University of Southern California Dissertations and Theses

The post-baccalaureate choices of racial and ethnic minority students with science and engineering majors

(USC Thesis Other)

The post-baccalaureate choices of racial and ethnic minority students with science and engineering majors

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

THE POST-BACCALAUREATE CHOICES OF RACIAL AND ETHNIC MINORITY
STUDENTS WITH SCIENCE AND ENGINEERING MAJORS

by

Araceli A. Espinoza

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
Doctor of Philosophy
(URBAN EDUCATION POLICY)

December 2013

Copyright 2013 Araceli A. Espinoza
ii
DEDICATION
Para mis padres, cuyos sacrificios han hecho mis sueños educacionales posibles.
To my parents whose many sacrifices have made my educational dreams possible.

iii
ACKNOWLEDGEMENTS

I could say this journey began four years ago, but it really began the moment I started
pre-school. Since then, many individuals in my life have contributed pieces to my “dissertation
puzzle.” Thank you.
I want to thank my parents (Angel and Maria Elena) and my siblings (Rocio, Angel
Jesus, and Lizet). Beyond providing love and support, they tolerated my moodiness and my
episodes of “writing hibernation.”
I am very fortunate to have friends and colleagues like Cecile Sam, Jonathan Mathis, and
Monica Esqueda. Their work ethic was my motivation and their friendship was my sustenance.
I am blessed to have met, and to have formed a wonderful relationship with, Eric. Thank
you for allowing me to vent, time and time again, and for reassuring me that it would get done.
During my time at USC, I met various faculty members who challenged me intellectually
and who took the time to invest in my professional development. Thank you.
To my committee chair and advisor, Dr. Darnell Cole, thank you for your guidance.
When we first met, you said you needed a graduate assistant who was willing to work hard. I
hope this dissertation has demonstrated my willingness.
I want to thank my committee members, Drs. John Slaughter and Azad Madni, for their
patience, support, and enthusiasm for my research.
To my mentor, Dr. Kristan Venegas: Thank you for believing in me. Thank you for your
(personal and professional) advice, for your encouragement, and for your generosity. Perhaps,
more importantly, thank you for being who I want to be “when I grow up.”
iv
Finally, I want to thank Diego, my pug. His silent, yet strong, support throughout this
process was invaluable. He kept me company at 2 a.m.—and when I fell asleep on the sofa he
kept me warm.

v
TABLE OF CONTENTS

Dedication ii
Acknowledgements iii
List of Tables vi
List of Figures viii
Abstract ix
Chapter One: Introduction 1
Chapter Two: Overview of the Literature 10
Chapter Three: Conceptual Model 22
Chapter Four: Study One 39
Conceptual Model and Application 39
Methods 46
Limitations 55
Findings 56
Discussion 60
Chapter Five: Study Two 63
Conceptual Model and Application 63
Methods 69
Limitations 75
Findings 76
Discussion 78
Chapter Six: Study Three 83
Conceptual Model and Application 84
Methods 93
Limitations 104
Findings 105
Discussion 112
Chapter Seven: Conclusion 118
References 127
Appendices 138
Appendix A 138
Appendix B 152

vi
LIST OF TABLES

Table 1 Science and Engineering Master’s Degrees Awarded to U.S. Citizens and
Permanent Residents in 2001 and 2010
3
Table 2 Science and Engineering Doctoral Degrees Awarded to U.S. Citizens and
Permanent Residents in 2001 and 2010
3
Table 3 Science and Engineering Master’s Degrees Awarded to Racial and
Ethnic Minority (REM) Women and Men in 2001 and 2010
4
Table 4 Science and Engineering Doctoral Degrees Awarded to Racial and
Ethnic Minority (REM) Women and Men in 2001 and 2010
4
Table 5 Overview of REM Undergraduate Students with S&E Majors for Study 1
and Study 2
47
Table 6 Overview of REM Undergraduate Students with Non-S&E Majors for
Study 1 and Study 2
48
Table 7 Results for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Students with S&E Majors
57
Table 8 Results for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Students with Non-S&E Majors
59
Table 9 Results for “Plan to Work Full-Time” for REM Undergraduate Students
with S&E Majors
77
Table 10 Results for “Plan to Work Full-Time” for REM Undergraduate Students
with Non-S&E Majors
79
Table 11 Overview of REM Undergraduate Women with S&E Majors 95
Table 12 Overview of REM Undergraduate Women with Non-S&E Majors 96
Table 13 Results for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Women with S&E Majors
107
Table 14 Results for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Women with Non-S&E Majors
108
Table 15 Results for “Plan to Work Full-Time” for REM Undergraduate Women
with S&E Majors
110
Table 16 Results for “Plan to Work Full-Time” for REM Undergraduate Women 111
vii
with Non-S&E Majors
Table A-1 Science and Engineering Majors for Study 1 and Study 2 138
Table A-2 Non-Science and Engineering Majors for Study 1 and Study 2 139
Table A-3 Non-Imputed Results for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Students with S&E Majors
141
Table A-4 Non-Imputed Results for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Students with Non-S&E Majors
142
Table A-5 Non-Imputed Results for “Plan to Work Full-Time” for REM
Undergraduate Students with S&E Majors
143
Table A-6 Non-Imputed Results for “Plan to Work Full-Time” for REM
Undergraduate Students with Non-S&E Majors
144
Table A-7 Science and Engineering Majors for Study 3 145
Table A-8 Non-Science and Engineering Majors for Study 3 146
Table A-9 Non-Imputed Results for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Women with S&E Majors
148
Table A-10 Non-Imputed Results for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Women with Non-S&E Majors
149
Table A-11 Non-Imputed Results for “Plan to Work Full-Time” for REM
Undergraduate Women with S&E Majors
150
Table A-12 Non-Imputed Results for “Plan to Work Full-Time” for REM
Undergraduate Women with Non-S&E Majors
151

viii
LIST OF FIGURES

Figure 1 Perna’s Conceptual Model of Student College Choice 37
Figure 2 Application of Perna’s Conceptual Model of Student College Choice for
“Plan to Attend Graduate/Professional School” for REM Undergraduate
Students with S&E and Non-S&E Majors
42
Figure 3 Variables and Coding for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Students with S&E and Non-Science and Majors
50
Figure 4 Reliability and Constituent Variables for “Social Capital Through
Student-Faculty Interactions”
52
Figure 5 Application of Perna’s Conceptual Model of Student College Choice for
“Plan to Work Full-Time” for REM Undergraduate Students with S&E
and Non-S&E Majors
66
Figure 6 Variables and Coding for “Plan to Work Full-Time” for REM
Undergraduate Students with S&E and Non-S&E Majors
71
Figure 7 Application of Perna’s Conceptual Model of Student College Choice for
“Plan to Attend Graduate/Professional School” for REM Undergraduate
Women with S&E and Non-S&E Majors
87
Figure 8 Application for Perna’s Conceptual Model of Student College Choice for
“Plan to Work Full-Time” for REM Undergraduate Women with S&E
and Non-S&E Majors
88
Figure 9 Variables and Coding for “Plan to Attend Graduate/Professional School”
for REM Undergraduate Women with S&E and Non-S&E Majors
98
Figure 10 Variables and Coding for “Plan to Work-Full Time” for REM
Undergraduate Women with S&E and Non-S&E Majors
99
Figure 11 Reliability and Constituent Variables for “College Social Capital” 101
Figure B-1 Science and Engineering Fields and Occupational Categories in SESTAT 152
Figure B-2 Majors Classified based on SESTAT: A Tool for Studying Scientists and
Engineers in the United States
161
Figure B-3 Selectivity of Private Four-Year Institutions 164

ix
ABSTRACT
Despite the need for individuals to pursue advanced degrees in science and engineering (S&E),
and despite the increased need for individuals to enter the S&E workforce, the student college
choice discussion does not often consider students’ post-baccalaureate choices. This dissertation
addresses two post-baccalaureate options—graduate education and full-time employment—with
a specific focus on the post-baccalaureate choices of racial and ethnic (REM) students with S&E
majors. The dissertation is composed of three studies. The first study focuses on the “graduate
education choice” of REM students, the second addresses the “full-time employment choice” of
REM students, and the third focuses on the “graduate education choice” and the “full-time
employment choice” of REM women. Moreover, because the focus of this dissertation is the
post-baccalaureate choices of REM students with S&E majors, its investigation is guided by
Laura Perna’s conceptual model of student college choice.
Keywords: science and engineering, racial and ethnic minority, post-baccalaureate choice
1
CHAPTER ONE: INTRODUCTION
Over the past 20 years, models of student college choice have examined why students
choose—or do not choose—to pursue a postsecondary education. Defined by Hossler, Braxton,
and Coopersmith (1989) as “a complex, multistage process during which an individual develops
aspirations to continue a formal education beyond high school, followed later by a decision to
attend a specific college, university or institution of advanced vocational training” (p. 234),
student college choice is perhaps more pertinent today than ever. Since his first inauguration,
President Obama and his administration have emphasized the importance of making the United
States (U.S.) the world leader in college degree attainment (Gonzalez, 2010). More specifically,
President Obama has called for the U.S. to generate an additional eight million college graduates
by 2020 (Gonzalez, 2010), in an effort to position America as the “best educated, most
competitive workforce in the world” (President Obama, 2009, as cited by Kanter, 2011).
The President’s call for a college-educated workforce is perhaps most pertinent within the
country’s science and engineering (S&E) sectors. By 2018, S&E occupations will account for
about 8.6 million jobs in the U.S. economy—an increase of 1.3 million from 2008—and will
require about 1.2 million employees with bachelor’s degrees (Carnevale, Smith, & Strohl, 2010).
To ensure the domestic workforce will possess the academic credentials necessary to satisfy
future job openings, the National Science Board (2010) has stressed the need to capitalize on
untapped talent by increasing the number of racial and ethnic minority (REM) individuals
pursuing baccalaureate degrees in S&E-related fields. In short, the National Science Board
assumes if more REM individuals enter and complete baccalaureate programs in S&E, more
REM individuals will enter the U.S. S&E workforce upon graduation.

2
It is unknown, however, if this direct link exists—that is, if REM individuals with S&E
baccalaureate degrees choose to pursue full-time employment immediately after graduation.
Moreover, the variety of factors that may contribute to REM individuals’ decisions to seek full-
time employment remains unclear. While S&E bachelor degree attainment is important,
procurement of advanced degrees is equally valuable. Of the 8.6 million S&E jobs projected by
2018, about 779,000 jobs will require employees with master’s degrees or higher (Carnevale et
al., 2010).
As illustrated in Table 1, current figures reveal slight increases in the number of master’s
degrees earned by REM individuals. In 2001, American Indian/Alaska Natives, Asian/Pacific
Islanders, Blacks, and Hispanics earned 476, 7,056, 6,174, and 4,113, respectively, of S&E
master’s degrees (National Science Foundation [NSF] & National Center for Science and
Engineering Statistics [NCSES], 2013). By 2010, American Indian/Alaska Natives,
Asian/Pacific Islanders, Blacks and Hispanics earned 629, 9,959, 10,292, and 7,379,
respectively, of S&E master’s degrees (NSF & NCSES, 2013).
The figures for S&E doctoral degree attainment are also modest, as demonstrated in
Table 2. In 2001, American Indian/Alaska Natives, Asian/Pacific Islanders, Blacks, and
Hispanics earned 78, 1,556, 748, and 815, respectively, of S&E doctoral degrees (NSF &
NCSES, 2013). By 2010, American Indian/Alaska Natives, Asian/Pacific Islanders, Blacks, and
Hispanics earned 103, 2,054, 973, and 1,169 respectively, of S&E doctoral degrees (NSF &
NCSES, 2013).
Notably, slightly more REM women earn S&E advanced degrees than do REM men. In
2001, REM women earned about 52% (i.e., 9,311) and REM men earned about 48% (i.e., 8,508)
of the 17,819 S&E master’s degrees attained by REM individuals, as shown in Table 3 (NSF &
3
NCSES, 2013). In 2010, REM women earned about 54.7% (i.e., 15,449) and REM men earned
about 45.3% (i.e., 12,810) of the 28,259 S&E master’s degrees attained by REM individuals
(NSF & NCSES, 2013).

Table 1

Science and Engineering Master’s Degrees Awarded to U.S. Citizens and Permanent Residents
in 2001 and 2010

Race/Ethnicity
Master’s Degrees
2001
Master’s Degrees
2010

Difference
American Indian or Alaska Native 476 629 153
Asian or Pacific Islander 7,056 9,959 2,903
Black 6,174 10,292 4,118
Hispanic 4,113 7,379 3,266
Other or Unknown 5,346 12,660 7,314
White 48,920 62,633 13,713

Total 72,085 103,552 31,467

Table 2

Science and Engineering Doctoral Degrees Awarded to U.S. Citizens and Permanent Residents
in 2001 and 2010

Race/Ethnicity
Doctoral Degrees
2001
Doctoral Degrees
2010

Difference
American Indian or Alaska Native 78 103 25
Asian or Pacific Islander 1,556 2,054 498
Black 748 973 225
Hispanic 815 1,169 354
Other or Unknown 1,044 1,993 949
White 13,020 14,278 1,258

Total 17,261 20,570 3,309

At the doctoral level, REM women earn marginally more degrees than do REM men.
Table 4 illustrates that, in 2001, REM men earned about 52% (i.e.,1,644) of the 3,197 S&E
doctoral degrees attained by REM individuals (NSF & NCSES, 2013). By 2010, REM women
earned about 53.5% (i.e., 2,298) of the 4,229 S&E doctoral degrees attained by REM individuals
(NSF & NCSES, 2013). Given the aforementioned figures, an investigation of factors
4
contributing to REM individuals’ decisions to pursue a graduate education seems particularly
relevant.

Table 3

Science and Engineering Master’s Degrees Awarded to Racial and Ethnic Minority (REM)
Women and Men in 2001 and 2010

Master’s Degrees
2001*
Master’s Degrees
2010*
REM Women
+
9,311 (52%) 15,449 (54.7%)
REM Men
+
8,508 (48%) 12,810 (45.35)
Total 17,819 28,259
* Degrees awarded to U.S. citizens and permanent residents
+ Women and men who identify as American Indian or Alaska Native, Asian or Pacific Islander,
Black, or Hispanic

Table 4

Science and Engineering Doctoral Degrees Awarded to Racial and Ethnic Minority (REM)
Women and Men in 2001 and 2010

Doctoral Degrees
2001*
Doctoral Degrees
2010*
REM Women
+
1,533 (48%) 2,298 (53.5%)
REM Men
+
1,664 (52%) 2,001 (46.5)
Total 3,197 4,299
* Degrees awarded to U.S. citizens and permanent residents
+ Women and men who identify as American Indian or Alaska Native, Asian or Pacific Islander,
Black, or Hispanic

Despite the need for individuals to pursue advanced degrees in S&E, and despite the
increased need for individuals to enter the S&E workforce, the student college choice discussion
does not often consider students’ post-baccalaureate choices. This dissertation addresses two
post-baccalaureate options—graduate education and full-time employment—with a specific
focus on the post-baccalaureate choices of REM students with S&E majors.
The dissertation is composed of three studies. The first study focuses on the “graduate
education choice” of REM students, the second addresses the “full-time employment choice” of
5
REM students, and the third focuses on the “graduate education choice” and the “full-time
employment choice” of REM women. Throughout these three studies, the following research
questions will be examined:
1. What factors inform the “graduate education choice” of REM undergraduate students
with S&E majors?
2. What factors inform the “full-time employment choice” of REM undergraduate students
with S&E majors?
3. What factors inform the “graduate education choice” REM undergraduate women with
S&E majors?
4. What factors inform the “full-time employment choice” of REM undergraduate women
with S&E majors?
Because the focus of this dissertation is the post-baccalaureate choices of REM students
with S&E majors, its investigation is guided by Laura Perna’s conceptual model of student
college choice—a model that is applied to all three studies. In Chapter Three, I describe Perna’s
conceptual model of student college choice, as well as the economic and sociological grounding
of the model.
In the section that follows, I provide a summary of other similarities that unite the three
studies in my dissertation.
Similarities Across All Three Studies
All three studies in this dissertation include a comparison group of REM undergraduate
students with non-S&E majors. This comparison group provides a reference point by which to
assess the ways in which the choices of REM students may be informed by their respective
majors. Moreover, in all three studies, I: utilize the 2003 Freshmen Survey and the 2007 College
6
Senior Survey, adhere to a consistent definition of REM students, and employ the same
classification of S&E majors.
Data
Since 1966, the Higher Education Research Institute (HERI) at the University of
California, Los Angeles (UCLA) has collected college student experience data, by way of the
Cooperative Institutional Research Program (CIRP). For each of the studies in this dissertation, I
utilized CIRP’s 2003 Freshmen Survey and its 2007 follow-up College Senior Survey. The
Freshmen Survey, which is typically administered during orientation and throughout the first few
months of classes, is designed to provide a snapshot of incoming students before those students
experience college. Key sections of the survey examine: student demographic characteristics,
established behaviors in high school, academic preparedness, concerns about financing college,
students’ values and goals, expectations of college, and interactions with peers and faculty.
The College Senior Survey is designed as an exit survey for graduating seniors and thus,
when used in conjunction with the Freshmen Survey, generates longitudinal data on students’
cognitive and affective development during college. The College Senior Survey covers a range
of college outcomes and post-college plans, including: academic achievement and engagement,
student-faculty interactions, students’ goals and values, students’ satisfaction with their college
experiences, and degree aspirations and career plans.
Definition of Racial and Ethnic Minority Students
The definition of “racial and ethnic minority” utilized by this dissertation includes
students who identify as: American Indian, Asian, Black, Latino, or Multiracial/Ethnic. The
“race/ethnicity” variable utilized for all three studies stems from the 2003 Freshmen Survey, in
which “race/ethnicity” categories were preset. I made the decision to include Asian students for
7
three reasons:
First: “Asian” is not a homogeneous group and includes Hmong, Laotian, Cambodian,
Vietnamese, Thai, Japanese, Korean, Filipino, Chinese, Pakistani, and Asian-Indian (Museus &
Kiang, 2009). Contrary to the “model minority myth
1
,” postsecondary degree attainment is not
the same for all individuals who identify as Asian. Southeast Asian-American populations hold
college degrees at rates lower than that of their East and South East Asian-American counterparts
(Museus & Kiang, 2009). Excluding Asian students from the definition of “racial and ethnic
minority” would insinuate that all Asian students are the same.
Second: Because the categories of the “race/ethnicity” variable were preset on the cited
Freshman Survey, it was not possible to disaggregate the “Asian” category and select the sub-
groups of students that would be considered as attaining college degrees at lower rates. Again,
excluding the “Asian” category completely would insinuate that all Asian students are the same.
Third: According to the National Science Foundation and the National Center for Science
and Engineering Statistics (2013), a “minority group” is a racial/ethnic group that encompasses a
small percentage of the U.S. population. American Indians or Alaska Natives, Asians, Blacks,
Hispanics, and Native Hawaiians or Other Pacific Islanders are minority groups (NSF & NCSES,
2013). Hence, including Asian students as part of my definition of REM students falls in line
with the definition proposed by the National Science Foundation and by the National Center for
Science and Engineering Statistics.
It should be noted that this dissertation focuses on “racial and ethnic minority” groups
and not on “underrepresented minority groups.” This distinction is important because Asian
students, while part of a racial and ethnic minority group, are often not considered

1
“The model minority [myth] is the notion that Asian-American students achieve universal and
unparalleled academic and occupational success” (Museus & Kiang, 2009, p. 6).
8
“underrepresented” in S&E fields (NSF & NCSES, 2013).
Definition of Science and Engineering Majors
To identify the S&E majors, I cross-referenced the “college major” variable from the
2007 College Senior Survey with SESTAT: A Tool for Studying Scientists and Engineers in the
United States. The latter, a study compiled by the National Science Foundation, organizes
educational majors into five main S&E educational fields: computer and mathematical sciences,
life and related sciences, physical and related sciences, social and related sciences, and
engineering. Each of these S&E educational fields is further defined within SESTAT by a minor
group discipline, and each minor group discipline is then further defined by sub-disciplines (see
Appendix B, Figure B-1). Educational fields that are not categorized into one of the five main
S&E educational fields are considered non-S&E majors. For the studies in this dissertation, I
utilized each S&E educational field and its corresponding minor group and sub-disciplines to
label S&E majors. I labeled majors that did not correspond to one of the five main S&E
educational fields as non-S&E majors (see Appendix B, Figure B-2).
It is important to note that one of the main S&E education fields is social and related
sciences. This main educational field consists of five minor groups of disciplines: economics,
political and related sciences, psychology, sociology, and anthropology and other social sciences
(see Appendix B, Figure B-1). Majors such as economics, political science and government,
anthropology, sociology, and ethnic studies (among others) are considered S&E majors. For the
studies in this dissertation, I labeled the aforementioned majors as S&E majors (see Appendix B,
Figure B-2).

