.
The results indicate that 2.22% of the state is Mexican-origin and this falls below the
relative average across the 44 states of 2.4% discussed in an earlier chapter.
89
The state level percent Mexican-origin concentration hides the fact that at smaller
geographical areas there are significant ethnic-geo-density-differences. For example, at
the county level Michigan has 83 counties ranging from 0.25% to 8.98% in their percent
of Mexicans in the county. Using the 2.4% threshold would mean that 15 counties
would be considered as part of the periphery areas instead of frontier areas. For
example, using the above stated threshold would mean that 18 of the 83 counties are
periphery areas and that the remaining 65 are frontier counties. Several counties—
Otawa, Lenawee, Saginaw, Van Buren, and Oceana—even have over 5% Mexican-
origin concentrations.
Another limitation concerns how the household social capital variable was
created. When creating this variable, the relationship of the child to the householder is
not ascertained. That is, if a child below the age of 5 was present in the household unit,
then the householder is assigned a value of “1” on the household social capital variable.
The non-core born-child may be a niece/nephew, grandchild, own or adopted child, and
even non-biologically related child among other potential relationship types. This may
not be such a big problem because—hypothetically—any relationship within the
household associated with a non-core region signifies a tie.
This means that those with an own biological child born outside the core are
given the same "network" connections as those who may have a non-biologically related
non-core-child in the household. The first scenario may be more appropriate for
measuring if the individual under investigation has a physical and direct connection to a
90
state outside the core while the latter may be measuring weaker non-core-network
connections.
Future Research
An important project in the future may undertake the same endeavor but instead
of using state level information for creating the Saenzian regions, it could use PUMA
level information using multilevel modeling. Creating the ethnic concentration estimates
for all the PUMAs requires a serious commitment and an explanation of the
sophisticated Bayesian hierarchical may be a tedious job, but such a project may help
show important associations in the variables being used in this project and even
demonstrate if context is important. This multileveling approach (using PUMA level
aggregations to create the Saenzian regions) may also find help from spatial modeling
with GIS software to determine a potentially different way of the designation of areas
comprising the core, periphery, and frontier.
An extensive evaluation of new destinations for Mexican-origin individuals may
also help illuminate research as to current migration trends that may be influence by the
post 9/11 anti-immigrant sentiment. Examining what regions have continually increased
in their Mexican-origin population may help ethnic population studies understand how
previously frontier regions are not becoming more like the periphery.
Another potential modification of this thesis for future study may be to include
Mexican-origin individuals who live in the Southwest and migrate crossing international
borders. The variables may indicate a different selectivity mechanism for those leaving
91
the core and moving to Alaska, Puerto Rico, Hawaii, and/or other non-U.S. countries.
Including such migrants and return-migrants may help understand if their migration has
any association with the capital hypotheses under investigation.
The study may be conducted using more recent data. This would help evaluate
how the different hypotheses operate in the current U.S. Mexican-origin population.
Recent data may be used to study their interregional migration using multilevel and
spatial analysis that includes other context level factors beyond the community social
capital measured used in this study.
For example, it may include an ethnic-concentration PUMA measure. What
percent of the area is composed of co-ethnics and those this exert any pull forces on their
migration or are there any significant interactions between this context measure and the
individual level human capital? Future researchers may also want develop an “economic
health measure” of the PUMA. These types of context measure may help ascertain if
and how different labor sectors at the place of residence and the place of destination
interact with individual-level factors to influence interregional migration though the
various human and social capitals used in the analysis with Saenzian concepts.
Implications
The research helps further migration theory by studying how general sociological
theories on interregional migration apply to the Mexican-origin population. The thesis
expands on the existing sociological literature by delineating how the different forms of
human and social capital influence the odds of Mexican-origin householders migrating
92
from the core region to the periphery and frontier region. The analysis only finds clear
and consistent support for the household social capital hypothesis.
Latinos/as are the largest minority group in the United States. Mexican-origin
people are the largest group under the Latina/o ethnic umbrella. Understanding how
various factors are associated with their movement out of the core region may further
inform policy makers and demographers on this ethnic population. In general, only 5
percent of the Southwest Mexican-origin population of 1995 migrated out of the core.
Most people stayed in their region of residence. It is important to understand how the
small group of people leaving influence their communities.
If poverty stricken PUMAs suffer brain-drain and having more educated
residents lowers poverty, then remedying poverty in poor communities may be more
challenging. The full sample model shows that college graduates have higher odds of
migrating out of the Southwest than those with less than some high school education,
while those between some high school and some college education had lower odds of
leaving the core than householders with less than a high school education. Thus, “mid-
range” educated householders (some high school to some college) may be more willing
to stay in the core region than their less than high school educated counterparts because
the former are more equipped for participation in potentially depleted local labor
markets.
The mid-rangers' apparent advantage may not be the motive for their non-
migration, because college graduates may be more equipped to act on the need to
increase their educational returns by migrating, while mid-range householders may be
93
more limited by their formal education skills. As a consequence, the less than high
school educated Mexican-origin householders may be “forced” to face the many
struggles and obstacles necessary in the migration process, whereas college graduates
have the “advantageous choice” of migrating in order to increase their returns on
education.