9
In the chapter that follows, I present an overview of the literature. In Chapter 3, I describe
Perna’s conceptual model of student college choice. In Chapter 4, I present the first study; in
Chapter 5, the second study; and in Chapter 6, the third study. In the final chapter of this
dissertation, I provide a summary of the key findings across the three studies and discuss the
implications of this dissertation.

10
CHAPTER TWO: OVERVIEW OF THE LITERATURE
In this chapter, I focus on three areas of literature: institutional selectivity, graduate
education, and employment. Throughout this review of the literature, I borrow from works that,
although not S&E specific, address factors contributing to the “graduate education choice” and
“full-time employment choice” of REM students with S&E majors.
For example, I include a section on institutional selectivity because institutional
selectivity overlaps and contributes to graduate education enrollment and employment. In the
section that follows I continue the discussion on institutional selectivity. Thereafter, I present the
literature on graduate education and employment.
Institutional Selectivity
Although a postsecondary institution may utilize a variety of measures to determine its
selectivity, a number of common indicators exist among these measures. Such indicators include
the high school class rankings of incoming students, the high school grade point averages (GPA)
of incoming freshmen, Scholastic Aptitude Test (SAT) scores, and the percentage of students
admitted to the institution (Alon & Tienda, 2008; Brewer, Eide, & Ehrenberg, 1999; Eide,
Brewer, & Ehrenberg, 1998). While it is typical for two or more of the aforementioned indicators
(for example, average SAT scores and the percentage of applicants admitted) to be coupled,
selectivity is often determined solely by an averaging of SAT scores (Bowen & Bok, 1998; Dale
& Krueger, 2002; Ethington & Smart, 1986; Mullen, Goyette, & Soares, 2003).
Empirical studies reveal a link between institutional selectivity and bachelor’s degree
completion. For instance, in his research on the college completion rates of minorities and
Whites during the 1970s, Astin (1985) found a direct relationship between the selectivity of an
academic institution and the degree completion rates of its REM students. More recently, Alon
11
and Tienda (2005) report that the probabilities of graduation for Black and Hispanic students are
higher at selective than at nonselective institutions. Moreover, Alon and Tienda (2005) indicate
that the racial and ethnic gap in graduation narrows as institutional selectively increases. Similar
to Alon and Tienda, Melguizo (2008) reports that, even after correcting for the sorting of
students into specific types of institutions, African American and Hispanic students benefit from
attending the most elite institutions. In regards to graduate education enrollment, Mullen,
Goyette, and Soares (2003) note that graduates of selective public or private research institutions
are more likely to continue their education after receiving a bachelor’s degree, irrespective of
college grades or majors. While Eide, Brewer, and Ehrenberg (1998) agree that institutional
selectivity matters, they stress that selectivity is significant in relation to institutional control.
They report that attending a selective private college significantly increases a student’s
probability of attending graduate school and more specifically, graduate school at a major
research institution (Eide et al, 1998).
The benefits associated with attending a selective institution also extend to individuals’
employment and economic returns (Bowen & Bok, 1998; Brewer et al., 1999; Carnevale et al.,
2010). Researchers like Dale and Krueger (2002) and Bowen and Bok (1998) propose that
individuals from low-income backgrounds and REM individuals benefit the most from attending
selective institutions. Dale and Krueger (2002) point out that students who attend more selective
colleges earn about the same as students of comparable ability who attend less selective schools.
However, students from low-income families earn more if they attend selective colleges (Dale &
Krueger, 2002). Bowen and Bok (1998) note that, when compared with African American
students who earn degrees from non-selective institutions, African American students who earn
their bachelor’s degrees from selective institutions are not only more likely to be employed, but
12
are also likely to earn more income over their lifetimes. After examining the 1976 matriculants
of the College and Beyond (C&B) survey, Bowen and Bok (1998) report that, on average,
African American women from selective schools earned 73% more than did all African
American women with bachelor’s degrees. This earning advantage was even greater for African
American men: African American men from selective schools earned 82% more than all African
American men with bachelor’s degrees (Bowen & Bok, 1998).
Graduate Education
Numerous factors help explain whether students will continue with their education
beyond a bachelor’s degree, including: parents’ income and education, a student’s sex and race
and ethnicity, college academic performance, and undergraduate debt. Regarding students with
S&E majors, the literature also highlights the roles of student-faculty interactions and
undergraduate research experience, as well as the influence of STEM
2
-specific opportunity and
support programs. In the subsections that follow, I describe each aforementioned factor and its
relation to graduate education enrollment.
Parental Income and Educational Level
Whereas parental income and educational level are associated with students’ decisions to
attend (and likelihood of graduating from) college, the influence of parental income and
educational level on students’ decisions to pursue post-baccalaureate studies is debated (Chen,
2009; Hossler, Schmit, & Vesper, 1999; Manski & Wise, 1983). On the one hand, prior research
indicates that parents’ backgrounds—e.g., incomes and educational levels—do not contribute to
students’ graduate enrollment (Mare, 1980; Stolzenberg, 1994). On the other hand, some studies
highlight the significance of parental background (Bowen & Bok, 1998; Ethington & Smart,

2
STEM is a shorthand widely used to refer to science, technology, engineering, and
mathematics. The acronym STEM and the abbreviation S&E are used interchangeably.
13
1986; Mullen et al., 2003; Nevill & Chen, 2007). Ethington and Smart (1986), for instance,
report that parents’ incomes and educational levels do influence graduate school attendance,
albeit indirectly. They suggest that, once students are in college, students’ academic and social
integration contribute to their decisions to pursue graduate studies more than their parental
background—but the influence of parental background does not disappear altogether (Ethington
& Smart, 1986). Moreover, the influence of parental background varies depending on the type of
graduate program the student intends to pursue (Mullen et al., 2003). Parental education is more
likely to have a strong influence on students’ matriculation into professional and doctoral
programs and a weaker influence on matriculation into master’s programs (Mullen et al., 2003).
Students with college-educated parents are also more likely to enroll in doctoral and first-
professional programs than are students whose parents have high school diplomas or less
(Mullen et al., 2003).
Student Sex, Race, and Ethnicity
Studies that focus on the graduate education pursuits of students are more prone to
consider a student’s sex than they are to address race and ethnicity. When race and ethnicity are
included in such studies, results are often not disaggregated, due to data limitations (e.g., the
number of REM students is too small for analysis). From the available literature, it is known that
women are more likely than men to enroll in master’s degree programs but less likely than men
to enroll in first-professional programs (Perna, 2004). When race and ethnicity are considered in
such examinations, and when differences in expected costs and benefits and financial and
academic resources are controlled, the likelihood of enrollment in a professional degree program
increases for African American women, but not for African American men (Perna, 2004).
For students with undergraduate S&E majors, the aspects that make graduate studies
14
appealing or unappealing depend upon the student’s sex. Undergraduate men with S&E majors,
who strive toward positions of authority and wealth, shy away from S&E graduate programs
(Sax, 2001). Undergraduate women with S&E majors are less likely to pursue S&E graduate
degrees if staying in physical proximity to family (Villarejo, Barlow, Kogan, Veazey, &
Sweeney, 2008) or raising a family is a high priority and if they are committed to helping others
or effecting social change (Sax, 2001).
Academic Performance
Academic performance—as measured by college GPA—serves as a key indicator as to
whether a student will continue with his or her studies after completion of a bachelor’s degree
(Ethington & Smart, 1986; Hearn, 1987; Mullen et al., 2003; Nevill & Chen, 2007; Pascarella &
Terenzini, 2005; Sax, 2001; Stolzenberg, 1994). Academic performance has a direct effect on
students’ likelihood of taking the Graduate Record Exam (GRE) (Stolzenberg, 1994), applying to
graduate school (Pascarella & Terenzini, 2005), and enrolling in a graduate program (Mullen et
al., 2003; Nevill & Chen, 2007). In fact, students with undergraduate GPAs of 3.5 or higher are
more likely both to enroll in a graduate program and to enroll in such a program full-time
(Neville & Chen, 2007).
Undergraduate Debt
Studies that consider the relationship between undergraduate debt and graduate school
enrollment offer evidence to suggest debt can serve as a negative or as an insignificant factor
(Choy & Gies, 1997; Malcom & Dowd, 2012; Millet, 2003; Weiler, 1991). On the negative side,
the costs associated with graduate school weigh more heavily upon students with undergraduate
debt, and the desire to avoid accruing additional debt serves as a reason for forgoing graduate
studies (Choy & Gies; Millet, 2003). Conversely, Weiler (1991) reports undergraduate debt is
15
insignificant as it pertains to graduate school enrollment. Debt, according to Weiler, is not a
deterrent for students with good academic standing and strong post-baccalaureate study
aspirations, because such students have a better understanding of the returns associated with
pursuing graduate studies. As such, Weiler proposes debt does not alter how students view the
costs and benefits associated with graduate school attendance.
The literature specific to S&E identifies debt as having a negative impact on post-
baccalaureate, graduate-level pursuits (Brazziel & Brazziel, 2001; Malcom & Dowd, 2009;
Seymour & Hewitt, 1997). Seymour and Hewitt (1997), for instance, indicate the best chance of
increasing the proportion of science undergraduates applying to graduate school is to address the
issue of financial aid, as the burden of debt is the strongest deterrent to graduate school
ambitions. In fact, if the goal is for STEM bachelor’s degree recipients to enroll in graduate
school immediately after bachelor’s degree completion, the ideal amount of undergraduate loan
debt is no debt at all (Malcom & Dowd, 2009). Brazziel and Brazziel (2001) point to debt as a
key reason why capable REM students decide to forego doctoral studies. REM students report
being nervous about paying off loans and other debts incurred during their baccalaureate studies
(Brazziel & Brazziel, 2001).
Malcom and Dowd (2012) also note differences in borrowing according to a student’s
race and ethnicity. Whereas African American and Latino STEM bachelor’s degree holders tend
to borrow to finance their undergraduate studies, Asian STEM bachelor’s degree holders are less
likely to borrow and are more likely to draw on parental support to pay for college (Malcom &
Dowd, 2012). Nonetheless, For African American, Asian, Latino, and White STEM bachelor’s
degree holders, undergraduate borrowing reduces the chances of attending graduate school
(Malcom & Dowd, 2012).
16
Student-Faculty Interactions, Undergraduate Research Experience, and STEM-Specific
Opportunity and Support Programs
Students who spend time with faculty members in and out of the classroom, and who
participate in research during their undergraduate careers, are more likely to pursue a graduate
education (Huss, Randall, Patry, Davis, & Hansen, 2002; MacLachlan, 2006; Sax, 2001;
Villarejo et al., 2008). For students with S&E majors, spending time with faculty members,
either by working on a professor’s research project or by serving as a teaching assistant, creates
direct opportunities to garner realistic notions of “life as academic scientists.” This hands-on
experience consequently encourages students to attend graduate school (Sax, 2001). Participation
in research activities is also a significant and direct predictor of perceived preparedness for
graduate school (Huss et al., 2002). In addition to attracting, retaining, and increasing the degree
aspirations of science majors (Kinkead, 2003; Lopatto, 2004), an undergraduate research
experience serves as a pathway to graduate studies (Russell, 2006; Strayhorn, 2010). Engaging in
undergraduate research increases students’ interest in doctoral programs (Russell, 2006) and
sustains or enhances their graduate degree aspirations (Russell, 2006; Strayhorn, 2010).
For REM students, student-faculty interactions and undergraduate research opportunities
are often facilitated through STEM-specific opportunity and support programs. Such programs
are designed to serve undergraduate REM students in the STEM fields and include the
Mathematics Workshop Program, the Meyerhoff Program, the Emerging Scholars Program, the
Biology Undergraduate Scholars Program, the Summer Undergraduate Research Program, and
the Undergraduate Research Opportunity Program, among others (Musesus, Palmer, Davis, &
Maramba, 2001). Although the services offered by these programs vary by institution, they
typically include financial support, mentoring, academic advisement, tutoring, career counseling,
17
and undergraduate research (Musesus et al., 2001). Empirical evaluation of some of these
programs—for instance, of the Meyerhoff Program, Biology Undergraduate Scholars Program,
and Summer Undergraduate Research Program—reveals that participation in such programs
positively influences students’ likelihood of graduate school attendance (Barlow & Villarejo,
2004; Maton, Hrabowski, & Schmitt, 2000; Maton, Sto Domingo, Stolle-McAllister,
Zimmerman, & Hrabowski, 2009; Walters, 1997).
Using longitudinal data from the Meyerhoff Program, Maton, Sto Domingo, Stolle-
McAllister, Zimmerman, and Hrabowski (2009) report that students who entered the Meyerhoff
Program between 1989 and 2003 were more likely to enroll in STEM PhD programs than were
students who had been accepted into the Meyerhoff Program but who had declined to participate.
Similarly, from her interviews with 14 alumni of the Summer Undergraduate Research Program,
Walters (1997) concludes that the interviewees’ participation in the Summer Undergraduate
Research Program had a positive influence on their decisions to pursue graduate studies in
STEM.
Employment
Beyond reports that provide statistics about the number of individuals employed in S&E-
related occupations
3
, there is a lack of research that considers what factors may contribute to the
S&E employment choices of individuals. Moreover, although the empirical studies of
undergraduate students with S&E majors explore career aspirations, such studies do not
necessarily address S&E full-time employment. The available literature on employment centers
around two key topics: major and employment congruence and employment and career

3
S&E-related occupations are those which employ a high proportion of individuals with training
in a S&E educational field and whose output directly or indirectly involves the production of
scientific ideas and new knowledge (Graham & Smith, 2005).
18
differences based upon an individual’s sex and race and ethnicity.
Major and Employment Congruence
Undergraduate major plays a role in employment. Researchers agree individuals with
certain majors are more likely to be employed in occupations related to their respective fields of
study (Choy & Bradburn, 2008; Robst, 2007). Using longitudinal data from the National Center
for Education Statistics, Choy and Bradburn (2008) examined the post-baccalaureate
employment status of the class of 1992-93 during the first 10 years after graduation. Findings
from their study revealed that individuals with career-oriented majors
4
, in comparison to
individuals with academic majors
5
, were more likely to find employment related to their
respective majors (Choy & Bradburn, 2008). A year after graduation, about 65% of individuals
with career-oriented majors reported they were employed in occupations closely related to their
majors. Conversely, 39% of individuals with academic majors reported their jobs were related to
their majors. Generally, during the 10 years after graduation, individuals with academic majors
were more likely to be spread out across more occupations than were those with career-oriented
majors who had stayed in occupations closely related to their undergraduate fields of study
(Choy & Bradburn, 2008). Individuals with career-oriented majors were also less likely to be
unemployed between 1994 and 1997, and again between 1997 and 2003 (Choy & Bradburn,
2008).
Additionally, researchers indicate major-employment congruence is linked to post-
baccalaureate studies, job satisfaction, and higher economic returns. As an individual invests

4
Career-oriented majors are defined as those that prepare students for employment in specific
occupational areas, including (but not limited to) business, education, health, engineering, and
computer science (Choy & Bradburn, 2008).
5
Academic majors include social and behavioral sciences, arts and humanities, biological
sciences, and mathematics/physical sciences (Choy & Bradburn, 2008).
19
more time in schooling, he or she is less likely to accept a job outside of his or her respective
field of study and, as such, is more likely to report his or her job is related to his or her field of
study (Robst, 2007). Individuals who perceive their jobs are related to their undergraduate
majors are more likely to be satisfied with their jobs, as major-employment congruence is
positively related to job satisfaction (Elton & Smart, 1988; Wolniak & Pascarella, 2005). The
consequence of being in a job outside of an individual’s respective degree field, however, is a
lower rate of return to schooling. Women who are not employed in the same fields as their
undergraduate degrees earn about 8.9% less than women who are employed in the same fields as
were studied for their undergraduate degrees (Robst, 2007). Men who are not employed in the
same fields as their undergraduate degrees earn about 10.2% less than men who are (Robst,
2007).
Employment and Career Differences Based on the Individual’s Sex, Race, and Ethnicity
The S&E workforce is composed of individuals who earned degrees over roughly three
decades (NSF & NCSES, 2013). Subsequently, White males are more likely than women and
REM individuals to be employed in S&E occupations (NSF & NCSES, 2013). In fact, women
and REM individuals constitute a lower percentage of the overall S&E workforce than of S&E
degree recipients who recently joined the workforce (NSF & NCSES, 2013). Moreover, women
and REM individuals’ participation in S&E occupations is lower than it is in the U.S. workforce
as a whole (NSF & NCSES, 2013). Among REM individuals, REM women comprise about 1 in
10 employed scientists and engineers (NSF & NCSES, 2013). Across all racial and ethnic
groups, female scientists and engineers who are not working or who are working part-time are
more likely than men to cite family responsibilities, whereas men are more likely to report
retirement, as the reason they are not working or are working part-time (NSF & NCSES, 2013).
20
There are also differences in regards to unemployment; unemployment rates are higher for REM
scientists and engineers than for White scientists and engineers (NSF & NCSES, 2013).
Research examining the career decisions of REM students with S&E majors typically
makes use of two expressions: “career commitment” and “career aspirations.” The phrase “career
commitment” is often used to describe students’ intentions to pursue occupations related to their
respective undergraduate S&E degrees (Moore, 2006). “Career commitment” refers to the
trajectory that spans from a student’s elementary to post-secondary schooling (Moore, 2006).
Along this trajectory, the factors that influence REM students’ to pursue bachelor’s degrees in
S&E and, subsequently, careers in S&E are: a strong interest in S&E, an aptitude for science and
mathematics, strong familial encouragement, and opportunities to participate in S&E-specific
support programs (Moore, 2006). When S&E-specific opportunity and support programs
provide students with research experience, students are exposed to environments conducive
toward careers in research (Villajero et al., 2008).
The phrase “career aspirations,” often used in association with “career commitment,”
refers to whether a student aspires to a career in S&E. Research in this area indicates students
pursuing S&E degrees do not have practical information about their career options, post-
graduation, and consequently abandon their career aspirations. In their interviews with
undergraduates with S&E majors, Seymour and Hewitt (1997) found undergraduate students
often reject S&E careers due to preconceptions based upon myths and stereotypes. Students
imagine S&E careers to have little job security, to require long hours, and to be intellectually dull
and repetitive (Seymour & Hewitt, 1997). As such, students need exposure to a broader
understanding of the work done by science professionals—an exposure that can be facilitated by
connecting students to practicing scientists in S&E related fields (Lewis & Collins, 2001).
21
When a students’ sex is considered in relation to his or her career aspirations, women
who abandon their S&E career aspirations typically do so because they are concerned with
aligning their education, career goals, and personal priorities (Seymour & Hewitt, 1997).
Women, more often than men, switch out of S&E majors if other majors appear to offer a greater
intrinsic interest, if other majors appear to offer a better overall educational experience, or
because the career options or lifestyles associated with S&E majors are seemingly less appealing
than the career options or lifestyles of people without S&E majors (Seymour & Hewitt, 1997). In
regards to “lifestyles,” for example, women who desire a “family-flexible job” are more likely to
abandon their S&E career aspirations (Frome, Alfeld, Eccles, & Barber, 2006). Conversely, if
men abandon their scientific career aspirations, they often do so because they believe there is a
lack of financial reward in S&E fields and a more lucrative future in business or law (Sax, 1994).