Policy makers trying to improve impoverished conditions among Mexican-origin
areas need to look for ways of keeping college educated individuals in their challenged
communities. This may help improve the economic condition of this growing ethnic
minority and by doing so allow a more stable integration mechanism that supports our
North American democratic ideals.
On the theory side, the findings highlight how the processual migration idea is on
a potentially fruitful path for scientific discovery. Of the three hypotheses investigated,
the household social capital seems to be the most stable predictor for interregional
migration among Mexican-origin householders living in the Southwest. Across nativity-
status models demonstrate that is a household member born outside the core is present;
there may be “strong” social network connections to the periphery or frontier region
influencing the decision to migrate. This variable seems to best capture the presence of
non-core migration network links. Using variables that account for actual links to
migrating social networks may help social demographers better specify how a
hypothetical social network measure is being conceptualized to frame the migration
networks link.
94
The main research question of how human capital and household social
capital and community social capital are associated with Mexican-origin interregional
migration from the Southwest into the other U.S. regions is answered after examining
three hypotheses. I find that human capital theory is supported in the full and foreign-
born samples and that the community social capital hypothesis is not supported. I also
find that the household social capital hypothesis is consistently supported even across
nativity-status groups.
An important conclusion I make is that using Saenzian concepts to trichotomize
the continental U.S. into three ethnic population based regions helps the various
variables used in statistical models to render us with nuanced and crucial associations.
For example, using Saenzian concepts allowed me to show that region-specific
household social capital matters in migration and that human capital matters more in
frontier migration than periphery migration. The project has explored Mexican-origin
interregional migration from the Southwest and finds that general sociological theories
apply.
95
REFERENCES
Acuña, Rodolfo. 1972. Occupied America: The Chicano's Struggle Toward Liberation.
Harper & Row Publishers, Canfield Press: San Francisco, California.
Acuña, Rodolfo. 2006. Occupied America: A History of Chicanos (4th addition). Austin
and London: University of Texas Press.
Aguilera, Michael B. and Douglas S. Massey. 2003. "Social Capital and the Wages of
Mexican Migrants: New Hypotheses and Test." Social Forces, Vol. 82(2):671-
701.
Alexander, John T. 1979. Spanish Surname Recent Migrant Families: Life Cycle,
Family, Socioeconomic and Housing Status. Palo Alto, California: R & E
Research.
Anaya, Rudolfo A. and Francisco Lomeli. 1989. Essays on the Chicano Homeland.
Albuquerque: Academia/ El Norte.
Anonymous. 1936. “Interstate Migration of Negroes.” Monthly Labor Review, Vol. 42:
976-978.
Arreola, D. D. 1985. “Mexican Americans.” In Ethnicity in Contemporary America: A
Geographical Appraisal, ed. Jesse O. McKee, 77-94. Dubuque, IA: Kendall/Hunt
Publishing.
Barrera, Mario. 1979. Race and Class in the Southwest: A Theory of Racial Inequality.
Notre Dame and London: University of Notre Dame Press.
Beale, Calvin L. 1973. “Migration Patterns of Minorities in the United States.” American
Journal of Agricultural Economics 55: 938-946.
Bean, Frank D., and Martha Tienda. 1988. “A Demographic Profile of the Hispanic
Population" Pp. 56-136 in The Hispanic Population of the United States, edited
by Frank Bean and Martha Tienda, New York: Russell Sage Foundation.
Becker, Gary S. 1964. Human Capital: A Theoretical and Empirical Analysis, with
Special Reference to Education, New York: National Bureau of Economic
Research.
Bernstein, Robert. 2007. More than 300 Counties Now "Majority-Minority. News
Released on Thursday, August 9,2007. U.S. Census Bureau News.
http://www.census.gov/Press-
Release/www/releases/archives/population/010482.html
Bishaw, Alemayehu and Jessica Semega. 2008. "Income, Earnings, and Poverty Data
From the 2007 American Community Survey." American Community Survey
Reports: U.S. Census Bureau. http://www.census.gov/prod/2008pubs/acs-09.pdf
Bogardus, Emory S. 1930. "The Mexican Immigrant and Segregation." The American
Journal of Sociology, Vol. 36(1):74-80.
Bogue, Donald J. 1949. “Subregional Migration as an Area of Research.” Social Forces,
Vol. 27: 392-400.
Bogue, Donald J., and Warren S. Thompson. 1949. “Migration and Distance.” American
Sociological Review, Vol. 14(2): 236-244.
Bogue, D. 1959. Internal Migration, Pp. 486-510 in The Study of Population Edited by
P. Hauser & O.D. Duncan. Chicago: University of Chicago Press.
96
Bohning, W.R. 1972. The Migration of Workers in the United Kingdom and the
European Community. London: Oxford University Press.
Borjas, George J. 1992b. “Self-Selection and Internal Migration in the United States.”
Journal of Urban Economics, Vol. 32(2): 159-185.
Borjas, George J. 2006. “Native Internal Migration and the Labor Market Impact of
Immigration.” Journal of Human Resources, Vol 41(2): 221-258.
Boswell, Thomas D. 1979. “The Growth and Proportional Distribution of the Mexican
Stock Population in the United States: 1910-1970.” Mississippi Geographer 7(1):
57-76.
Bracanovic, Tomislav. 2006. "May we Take Sociobiology Seriously? Neutralizing some
Objections." Drustvena Istrazivanja, Vol. 15(3): 429-452.
Bright, Margaret L., and Dorothy S. Thomas. 1941. “Interstate Migration and
Intervening Opportunities.” American Sociological Review, Vol. 6(6): 773-783.
Brunton, Anne. 1973. “The Chicano Migrant.” Pp. 485-497 in Introduction to Chicano
Studies, edited by Livie Isauro Duran and H. Russell Bernard. New York: The
MacMilliam Company.
Bushee, Alice H. 1923. "Spanish Influence in the Southwest." Hispania, Vol. 6(3):148-
157.
Calavita, Kitty. 1992. Inside the State: The Bracero Program, Immigration and the
I.N.S. New York: Routledge.
Cai, Qian. 2007. “New Techniques in Small Area Population Estimates by Demographic
Characteristics.” Population Research and Policy Review, 26: 203-218.
Camarota, Steven A. 2007. "Immigrant Employment Gains and Native Losses, 2000-
2004." Pages 139-156 in Debating Immigration, edited by Carol M. Swain. New
York: Cambridge University Press.
Cameron, Christopher David Ruiz. 2000. "One Hundred Fifty Years of Solitude:
Reflections on the End of the History Academy's Dominance of Scholarship on
the Treaty of Guadalupe Hidalgo." Bilingual Review, Vol.25(1):1-22.
Cardenas, Gilbert. 1976. “’Los Desarraigados’: Chicanos in the Midwestern Region of
the United States.” Aztlan: International Journal of Chicano Studies Summer,
Research 7: 153-186.
Cardenas, Gilbert. 1977. “Mexican Labor: A View To Conceptualizing the Effects of
Migration, Immigration and the Chicano Population in the United States.” Pp.
159-181 in Cuantos Somos: a Demographic Study of the Mexican American
Population, edited by Charles H. Teller, Leo F. Estrada, Jose Hernandez and
David Alvirez, eds Publisher: Austin: Center for Mexican American Studies,
University of Texas.
Cebula, Richard J. 1974. "Interstate Migration and the Tiebout Hypothesis: An analysis
According to Race, Sex and Age." Journal of the American Statistical
Association, Vol. 69(348):876-879.
U.S. Census Bureau. 2006. Facts for Features September 5 Newsroom: CB06-ff.14.
U.S. Census Bureau. 2007. Presentation created by: Ethnicity and Ancestry Branch
Population Division. Anna M. Owens:
97
http://www.census.gov/population/www/socdemo/hispanic/files/Internet_Hispani
c_in_US_2006.pdf.
U.S. Census Bureau. 2007. Summary File 3: 2000 Technical Documentation.
http://www.census.gov/prod/cen2000/doc/sf3.pdf.
U.S. Census Bureau. 2008. Public Use Microdata Sample: 2000 Census of Population
and Housing. PUMA/16-US(RV).
http://www.census.gov/prod/cen2000/doc/pums.pdf
Chaves, Maria L., Brian Wampler, and Ross E. Burkhart. 2006. "Left Out: Trust and
Social Capital Among Migrant Seasonal Farmworkers." Social Science
Quarterly, Vol. 87(5): 1012-1029.
Coleman, James S. 1988. “Social Capital in the Creation of Human Capital.” American
Journal of Sociology 94:S95-S120.
Crowder, Kyle, and Scott J. South. 2005. "Race, Class, and Changing Patterns of
Migration between Poor and Nonpoor Neighborhoods." American Journal of
Sociology, Vol. 110(6):1715-1763.
Crowley, Martha, Daniel T. Lichter, and Zhenchao Qian. 2006. “Beyond Gateway
Cities: Economic Restructuring and Poverty Among Mexican Immigrant
Families and Children.” Family Relations Vol. 55(3): 345-360.
Curran, Sara R. et.al. 2006. "Mapping Gender and Migration in Sociological
Scholarship: Is it Segregation or Integration?" International Migration Review,
Vol. 40(1):199-223.
Dalaker, Joseph. 2001. Poverty in the United States: 2000. U.S. Census Bureau, Current
Population Reports, P60-214.
Davila, Alberto and Rogelio Saenz. 1990. “The Effect of Maquiladora Employment on
the Monthly Flow of Mexican Undocumented Immigration to the U.S., 1978-
1982.” International Migration Review Vol. 24: 96-107.
Davis, Kingsley. 1963. “The Theory of Change and Response in Modern Demographic
History.” Population Index 29:345-366.
de Leeuw, Jan. 2002. “Introduction.” Pp. xix - xxii in Stephen W. Raudenbush and
Anthony S. Bryk, Hierarchical Linear Models: Applications and Data Analysis
Methods. Second Edition. Thousand Oaks, CA: Sage Publications.
Donato, Katharine M. et. al. 2006. "A Glass Half Full? Gender in Migration Studies."
International Migration Review, Vol. 40(1):3-26.
Domina, Thurston. 2006. "What Clean Break?: Education and Nonmetropolitan
Migration Patterns, 1989-2004." Rural Sociology, Vol. 71(3): 373-398.