22
CHAPTER THREE: CONCEPTUAL MODEL
In this chapter, I provide an overview of the conceptual grounding I use in the three
studies presented in this dissertation. As discussed in the introduction, I approach REM students’
post-baccalaureate choices from a “student-college choice” perspective. Two theoretical
perspectives have generally guided the research on student college choice: an economic model of
human capital and a sociological model of status attainment (Hossler et al., 1989; Paulsen, 1990).
A human capital approach to college choice emphasizes that students will decide to attend
college if the expected benefits of the education outweigh its expected costs (Hossler et al., 1989;
Paulsen 1990). Traditional sociological status attainment models, however, typically focus on the
effects of a student’s socioeconomic status on his or her educational aspirations (Hossler et al.,
1989; Paulsen, 1990).
More recently, scholars have expanded prior models of student college choice and have
created “integrated models” that incorporate the sociological concepts of habitus, cultural capital,
and social capital (McDonough, 1997; Paulsen & St. John, 2002; Perna, 2006). For example,
Paulsen and St. John (2002) incorporate the notion of habitus into their “student choice
construct.” Likewise, in her “model of college choice,” Perna (2006) includes habitus, cultural
capital, and social capital. These integrated models address the strengths and limitations of the
economic model of human capital, as well as the strengths and limitations of the sociological
concepts of habitus, cultural capital, and social capital. For instance, while the human capital
model approach to college choice offers a framework by which to understand college choice, it
does not examine the nature of information available to the decision-maker (Manski, 1993).
Similarly, while using cultural capital and social capital to study college choice can provide
information about the ways in which individuals gather information, these concepts do not offer
23
methods of understanding how individuals make their decisions based on the information
gathered (Manski, 1993). As such, the advantage of integrated models—such as those of Paulsen
and St. John (2002) and Perna (2006)—is two-fold. First, the researchers consider that college
choice reflects an individual’s situated context. Second, the researchers consider there are
multiple routes leading to college enrollment—and that these routes vary by race and ethnicity,
socioeconomic status, age, and other factors (Paulsen & St. John, 2002; Perna, 2006).
In this dissertation, I utilize Perna’s conceptual model of student college choice to
examine the post-baccalaureate choices—“graduate education choice” and “full-time
employment choice”—of REM undergraduate students with S&E majors. While Perna’s model
is relatively new and is primarily intended to explore the college choices of students making the
transition from high school to college, Perna (2006) notes her model can also be used for parallel
processes, such as the decision of a bachelor’s degree recipient to pursue a graduate education. In
addition to using Perna’s model to examine “graduate education choice,” I propose the same
model can be extended to examine “full-time employment choice.” Given that Perna’s model
draws from the economic model of human capital, as well as from the sociological concepts of
habitus, cultural capital, and social capital, the section that follows provides an overview of these
concepts as well as the strengths and limitations of these concepts. Thereafter, a description of
Perna’s conceptual model of student college choice is presented.
Economic and Sociological Grounding of Perna’s Model
Human Capital
At the center of Perna’s conceptual model of student college choice lies the economic
model of human capital investment. The notion of human capital stems, in large part, from the
economic research of the 1960s (Becker, 1962; Schultz, 1961). As defined by Douglass (1997)
24
and supported by most economists, human capital “consists of the acquired energy, motivation,
skills, and knowledge possessed by human beings, which can be harnessed over a period of time
to the task of producing goods and services” (p. 362). It includes competencies learned in the
home, through on-the-job training, and through a formal education (Becker, 1962; Douglass,
1997; Schultz, 1961). From a human capital perspective, then, differences in the productivity
levels of individuals are due to the investments individuals make in their respective personal
development; by investing in their personal development, individuals invest in their human
capital and, consequently, enhance their productivity (Becker, 1962; Douglass, 1997; Schultz,
1961). An education is considered the most valuable of human capital investments (Becker,
1962). Hence, economists refer to an investment in a postsecondary education as an investment
in human capital (DesJardins & Toutkoushian, 2005).
According to the economic model of human capital, when deciding whether or not to
invest time and money into a college education, an individual will weigh the value of expected
future benefits against the expected costs of the education (Becker, 1962, 1993; Douglass, 1997).
In the process of calculating the total expected benefits of a college education, it is assumed an
individual will act rationally, in such a way that will maximize that individual’s utility, and that
the individual will consider the monetary and non-monetary benefits and costs of attending
college (Becker, 1993). The long-term monetary and non-monetary benefits associated with
investing in a post-secondary education include: better health, a longer life, higher lifetime
earnings, informed purchases, and lower probabilities of unemployment (Leslie & Brinkman,
1988). The costs of investing in a college education include direct costs (e.g., tuition and fees for
housing and books) and the indirect costs of foregone earnings and leisure time (Becker, 1993).

25
Aligned with expected costs and benefits are the demand for human capital and the
supply of resources for investing in human capital. Whether or not an individual decides to invest
in a college education is influenced, in part, by the forces that shape demand for human capital
and by the supply of resources available for investment in human capital (Becker, 1993; Ellwood
& Kane, 2000). In relation to pursuing a college education, proponents of the human capital
model propose that differences reflect variations across individuals according to the respective
academic preparations and achievements of those individuals (Paulsen, 2001). Differences in the
supply of resources available to pay for the costs of a college education are expected to reflect
differences in the accessibility of parental income and financial aid (Ellwood & Kane, 2000;
Paulsen, 2001). Hence, individuals who are more likely to invest in a postsecondary education—
and, consequently, more likely to invest in their human capital—are those who are academically
prepared and who have financial resources (Paulsen, 2001).
Strengths and limitations of human capital. The primary strength of human capital is
that it carries a considerable amount of explanatory power regarding the prediction of the effects
of changes in monetary benefits and costs on student enrollment behavior (Paulsen, 2001). As
such, a key contribution of human capital approaches to college choice is a focus on the effects
of finances (including tuition, family income, and financial aid) on enrollment (Ellwood & Kane,
2000; Manski & Wise, 1983). The primary limitation of the economic model of human capital,
however, is that it assumes rational behavior: “The conventional definition of rational behavior
usually holds that individuals have a well-defined set of preferences and, when faced with a set
of choices, they will choose the option that maximizes their satisfaction” (DesJardins &
Toutkoushian, 2005, p. 191).

26
In short, human capital assumes a rational choice model in which individuals’ actions—
assumed to be intentional and goal-oriented—are instrumental and geared toward a particular
end, and in which individuals have the ability to weigh options to make the best decisions
(DesJardins & Toutkoushian, 2005). In relation to college choice, human capital assumes
individuals will have the information necessary for rational action when weighing the expected
benefits and costs of attending college. Economists note that rationality does not require the
decision-maker to have perfect information, but rather that he or she tries to make a decision
based upon the information at his or her disposal (DesJardins & Toutkoushian, 2005).
Economists further establish that the key to evaluating rationality is to determine whether or not
the individual acts in a manner consistent with his or her preferences (DesJardins &
Toutkoushian, 2005). However, as DesJardins and Toutkoushian (2005) point out, economists
“take preferences as given and do not delve into how [such preferences] are formed or why they
differ across individuals” (p. 211).
Moreover, it is not sound to presume that all students absorb or utilize information in the
same way (Tierney & Venegas, 2007), are well informed about the costs and economic benefits
of investing in a college education (Perna, 2006), or have identical perceptions of the economic
benefits and costs associated with a postsecondary education (Paulsen, 2001). For instance,
Paulsen (2001) explains that factors including socioeconomic status and background, early home
and school environments, and access to information about financial aid and employment
opportunities can result in varying perspectives regarding the economic benefits and costs of a
postsecondary education.

27
Cultural Capital, Habitus, and Social Capital
Perna’s conceptual model of student college choice integrates the sociological concepts
of cultural capital, habitus, and social capital. The sociological concepts of cultural capital and
habitus are attributed to the French sociologist Pierre Bourdieu. While the development of social
capital is also attributed to Bourdieu, the most frequently applied conceptualization of social
capital in educational research is that of James Coleman
6
. In this subsection, an overview of the
concepts of cultural capital and habitus as proposed by Bourdieu is presented, as well as an
overview of social capital as proposed by Coleman.
Cultural capital and habitus. With his conceptualization of cultural capital, Bourdieu
aimed to demonstrate an individual’s culture could act as a form of capital and, consequently, as
a “power source” in social settings (Swartz, 1997, p. 75). Hence, cultural capital is related to the
class-based socialization of culturally relevant skills, abilities, preferences, tastes, or norms that
act as forms of currency in the social realm (Bourdieu, 1979/1984). According to Bourdieu,
cultural capital is primarily acquired through an individual’s family and through a formal
education (Bourdieu, 1979/1984). However, it is more difficult to acquire cultural capital
through an education, alone, than through the pairing of these factors. Bourdieu (1979/1984)
states:
Total, early, imperceptible learning, performed within the family from the earliest days of
life and extended by scholastic learning which presupposes and completes it, differs from
belated, methodical learning...It confers the self-certainty which accompanies the
certainty of possessing culture legitimacy, and the ease which is the touchstone of
excellence; it produces the paradoxical relationship to culture made up of self-confidence

6
Perna’s model uses Coleman’s conceptualization of social capital.

28
amid (relative) ignorance and of casualness amid familiarity, which bourgeois families
hand down to their offspring as if it were an heirloom. (p. 66)
Hence, cultural capital transmitted at birth and reinforced through an education—as compared to
cultural capital gained only through an education—bestows upon the individual the privilege and
certainty of possessing society’s legitimate culture.
According to Bourdieu, there are three forms of cultural capital: 1) embodied, 2)
objectified, and 3) institutionalized (Bourdieu, 1986). In the embodied state, cultural capital is a
“form of long-lasting dispositions of the mind and body,” that is, what the individual knows and
can do (Bourdieu, 1986, p. 243). Embodied capital requires an investment of time and cannot be
separated from the individual (Bourdieu, 1986). Cultural goods (e.g., paintings, books,
instruments) represent the objectified state of cultural capital (Bourdieu, 1986). Objectified
cultural capital is appropriated materially with economic capital and symbolically via the
embodied state (Bourdieu, 1986). Educational qualifications represent the institutionalized state
of cultural capital. Institutionalized cultural capital creates a “certificate of cultural competence
which confers on its holder a conventional, constant, legally guaranteed value with respect to
culture” (Bourdieu, 1986, p. 248). For example, when a college confers a degree, the individual’s
embodied cultural capital (e.g., perceived academic qualifications) takes on an objective value:
the academic degree represents objectified cultural capital granted through institutionalized
cultural capital.
Using a metaphor of a poker game, cultural capital represents the cards that an individual
can play. Some cards are automatically dealt to an individual based on his or her background,
while other cards are requested by the individual or are exchanged, depending upon the
individual’s education (Winkle-Wagner, 2010). In a fair game of poker, each round of the game
29
determines which cards are valuable. In the world of cultural capital, however, it is maintained
that some people are always given the hand of cards they need. Additionally, cultural capital
alone does not help explain the “rules” of the game or how the game is played (Bourdieu,
1979/1984; Winkle-Wagner, 2010). As such, it is necessary to discuss cultural capital as it
relates to Bourdieu’s concepts of field and habitus.
Bourdieu’s concept of field pertains to the space in which cultural competence—or
knowledge of particular tastes, dispositions, or norms—is both produced and given a price
(Bourdieu, 1979/1984). The field determines the properties, internalized as dispositions and
objectified as economic or cultural goods, that are valid, active, or pertinent in a given social
setting (Bourdieu, 1979/1984). A field is not universal—many fields subsist—and it is only
within a particular field that cultural capital holds value, produces an effect, or even exists
(Bourdieu, 1979/1984). As in a game, each field has its own rules or systems of valuation that
determine the conditions of entry (e.g., educational credentials, particular mannerisms or tastes,
economic capital) and the social relations within the field (e.g., who is valued or recognized,
whose cultural norms are recognized or rewarded) (Bourdieu, 1979/1984; Winkle-Wagner,
2010). Even the rules themselves can differ from game to game, a fact that dramatically alters the
value of the hand an individual is dealt and influences whether or not that individual can
exchange cards (Bourdieu, 1979/1984; Winkle-Wagner, 2010). Hence, an individual’s cultural
capital might be useful in one field and essentially meaningless in another (Bourdieu, 1979/1984;
Winkle-Wagner, 2010).
Habitus, as defined by Bourdieu (1977), is “a set of durable and transposable dispositions
that mediate between objective structures of social relations and individual subjective behavior”
(p. 72). Dispositions are the schemes of perception, thought, and action individuals acquire in
30
response to the structures they encounter (e.g., society, family, and the educational system)
(Bourdieu, 1977). Although socialization toward a particular habitus begins in early childhood, it
continues into adulthood, as individuals internalize the “rules” that govern the field of interaction
and their places within that field (Bourdieu, 1977; Swartz, 1997). In a continuance of the poker
game metaphor, if cultural capital represents the cards, and if the field is the setting in which the
game is played, habitus is the approach an individual takes to playing his or her hand of cards.
Habitus, then, relates to the cultural capital an individual recognizes as being available in social
settings (Bourdieu, 1977; Winkle-Wagner, 2010).
Strengths and limitations of cultural capital. A notable strength of Bourdieu’s theory of
cultural capital is that it expands the very notion of “capital” to mean not only a tangible,
physical property, but also an intangible currency attainable through an individual’s social origin
and education (Bourdieu, 1979/1984, 1986). With his definition of cultural capital, Bourdieu
(1986) also moves beyond the notion of human capital, noting that human capital ignores the
scholastic yield from an education depends upon the cultural capital previously invested by a
family. Hence, cultural capital theory considers the role of family background, whereas human
capital does not.
The second strength of Bourdieu’s cultural capital theory is that it recognizes the role of
institutions in maintenance of a status quo. For instance, according to Bourdieu (1979/1984),
educational institutions reward students who are already equipped with cultural capital and
legitimize that capital by creating the appearance that the reproduction of social hierarchies is
based on a student’s gifts, merits, or skills.
A third strength of Bourdieu’s cultural capital is that it, and the larger framework in
which it is rooted (i.e., field and habitus), allows for examination of not only the individual and
31
the group, but also of the group and the social structure that surrounds it (Winkle-Wagner, 2010).
Overall, when used appropriately, cultural capital is a valuable theoretical source, specifically in
respect to research related to class issues, social stratification, or an understanding of how
inequality is perpetuated (Winkle-Wagner, 2010).
Despite its numerous strengths, Bourdieu’s theory of cultural capital is burdened by five
limitations. First, an economic metaphor is implicit in the notion of cultural capital and assumes
that an individual’s primary motivation is the exchange of goods, services, or money—or, in the
case of cultural capital, the exchange of cultural knowledge, competencies, and skills (Winkle-
Wagner, 2010). In other words, an economic metaphor builds from a “means-to-ends”
assumption that is individualistic and implies cultural capital is about individuals’ maximizing
their options (Winkle-Wagner, 2010).
Second, Bourdieu’s cultural capital theory is rooted in a problematic Marxist dominant-
subordinate class dichotomy. Consequently, cultural capital is a class-based theory: individuals
with strong cultural capital are those from the upper class who share a common culture, and
those with weaker cultural capital are from the subordinate class (Winkle-Wagner, 2010). Third,
Bourdieu’s work implies an individual’s agency is limited because of his or her unconscious
acceptance of the existing structures (Winkle-Wagner, 2010). Fourth, Bourdieu’s theory does not
refer to race, ethnicity, or sex. Throughout his work (e.g., Distinction: A Social Critique of the
Judgment of Taste), an assumption lingers that participants are homogenous.
Lastly, if misused, cultural capital can be a deficiency model. Winkle-Wagner (2010)
notes that Bourdieu’s emphasis on explaining the cultural capital of the majority group is most
often adapted in educational research. Applying this perspective results in a deficiency approach,
emphasizing the cultural capital that non-majority groups do not have.
32
Social capital. Drawing upon a base of rational choice theory, Coleman (1988) defines
social capital as “not a single entity but a variety of different entities, with two elements in
common: they all consist of some aspect of social structures, and they facilitate certain actions of
actors—whether persons or corporate actors—within the structure” (p. S98). Hence, social
capital, as conceptualized by Coleman, is defined by its function and is inherent in the structure
of relations between and among actors (Dika & Singh, 2002). Because social capital exists in the
relations among individuals, it is intangible, and, like physical capital and human capital, can
facilitate productivity (Coleman, 1988, 1990).
According to Coleman (1988), there are three forms of social capital: 1) obligations and
expectations, 2) information channels, and 3) norms and effective sanctions. The first form of
social capital, obligations and expectations, is the “credit slips” individuals’ exchange among
each other (Coleman, 1990). For instance, if Miguel does a favor for Enrique and trusts Enrique
to reciprocate in the future, this establishes an expectation in Miguel that he will be able redeem
his “credit slip”—and an obligation on the part of Enrique to honor the “credit slip” (Coleman,
1988). This form of social capital depends on trustworthiness—the belief that obligations will be
repaid, the extent to which obligations are held, and the actual tendency of individuals in the
social environment to give and ask for help (Coleman, 1988, 1990).
Information channels—the second form of Coleman’s social capital—refers to the
gathering and sharing of information through social relations (Coleman, 1988, 1990).
Information is important because it provides a basis for action, but the gathering of information
is costly. At minimum, such gathering requires attention, which is not always in supply, but
individuals in a social structure can depend upon one another for information (Coleman, 1988).
If, for example, Enrique reads the sports section of the newspaper every day, but Miguel does
33
not, Miguel can turn to Enrique for information about sports.
The third form of social capital articulated by Coleman—norms and effective sanctions—
is the shared standards of behavior and values of a social structure (Coleman, 1990). When a
norm exists and is effective, it becomes a form of social capital because it can, for example,
inhibit crime and increase the ability to walk freely and safely in a city at night (Coleman, 1988).
Overall, Coleman presents social capital as positive social control for which expectations and
obligations, information channels, and norms are types of social capital (formed through social
relations) and are to be used by individuals within the social structure (Dika & Singh, 2002;
Portes, 1998).
Coleman (1990) acknowledges that, just as social capital can be created, it can also be
destroyed. Moreover, he identifies closure, stability, and ideology as key factors associated with
these processes of creation and destruction (Coleman, 1990). Closure refers to closure in the
social network, a necessary process that facilitates the emergence of norms and trust. Following
closure, norms can develop that keep those within the system from imposing externalities on one
another. In contrast, a lack of closure makes it impossible for norms to develop (Coleman,
1990).
To maintain social capital, the social structure also needs stability (Coleman, 1990).
Disruptions to the social organization—such as an individual’s mobility—can be destructive to
the structure itself and, consequently, to the social capital that depends upon the social structure
(Coleman, 1990).
Lastly, depending on the type of ideology, social capital can be produced or damaged.
For instance, an ideology centered on the collective helps create social capital because such an
ideology encourages an individual to act in the interests of something or someone other than
34
himself or herself (Coleman, 1990). Conversely, according to Coleman (1990), an ideology of
self-sufficiency inhibits the creation of social capital because such an ideology promotes
independence rather than reciprocal assistance.
Strengths and limitations of social capital. The primary strength of Coleman’s theory of
social capital is its orientation. That is, Coleman (1988, 1990) depicts social capital as positive
individual or collective action generated by networks of relationships, reciprocity, trust, and
social norms. Another strength of this theory is that Coleman (1988, 1990) does not equate social
capital with class, perhaps assuming that an individual’s class does not influence his or her social
capital. In fact, Coleman’s (1988) work supports the idea that parents of any social class can
accumulate social capital for their children if those parents adopt the appropriate parenting
norms—for example, being involved at their children’s schools.
Finally, Coleman’s theory of social capital carries a component of public good. Coleman
notes that—unlike physical capital and human capital—the forms of social capital are not private
goods (Coleman, 1988). Coleman (1988) states that the “social structures that make possible
social norms, and the sanctions that enforce them, do not benefit primarily the person or persons
whose efforts would necessary bring them about, but benefit all those who are a part of such
structure” (p. S116).
Nonetheless, Coleman’s theory of social capital is not without limitations. The first of
these limitations relates to the theorist’s focus on the positive effect of inclusion in a social group
without consideration of associated negative implications (Portes & Landolt, 1996; Portes,
1998). Such negative consequences of social capital include: the exclusions of outsiders, excess
claims on group members, restrictions of an individual’s freedom, and downward leveling norms
(Portes, 1998). In short, the same strong ties that can benefit members of a social group can also
35
help exclude others from access to that group. As noted by Portes (1998), not only is closure
needed for social capital to exist, it can create excess claims on group members by preventing,
for example, the success of business initiatives outside of the group. Similarly, being a member
of a group can restrict an individual’s freedom because group membership requires conformity
(Portes, 1998).
In regards to downward leveling norms, Portes (1998) refers to situations in which
individuals are united by “a common experience of adversity and opposition to mainstream
society” (p. 17). In such a situation, individual success would undermine the unity of the group
because group membership is based on the idea that such success is impossible. The result,
according to Portes (1998), “is downward leveling norms that operate to keep members of a
downtrodden group in place and force the more ambitious to escape from it” (p. 17).
Another limitation of Coleman’s theory is his “top-down” view of the family-child
relationship (Dika & Singh, 2002). Coleman’s work supports the notion that it is the
responsibility of the family to adopt certain norms in order to advance a child’s social capital.
Such a view ignores agency—the possibility of the child becoming an adolescent capable of
accessing his or her own social capital (Dika & Singh, 2002). Coleman’s work—in relation to
the family’s adoption of norms—also emphasizes parental involvement, but ignores the ways in
which race and social class influence parental involvement (Dika & Singh, 2002).