Durand, Jorge, Edward Telles, and Jennifer Flashman. 2006. “The Demographic
Foundations of the Latino Population.” Pp. 66-99 in Hispanics and the future of
America, edited by Marta Tienda and Faith Mitchell.
Ellis, Mark and Jamie Goodwin-White. 2006. “Generation Internal Migration in the
U.S.: Dispersion from States of Immigration?” International Migration Review,
Vol. 40(4): 899-926.
Ellis, Mark, Richard Wright, and Virginia Parks. 2006. "The Immigrant Household and
Spatial Assimilation: Partnership, Nativity, and Neighborhood Location." Urban
Geography, Vol. 27(1): 1-19.
98
Entwisle, Barbara, Katherine Faust, Ronald R. Rindfuss, and Toshiko Kaneda. 2007.
"Networks and Contexts: Variation in the Structure of Social Ties." American
Journal of Sociology, Vol. 112(5): 1495-1533.
Estrada, Leobardo F. 1976. “A Demographic Comparison of the Mexican Origin
Population in the Midwest and Southwest.” Aztlan 7: 203-234.
Feliciano, Cynthia. 2005. “Does Selective Migration Matter? Explaining Ethnic
Disparities in Educational Attainment among Immigrants’ Children.”
International Migration Review, Vol. 39(4): 841-871.
Fernandez, Leticia, Cheryl Howard and Jon Amastae. 2007. “Education, Race/Ethnicity
and Out-migration from a Border City.” Population Research and Policy Review,
26:103-124.
Foulkes, Matt, and K. Bruce Newbold. 2000. “Migration Propensities, Patterns, and the
Role of Human Capital: Comparing Mexican, Cuban, and Puerto Rican Interstate
Migration, 1985–1990” Professional Geographer, 52(1) 2000, pages 133–145
Fotheringham, Stewart A., Tony Champion, Colin Wymer, and Mike Coombes. 2000.
“Measuring Destination Attractivity: A Migration Example.” International
Journal Of Population Geography, Vol. 6: 391-421.
Franklin, Rachel S. 2003. Domestic Migration Across Regions, Divisions, and States:
1995-2000. U.S. Census Bureau Special Reports. CENSR-7.
Frey, William H. and Kao-le Liaw. 1998. “Immigrant Concentration and Domestic
Migrant Dispersal: Is Movement to Nonmetropolitan Areas “With Flight”?”
Professional Geographer, 50(2): 215-232.
Frey, William H. and Kao-Lee Liaw. 2005. “Migration within the United States: Role of
Race-Ethnicity.” A Wharton Paper on Urban Affairs from the Brookings
Institution.
Fulton, John A. 2007. Differences Between African-Americans and Whites in Migration
to The United States South, 1995-2000. PhD dissertation from the University of
Wisconsin-Madison.
Garcia, John A. 1981. “Hispanic migration: Where they are moving and why.” Agenda:
A Journal of Hispanic Issues 2:14-17.
Galarza, Ernesto. 1981. "Mexicans in the Southwest: A Culture in Process." Campo
Libre--Jounal of Chicano Studies, Vol. 1(1):87-118.
Galarza, Ernesto, Hernan Gallegos, and Julian Samora. 1970. Mexican-Americans in the
Southwest. McNally & Loftin Publishers: Santa Barbara, California.
Gale, Stephen. 1973. “Explanation Theory and Models of Migration.” Economic
Geography, Vol. 19(3): 257-274.
Ge Lin, Christiadi. 2006. "Examining Geographic and Occupational Mobility: A
Loglinear Modeling Approach." Papers in Regional Science, Vol. 85(4):505-522.
Gilroy, Marilyn. 2007. “Hispanics Find Jobs that Shift Migration.” The Hispanic
Outlook in Higher Education, 49-53.
Giordono, Leanne Schroeder. 2000. “Out-of-State vs. In-State Migration in the United
States.” Journal of Public and International Affairs Vol.11: 157-84.
99
Goetz, Stephan J., and Anil Rupasingha. 2003. "The Returns on Higher Education:
Estimates for the 48 contiguous States." Economic Development Quarterly, Vol.
17(4):337-351.
Golash-Boza, Tanya. 2005. "Assessing the Advantages of Bilingualism for the Children
of Immigrants." International Migration Review, Vol. 39(3):721-753.
Goldstein, Harvey, William Browne, and Jon Rasbash. 2002. “Partitioning Variation in
Multilevel Models.” Understanding Statistics, Vol. 1: 223-231.
Gonzalez, Nancie L. 1967. "The Spanish Americans of New Mexico, a Distinctive
Heritage." University of California Press, Los Angeles, California, September.
Gouveia, Lourdes, and Rogelio Saenz. 2000. “Global Forces and Latino Population
Growth in the Midwest: A regional and Subregional Analysis.” Great Plains
Research 10: 305-328.
Granovetter, Mark S. 1973. “The Strength of Weak Ties.” American Journal of
Sociology, Vol. 78(6): 1360-1380.
Granovetter, Mark S. 1983. “The Strength of Weak Ties: A Network Theory Revisited.”
Social Theory, Vol. 1: 201-223.
Grebler, Leo, Joan W. Moore, and Ralph C. Guzman. 1970. “The Mexican-American
People: The Nation’s Second Largest Minority.” New York, Free Press.