36
Perna’s Conceptual Model of Student College Choice
At the center of Perna’s (2006) model is the human capital perspective, in which college
choice is based upon the weighing of expected benefits against expected costs (see Figure 1).
The human capital model proposes college choice is based upon the weighing of expected
benefits of a collegiate education against its expected costs (Becker, 1993; Perna, 2006). Those
expected costs include the costs of attendance and forgone earnings, while the expected benefits
include monetary and non-monetary benefits (Perna, 2006). Aligned with expected costs and
benefits are the demand for human capital and the supply of resources for investment in human
capital. Differences in demand are based upon individuals’ academic preparation and
achievement, whereas differences in supply reflect the availability of resources —such as family
income and financial aid—to pay for college (Ellwood & Kane, 2000; Perna, 2006; Paulsen,
2001).
Moving beyond the human capital model perspective, Perna’s model specifies that an
individual’s calculations of the expected costs and benefits are nested within four contextual
layers that shape an individual’s choice: 1) the individual’s habitus, 2) the school and community
contexts, 3) the higher education context, and 4) the broader social, economic, and policy
contexts (see Figure 1).
The first layer—an individual’s habitus—reflects an individual’s demographic
characteristics, including sex, race or ethnicity, and socioeconomic status, as well as his or her
cultural capital and social capital (Perna, 2006). For the second layer (school and community
context), Perna borrows McDonough’s (1997) notion of “organizational habitus” to address the
manner in which social structures and resources facilitate or impede an individual’s college
choice. For instance, research by Stanton-Salazar (1997) shows institutional structures
37
Figure 1. Perna’s Conceptual Model of Student College Choice

Adapted from “Studying College Access and Choice: A Proposed Conceptual Model,” by
L.W. Perna (2006) in J.C. Smart (ed.), Higher Education: Handbook of Theory and Research,
Vol. 21 p. 117. Copyright 2006 by Springer.

!
!
!
!
!
!
!
!
!
Social, economic, & policy context (layer 4)
Demographic characteristics
Economic characteristics
Public policy characteristics
Higher education context (layer 3)
Marketing and recruitment
Location
Institutional characteristics
Habitus (layer 1)

Demographic characteristics
Gender
Race/Ethnicity

Cultural Capital
Cultural knowledge
Value of college attainment

Social Capital
Information about college
Assistance with college processes

School and community context (layer 2)
Availability of resources
Types of resources
Structural supports and barriers
Demand for higher education
Academic preparation
Academic achievement

Supply of resources
Family income
Financial aid

! !
Expected benefits
Monetary
Non-monetary

Expected costs
College costs
Foregone earnings

! !

College
Choice
38
limit the extent to which working-class minority students can develop “trusting” relationships
with institutional agents, such as counselors, teachers, and middle-class peers. Layer two, then,
represents the availability of resources, types of resources, and structural supports and barriers
within the school and community context (Perna, 2006).
The higher education context—the third layer—represents the role post-secondary
institutions play in shaping college choice. According to Perna (2006), higher education
institutions—through the information they provide to students and families, their determinations
regarding which applicants can enroll, and their control over the number of available enrollment
slots—can influence the college choice process. Perna, Steele, Woda, Hibbert (2005), for
example, speculate that an increase in population growth may increase demand for higher
education and, consequently, limit the availability of enrollment slots. In such a situation,
institutions may increase academic requirements or increase tuition, actions that are more likely
to have a negative impact on low-income and racial and ethnic minority students (Perna et al.,
2005).
Finally, the fourth layer accounts for the broader social, economic, and political contexts
that can indirectly and directly influence an individual’s college choice, such as demographic
changes, unemployment rate, and federal policies on financial aid (Perna, 2006). In sum, with
her model, Perna (2006) suggests that, although college choice is based on a comparison of the
expected benefits and the expected costs of enrolling, assessments of the benefits and costs are
shaped not only by the demand and supply of resources with which to pay for college, but also
by an individual’s habitus and (directly and indirectly) by the school and community context, the
higher education context, and the social, economic, and policy contexts.

39
CHAPTER FOUR: STUDY ONE
While REM individuals’ shares of S&E bachelor’s degrees have been rising since 1991,
current figures reveal slight increases in the numbers of master’s degrees and doctoral degrees
earned by REM individuals (see Table 1, Table 2, Table 3, Table 4) (NSF & NCSES, 2013).
Advanced degree attainment is important because, of the 8.6 million S&E jobs projected by
2018, about 779,000 jobs will require employees with master’s degrees or higher (Carnevale et
al., 2010). Research focusing on the post-baccalaureate choices of REM individuals with S&E
majors, however, is limited.
In this chapter, I present the first study, which focuses on the “graduate education choice”
of REM undergraduate students with S&E majors. I also examine a comparison group of REM
undergraduate students with non-S&E majors. Accordingly, the research questions I address in
this study are:
I. What factors inform the “graduate education choice” of REM undergraduate students
with S&E majors?
II. What factors inform the “graduate education choice” of REM undergraduate students
with non-S&E majors?
In the section that follows, I provide an overview of the conceptual model that guides this
study. Thereafter, I present the methods and the findings.
Conceptual Model and Application
Perna’s (2006) conceptual model of student college choice serves as the conceptual
grounding for this study. While her model is intended to explore the college choices of students
making the transition from high school to college, Perna notes that her model can also be used
for parallel processes, such as the transition from undergraduate studies to graduate studies.
40
Perna’s model draws from the economic model of human capital, as well as from the
sociological concepts of habitus, cultural capital, and social capital (see Figure 1). In the
subsections that follow, I describe how Perna’s model is applied in this study.
Center of the Model: Human Capital
At the center of Perna’s model of student college choice is the economic model of human
capital investment. The human capital model proposes that college choice is based upon the
weighing of expected benefits against expected costs (Becker, 1993; Perna, 2006). The expected
costs of investing in a college education include direct costs, such as tuition and fees for housing
and books, and the indirect costs of foregone earnings and leisure time (Becker, 1993). The
expected benefits—long-term monetary and non-monetary—associated with investing in a
postsecondary education include: better health, a longer life, higher lifetime earnings, informed
purchases, and lower probabilities of unemployment (Leslie & Brinkman, 1988). The human
capital model proposes that, if the individual determines the benefits outweigh the expected
costs, the individual will make the choice to attend college (Becker, 1962, 1993; Douglass,
1997).
Aligned with expected costs and benefits are the demand for human capital and the
supply of resources for investment in human capital. That is, the human capital model recognizes
differences in college choice are connected to the forces that shape the demand for human capital
and to the supply of resources available for investment in human capital (Becker, 1993; Ellwood
& Kane, 2000). In the human capital model, differences in the demand for a post-secondary
education are expected to reflect variations across individuals according to their academic
preparations and achievements (Paulsen, 2001). Differences in the supply of resources available
to pay for the costs of a college education are expected to reflect differences in the accessibility
41
of parental income and financial aid (Ellwood & Kane, 2000; Paulsen, 2001). Hence, individuals
who are more likely to make the choice to attend college are those who are academically
prepared and who have greater personal financial resources (Paulsen, 2001).
To account for the human capital aspect of Perna’s model of student college choice, this
study includes seven variables as proxies for expected benefits, demand, and supply of resources
(see Figure 2). Four variables are included to represent expected benefits, and three of these
variables stand as non-monetary benefits. The three non-monetary variables are: “helping others
who are in difficulty,” “becoming an authority in my field,” and “becoming successful in a
business of my own.” The fourth variable (i.e., “being very well-off financially”) represents a
monetary benefit.
Demand-oriented variables include undergraduate GPA and “preparedness for
graduate/advanced education.” Given that human capital theory proposes that the individuals
who are more likely to make the choice to attend college are those who are academically
prepared (Paulsen, 2001), I assume for the purposes of this study that REM individuals with
competitive undergraduate GPAs—and who believe their undergraduate studies prepared them
for a graduate education—will be more likely to make the choice to attend graduate/professional
school.
Finally, “undergraduate student loan amount” serves as a variable representing supply of
resources. I propose that having a low debt amount (or no debt at all) is considered a supply of
resources. As noted by Malcom and Dowd (2011), if the goal is for STEM bachelor’s degree
recipients to enroll in graduate school immediately after completion of their bachelor’s degrees,
the ideal amount of undergraduate loan debt is no debt at all (Malcom & Dowd, 2009).

42
Figure 2. Application of Perna’s Conceptual Model of Student College Choice for “Plan to
Attend Graduate/Professional School” for REM Undergraduate Students with S&E and Non-
S&E Majors

Habitus (layer 1)

Demographic Characteristics
 Student’s Sex

Cultural Capital
 Parents’ Levels of Education

Social Capital
 Participated in an Internship
 Participated in a Racial/Ethnic
Student Organization
 Social Capital through Student-
Faculty Interactions
Demand for Higher
Education

 Undergraduate GPA
 Preparedness for
Graduate/Advanced
Education

Supply of Resources
 Undergraduate Loan
Amount
Expected Benefits

 Helping Others in
Difficulty
 Being Very Well-
Off Financially
 Becoming an
Authority in my
Field
 Becoming
Successful in a
Business of my
Own

Graduate
Education
Choice
Higher Education Context (layer 3)

 Institutional Selectivity

43
Four Contextual Layers
Moving beyond the human capital model perspective, Perna’s model specifies that an
individual’s calculations of the expected costs and benefits are nested within four contextual
layers. Specifically, Perna proposes these four contextual layers shape an individual’s choice of
whether to pursue a postsecondary education. The layers
7
are: 1) the individual’s habitus, 2) the
school and community contexts, 3) the higher education context, and 4) the broader social,
economic, and policy contexts. For this study, I consider the first and third layers (see Figure 2).
Habitus. The first layer, an individual’s habitus, reflects an individual’s demographic
characteristics, including sex, race and ethnicity, and socioeconomic status, as well as his or her
cultural capital and social capital (Perna, 2006). Cultural capital is related to the class-based
socialization of culturally relevant skills, abilities, preferences, tastes, or norms, which acts as a
form of currency in the social realm
8
(Bourdieu, 1979/1984). As noted by Bourdieu (1979/1984),
cultural capital is primarily acquired through an individual’s family. Social capital exists in the
relations between individuals and, like physical and human capital, can facilitate productivity
9

(Coleman, 1988). According to Coleman (1988, 1990) one form of social capital is “information
channels,” which refers to the gathering and sharing of information through social relations
(Coleman, 1988, 1990).
To account for the habitus layer, five variables are considered (see Figure 2). The first
variable, student’s sex, is included because prior research indicates differences, based upon sex,
in the graduate education pursuits of REM individuals (Perna, 2004). The next variable, parents’
levels of education, is used as a proxy for cultural capital. Given that cultural capital is primarily

7
For an overview of the layers, see Chapter 3.
8
For an overview of cultural capital, see Chapter 3.
9
For an overview of social capital, see Chapter 3.
44
acquired through family (Bourdieu, 1979/1984), I hypothesize that parents’ levels of education
bestow relevant skills that inform an individual’s “graduate education choice.” As noted by
Ethington and Smart (1986), when an undergraduate student is determining whether he or she
will pursue graduate studies, the influence of parental background does not disappear.
The three remaining variables are participation in an internship program, participation in
a racial/ethnic student organization, and a composite variable, “social capital through student-
faculty interactions.” These three variables stand as proxies for social capital. Given that
“information channels” (i.e., social capital) can be summarized as the gathering and sharing of
information through social relations (Coleman, 1988, 1990), I assume that, by participating in an
internship, being a member of a racial/ethnic student organization, and by interacting with
faculty members, REM individuals are exposed to social relations through which they can gather
information about graduate school. This assumption is supported by Sax (2001), who notes that,
for students with S&E majors, spending time with faculty members—either by working on a
professor’s research project or by serving as a teaching assistant—creates direct opportunities for
students to garner realistic notions of “life as academic scientists” and, consequently, encourages
students to attend graduate school.
It is important to note that, because Coleman’s theory of social capital limits the notion of
agency
10
, I adopt the notion that “individuals have a degree of agency to shape their realities
(Tierney & Venegas, 2006, p. 1690). In other words, I adopt a human agency perspective
towards social capital and believe it is possible for a REM student to garner his or her own social
capital; through interactions with peers and faculty members, a REM student can access

10
Coleman’s work supports the notion that it is the responsibility of the family to adopt certain
norms to advance the child’s social capital. Such a view ignores agency—that is, the possibility
of the child becoming an adolescent capable of accessing his or her own social capital (Dika &
Singh, 2002).
45
information that can ultimately inform his or her “graduate education choice.”
Higher education context. The third of Perna’s contextual layers, the higher education
context, represents the role postsecondary institutions play in shaping college choice. For this
study, in particular, institutional type and control, as well as institutional selectivity, are
considered. In regards to institutional type and control, the data are filtered to consider only four-
year, private, predominately White institutions. In regards to institutional selectivity, scale
variable
11
of median SAT scores and/or ACT composite scores is included to represent
institutional selectivity (see Figure 2). As indicated in the literature, graduates of selective
research institutions are more likely to continue their education after receiving bachelor’s
degrees (Mullen et al., 2003). The research also indicates that REM students who attend
selective, as opposed to non-selective, institutions are more likely to complete bachelor’s degrees
(Alon & Tienda, 2005; Melguizo, 2008). Bachelor degree completion is, of course, important
because REM students must first complete undergraduate studies before considering graduate
school. The intersection between institutional selectivity and control is also important. Eide,
Brewer, and Ehrenberg (1998) note that attending a selective private college not only increases
the probability of attending graduate school, but also the likelihood of attending graduate school
at a major research institution.
In sum, in this study, I utilize Perna’s integrated model of student college choice to
examine the factors that inform the graduate education choices of REM undergraduate students
with S&E majors. Specifically, I focus on the human capital aspect, the habitus layer, and the

11
This variable is the median SAT scores and/or ACT composite scores of the entering class as
reported to the Integrated Postsecondary Educational Data System (IPEDS). The median SAT is
based on a combination of verbal and math scores (i.e., verbal + math). See Figure B-3 in
Appendix B.

46
higher education context layer (see Figure 2). The section that follows provides an overview of
the methods utilized in this study.
Methods
12

Samples
This study utilized two samples: a sample of REM undergraduate students who reported
S&E majors and a comparison group sample of REM undergraduate students who reported non-
S&E majors. In both samples, all students had attended private, predominately White, four-year
institutions and, by June of 2003, had earned bachelor’s degrees. A total of 446 students reported
S&E majors. Of these, 65.5% were women, 1.3% identified as American Indian, 37.9% as Asian,
17.5% as Black, 18.4% as Latino, and 24.9% as multiracial/multiethnic
13
. The majority of
students with S&E majors reported undergraduate GPAs of “B+” (65.1%), parents with college
degrees or higher (66.4%), and estimated parental incomes of less than $60,000 (50.2%). The top
three S&E majors
14
reported were: psychology (17.9%), political science (16.4%), and biology
(10.3%) (see Table 5).
A total of 518 students reported non-S&E majors. Of these students, 70.8% were women,
1.2% identified as American Indian, 26.3% as Asian, 20.1% as Black, 23.7% as Latino, and
28.8% as multiracial/multiethnic. The majority of students with non-S&E majors reported
undergraduate GPAs of “B+” (58.6%), parents with college degrees or higher (59.8%), and
estimated parental incomes of less than $60,000 (51.1%). As shown in Table 6, the top three
non-S&E majors
15
reported were: English (10.6%), history (8.7%), and other arts and humanities
(7.3%).

12
For an overview of the data, see p. 6.
13
The student marked “two or more races/ethnicities” as his or her race or ethnicity.
14
For the full list of S&E majors, see Table A-1 in Appendix A.
15
For the full list of non-S&E majors, see Table A-2 in Appendix A.
47
Table 5

Overview of REM Undergraduate Students with S&E Majors for Study 1 and Study 2 (N=446)

Characteristic

Percent

Race/Ethnicity

American Indian

1.3

Asian 37.9

Black

17.5

Latino

18.4

Two or More Races/Ethnicities

24.9

Sex
Female

65.5

Male

34.5
Parents’ Education
Some College or Less

33.6
College Degree or Higher

66.4

Parents’ Income

$59,999 or less

50.2

$60,000 to $999,999

21.8

$100,000 or more

28

Undergraduate GPA
B or Lower

34.9

B+ or Higher 65.1

S&E Majors (Top 3)*

Psychology 17.9

Political Science

16.4

Biology (General)

10.3

* For the full list of majors see Table A-1 in Appendix A
48
Table 6

Overview of REM Undergraduate Students with Non-S&E Majors for Study 1 and Study 2
(N=518)

Characteristic

Percent

Race/Ethnicity
American Indian 1.2
Asian 26.3
Black 20.1
Latino 23.7
Two or More Races/Ethnicities 28.8
Sex
Female 70.8
Male 29.2
Parents’ Education
Some College or Less 40.2
College Degree or Higher 59.8
Parents’ Income
$59,999 or less 51.1
$60,000 to $999,999 23.6
$100,000 or more 25.2
Undergraduate GPA
B or Lower 41.4
B+ or Higher 58.6
Non-S&E Majors (Top 3)*
English 10.6
History 8.7
Other Arts & Humanities

7.3

* For the full list of majors, see Table A-2 in Appendix A.

49
Variables
With the exception of student’s sex, parental levels of education, and institutional
selectivity, all variables were from the 2007 College Senior Survey. For both samples, variables
were included to account for Perna’s model of student college choice—specifically, human
capital, layer one, and layer three (see Figure 3). To represent human capital, variables were
included to account for expected benefits, demand for higher education, and supply of resources.
Variables accounting for expected benefits included: “helping others who are in difficulty,”
“being very well-off financially,” “becoming an authority in my field,” and “becoming
successful in a business of my own.”
The variables addressing demand for higher education included undergraduate GPA and
preparedness for graduate or advanced education. To represent supply of resources,
undergraduate student loan amount was also included. Five additional variables were included to
represent habitus. These variables accounted for demographic characteristics, cultural capital,
and social capital. The habitus-oriented variables were: student’s sex, parents’ levels of
education, participation in an internship program, participation in a racial/ethnic student
organization, and “social capital through student-faculty interactions,” a composite variable (see
Figure 3). To represent the higher education context, institutional selectivity was included. The
dependent variable was binary—“do not plan to attend graduate/professional school” and “plan
to attend graduate/professional school”—and was from the 2007 College Senior Survey (see
Figure 3).
Analyses
For both samples (REM students with S&E majors and REM students with non-S&E
majors), statistical analyses consisted of descriptive statistics, exploratory factor analysis,

50
Figure 3. Variables and Coding for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Students with S&E and Non-S&E Majors

Variable

Coding

Human Capital—Expected Benefits

Helping Others Who Are in Difficulty 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Being Very Well-Off Financially 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Becoming an Authority in My Field 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Becoming Successful in a Business of my Own 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Human Capital—Demand for Higher Education
Undergraduate GPA 0 = D, C, C+, B-, B
1 = B+, A-, A or A+
Preparedness for Graduate/Advanced Education 0 = Much Weaker, Weaker, No Change
1 = Stronger, Much Stronger
Human Capital—Supply of Resources
Undergraduate Loan Amount Scale Variable — Less than $10,000 to
$2000,000 or More
Habitus (Layer 1)—Demographic Characteristics

Student’s Sex 0 = Male 1 = Female
Habitus (Layer 1)—Cultural Capital
Parents’ Levels of Education 0 = Grammar School or Less, Some High
School, High School Graduate, Some
College
1 = College Degree, Some Graduate
School, Graduate Degree
Habitus (Layer 1)—Social Capital
Participated in an Internship Program 0 = No 1 = Yes
Participated in a Racial/Ethnic Student Organization 0 = No 1 = Yes
Student-Faculty Interactions (  =.685 &  =.634)
 Faculty Provided Advice and Guidance about
Educational Programs
 Faculty Provided Opportunity to Work on a
Research Project
 Faculty Provided a Letter of Recommendation

0 = Not At All, Occasionally
1 = Frequently
Higher Education Context (Layer 3)
Institutional Selectivity Scale Variable — 800 to 1600
Dependent

Plan to Attend Graduate/Professional School
0 = No 1 = Yes

51
missing value analysis and multiple imputation, screening for multicollinearity, logistic
regression, and testing for linearity of the logit. Descriptive statistics were used to garner
summaries of the samples (see Tables 5 & 6).
Exploratory factor analysis. The variables considered for the composite variable,
“social capital through student-faculty interactions,” were subjected to principle component
analysis (PCA). Prior to PCA, suitability of data for factor analysis was assessed. For both
samples, inspection of the correlation matrixes revealed the presence of coefficients of .3 and
above. The Kaiser-Meyer-Olkin values were .64 for the sample of REM students with S&E
majors and .612 for the sample of REM students with non-S&E majors. These values met the
recommended value of .6 (Pallant, 2011), and the Barlett’s Tests of Sphericity (Pallant, 2011)
reached statistical significance for both samples, supporting the factorability of correlation
matrixes.
PCA for the sample of REM students with S&E majors revealed the presence of only one
component with eigenvalues exceeding 1 and explaining 61.4% of the variance. Because only
one component was extracted, rotation was not plausible. The component matrix revealed that
the variables faculty provided advice and guidance about educational program (.836), faculty
provided opportunity to work on a research project (.778), and faculty provided letter of
recommendation (.734) loaded strongly. Reliability statistics for the component showed a
Cronbach’s Alpha on standardized items of .685 (see Figure 4). PCA for the sample of REM
students with non-S&E majors also revealed the presence of only one component with
eigenvalues exceeding 1 and explaining 57.8% of the variance. The component matrix revealed
that, as before, the variables faculty provided advice and guidance about educational program
(.829), faculty provided opportunity to work on research project (.722), and faculty provided
52
letter of recommendation (.726) loaded strongly. Reliability statistics for the component showed
a Cronbach’s Alpha on standardized items of .634 (see Figure 4).

Figure 4. Reliability and Constituent Variables for “Social Capital Through Student-Faculty
Interactions”

Reliability

Constituent Variables

 = .685 REM undergraduate students with
S&E majors

 = .634 REM undergraduate students with
non-S&E majors

 Faculty Provided Advice and Guidance
about Educational Program
 Faculty Provided Opportunity to Work on
a Research Project
 Faculty Provided a Letter of
Recommendation

Missing value analysis and multiple imputation. Missing value analyses were
conducted for both samples to verify the extent of missing data. These analyses revealed random
missing data patterns, a discovery that supported the implementation of multiple imputation to
compensate for the missing values (Allison, 2009). Multiple imputation was selected because,
unlike mean value replacement, it provides an accurate estimate of missing data—and, like
maximum likelihood, multiple imputation estimates are consistent and asymptotically normal
(Allison, 2009). Moreover, multiple imputation can be applied to any type of data or model, an
advantage that is not offered by maximum likelihood (Allison, 2009).
Using IBM SPSS 20, the fully conditional specification approach of multiple imputation
was used to impute missing values. However, missing values for the variables student’s sex and
parents’ educational levels were not imputed and were deleted prior to imputation. As
recommended, during the imputation process, five copies of the data were created, each with
missing values imputed; five data sets are enough to reveal parameter estimates that are close to
being fully efficient (Allison, 2009). The results showed the original data set and the pooled
53
results across the five imputed data sets.
Multicollinearity. To screen for multicollinearity, a linear regression for each sample
was run, and collinearity diagnostics were examined. Menard (1995) suggests tolerance values of
less than 0.1 indicate a collinearity problem. Similarly, Myers (1990) suggests a VIF value
greater than 10 is cause for concern. The collinearity diagnostics for both linear regressions
revealed no major collinearity between the independent variables. Both linear regressions
showed tolerance values greater than 0.1 and VIF values less than 10. Hence, there were no
major multicollinearity issues.
Logistic regression. Logistic regression was selected because it allows the prediction of
a discrete outcome—for this study, “do not plan to attend graduate/professional school” and
“plan to attend graduate/professional school”—and it predicts the category of outcome for
individual cases using the most parsimonious model (Field, 2009; Tabachnick & Fidell, 2007).
Because two continuous variables were included (i.e., undergraduate loan amount and
institutional selectivity, as shown in Figure 3), it was necessary to check that each continuous
variable was linearly related to the log of the outcome variable (i.e., “plan to attend
graduate/professional school”) (Field, 2009). To test for linearity, a logistic regression for each
sample was run with the interactions between each continuous variable and the log of that
variable—e.g., Selectivity x LnSelectivity, where Ln is the natural log transformation (Field,
2009). The results of both regressions indicated the interaction terms were not significant and,
consequently, that the assumption of linearity of the logit was met for the two continuous
variables (Field, 2009).
Two binary logistic regressions were conducted, one for REM students with S&E majors
and the other for REM students with non-S&E majors. All variables were simultaneously entered
54
for both regressions. Concerning the logistic regression of REM students with S&E majors, the
following functional form expressed the relationship between the dependent variable and the
independent variables:
REM undergraduate students with S&E majors grad
gi
= f ([HD
i
WF
i
AF
i
SB
i
UG
i
PG
i
UA
i
SX
i

PE
i
SC
i
PI
i
SO
i
IS
i
], u
i
)
Where students’ grad
gi
= 0 do not plan to attend graduate/professional school, 1 plan to
attend graduate/professional school. HD
i
= helping others who are in difficulty; WF
i
=
being very well-off financially; AF
i
= becoming an authority in my field; SB
i
= becoming
successful in a business of my own; UG
i
= undergraduate GPA; PG
i
= preparedness for
graduate/advanced education; UA
i
= undergraduate loan amount; SX
i
= student’s sex; PE
i

= parents’ levels of education; SC
i
= social capital through student-faculty interactions;
PI
i
= participated in internship program; SO
i
= participated in racial/ethnic student
organization; IS
i
= institutional selectivity; and u
i
= a stochastic error term.
Hence, the following was the logit model, where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
HD
i
+ b
3
WF
i
+...+ b
7
PG
i
+ b
8
UA
i
… + b
13
SO
i
b
14
IS
i
+ u
i

For the logistic regression of REM students with non-S&E majors, the following
functional form expressed the relationship between the dependent variable and the independent
variables:
REM undergraduate students with non-S&E majors grad
gi
= f ([HD
i
WF
i
AF
i
SB
i
UG
i
PG
i
UA
i
SX
i
PE
i
SC
i
PI
i
SO
i
IS
i
], u
i
)
Where students’ grad
gi
= 0 do not plan to attend graduate/professional school, 1 plan to
attend graduate/professional school. HD
i
= helping others who are in difficulty; WF
i
=
being very well-off financially; AF
i
= becoming an authority in my field; SB
i
= becoming
55
successful in a business of my own; UG
i
= undergraduate GPA; PG
i
= preparedness for
graduate/advanced education; UA
i
= undergraduate loan amount; SX
i
= student’s sex; PE
i

= parents’ level of education; SC
i
= social capital through student-faculty interactions; PI
i

= participate in internship program; SO
i
= participated in racial/ethnic student
organization; IS
i
= institutional selectivity; and u
i
= a stochastic error term.
Hence, the following was the logit model, where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
HD
i
+ b
3
WF
i
+...+ b
7
PG
i
+ b
8
UA
i
…+b
13
SO
i
+b
14
IS
i
+ u
i

Limitations
As with any study, there are limitations to address. First: data from the 2003 Freshmen
Survey and 2007 College Senior Survey were self-reported. Similarly, the dependent variable
(“do not plan to attend graduate school” and “plan to attend graduate school”) only provides
information about a student’s “intended plans” to choose or not to choose graduate school. This
variable does not confirm whether a student actually chose, or did not choose, graduate school.
Second: this study treated REM students as a homogeneous group. Due to available
sample sizes, it was not possible to conduct separate regression analyses—for example, a
regression analysis of African American or Latino students.
Third: in both samples, more than half of total participants were women, and the samples
did not include students of non-traditional age or students who had transferred from two-year
institutions. Finally, despite its advantages, multiple imputation produces different results every
time it is used (Allison, 2009). This is because the “imputed values are random draws rather than
deterministic quantities,” (Allison 2009, p. 81).
56
Findings
16

REM Undergraduate Students With S&E Majors
The results of the logistic regression for REM students with S&E majors indicated the
final block was statistically reliable in distinguishing between “do not plan to attend
graduate/professional school” and “plan to attend graduate/professional school.” The inferential
goodness-of-fit test—the Hosmer–Lemeshow (H–L)—yielded a X
2
(8) of 5.31 and was
insignificant (p >.05), indicating that the model fit the data well; the null hypothesis of a good
model fit to data was plausible. The Cox and Snell R
2
and the Nagelkerke R
2
were .12 and .17,
respectively (see Table 7).
Wald statistics of the variables included to represent human capital indicated “helping
others who are in difficulty” (X
2
(1) 4.67, p<.05) and “preparedness for graduate or advanced
education” (X
2
(1) 6.09, p<.01) significantly predicted “plan to attend graduate/professional
school.” Respectively, odds ratios demonstrated that students who reported their preparedness
for graduate or advanced education as “stronger/much stronger” than when they had begun
college were 3.2 times more likely to report they planned to attend graduate/professional school.
Conversely, the odds ratio for “helping others who are in difficulty” was below 1; students who
reported that “helping others who are in difficulty” was “very important/essential” were .5 times
less likely to report they planned to attend graduate school.
For the habitus component (i.e., layer one), Wald statistics specified an individual’s sex
(X
2
(1) 3.86, p<.05), and “social capital through student-faculty interactions” (X
2
(1) 13.78,
p<.01), were significant. Odds ratios showed women were about 2 times more likely to report
they planned to attend graduate school. Similarly, students who reported frequent social capital

16
The findings reported are the pooled results across the five imputed data sets. For the non-
imputed results, see Table A-3 and Table A-4 in Appendix A.
57
Table 7

Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Students with S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others in Difficulty -.685 .317 4.67 .031* .504 .271 .938

Being Very Well-Off Financially .250 .266 .883 .349 1.28 .761 2.16

Becoming an Authority in my Field .309 .268 1.33 .248 1.36 .806 2.30

Becoming Successful in a Business of my Own -.430 .262 .097 .101 .651 .389 1.09

Undergraduate GPA .042 .253 .027 .867 1.04 .635 1.71

Preparedness for Graduate/Advanced Education 1.16 .470 6.09 .013** 3.20 1.28 8.05

Undergraduate Loan Amount -.020 .117 .029 .864 .980 .776 1.24

Student’s Sex .495 .252 3.86 .050* 1.64 1.00 2.69

Parents’ Levels of Education -.206 .243 .718 .396 .814 .505 1.31

Participated in Internship Program -.016 .247 4.19 .950 .984 .607 1.60

Participated in a Racial/Ethnic Student Organization -.344 .228 2.27 .132 .709 .453 1.11

Social Capital Through Student-Faculty Interactions .928 .250 13.78 .000*** 2.53 1.55 4.13

Institutional Selectivity -.003 .001 9.00 .012** .997 .995 .999

Constant 1.40 2.01 .485 .490

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.31 8 .724

Omnibus Test of Model 56.53 13 .000***

-2 Log Likelihood 479.83 +

Cox & Snell R
2
.119

Nagelkerke R
2
.170

* p<.05; **p<.01; ***p<.000 + 536.36 with only constant in model
58
interactions with faculty were 2.5 times more likely to report they planned to attend
graduate/professional school.
Finally, for the higher education context (i.e., layer three), the Wald statistic for
institutional selectivity (X
2
(1) 9.0, p<.01) was significant. The odds ratio for institutional
selectivity (.997) indicated that, as institutional selectivity decreased, the odds of planning to
attend graduate/professional school decreased (see Table 7).
REM Undergraduate Students With Non-S&E Majors
The results of the logistic regression for REM students with non-S&E majors indicated
the model was statistically reliable in distinguishing between “do not plan to attend
graduate/professional school” and “plan to attend graduate/professional school.” The inferential
goodness-of-fit test—the Hosmer–Lemeshow (H–L)—yielded a X
2
(8) of 3.58 and was
insignificant (p >.05), indicating the model fit the data well. The Cox and Snell R
2
and the
Nagelkerke R
2
were .7 and .11, respectively (see Table 8).
Wald statistics indicated three variables were significant. Two of those variables were
undergraduate GPA (X
2
(1) 4.98, p<.05) and “preparedness for graduate/advance education”
(X
2
(1) 7.18, p<.01), which were included to represent human capital. Odds ratios revealed that
students who reported GPAs of B+ or higher and that their “preparedness for graduate/advance
education” was “stronger/much stronger” than it was when they had begun college, were about 2
and 4 times, respectively, more likely to report they planned to attend graduate/professional
school.
The third significant variable, as revealed by Wald statistics, was “social capital through
student-faculty interactions” (X
2
(1) 7.15, p<.01), which was included as part of the habitus
component (i.e., layer one). Students who reported frequent “social capital interactions” with
59
Table 8

Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Students with Non-S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others in Difficulty .143 .328 .190 .663 1.15 .606 2.19

Being Very Well-Off Financially .212 .266 .635 .426 1.24 .733 2.08

Becoming an Authority in my Field .166 .260 .407 .524 1.18 .709 1.96

Becoming Successful in a Business of my Own -.304 .259 .587 .240 .738 .444 1.23

Undergraduate GPA .587 .263 4.98 .026* 1.80 1.07 3.01

Preparedness for Graduate/Advanced Education 1.45 .541 7.18 .007** 4.28 1.48 12.34

Undergraduate Loan Amount -.240 .155 2.39 .132 .787 .573 1.08

Student’s Sex -.317 .254 1.32 .213 .728 .442 1.20

Parents’ Levels of Education .037 .243 .023 .879 1.04 .645 1.67

Participated in Internship Program -.152 .243 .391 .531 .859 .534 1.38

Participated in a Racial/Ethnic Student Organization .146 .238 .376 .541 1.16 .725 1.85

Social Capital Through Student-Faculty Interactions .679 .254 7.15 .008** 1.97 1.20 3.24