Green, Judith L., Gregory Camilli, and Patricia B. Elmore with Audra Sjukauskaite and
Elizabeth Grace. 2006. Handbook of complementary Methods in Education
Research. Lawrence Erlbaum Associates, Mahwah: New Jersey.
Greenwood, Michael J. 1975. "A Simultaneous-Equations Model of Urban Growth and
Migration." Journal of the America Statistical Association, Vol. 70(352):797-
810.
Greenwood, Michael J. 1985. “Human Migration: Theory, Models, and Empirical
Studies.” Journal of Regional Science Vol.25: 521-44.
Greenwood, Michael J. and Patrick J. Gormely. 1971. "A Comparison of the
Determinants of White and Nonwhite Interstate Migration." Demography, Vol.
8(1):141-155.
Griswold, Richard del Castillo. 1984. La Familia: Chicano Families in the Urban
Southwest, 1848 to the Present. Notre Dame, Ind.: University of Notre Dame
Press.
Griswold, Richard del Castillo. 1990. The Treaty of Guadalupe Hidalgo: A Legacy of
Conflict. University of Oklahoma Press: Norman,.
Gurak, Douglas T., and Fe Caces. 1992. “Migration Networks and the Shaping of
Migration Systems” Pp. 150-177, in International Migration Systems: A Global
Approach, by Mary Kritz, Lin Lean Lim, and Hania Zlotnix. Oxford University
Press, USA.
Gurak, Douglas T. and Mary M. Kritz. 2000. “The Interstate Migration of U.S.
Immigrants: Individual and contextual Determinants.” Social Forces, Vol. 78(3):
1017-1039.
Gutierrez, Jose Angel. 1998. The Making of a Chicano Militant: Lessons from Cristal.
Madiso, Wisconsin: University of Wisconsin Press.
100
Guzman, Betsy and Eileen Diaz McConnell. 2002. “The Hispanic Population: 1990-
2000 Growth and Change.” Population Research and Policy Review Vol. 21:
109-128.
Gwin, J. Blaine. 1921. "Immigration Along Our Southwest Border." Annals of the
American Academy of Political and Social Science. Vol. 93:126-130.
Hacking, Ian. 2005. "Why Race Still Matters." Daedalus, Vol. 134(1):102-116.
Hamilton, C. Horace. 1959. "Educational Selectivity of Net Migration from the South."
Social Forces, Vol. 38(1):33-42.
Hawley, Amos H. 1950. Human Ecology: A Theory of Community Structure. New York:
Ronald Press.
Hawley, Amos H. 1968. “Human Ecology.” Pages 328-337 in International
Encyclopedia of the Social Sciences, Edited by D. L. Sills. New York: Crowell,
Collier and McMillan.
Hawley, Amos. H. 1984. “Human Ecological and Marxian Theories.” The American
Journal of Sociology, Vol. 89(4):904-917.
Henrie, Christopher J. and David A. Plane. 2008. “Exodus from the California Core:
Using Demographic Effectiveness and Migration Impact Measures to Examine
Population Redistribution Within The Western United States.” Population
Research Policy Review, Vol. 27:43-64.
Herbele, Rudolph. 1938. “The Causes of the Rural-Urban Migration: A Survey of
Swansea.” American Journal of Sociology Vol. 43: 932-950.
Hernandez, Luis F. 1975. Aztlan: The Southwest and its Peoples. rochelle Park, N.J.:
Hayden Book Co.
Hiller, Harry H., and Kendall S. McCaig. 2007. “Reassessing the role of partnered
women in migration decision-making and migration outcomes.” Journal of
Social and Personal Relationships Vol. 24(3): 457–472.
Hobbs, A. H. 1942. “Specificity and Selective Migration.” American Sociological
Review, Vol. 7(6): 772-781.
Johnson, Kenneth M. et. al. 2005. "Temporal and Spatial Variation in Age-Specific Net
Migration in the United States." Demography, Vol. 42(4): 791-812.
Jones, Robert C. 1948. “Ethnic Family Patterns: The Mexican Family in the United
States.” The American Journal of Sociology, Vol. 53(6): 450-452.
Katz, Eliakim, and Hillel Rapoport. 2005. "On Human Capital Formation with Exit
Options." Journal of Population Economics, Vol. 18:267-274.
Katz, Fred E. 1966. "Social Participation and Social Structure." Social Forces, Vol.
45(2):199-210.
Kleiner, Morris M., Robert S. Gay, and Karen Greene. 1982. “Licensing, Migration, and
Earnings: Some Empirical Insights.” Policy Studies Review,Vol. 1(3): 510-522.
Kritz, Mary M., and June Marie Nogle. 1994. “Nativity Concentration and Internal
Migration among the Foreign-Born.” Demography Vol. 31: 509-24.
Lee, Everett S. 1966. “A Theory of Migration.” Population Association of America, Vol.
3(1): 47-57.
Lee, Everett S., and Anne S. Lee. 1960. “Internal Migration Statistics for the United
States.” Journal of the American Statistical Association, Vol. 55(292): 664-697.
101
Lee, Everett S. 1966. “A Theory of Migration.” Demography Vol. 3: 47-57.
Lekwa, Verl L., Tom W. Rice, and Matthew V. Hibbing. 2007. "The Correlates of
Community Attractiveness." Environment and Behavior, Vol. 39(2):198-216.
Litwak, Eugene. 1960. “Geographic Mobility and Extended Family Cohesion.”
American Sociological Review, Vol. 25(3): 385-394.
Long, Larry. 1992. “Changing Residence: Comparative Perspectives on its Relationship
to Age, Sex, and Marital Status.” Population Studies, Vol. 46(1): 141-158.
Lumsden, Charles J. and Edward O. Wilson. 2005. Genes, Mind, and Culture: The
Coevolutionary Process. Singapore: World Scientific.
McWilliams, Carey. 1949. North from Mexico: The Spanish-Speaking People of the
United States, The Poeple of America Series, Edited by Louis Adamic.
Philadelphia and New York: J. B. Lippincott Co.
Mahler, Sarah J. and Patricia R. Pessar. 2006.
Gender Matters: Enthnographers Bring Gender from the Periphery toward the
Core of Migration Studies.” International Migration Review, Vol. 40(1): 27-63.
Marks, Carole. 1983. “Lines of Communication, Recruitment Mechanisms, and the
Great Migration of 1916-1918.” Social Problems 31(1):73-83.
Martinez-Brawley, Emilia E. and Paz M-B. Zorita. 2006. “Language, Identity and
Empowerment: The Case of Spanish in the Southwest.” Journal of Ethnic &
Cultural Diversity in Social Work, Vol. 15(1/2): 81-95.
Massey, Douglas S. 1987. “Understanding Mexican Migration to the United States.”
American Journal of Sociology 92:1372-1403.
Massey, Douglas S. 1990. “Social Structure, Household Strategies, and the Cumulative
Causation of Migration.” Population Index 56:3-26.
Massey, Douglas S. and Kristin E. Espinosa. 1997. “What's Driving Mexico-U.S.
Migration?: A Theoretical, Empirical, and Policy Analysis.” American Journal of
Sociology 102:939-999.
McHugh, Kevin E. 1989. “Hispanic Migration and Population Redistribution in the
United States.” Professional Geographer, 41(1), pp. 429-439.
Menjivar, Cecilia. 2000. Fragmented Ties: Salvadoran Immigrant Networks in America.
Berkeley: University of California Press.
Michel, Philippe, Anne Perrot, Jacques-Francois Thisse. 1996. "Interregional
Equilibrium with Heterogeneous Labor." Journal of Population Economics, Vol.
9:95-113.
Moore, Joan W and (with Alfredo Cuellar). 1970a. Mexican Americans. Englewood
Cliffs, N.J.: Prentice-Hall.
Moore, Joan W. 1970b. “Colonialism: The Case of the Mexican Americans.” Social
Problems, 17(4), pp. 463-472.
Muñoz, Carlos. 1989. Youth, Identity, Power: The Chicano Movement. New York:
Verso Books.
Murguia, Edward. 1975. Assimilation, Colonialism and the Mexican American People.
Austin: University of Texas, Center for Mexican American Studies.
Nostrand, Richard L. 1975. "Mexican Americans Circa 1850." Annals of the Association
of American Geographers, Vol. 65(3):378-390.
102
Pachon, Harry P. and Joan W. Moore. 1981. "Mexican Americans." Annals of the
American Academy of Political and Social Science, 545:111-124.
Parades, Americo. 1958. With His Pistol In His Hand: A Border Ballad and Its Hero.
Autin: University of Texas Press.
Pantoja, Adrian D., Cecilia Menjívar and Lisa Magaña. 2008. "The Spring Marches of
2006: Latinos, Immigration, and Political Mobilization in the 21st Century."
American Behavioral Scientist, Vol. 52(4):499-506.
Perrigo, Lynn I. 1971. The American Southwest: Its People and Cultures. Albuquerque:
University of New Mexico Press.
Perry, Marc J. 2003. State-to-State Migration Flows: 1995 to 2000. U.S. Census Bureau,
2000 special report, CENSR-8.
Pingle, Jonathan F. 2007. "A Note on Measuring Internal Migration in the United
States." Economic Letters, Vol. 94:38-42.
Poston, Dudley L. Jr. and W. Parker Frisbie. 2006. “Ecological Demography.” Pages
601-623 in Handbook of Population, edited by Dudley L. Poston and Michael
Micklin, New York: Springer.
Rapoport, Anotol. 1953. “Spread of Information through a Population with
Sociostructural Bias: I. Assumption of Transitivity.” Bulletin of Mathematical
Biophysics, Vol. 15: 523-533.
Ravenstein, E. G. 1876. “The Birthplaces of the People and The Laws of Migration.”
The Geographical Magazine 111.
Ravenstein, E. G. 1885. “The Laws of Migration.” Journal of the Royal Statistical
Society Vol. 48: 167-235.
Ravenstein, E. G. 1889. “The Laws of Migration: Second Paper.” Journal of the Royal
Statistical Society Vol. 52: 241-305.
Redstone, Ilana Akresh. 2006. "Occupational Mobility Among Legal Immigrants to the
United States." International Migration Review, Vol. 40(4):854-884.