Institutional Selectivity .000 .001 0 .728 .997 .998 1.00

Constant -1.62 2.39 .459 .502

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 3.58 8 .893

Omnibus Test of Model 36.15 13 .001***

-2 Log Likelihood 477.80 +

Cox & Snell R
2
.7

Nagelkerke R
2
.11

* p<.05; **p<.01; ***p<.000 + 513.95 with only constant in model
60
faculty were about 2 times more likely to report they planned to attend graduate/professional
school (see Table 8).
Discussion
Three primary discussion points arise from the results of the analyses. The first
discussion point concerns the human capital aspect of Perna’s model. In the economic model of
human capital, college choice is based on weighing the expected benefits of collegiate education
against its expected costs (Becker, 1993; Perna, 2006). Of the variables included to account for
expected benefits, only one was significant: “helping others who are in difficulty.” More
specifically, this variable was only significant for REM students with S&E majors. However, this
variable carried a negative effect. REM students with S&E majors who reported that “helping
others who are in difficulty” was “very important/essential” were .5 times less likely to report
they planned to make the choice to attend graduate/professional school. It could be that students
do not perceive graduate school as a means through which they can help others.
The demand for human capital is aligned with expected benefits. To account for demand,
the variables included were: undergraduate GPA and “preparedness for graduate/advanced
education.” I hypothesized that REM individuals with competitive undergraduate GPAs and who
believed their undergraduate studies prepared them for a graduate education would be more
likely to make the choice to attend graduate school. The results aligned with this hypothesis.
REM undergraduate students with S&E majors and REM undergraduate students with non-S&E
majors who reported their preparedness for graduate/advanced education was “stronger/much
stronger” than it was when they had begun college were 3.2 and 4 times, respectively, more
likely to report they planned to make the choice to attend graduate/professional school. In respect
to undergraduate GPA, this variable was also significant, but only for REM students with non-
61
S&E majors. REM students with non-S&E majors who reported GPAs of B+ or higher were
about 2 times more likely to report they planned to make the choice to attend
graduate/professional school. This finding is consistent with previous research that indicates that
academic performance—as measured by college GPA—has a direct effect on students’
likelihood of taking the GRE (Stolzenberg, 1994), applying to graduate school (Pascarella &
Terenzini, 2005), and enrolling in graduate programs (Mullen et al., 2003; Nevill & Chen, 2007).
The second discussion point illustrates a relationship to the first layer of Perna’s model.
The first layer, an individual’s habitus, reflects demographic characteristics including sex, race,
and ethnicity, and socioeconomic status, as well as his or her cultural capital and social capital
(Perna, 2006). Of the five variables included to account for habitus, two were significant:
student’s sex and “social capital through student-faculty interactions.” In regards to the former,
REM women with S&E majors were about 2 times more likely—than REM men with S&E
major—to report they planned to make the choice to attend graduate/professional school. In
regards to the latter, given that “information channels” (i.e., social capital) concern the gathering
and sharing of information through social relations (Coleman, 1988/1990) I assumed that, by
interacting with faculty members, REM individuals would be exposed to social relations through
which they could gather information about graduate school. The findings aligned with this
inference. REM students with S&E majors and REM students with non-S&E majors who
reported frequent interactions with faculty were 2.5 and 2 times, respectively, more likely to
report they planned to make the choice to attend graduate/professional school.
It is important to note the variable “social capital through student-faculty interactions”
was a composite variable, which consisted of three individual variables: faculty provided advice
and guidance about educational programs, faculty provided opportunity to work on research
62
projects, and faculty provided letter of recommendation. Given that this composite variable was
significant for REM students with S&E majors and for REM students with non-S&E majors, this
finding indicates that interactions with faculty members are important for REM students who
attend private, predominately White, four-year institutions, regardless of students’ academic
majors.
The final discussion point concerns the third layer of Perna’s model. This layer, the
higher education context, represents the role postsecondary institutions play in shaping college
choice (Perna, 2006). For this study in particular, institutional type and control—as well as
institutional selectivity—were considered. Across both samples, all students had attended
private, predominately White, four-year institutions. And while both logistic regressions
included institutional selectivity, the results revealed institutional selectivity was only significant
for REM students with S&E majors. That significance, however, was negative. The odds ratio
for institutional selectivity (.997) indicated that, as institutional selectivity decreased, the odds of
planning to make the choice to attend graduate/professional school decreased. This finding is
noteworthy for two reasons. First, as supported by prior research (Eide et al., 1998; Mullen et al.,
2003), it confirms institutional selectivity influences graduate school attendance—or, within the
context of this study, whether students will (or will not) plan to make the choice to attend
graduate school. Second, while prior research indicates REM individuals who attend selective
institutions are more likely to complete bachelor’s degrees (Alon & Tienda, 2005; Melguizo,
2008), this finding highlights that—for REM students with S&E majors—institutional selectivity
also matters, as such selectivity relates to students’ intended graduate education choices.
63
CHAPTER FIVE: STUDY TWO
As noted in the introduction, by 2018, S&E occupations will account for about 8.6
million jobs in the U.S. economy (Carnevale et al., 2010). These S&E jobs will require about 1.2
million employees with bachelor’s degrees (Carnevale et al., 2010). To fill these jobs, it is
necessary to increase the number of REM individuals pursuing baccalaureate degrees in S&E-
related fields. However, it is unknown whether REM individuals who graduate with S&E
baccalaureate degrees choose to pursue full-time employment. Moreover, it is unclear what
factors contribute to REM individuals’ decisions to seek full-time employment.
In this chapter, I present the second study, which focuses on the “full-time employment
choice” of REM undergraduate students with S&E majors. I also examine a comparison group of
REM undergraduate students with non-S&E majors. Accordingly, the research questions I
address in this study are:
I. What factors inform the “full-time employment choice” of REM undergraduate
students with S&E majors?
II. What factors inform the “full-time employment choice” of REM undergraduate
students with non-S&E majors?
In the section that follows, I provide an overview of the conceptual model that guides this
study. Thereafter, I present the methods and the findings.
Conceptual Model and Application
Perna’s (2006) conceptual model of student college choice serves as the conceptual
grounding for this study. Perna’s model draws from the economic model of human capital, as
well as from the sociological concepts of habitus, cultural capital, and social capital (see Figure
1). Perna notes her model can be used for parallel processes, such as the transition from
64
undergraduate studies to graduate studies. I propose that the model can also be extended to
examine employment choice. In the subsections that follow, I describe how I apply Perna’s
model in this study.
Center of the Model: Human Capital
At the center of Perna’s model of student college choice is the economic model of human
capital investment. This model proposes that college choice is based on weighing the expected
benefits of education against its expected costs (Becker, 1993; Perna, 2006). The expected costs
of investing in a college education include direct costs (such as tuition and fees for housing and
books) and indirect costs of foregone earnings and leisure time (Becker, 1993). The expected
benefits—long-term monetary and non-monetary—associated with investing in a postsecondary
education include: better health, a longer life, higher lifetime earnings, informed purchases, and
lower probabilities of unemployment (Leslie & Brinkman, 1988). The human capital model
proposes that, if the individual determines the benefits of a college education outweigh its
expected costs, then the individual will choose to attend college (Becker, 1962, 1993; Douglass,
1997). Aligned with expected costs and benefits are the demand for human capital and the supply
of resources for investment in human capital. That is, the human capital model recognizes that
differences in college choice are also connected to the forces that shape both the demand for
human capital and the supply of resources to invest in human capital (Becker, 1993; Ellwood &
Kane, 2000).
In the human capital model, differences in the demand for a post-secondary education are
expected to reflect variations across individuals according to the individuals’ respective
academic preparation and achievement (Paulsen, 2001). Differences in the supply of resources
available to pay for the costs of a college education are expected to reflect differences in the
65
accessibility of parental income and financial aid (Ellwood & Kane, 2000; Paulsen, 2001).
Hence, individuals who are more likely to choose to attend college are those who are
academically prepared and who have greater personal financial resources (Paulsen, 2001).
For this study, to account for the human capital aspect of Perna’s model, six variables are
included as proxies for expected benefits, demand, and supply of resources (see Figure 5).
Concerning expected benefits, three variables are included—two of which represent non-
monetary benefits. These non-monetary benefit variables are: “raising a family,” and “becoming
successful in a business of my own.” The third variable, “being very well-off financially,”
represents a monetary benefit. Concerning demand, the included variables are undergraduate
GPA and “preparedness for employment after college.”
Given that human capital theory proposes that individuals who are more likely to choose
to attend college are those who are academically prepared (Paulsen, 2001), I assume for the
purposes of this study that REM students with GPAs of “B” or lower will be more likely to
choose to work full-time. That is, students with below-average GPAs may perceive the choice of
working full-time as the realistic choice, compared with pursuing continued education. I also
assume REM individuals who believe their undergraduate studies prepared them for employment
after college will be more likely to make the choice to work full-time.
Finally, concerning supply of resources, the variable considered is “undergraduate loan
amount.” I propose that having a low debt amount (or no debt at all) is considered a supply of
resources. I hypothesize, however, that REM students with increasing loan amounts will be more
likely to make the choice to work full-time. Students with loans are conceivably more prone to
choosing to work full-time in order to begin the process of repayment.

66
Figure 5. Application of Perna’s Conceptual Model of Student College Choice for “Plan to Work
Full-Time” for REM Undergraduate Students with S&E and Non-S&E Majors

Habitus (layer 1)

Demographic Characteristics
 Student’s Sex

Cultural Capital
 Parents’ Levels of Education

Social Capital
 Participated in an Internship
 Met with an Advisor/Counselor
About Career Plans
 Faculty Provided Help in Achieving
Professional Goals
 Faculty Provided an Opportunity to
 Work on a Research Project
 Faculty Provided a Letter of
Recommendation

Demand for Higher
Education

 Undergraduate GPA
 Preparedness for
Employment after
College

Supply of Resources
 Undergraduate Loan
Amount
Expected Benefits

 Raising a Family
 Being Very Well-
off Financially
 Becoming
Successful in a
Business of my
Own

Full-Time
Employment
Choice
Higher Education Context (layer 3)

 Institutional Selectivity

67
Four Contextual Layers
Moving beyond the human capital model perspective, Perna’s model contends that an
individual’s calculations of the expected costs and benefits are nested within four contextual
layers. Specifically, Perna proposes that these four contextual layers shape an individual’s
choice. The layers
17
are: 1) the individual’s habitus, 2) the school and community contexts, 3) the
higher education context, and 4) the broader social, economic, and policy contexts. For this
study, I consider the first and third layers (see Figure 5).
Habitus. The first layer, an individual’s habitus, reflects demographic characteristics,
including sex, race, and ethnicity, and socioeconomic status, as well as his or her cultural capital
and social capital (Perna, 2006). Cultural capital is related to the class-based socialization of
culturally relevant skills, abilities, preferences, tastes, or norms that act as forms of currency in
the social realm
18
(Bourdieu, 1979/1984). As noted by Bourdieu (1979/1984), cultural capital is
primarily acquired through an individual’s family. Social capital exists in the relations between
individuals and, like physical and human capital, can facilitate productivity
19
(Coleman, 1988).
According to Coleman (1988, 1990) one form of social capital is “information channels,” the
gathering and sharing of information through social relations (Coleman, 1988, 1990).
To account for the habitus layer, seven variables are considered (see Figure 5). The first
variable, student’s sex, is included because prior research indicates that differences exist, based
upon sex, concerning full-time employment of REM individuals (NSF & NCSES, 2013). The
next variable, parents’ levels of education, is used as a proxy for cultural capital. Given that
cultural capital is primarily acquired through family (Bourdieu, 1979/1984), I postulate that

17
For an overview of the layers, see Chapter 3.
18
For an overview of cultural capital, see Chapter 3.
19
For an overview of social capital, see Chapter 3.
68
parents’ levels of education bestow relevant skills that inform an individual’s full-time
employment choice. The remaining five variables are proxies for social capital: participation in
an internship program, “met with an advisor/counselor about career plans,” “faculty provided
help in achieving professional goals,” “faculty provided an opportunity to work on a research
project,” and “faculty provided a letter of recommendation.” Given that “information channels”
(i.e., social capital) are the gathering and sharing of information through social relations
(Coleman, 1988, 1990), I assume participation in an internship and interactions with staff and
faculty members exposes REM students with S&E majors to social relations through which they
can gather information about their employment options. As Lewis and Collins (2001) indicate,
students need exposure to a broader understanding of the work done by science professionals. It
is reasonable to infer that, through the aforementioned social relations, REM students may gather
a broader understanding of S&E careers—an understanding which may, in turn, inform their
employment choices.
It is important to note that, because Coleman’s theory of social capital limits the notion of
agency
20
, I work from the assumption that “individuals have a degree of agency to shape their
realities” (Tierney & Venegas, 2006, p. 1690). That is, I adopt a human agency-oriented
perspective towards social capital. From this perspective, I think it is possible for a REM student
to garner his or her own social capital.
Higher education context. The third layer, the higher education context, represents the
role postsecondary institutions play in shaping college choice. For this study in particular,
institutional type and control—as well as institutional selectivity—are considered. The data are

20
Coleman’s work supports the notion that it is the responsibility of the family to adopt certain
norms to advance the child’s social capital. Such a view ignores agency—that is, the possibility
of the child becoming an adolescent capable of accessing his or her own social capital (Dika &
Singh, 2002).
69
filtered to consider only four-year, private, predominately White institutions, and a scale
variable
21
of median SAT scores and/or ACT composite scores is included to represent
institutional selectivity (see Figure 5). The research indicates REM students who attend selective,
as opposed to non-selective, institutions are more likely to complete bachelor’s degrees (Alon &
Tienda, 2005; Melguizo, 2008). Moreover, the benefits associated with attending a selective
institution extend to individuals’ employment and economic returns (Bowen & Bok, 1998;
Brewer et al., 1999).
In sum, in this study, I utilize Perna’s integrated model of student college choice to
examine factors that inform the “full-time employment choice” of REM undergraduate students
with S&E majors. More specifically, I focus on the human capital aspect, the habitus layer, and
the higher education context layer (see Figure 5). The section that follows provides an overview
of the methods.
Methods
22

Sample
This study utilized two samples—a sample of REM students who reported S&E majors
and a comparison group sample of REM students who reported non-S&E majors. Across both
samples, all students had attended private, predominately White, four-year institutions and, by
June of 2003, had earned bachelor’s degrees. A total of 446 students reported S&E degrees. Of
these students, 65.5% were women, 1.3% identified as American Indian, 37.9% as Asian, 17.5%

21
This variable is the median SAT scores and/or ACT composite scores of the entering class as
reported to the Integrated Postsecondary Educational Data System (IPEDS). The median SAT is
based on a combination of verbal and math scores (i.e., verbal + math). See Figure B-3 in
Appendix B.
22
For an overview of the data, see p. 6
70
as Black, 18.4% as Latino, and 24.9% as multiracial/multiethnic
23
. The majority of students with
S&E majors reported undergraduate GPAs of “B+” (65.1%), parents with college degrees or
higher (66.4%), and estimated parental incomes of less than $60,000 (50.2%). The top three S&E
majors
24
reported were: psychology (17.9%), political science (16.4%), and biology (10.3%) (see
Table 5).
A total of 518 students reported non-S&E majors. Of these students, 70.8% were women,
1.2% identified as American Indian, 26.3% as Asian, 20.1% as Black, 23.7% as Latino, and
28.8% as multiracial/multiethnic. The majority of students with non-S&E majors reported
undergraduate GPAs of “B+” (58.6%), parents with college degrees or higher (59.8%), and
estimated parental incomes of less than $60,000 (51.1%). The top three non-S&E majors
25

reported were: English (10.6%), history (8.7%), and other arts and humanities (7.3%) (see Table
6).
Variables
With the exception of student’s sex, parents’ levels of education, and institutional
selectivity, all variables were from the 2007 College Senior Survey. Across both samples,
variables were included to account for Perna’s model of student college choice—specifically,
human capital, layer one, and layer three (see Figure 6). To represent human capital, variables
were included to account for expected benefits, demand for higher education, and supply of
resources. The variables representing expected benefits included “raising a family,” “being very
well-off financially,” and “becoming successful in a business of my own.” The variables
included for demand for higher education were undergraduate GPA and “preparedness for

23
The student marked “two or more races/ethnicities” as his or her race or ethnicity.
24
For the full list of S&E majors, see Table A-1 in Appendix A.
25
For the full list of non-S&E majors, see Table A-2 in Appendix A.
71
Figure 6. Variables and Coding for “Plan to Work Full-Time” for REM Undergraduate Students
with S&E and Non-S&E Majors

Variable

Coding

Human Capital—Expected Benefits
Raising a Family 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Being Very Well-Off Financially 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Becoming Successful in a Business of my Own 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Human Capital—Demand for Higher Education
Undergraduate GPA 0 = D, C, C+, B-, B
1 = B+, A-, A or A+
Preparedness for Employment After College 0 = Much Weaker, Weaker, No Change
1 = Stronger, Much Stronger
Human Capital—Supply of Resources

Undergraduate Loan Amount
Scale Variable — Less than $10,000 to
$2000,000 or More
Habitus (Layer 1)—Demographic Characteristics
Student’s Sex 0 = Male 1 = Female
Habitus (Layer 1)—Cultural Capital

Parents’ Levels of Education
0 = Grammar School or Less, Some High
School, High School Graduate, Some
College
1 = College Degree, Some Graduate
School, Graduate Degree
Habitus (Layer 1)—College Social Capital

Participated in Internship Program
0 = No 1 = Yes

Met with an Advisor/Counselor About Career Plans

0 = Not At All, Occasionally
1 = Frequently

Faculty Provided Help in Achieving Professional
Goals

0 = Not At All, Occasionally
1 = Frequently

Faculty Provided an Opportunity to Work on a
Research Project

0 = Not At All, Occasionally
1 = Frequently
Faculty Provided a Letter of Recommendation

0 = Not At All, Occasionally
1 = Frequently
Higher Education Context (Layer 3)
Institutional Selectivity Scale Variable — 800 to 1600
Dependent

Plan to Work Full-Time
0 = No 1 = Yes

72
employment after college.” Undergraduate loan amount was included to represent supply of
resources. Five variables—accounting for demographic characteristics, cultural capital, and
social capital—were included. These variables were: sex, parents’ levels of education,
participation in an internship program, “met with an advisor/counselor about career plans,”
“faculty provided help in achieving professional goals,” “faculty provided opportunity to work
on a research project,” and “faculty provided a letter of recommendation.” Institutional
selectivity was included to represent the higher education context. The dependent variable was
binary—“do not plan to work full-time” and “plan to work full-time”—and was from the 2007
College Senior Survey (see Figure 6).
Analyses
For both samples—REM students with S&E majors and REM students with non-S&E
majors—statistical analyses consisted of descriptive statistics, missing value analysis and
multiple imputation, screening for multicollinearity, logistic regression, and testing for linearity
of the logit. Descriptive statistics were used to garner summaries of the samples (see Tables 5 &
6).
Missing value analysis and multiple imputation. For both samples, missing value
analyses were conducted to verify the extent of missing data. These analyses revealed random
missing data patterns, a discovery which supported the implementation of multiple imputation
(Allison, 2009). As such, multiple imputation was used to compensate for the missing values.
Multiple imputation was selected because (unlike mean value replacement) it provides an
accurate estimate of missing data. Multiple imputation also (like maximum likelihood) yields
consistent and asymptotically normal estimates (Allison, 2009). Moreover, multiple imputation
can be applied to any type of data or model, an advantage that is not offered by maximum
73
likelihood (Allison, 2009). Using IBM SPSS 20, the fully conditional specification approach of
multiple imputation was used to calculate missing values. However, missing values for the
variables student’s sex and parents’ educational levels were not imputed and were deleted prior
to imputation. As recommended, during the imputation process, five copies of the data were
created, each with missing values imputed; five data sets are enough to get parameter estimates
that are close to being fully efficient (Allison, 2009). The results showed the original data set and
the pooled results across the five imputed data sets.
Multicollinearity. To screen for multicollinearity, a linear regression for each sample
was run, and collinearity diagnostics were examined. Menard (1995) suggests that tolerance
values less than 0.1 indicate a collinearity problem. Similarly, Myers (1990) suggests that a VIF
value greater than 10 is cause for concern. The collinearity diagnostics for both linear regressions
revealed no major collinearity between the independent variables. Both linear regressions
showed tolerance values greater than 0.1 and VIF values less than 10. Hence, there were no
major multicollinearity issues.
Logistic regression. Logistic regression was selected because it allows the prediction of
a discrete outcome—for this study, “do not plan to work full-time” and “plan to work full-
time”—and predicts the category of outcome for individual cases using the most parsimonious
model (Field, 2009; Tabachnick & Fidell, 2007). Because two continuous variables were
included (i.e., undergraduate loan amount and institutional selectivity, as shown in Figure 6) it
was necessary to check that each one was linearly related to the log of the outcome variable (i.e.,
plan to work full-time) (Field, 2009). To test for linearity, a logistic regression for each sample
was run with the interactions between each continuous variable and the log of that variable—
e.g., Selectivity x LnSelectivity, where Ln is the natural log transformation (Field, 2009). The
74
results of both regressions indicated the interaction terms were not significant and, consequently,
that the assumption of linearity of the logit was met for the two continuous variables (Field,
2009).
Two binary logistic regressions were conducted, one for REM students with S&E majors
and the other for REM students with non-S&E majors. All variables were simultaneously entered
for both regressions. For the logistic regression of REM students with S&E majors, the following
functional form expressed the relationship between the dependent variable and the independent
variables:
REM undergraduate students with S&E majors work
gi
= f ([RF
i
WF
i
SB
i
UG
i
PE
i
UA
i
SS
i
PE
i
PI
i

MC
i
FG
i
FR
i
FL
i
IS
i
], u
i
)
Where work
gi
= 0 do not plan to work full-time, 1 plan to attend to work full time. RF
i
=
raising a family; WF
i
= being very well-off financially; SB
i
= becoming successful in a
business of my own; UG
i
= undergraduate GPA; PE
i
= preparedness for employment
after college; UA
i
= undergraduate loan amount; SS
i
= student’s sex; PE
i
= parents’ levels
of education; PI
i
= participated in internship program; MC
i
= met with an
advisor/counselor about career plans; FG
i
= faculty provided help in achieving
professional goals; FR
i
= faculty provided opportunity to work on a research project; FL
i
= faculty provided a letter of recommendation; IS
i
= institutional selectivity; and u
i
= a
stochastic error term.
Hence, the following is a logit model where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
RF
i
+ b
3
WF
i
+...+ b
7
UA
i
…+b
10
PI
i
+b
11
MC
i
+…b
15
IS
i
+ u
i

For the logistic regression of REM students with non-S&E majors, the following
functional form expressed the relationship between the dependent and the independent variables:
75
REM undergraduate students with non-S&E majors work
gi
= f ([RF
i
WF
i
SB
i
UG
i
PE
i
UA
i
SS
i

PE
i
PI
i
MC
i
FG
i
FR
i
FL
i
IS
i
], u
i
)
Where work
gi
= 0 do not plan to work full-time, 1 plan to attend to work full time. RF
i
=
raising a family; WF
i
= being very well-off financially; SB
i
= becoming successful in a
business of my own; UG
i
= undergraduate GPA; PE
i
= preparedness for employment
after college; UA
i
= undergraduate loan amount; SS
i
= student’s sex; PE
i
= parents’ levels
of education; PI
i
= participated in internship program; MC
i
= met with an
advisor/counselor about career plans; FG
i
= faculty provided help in achieving
professional goals; FR
i
= faculty provided opportunity to work on a research project; FL
i
= faculty provided a letter of recommendation; IS
i
= institutional selectivity; and u
i
= a
stochastic error term.
Hence, the following is a logit model where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
RF
i
+ b
3
WF
i
+...+ b
7
UA
i
…+b
10
PI
i
+b
11
MC
i
+…b
15
IS
i
+ u
i

Limitations
As with any study, there are limitations to address. First: data from the 2003 Freshmen
Survey and 2007 College Senior Survey were self-reported. Similarly, the dependent variable
(“do not plan to work full-time” and “plan to work full-time”) only provides information about a
student’s “intended plans” to choose or not to choose to work full-time. It does not confirm
whether a student actually chose or did not choose to work full-time.
Second: REM students were treated as a homogeneous group. That is, because of the
sample sizes, it was not possible to conduct separate regression analyses—for example, a
regression analysis for African American students or for Latino students.
Third: in both samples, more than half of participants were women, and the samples did
76
not include students of a non-traditional age or students who had transferred from two-year
institutions. Finally, despite its advantages, multiple imputation produces different results every
time it is used (Allison, 2009). This is because the “imputed values are random draws rather than
deterministic quantities,” (Allison 2009, p. 81).
Findings
26

REM Undergraduate Students With S&E Majors
The results of the logistic regression for REM students with S&E majors indicated the
final block was statistically reliable in distinguishing between “do not plan work full-time” and
“plan to work full time.” The inferential goodness-of-fit test—the Hosmer–Lemeshow (H–L)—
yielded a X
2
(8) of 12.49 and was insignificant (p >.05), indicating the model fit the data well; the
null hypothesis of a good model fit to data was plausible. The Cox and Snell R
2
and the
Nagelkerke R
2
were .11 and .15, respectively (see Table 9).
Wald statistics of the variables included to represent human capital indicated “raising a
family” (X
2
(1) 4.06, p<.05) and “preparedness for employment after college” (X
2
(1) 6.06, p<.01)
significantly predicted “plan to work full-time.” Respectively, odds ratios demonstrated students
who reported “raising a family” was “very important/essential” were about 2 times more likely to
report they planned to work full-time. Similarly, students who reported their “preparedness for
employment” was “stronger/much stronger” than when they began college were 2 times more
likely to report they planned to work-full-time.
Concerning the habitus layer, Wald statistics specified parents’ levels of education (X
2
(1)
5.80, p<.01), “faculty provided an opportunity to work on a research project” (X
2
(1) 3.98,
p<.05), and “faculty provided a letter of recommendation” (X
2
(1) 16.61, p<.000), were

26
The findings reported are the pooled results across the five imputed data sets. For the non-
imputed results, see Table A-5 and Table A-6 in Appendix A.
77
Table 9

Results for “Plan to Work Full-Time” for REM Undergraduate Students with S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Raising a Family .502 .249 4.06 .044* 1.65 1.01 2.69

Being Very Well-Off Financially .113 .243 .216 .640 1.12 .696 1.80

Becoming Successful in a Business of my Own .301 .251 1.44 .230 1.35 .826 2.21

Undergraduate GPA .183 .241 .576 .448 1.20 .748 1.93

Preparedness for Employment After College .724 .294 6.06 .014** 2.06 1.16 3.67

Undergraduate Loan Amount .179 .133 1.81 .204 1.20 .896 1.59

Student’s Sex .031 .228 .018 .893 1.03 .659 1.61

Parents’ Levels of Education -.573 .238 5.80 .016* .564 .354 .899

Participated in Internship Program .222 .237 .877 .348 1.25 .784 1.99

Met with an Advisor/Counselor About Career Plans -.305 .296 1.06 .303 .737 .412 1.32

Faculty Provided Help in Achieving Professional
Goals
.209 .300 .485 .486 1.23 .684 2.22

Faculty Provided an Opportunity to Work on a
Research Project
.527 .264 3.98 .046* 1.69 1.01 2.84

Faculty Provided a Letter of Recommendation -1.08 .266 16.61 .000*** .338 .201 .570

Institutional Selectivity .002 .001 4.0 .125 1.00 .999 1.00

Constant -4.23 2.34 3.27 .094

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 12.49 8 .131

Omnibus Test of Model 53.23 14 .000***

-2 Log Likelihood 542.45 +

Cox & Snell R
2
.112

Nagelkerke R
2
.153

* p<.05; **p<.01; ***p<.000 + 595.68 with only constant in model
78
significant. The odds ratios showed students who reported faculty frequently provided
opportunities to work on research projects were about 2 times more likely to report they planned
to work full-time. Conversely, the odds ratios for parents’ level of education and “faculty
provided a letter of recommendation” were below 1, indicating the odds of being classified as
“plan to work full-time” decreased by the respective ratio. That is, students who reported one or
more of their parents had a college degree or higher, and students who reported that faculty
frequently provided a letter of recommendation, were respectively .5 times and .3 times less
likely to report they planned to work full time (see Table 9).
REM Undergraduate Students With Non-S&E Majors
The results of the logistic regression for REM students with non-S&E majors indicated
the model was statistically reliable in distinguishing between “do not plan to work full-time” and
“plan to work full-time.” The inferential goodness-of-fit test—the Hosmer–Lemeshow (H–L)—
yielded a X
2
(8) of 5.83 and was insignificant (p >.05), indicating the model fit the data well. The
Cox and Snell R
2
and the Nagelkerke R
2
were .08 and .11, respectively. However, Wald statistics
indicated only one variable was significant (see Table 10). This variable was “participated in an
internship program” (X
2
(1) 6.40, p<.01), which was included to represent the habitus layer (i.e.,
layer one). Odds ratios revealed students who reported they participated in an internship program
were about 2 times more likely to report they planned to work full-time (see Table 10).
Discussion
Although the results of the logistic regressions indicated both models were statistically
reliable in distinguishing between “do not plan to work full-time” and “plan to work full-time,”
the results indicated the model for REM students with S&E majors had more significant
variables. That is, the variables included to represent the human capital aspect, layer one, and
79
Table 10

Results for “Plan to Work Full-Time” for REM Undergraduate Students with Non-S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Raising a Family -.070 .272 .066 .797 .932 .546 1.59

Being Very Well-Off Financially .116 .236 .241 .622 1.12 .707 1.78

Becoming Successful in a Business of my Own .015 .233 4.14 .948 1.01 .643 1.60

Undergraduate GPA -.449 .242 3.44 .064 .638 .397 1.03

Preparedness for Employment After College -.002 .293 .046 .995 .998 .562 1.77

Undergraduate Loan Amount .171 .133 1.65 .205 1.19 .908 1.55

Student’s Sex .237 .233 1.03 .309 1.27 .803 2.00

Parents’ Levels of Education -.264 .222 1.41 .234 .768 .497 1.19

Participated in Internship Program .554 .219 6.40 .011** 1.74 1.13 2.67

Met with an Advisor/Counselor About Career Plans .280 .293 .913 .338 1.32 .745 2.35

Faculty Provided Help in Achieving Professional
Goals
-.265 .306 .750 .389 .767 .416 1.41

Faculty Provided Opportunity to Work on a Research
Project
-.389 .268 2.11 .147 .678 .400 1.15

Faculty Provided Letter of Recommendation -.471 .278 2.87 .093 .624 .360 1.08

Institutional Selectivity -.002 .001 4.00 .090 .998 .996 1.00

Constant -1.83 2.12 .745 .392

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.83 8 .666

Omnibus Test of Model 42.98 14 .000***

-2 Log Likelihood 555.51 +

Cox & Snell R
2
.080

Nagelkerke R
2
.116

* p<.05; **p<.01; ***p<.000 + 598.49 with only constant in model
80
layer three of Perna’s model were more pertinent for REM students with S&E majors than for
REM students with non-S&E majors. In regards to the model for REM students with non-S&E
majors, the significant variable was “participated in an internship program.” REM students with
non-S&E majors who participated in internship programs during their undergraduate studies
were 2 times more likely to report that, upon graduation, they planned to make the choice to
work full-time.
The variable “participated in an internship program” —among other variables—was
included to account for habitus (i.e., layer one). Specifically, “participated in an internship
program” was included as a proxy for social capital. I assumed that, by participating in an
internship, REM students with S&E majors would be exposed to social relations through which
they could gather information about their employment options. The findings, however, revealed
that “participating in an internship program” was only significant for REM students with non-
S&E majors. Hence, by participating in an internship program, REM students with non-S&E
majors were able to inform their choice of working full-time after college.
In regards to the model for REM students with non-S&E majors, the findings revealed
that five variables were significant. Two of these significant variables were “raising a family”
and “preparedness for employment after college,” which were included to account for human
capital. In the economic model of human capital, college choice is based on weighing the
expected benefits of a collegiate education against the expected costs (Becker, 1993; Perna,
2006). The variable “raising a family” was included to represent an expected benefit of making
the choice to work full-time. REM students with S&E majors who reported “raising a family”
was “very important/essential” were 2 times more likely to report they planned to make the
choice to work full-time. In this study, REM students with S&E majors associated the choice of
81
working-full time after college with the benefit of raising a family. Aligned with expected costs
and benefits are the demand for human capital and the supply of resources for investing in
human capital (Becker, 1993; Ellwood & Kane, 2000).
The variable “preparedness for employment after college” was included to represent the
demand for human capital. As indicated by human capital theory, differences in demand for a
postsecondary education reflect variations across individuals according to their respective
academic preparation and achievement (Paulsen, 2001). Within the context of this study, I
imagined differences in demand for working full-time would reflect variations across REM
students with S&E majors, according to whether the students perceived that their undergraduate
studies prepared them for full-time employment after college. Correspondingly, the findings
indicated REM students with S&E majors who, during their final year of undergraduates studies,
reported their preparedness for employment was “stronger/much stronger” than it had been when
they began college were 2 times more likely to report they planned to make the choice to work
full-time.
The remaining three significant variables were: “parents’ educational levels,” “faculty
provided an opportunity to work on a research project,” and “faculty provided a letter of
recommendation.” These variables—among others—were included to account for layer one (i.e.,
habitus) of Perna’s model. The “parents’ educational levels” variable was a proxy for cultural
capital. Given that cultural capital is primarily acquired through family (Bourdieu, 1979/1984), I
postulated that parents’ levels of education imbue an individual with relevant skills that inform
the individual’s full-time employment choice. The findings revealed REM students with S&E
majors whose parents’ had college degrees or higher were about .5 times less likely to plan to
make the choice to work full-time after college. It is possible that REM students with parents
82
who have a college degree or higher are exposed to information about their choices beyond full-
time employment. That is, parents with a college education are perhaps better equipped to
provide advice about the post-baccalaureate choices—besides full-time employment—associated
with an S&E degree.
The variables “faculty provided an opportunity to work on a research project” and
“faculty provided a letter of recommendation” were included as proxies for social capital. Given
that “information channels” (i.e., social capital) are the gathering and sharing of information
through social relations (Coleman, 1988, 1990) I assumed that by interacting with faculty
members, REM students with S&E majors would be exposed to social relations through which
they could gather information about full-time employment. REM students who reported faculty
frequently provided an opportunity to work on a research project were 2 times more likely to
report they planned to make the choice to work full-time after college. While prior research
indicates an undergraduate research experience serves as a pathway to graduate studies (Russell,
2006; Strayhorn, 2010), the aforementioned finding indicates that having an opportunity to work
on a research project also helps inform REM students’ full-time employment choice. Conversely,
REM students with S&E majors who reported faculty frequently provided a letter of
recommendation were .3 times less likely to report they planned to make the choice to work full-
time. It could be that the letter of recommendation from a faculty member provided an
alternative post-baccalaureate choice.
83
CHAPTER SIX: STUDY THREE
Research continues to provide insight on what repels, attracts, and retains REM women
in S&E baccalaureate programs (Cole & Espinoza, 2009; Huang, Taddese, & Walter, 2000;
Zhao, Carini & Kuh 2005). However, this research does not always consider REM women’s
post-baccalaureate choices and what factors contribute to those choices. Current figures illustrate
that REM women are earning slightly more S&E advanced degrees than are their REM male
counterparts (NSF & NCSES, 2013). As shown in Table 3, in 2001, women earned about 52%
(i.e., 9,311) and men earned about 48% (i.e., 8,508), of the 17,819 S&E master’s degrees earned
by REM individuals (NSF & NCSES, 2013). Table 3 also illustrates that, in 2010, women earned
about 54.7% (i.e., 15,449) and men earned about 45.3% (i.e., 12,810), of the 28,259 S&E
master’s degrees earned by REM individuals (NSF & NCSES, 2013).
At the doctoral level, there has been a shift towards women earning marginally more
degrees than men. Table 4 shows that, in 2001, REM men earned about 52% (i.e., 1,644) of the
3,197 S&E doctoral degrees earned by REM individuals—and by 2010, REM women earned
about 53.5% (i.e., 2,298) of the 4,229 S&E doctoral degrees earned by REM individuals (NSF &
NCSES, 2013). In respect to employment, in general, women’s participation in S&E occupations
is lower than it is in the U.S. workforce as a whole (NSF & NCSES, 2013). REM women,
specifically, comprise about 1 in 10 employed scientists and engineers (NSF & NCSES, 2013).
While the aforementioned data is informative, they do not provide insight on what factors
contribute to the post-baccalaureate choices of REM undergraduate women with S&E majors.
In this chapter, I present the third study, which focuses on the “graduate education
choice” and the “full-time employment choice” of REM undergraduate women with S&E
majors. I also examine a comparison group of REM undergraduate women with non-S&E
84
majors. Accordingly, the research questions I address in this study are:
I. What factors inform the “graduate education choice” of REM undergraduate
women with S&E majors?
II. What factors inform the “graduate education choice” of REM undergraduate
women with non-S&E majors?
III. What factors inform the “full-time employment choice” of REM undergraduate
women with S&E majors?
IV. What factors inform the “full-time employment choice” of REM undergraduate
women with non-S&E majors?
In the section that follows, I provide an overview of the conceptual model that guides this
study. Thereafter, I present the methods and the findings.
Conceptual Model and Application
Perna’s (2006) conceptual model of student college choice serves as the conceptual
grounding for this study. While her model is intended to explore the college choices of students
making the transition from high school to college, Perna notes it can also be used to examine
parallel processes, such as the transition from undergraduate studies to graduate studies.
Moreover, I propose the Perna model can be extended to examine “employment choice.” As I
address the “graduate education choice” and the “full-time employment choice” of REM women
in this study, I apply Perna’s model to each choice—that is, I include variables both to account
for the “graduate education choice” and to account for the “full-time employment choice.” In the
subsections that follow, I describe how I conceptualized the model for each “post-baccalaureate
choice.”

85
Center of the Model: Human Capital
At the center of Perna’s model of student college choice is the economic model of human
capital investment (see Figure 1). The human capital model proposes college choice is based on
weighing the expected benefits of a collegiate education against its expected costs (Becker, 1993;
Perna, 2006). The expected costs of investing in a college education include direct costs (such as
tuition and fees for housing and books) and indirect costs of foregone earnings and leisure time
(Becker, 1993). The expected benefits—long-term monetary and non-monetary—associated with
investing in a postsecondary education include: better health, a longer life, higher lifetime
earnings, informed purchases, and lower probabilities of unemployment (Leslie & Brinkman,
1988). The human capital model proposes that, if the individual determines the benefits outweigh
the expected costs, the individual will make the choice to attend college (Becker, 1962, 1993;
Douglass, 1997).
Aligned with expected costs and benefits are the demand for human capital and the
supply of resources for investment in human capital. That is, the human capital model recognizes
differences in college choice are connected to the forces that shape the demand for human capital
and the supply of resources to invest in human capital (Becker, 1993; Ellwood & Kane, 2000). In
the human capital model, differences in the demand for a post-secondary education are expected
to reflect variations across individuals, according to the individuals’ respective academic
preparation and achievement (Paulsen, 2001). Differences in the supply of resources available to
pay for the costs of a college education are expected to reflect differences in the accessibility of
parental income and financial aid (Ellwood & Kane, 2000; Paulsen, 2001). Hence, individuals
who are more likely to choose to attend college are those who are academically prepared and
who have greater personal financial resources (Paulsen, 2001).
86
“Graduate education choice.” Concerning the “graduate education choice,” to account
for the human capital aspect of Perna’s model of student college choice, six variables are
included as proxies for expected benefits, demand, and supply of resources (see Figure 7). Three
variables are included to represent expected benefits. These variables, considered non-monetary
benefits of choosing to attend graduate school, are: “helping others who are in difficulty,”
“obtaining recognition from my colleagues for contributions to my special field,” and “becoming
successful in a business of my own.”
For demand, the variables included are undergraduate GPA and “preparedness for
graduate/advanced education.” Given that human capital theory proposes individuals who are
more likely to choose to attend college are those who are academically prepared (Paulsen, 2001),
for this study I assume REM women with competitive undergraduate GPAs—and who believe
that their undergraduate studies prepared them for a graduate education—will be more likely to
choose to attend graduate/professional school.
Finally, concerning supply of resources, the variable considered is “undergraduate
student loan amount.” I propose having a low debt amount (or no debt at all) is considered a
supply of resources. As noted by Malcom and Dowd (2011), if the goal is for STEM bachelor’s
degree recipients to enroll in graduate school immediately after bachelor’s degree completion,
the ideal amount of undergraduate loan debt is no debt at all (Malcom & Dowd, 2009).
“Full-time employment choice.” Concerning “full-time employment choice,” to account
for the human capital aspect of Perna’s model, six variables are included as proxies for expected
benefits, demand, and supply of resources (see Figure 8). Three variables are included to
represent expected benefits, two of which represent non-monetary benefits. These non-monetary
expected benefit variables are: “helping others who are in difficulty” and “raising a family.”
87
Figure 7. Application of Perna’s Conceptual Model of Student College Choice for “Plan to
Attend Graduate/Professional School” for REM Undergraduate Women with S&E and Non-S&E
Majors

Habitus (layer 1)

Cultural Capital
 Parents’ Levels of Education

Social Capital
 Participated in an Internship
 College Social Capital

Demand for Higher
Education

 Undergraduate GPA
 Preparedness for
Graduate/Advanced
Education

Supply of Resources
 Undergraduate Loan
Amount
Expected Benefits

 Helping Others
Who are in
Difficulty
 Obtaining
Recognition From
My Colleagues
For Contributions
to My Special
Field
 Becoming
Successful in a
Business of my
Own

Graduate
Education
Choice
Higher Education Context (layer 3)

 Institutional Selectivity

88
Figure 8. Application for Perna’s Conceptual Model of Student College Choice for “Plan to
Work Full-Time” for REM Undergraduate Women with S&E and Non-S&E Majors

Habitus (layer 1)

Cultural Capital
 Parents’ Levels of Education

Social Capital
 Faculty Provided Advice and Guidance
About Educational Program
 Faculty Provided a Letter of
Recommendation
 Faculty Provided an Opportunity to
Work on a Research Project

Demand for Higher
Education

 Undergraduate GPA
 Preparedness for
Employment after
College

Supply of Resources
 Undergraduate Loan
Amount
Expected Benefits

 Helping Others
Who Are in
Difficulty
 Raising a Family
 Being Very Well-
off Financially

Full-Time
Employment
Choice
Higher Education Context (layer 3)

 Institutional Selectivity

89
The third variable, “being very well-off financially,” represents a monetary benefit.
For demand, the variables include undergraduate GPA and “preparedness for
employment after college.” Given that human capital theory proposes individuals who are more
likely to make the choice to attend college are those who are academically prepared (Paulsen,
2001), for this study I assume REM women with GPAs of “B” or lower will be more likely to
choose to work full-time. That is, women with GPAs that are lower than average may perceive
working full-time as the realistic choice. I also assume REM women who believe their
undergraduate studies prepared them for employment after college will be more likely to make
the choice to work full-time.
Finally, concerning supply of resources, the variable considered is “undergraduate loan
amount.” I propose having a low debt amount (or no debt at all) is considered a supply of
resources. I hypothesize, however, that REM women with increasing loan amounts will be more
likely to make the choice to work full-time. Students with loans are conceivably more prone to
choose to work full-time in order to begin the process of repayment.
Four Contextual Layers
Moving beyond the human capital model perspective, Perna’s model specifies an
individual’s calculations of the expected costs and benefits are nested within four contextual
layers that shape an individual’s choice. These layers
27
are: 1) the individual’s habitus, 2) the
school and community contexts, 3) the higher education context, and 4) the broader social,
economic, and policy contexts. For this study, I consider the first and third layers (see Figures 7
& 8).