Reichert, Joshua S. 1981. “The Migrant Syndrome: Seasonal U.S. Wage Labor and
Rural Development in Central Mexico.” Human Organization Vol. 40:56-66.
Ritchey, P. Neal. 1976. "Explanations of Migration." Annual Review of Sociology, Vol.
2:363-404.
Rodriguez, Clara E., Cecilia Menjivar, Havidan Rodriguez, and Rogelio Saenz. 2008.
Latinas/os in the United States: Changing the Face of America. Springer: New
York.
Ross, Frank A. and Andre G. Truxal. 1931. "Primary and Secondary Aspects of
Interstate Migrations." The American Journal of Sociology, Vol. 37(3):435-444.
Romo, Ricardo. 1978. “The Urbanization of Southwestern Chicanos in the Early
Twentieth Century.” Pp 183-207 in New Directions in Chicano Scholarship,
edited by Ricardo Romo and Raymund Paredes. La Jolla: Chicano Studies
Program, University of California, San Diego, California.
Sjaasta, Larry A. 1962. “The Costs and Returns of Human Migration.” Journal of
Political Economy, Vol. 70(5), Part 2: 80-93.
103
Saenz, Rogelio. 1989a. “Mexican American Interstate Migration Flows among Arizona,
California, Colorado, New Mexico, and Texas.” Sociology and Social Research
73: 153-17.
______. 1989b. “Selectivity of Mexican American Intraregional Migration in the
Southwest.” Hispanic Journal of Behavioral Sciences 11 (2): 148-155.
Saenz, Rogelio. 1991. “Interregional migration patterns of Chicanos: the Core,
Periphery, and Frontier.” Social Science Quarterly 72 (1):135-48.
Saenz, Rogelio. 1997. “Ethnic Concentration and Chicano Poverty: A Comparative
Approach.” Social Science Research, Vol. 26: 205-228.
Saenz, Rogelio and Alberto Davila. 1992. “Chicano Return Migration to the Southwest:
An Integrated Human Capital Approach.” International Migration Review, Vol.
26(4): 1248-1266.
Saenz, Rogelio, Katharine M. Donato, Lourdes Gouveia, and Cruz Torrez. 2003.
"Latinos in the South: A Glimpse of Ongoing Trends and Research." Southern
Rural Sociology, Vol.19(1):1-19.
Saenz, Rogelio, and Robert N. Anderson. 1994. “The Ecology of Chicano Interstate Net
Migration, 1975-1980.” Social Science Quarterly Vol. 75(1): 37-52.
Saenz, Rogelio and Cynthia Cready. 1997. “The Southwest-Midwest Mexican American
Migration Flows, 1985-1990.” JSRI Research Report #20, The Julian Samora
Research Institute, Michigan State University, East Lansing, Michigan.
Saenz, Rogelio. 1999. “Mexican Americans.” Pp. 209-29 in The Minority Report: An
Introduction to Racial, Ethnic, and Gender Relations, 3rd edition, edited by
Anthony G. Dworkin and Rosalind J. Dworkin. Fort Worth, TX: Holt, Rinehart,
and Winston, Inc.
Saenz, Rogelio. 2004. Latinos and the Changing Face of America, a report in The
American People Census 2000 Series, edited by R. Farley and J. Haaga. New
York and Washington, DC: Russell Sage Foundation and Population Reference
Bureau.
Saenz, Rogelio, Cynthia M. Cready, and Maria Cristina Morales. 2007. “Adios Aztlan:
Mexican American Outmigration from the Southwest.” Pp. 189-213 in The
Sociology of Spatial Inequality, edited by L. M. Lobao, G. Hooks, and A. R.
Tickamyer. Albany: SUNY.
Samora, Julian (with the assitance of Jorge Bustamante and Gilbert Cardenas). 1971. Los
Mojados: The Wetback Story. Notre Dame, Inc. & London: University of Notre
Dame Press.
Samora, Julian and Patricia Vandel Simon. 1993. A History of the Mexican-American
People. Notre Dame: University of Notre Dame Press.
Schultz, T. W. 1961. “Investment in Human Capital.” American Economic Review 1-17.
Schwartz, Aba. 1976. “Migration, Age and Education.” Journal of Political Economy,
Vol. 70(5): 701-720.
Shaw, Paul R. 1975. Migration Theory and Fact. Philadelphia: Regional Science
Research Institute, Pg. 17-36.
Shockley, John S. 1974. Chicano Revolt in a Texas Town. Notre Dame, IN: Notre Dame
University Press.
104
Sly, David F. 1972. "Migration and the Ecological Complex." American Sociological
Review, Vol. 37(5):615-628.
Smith-Buani, Josepth Augustine. 2001. State-to-State Migration During 1975-1980 and
1985-90: An Analysis by Age, Race, Sex, and Education. Urban Studies PhD
dissertation from Portland State University.
Stone, Brad Lowell. 2008. "The Most Unique of all Unique Species." Published on-line
in Sociology, Vol. 45:146-151.
Stouffer, Samuel A. 1940. “Intervening Opportunities: A Theory Relating Mobility and
Distance.” American Sociological Review, Vol. 5(6): 845-867.
Stouffer, Samuel A. 1960. “Intervening Opportunities and Competing Migrants.”
Journal of Regional Science, Vol. 2(1): 1-26.
Sweeney, Stuart H. and Harvey A. Goldstein. 2005. "Accounting for Migration in
Regional Occupational Employment Projections." The Annals of Regional
Science, vol. 39:297-316.
Takahashi, Lois M. and Michelle G. Magalong. 2008. “Disruptive Social Capital:
(Un)Healthy socio-spatial Interactions among Filipino men Living with
HIV/AIDS.” Health & Place, Vol. 14: 182-197.
Taeuber, Conrad, and Irene B. Taeuber. 1938. "Short Distance Interstate Migrations."
Social Forces, Vol. 16(4): 503-506.
Thomas, Dorothy S. 1938. Research Memorandum on Migration Differentials. New
York: Social Science Research Council.
Tilly, Charles. 1990. "Transplanted Networks." Pages 79-95 in Immigration
Reconsidered: History, Sociology, and Politics, edited by Virginia Yans-
McLaughlin. New York: Oxford University Press.
Tilly, Charles. 2007. "Trust Networks in Transnational Migration." Sociological Forum,
Vol. 22(1):3-25.
Tilly, Charles and C. Harold Brown. 1967. "On Uprooting, Kinship, and the Auspices of
Migration." International Journal of Comparative Sociology, Vol. 8(2):139-164.
Trujillo, Armando. 2005. “Politics, School Philosophy, and Language Policy: The Case
of Crystal City Schools.” Educational Policy 19:621-654.
Valdivia, Corinne, et. al. 2008. "The Impact of Networks and the Context of Reception
on Asset Accumulation Strategies of Latino Newcomers in New Settlement
Communities on the Midwest." American Journal of Agricultural and
Economics, Vol.5 :1319-1325.
Wall, Ellen, Gabriele Ferrazzi, and Frans Schryer. 1998. "Getting the Goods on Social
Capital." Rural Sociology, Vol. 62(2):300-322.
Watson, James B. and Julian Samora. 1954. "Subordinate Leadership in a Bicultural
Community: An Analysis." American Sociological Review, Vol. 19(4):416-421.
Wattenberg, William W. "Attitude Toward community Size as Evidenced by Migratory
Behavior, 1935-1940." Social Forces, Vol.26(4):437-442.
Weber, David J. 1982. The Mexican Frontier, 1821-1846: The American Southwest
under Mexico. Albuquerque: University of New Mexico Press.
105
Weeks, John R. 1979. “Cultural Demography of Midwestern Chicano Communities.” Pg
229-251 in The Chicano Experience, edited by Stanley A. West and June
Macklin. Boulder, CO: Westview Press.
Weeks, John R. 1996. Population: An Introduction to Concepts and Issues. Wadsworth
Publishing Company: Belmont, California.
Westefeld, Albert. 1940. "The Distance Factor in Migration." Social Forces, Vol.
19(2):213-218.
White, Katherine J. C, Kyle Crowder, Stewart E. Tolnay, and Robert M. Adelman. 2005.
“Race, Gender, and Marriage: Destination Selection during the Great Migration.”
Demography, Vol. 42(2): 215-241.
White, S. 1980. “A Philosophical Dichotomy in Migration Research.” Professional
Geography, 32: 6-13.
Wilson, Beth A., E. Helen Berry, and Michael B. Toney. 2008. “Breaking New Ground:
A Longitudinal Comparison of Onward Migration by Hispanics, Blacks and
Whites in the U.S.” Population and Society, Vol. 4(2):1-27.
Wilson, Beth A., E. Helen Berry, Michael B. Toney, Young-Taek Kim, and John B.
Cromartie. 2008. “A Panel Based Analysis of the Effects of Race/Ethnicity and
Other individual Level Characteristics at Leaving on Returning.” Population
Research Policy Review, published on-line on October 15,2008.
Wilson, Edward O. 1975. Sociobiology: The New Synthesis. Cambridge, Massachusetts:
Belknap Press of Harvard University Press.
Wolf, Douglas A., and Charles F. Longino. 2005. "Our "Increasingly Mobile Society"?
The Curious Persistence of a False Belief." The Gerontologist, Vol. 45(1):5-11.
Wolpert, Julian. 1967. “Distance and Directional Bias in Inter-Urban Migratory
Streams.” Annals of the Association of American Geographer, Vol. 57: 605-616.
Yankow, Jeffrey J. 2003. “Migration, Job Change, and Wage Growth: A New
Perspective on the Pecuniary Return to Geographic Mobility.” Journal of
Regional Science, Vol. 43(3): 483-516.
Yezer, Anthony M., and Lawrence Thurston. 1976. “Migration Patterns and Income
Change: Implications for the Human Capital Approach to Migration.” Southern
Economic Journal, Vol. 42(4): 693-702.
Zuniga, Victor and Ruben Hernandez-Leon. 2005. New Destinations: Mexican
Immigration in The United States. Russell Sage Foundation.
106
VITA
Name: Carlos Siordia
Address: 805 Navarro Dr. Apt. D
College Station, TX 77845
Email Address: csiordia@tamu.edu
Education: B.A., double major in Sociology & Psychology
The University of Texas-Pan American, 2006