27
For an overview of the layers, see Chapter 3.
90
Habitus. The first layer, an individual’s habitus, reflects demographic characteristics,
including sex, race and ethnicity, and socioeconomic status, as well as his or her cultural capital
and social capital (Perna, 2006). Cultural capital is related to the class-based socialization of
culturally relevant skills, abilities, preferences, tastes, or norms that act as a form of currency in
the social realm
28
(Bourdieu, 1979/1984). As noted by Bourdieu (1979/1984), cultural capital is
primarily acquired through an individual’s family. Social capital exists in the relations between
individuals and—similar to physical and human capital—can facilitate productivity
29
(Coleman,
1988). According to Coleman (1988, 1990) one form of social capital is “information channels,”
a term that refers to the gathering and sharing of information through social relations (Coleman,
1988, 1990).
“Graduate education choice.” Concerning the “graduate education choice,” to account
for the habitus layer, three variables are considered (see Figure 7). The first variable, parents’
levels of education, is used as a proxy for cultural capital. Given that cultural capital is primarily
acquired through family (Bourdieu, 1979/1984), I hypothesize that parents’ levels of education
bestow relevant skills that inform an individual’s graduate education choice. As noted by
Ethington and Smart (1986), when an undergraduate student is determining whether he or she
will pursue graduate studies, the influence of parental background does not disappear.
The remaining two variables are “participated in an internship program” and a composite
variable, “college social capital.” These variables are proxies for social capital. In particular, the
composite “college social capital” represents interactions with faculty members (see Figure 7).
Given that “information channels” (i.e., social capital) are the gathering and sharing of
information through social relations (Coleman, 1988, 1990) I assume that, by participating in an

28
For an overview of cultural capital, see Chapter 3.
29
For an overview of social capital, see Chapter 3.
91
internship and by interacting with faculty members, REM women are exposed to social relations
through which they can gather information about graduate school. Research performed by Sax
(2001) supports this hypothesis, noting that spending time with faculty members (either by
working on a professor’s research project or serving as a teaching assistant) creates direct
opportunities for students with S&E majors to garner realistic notions of “life as academic
scientists”—notions which, consequently, encourage students to attend graduate school.
“Full-time employment choice.” Concerning the “full-time employment choice,” to
account for the habitus layer, four variables are considered (see Figure 8). The first, “parents’
level of education,” is used as a proxy for cultural capital. Given that cultural capital is primarily
acquired through family (Bourdieu, 1979/1984), I postulate that parents’ levels of education
bestow relevant skills that inform an individual’s full-time employment choice. The remaining
three variables are proxies for social capital: “faculty provided advice and guidance about
educational program,” “faculty provided a letter of recommendation,” and “faculty provided an
opportunity to work on a research project.” Given that “information channels” (i.e., social
capital) are the gathering and sharing of information through social relations (Coleman, 1988,
1990) I assume that, by interacting with faculty members, REM women with S&E majors are
exposed to social relations through which they can gather information about their employment
options. As indicated by Lewis and Collins (2001), students need exposure to a broader
understanding of the work done by science professionals. It is reasonable to infer that, through
the aforementioned social relations, REM women may gather a broader understanding of such
work, which can, in turn, inform their employment choices.
It is important to note that, because Coleman’s theory of social capital limits the notion of
92
agency
30
, I adopt the notion that “individuals have a degree of agency to shape their realities”
(Tierney & Venegas, 2006, p. 1690). That is, I adopt a human agency-oriented perspective
towards social capital. From this perspective, I think it is possible for REM women to garner
their own social capital.
Higher education context. The third layer, the higher education context, represents the
role post-secondary institutions play in shaping college choice. For the “graduate education
choice” and the “full-time employment choice,” institutional type and control, as well as
institutional selectivity, are considered. In regards to institutional type and control, the data are
filtered to consider only four-year, private, predominately White institutions. In regards to
institutional selectivity, a scale variable
31
of median SAT scores and/or ACT composite scores is
included to represent institutional selectivity (see Figures 7 & 8).
As indicated in the literature, graduates of selective research institutions are more likely
to continue their education after receiving bachelor’s degrees (Mullen et al., 2003). For REM
students, specifically, the research indicates REM students who attend selective, as opposed to
non-selective, institutions are more likely to complete bachelor’s degrees (Alon & Tienda, 2005;
Melguizo, 2008). Bachelor’s degree completion is, of course, important because REM students
must complete their undergraduate studies before considering the choice between graduate
school and full-time employment.

30
Coleman’s work supports the notion that it is the responsibility of the family to adopt certain
norms to advance the child’s social capital. Such a view ignores agency—that is, the possibility
of the child becoming an adolescent capable of accessing his or her own social capital (Dika &
Singh, 2002).
31
This variable is the median SAT scores and/or ACT composite scores of the entering class as
reported to the Integrated Postsecondary Educational Data System (IPEDS). The median SAT is
based on a combination of verbal and math scores (i.e., verbal + math). See Figure B-3 in
Appendix B.

93
Moreover, the intersection between institutional selectivity and control is important. Eide,
Brewer, and Ehrenberg (1998) note that attending a selective private college not only increases
the probability of attending graduate school, it also increases the likelihood of attending graduate
school at a major research institution. Similarly, the benefits associated with attending a selective
institution extend to individuals’ employment and economic returns (Bowen & Bok, 1998;
Brewer et al., 1999).
In sum, in this study, I utilize Perna’s integrated model of student college choice to
examine the factors that inform the “graduate education choice” and the “full-time employment
choice” of REM undergraduate women with S&E majors. Specifically, I focus on the human
capital aspect, the habitus layer, and the higher education context layer (see Figures 7 & 8). The
section that follows provides an overview of the methods.
Methods
32

Given that in this study I address the “graduate education choice” and the “full-time
employment choice” of REM women, I conducted four analyses (two analyses per “choice”).
Concerning the “graduate education choice,” one analysis focused on REM women with S&E
majors and another on REM women with non-S&E majors. Similarly, concerning the “full-time
employment choice,” one analysis focused on REM women with S&E majors and another on
REM women with non-S&E majors. In the subsections that follow, I describe the samples and
analyses.
Samples
Two samples were utilized for this study: a sample of REM women who reported S&E
majors and a comparison group sample of REM women who reported non-S&E majors. Across

32
For an overview of the data, see p. 6
94
both samples, all students had attended private, predominately White, four-year institutions and,
by June of 2003 had earned bachelor’s degrees. A total of 292 female students across the
samples reported S&E majors. Of these students, 1.7% identified as American Indian, 35.3% as
Asian, 17.8% as Black, 18.5% as Latino, and 26.7% as multiracial/multiethnic
33
. As displayed in
Table 11, the majority of female students with S&E majors reported undergraduate GPAs of
“B+” (65.6%), parents with college degrees or higher (65.1%), and estimated parental incomes
of less than $60,000 (49.1%). The top three S&E majors
34
reported were: psychology (24.0%),
political science (14.0%), and biology (12.0%).
A total of 367 female students reported non-S&E majors. Of these students, 1.1%
identified as American Indian, 28.3% as Asian, 19.3% as Black, 21.0% as Latino, and 30.2% as
multiracial/multiethnic. The majority of female students with non-S&E majors reported under-
graduate GPAs of “B+” (78.2%), parents with college degrees or higher (58.6%), and an
estimated parental income of less than $60,000 (52.6%). As illustrated in Table 12, the top three
non-S&E majors
35
reported were: English (12.0%), art, fine, and applied (9.0%), and other arts
and humanities (8.7%).
Variables
36

The analyses for “graduate education choice” and the analyses for “full-time employment
choice” included variables to account for Perna’s model of student college choice. Specifically,
variables were included for human capital, layer one, and layer three.

33
The student marked “two or more races/ethnicities” as her race/ethnicity.
34
For the full list of majors, see Table A-7 in Appendix A.
35
For the full list of majors, see Table A-8 in Appendix A.
36
With the exception of variables concerning student’s sex, parents’ levels of education, and
institutional selectivity (all of which were from the 2003 Freshmen Survey), all variables were
from the 2007 College Senior Survey.
95
Table 11

Overview of REM Undergraduate Women with S&E Majors (N=292)

Characteristic

Percent

Race/Ethnicity

American Indian

1.7

Asian 35.3

Black

17.8

Latino

18.5

Two or More Races/Ethnicities

26.7

Parents’ Education
Some College or Less

34.9
College Degree or Higher

65.1

Parents’ Income

$59,999 or less

49.1

$60,000 to $999,999

22.5

$100,000 or more

28.4

Undergraduate GPA
B or Lower

34.4

B+ or Higher 65.6

S&E Majors (Top 3)*

Psychology 24.0

Political Science

14.0

Biology (General)

12.0

*For the full list of majors, see Table A-7 in Appendix A.

96
Table 12

Overview of REM Undergraduate Women with Non-S&E Majors (N=367)

Characteristic

Percent

Race/Ethnicity
American Indian 1.1
Asian 28.3
Black 19.3
Latino 21.0
Two or More Races/Ethnicities 30.2
Parents’ Education
Some College or Less 41.4
College Degree or Higher 58.6
Parents’ Income
$59,999 or less 52.6
$60,000 to $999,999 22.9
$100,000 or more 24.5
Undergraduate GPA
B or Lower 21.8
B+ or Higher 78.2
Non-S&E Majors (Top 3)*
English 12.0
Art, Fine, & Applied 9.0
Other Arts & Humanities 8.7

*For the full list of majors, see Table A-8 in Appendix A.

97
Graduate education choice. To represent human capital, variables were included to
account for expected benefits, demand for higher education, and supply of resources (see Figure
9). Concerning expected benefits, the variables included were: “helping others who are in
difficulty,” “obtaining recognition from my colleagues for contributions to my special field,” and
“becoming successful in a business of my own.” The variables included concerning demand for
higher education were undergraduate GPA and preparedness for graduate/advanced education.
Undergraduate loan amount was included to represent supply of resources, Additionally,
five variables—accounting for cultural capital and social capital—were included to represent
habitus (i.e., layer one). These variables were: parents’ levels of education, participation in an
internship program, and “college social capital,” a composite variable (see Figure 9).
Additionally, institutional selectivity was included to represent the higher education context (i.e.,
layer three). The dependent variable was binary (“do not plan to attend graduate/professional
school” and “plan to attend graduate/professional school”) and was from the 2007 College
Senior Survey (see Figure 9).
Full-time employment choice. To represent human capital, variables were included to
account for expected benefits, demand for higher education, and supply of resources (see Figure
10). Concerning expected benefits, the variables included were “helping others who are in
difficulty,” “raising a family,” and “being very well-off financially.” The variables included
concerning demand for higher education were undergraduate GPA and “preparedness for
employment after college.” To represent supply of resources, undergraduate loan amount was
included.
Four variables—accounting for cultural capital and social capital—were included to
represent habitus (i.e., layer one). These variables were parents’ levels of education, “faculty
98
Figure 9. Variables and Coding for “Plan to Attend Graduate/Professional School” for REM
Undergraduate Women with S&E Non-S&E Majors

Variable

Coding

Human Capital—Expected Benefits

Helping Others Who are in Difficulty 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Obtaining Recognition From My Colleagues
For Contributions to My Special Field

0 = Not Important, Somewhat Important
1 = Very Important, Essential
Becoming Successful in a Business of My Own

0 = Not Important, Somewhat Important
1 = Very Important, Essential
Human Capital—Demand for Higher Education
Undergraduate GPA 0 = D, C, C+, B-, B
1 = B+, A-, A or A+

Preparedness for Graduate/Advanced Education 0 = Much Weaker, Weaker, No Change
1 = Stronger, Much Stronger

Human Capital—Supply of Resources
Undergraduate Loan Amount Scale Variable — Less than $10,000 to
$2000,000 or More

Habitus (Layer 1)—Cultural Capital

Parents’ Levels of Education 0 = Grammar School or Less, Some High
School, High School Graduate, Some
College
1 = College Degree, Some Graduate School,
Graduate Degree

Habitus (Layer 1)—Social Capital
Participated in Internship Program 0 = No 1 = Yes
College Social Capital (  =.735 &  =.704)
 Faculty Provided Emotional Support &
Encouragement
 Faculty Provided Advice & Guidance about
Educational Program
 Faculty Provided Opportunity to Work on a
Research Project
 Faculty Provided Letter of
Recommendation
0 = Not At All, Occasionally
1 = Frequently
Higher Education Context (Layer 3)
Institutional Selectivity

Scale Variable — 800 to 1600

Dependent

Plan to Attend Graduate/Professional School 0 = No 1 = Yes

99
Figure 10. Variables and Coding for “Plan to Work Full-Time” for REM Undergraduate Women
with S&E Majors and Non-S&E Majors

Variable

Coding

Human Capital—Expected Benefits
Helping Others Who Are in Difficulty 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Raising a Family 0 = Not Important, Somewhat Important
1 = Very Important, Essential

Being Very Well-Off Financially 0 = Not Important, Somewhat Important
1 = Very Important, Essential
Human Capital—Demand for Higher Education
Undergraduate GPA 0 = D, C, C+, B-, B
1 = B+, A-, A or A+

Preparedness for Employment After College
0 = Much Weaker, Weaker, No Change
1 = Stronger, Much Stronger
Human Capital—Supply of Resources
Undergraduate Loan Amount Scale Variable — Less than $10,000 to
$2000,000 or More
Habitus (Layer 1)—Cultural Capital
Parents’ Levels of Education 0 = Grammar School or Less, Some High
School, High School Graduate, Some
College
1 = College Degree, Some Graduate School,
Graduate Degree
Habitus (Layer 1)—Social Capital
Faculty Provided Advice and Guidance About
Educational Program

0 = Not At All, Occasionally
1 = Frequently

Faculty Provided a Letter of Recommendation

0 = Not At All, Occasionally
1 = Frequently
Faculty Provided Opportunity to Work on a
Research Project

0 = Not At All, Occasionally
1 = Frequently
Higher Education Context (Layer 3)

Institutional Selectivity

Scale Variable — 800 to 1600
Dependent

Plan to Work Full-Time

0 = No 1 = Yes

100
provided advice and guidance about educational programs,” “faculty provided a letter of
recommendation,” and “faculty provided opportunity to work on a research project.” Institutional
selectivity was included to represent the higher education context (i.e., layer three). The
dependent variable was binary (“do not plan to work full-time” and “plan to work full-time”) and
was from the 2007 College Senior Survey (see Figure 10).
Analyses
Statistical analyses consisted of descriptive statistics, exploratory factor analysis, missing
value analysis and multiple imputation, screening for multicollinearity, logistic regression, and
testing for linearity of the logit. Exploratory factor analysis was only conducted for the analyses
that focused on “graduate education choice.” Descriptive statistics were used to garner
summaries of the samples (see Tables 11 & 12).
Exploratory factor analysis. The variables considered for the composite variable
“college social capital”
37
were subjected to principle component analysis (PCA). Prior to PCA,
suitability of data for factor analysis was assessed. Across both samples, inspection of the
correlation matrixes revealed the presence of coefficients of .3 and above. The Kaiser-Meyer-
Olkin values were .753 for the sample of REM women with S&E majors and .729 for the sample
of REM women with non-S&E majors. These values met the recommended value of .6 (Kaiser,
1974), and the Barlett’s Tests of Sphericity (Bartlett, 1954) reached statistical significance for
both samples, supporting the factorability of correlation matrixes. PCA for the sample of REM
women with S&E majors revealed the presence of only one component with eigenvalues
exceeding 1 and explaining 55.8% of the variance. Because only one component was extracted,
rotation was not plausible. The component matrix revealed that the variables faculty provided

37
This composite variable was only considered for the analyses that focused on “graduate
education choice.”
101
emotional support and encouragement (.737), faculty provided advice and guidance about
educational programs (.809), faculty provided opportunity to work on a research project (.763),
and faculty provided letter of recommendation (.671) loaded strongly. Reliability statistics for the
component showed a Cronbach’s Alpha on standardized items of .735 (see Figure 11). PCA for
the sample of REM women with non-S&E majors also revealed the presence of only one
component with eigenvalues exceeding 1 and explaining 53.1% of the variance. The component
matrix revealed that, as before, the variables faculty provided emotional support and
encouragement (.760), faculty provided advice and guidance about educational program (.804),
faculty provided opportunity to work on research project, (.650) and faculty provided letter of
recommendation (.692) loaded strongly. Reliability statistics for the component showed a
Cronbach’s Alpha on standardized items of .704 (see Figure 11).

Figure 11. Reliability and Constituent Variables for “College Social Capital”

Reliability

Constituent Variables

 = .735 REM undergraduate women with
S&E majors

 = .704 REM undergraduate women with
non-S&E majors

 Faculty Provided Emotional Support &
Encouragement
 Faculty Provided Advice and Guidance
about Educational Program
 Faculty Provided Opportunity to Work on
a Research Project
 Faculty Provided a Letter of
Recommendation

Missing value analysis and multiple imputation. For both samples, missing value
analyses were conducted to verify the extent of missing data. These analyses revealed random
missing data patterns, a discovery that supported the implementation of multiple imputation
(Allison, 2009). As such, multiple imputation was used to compensate for the missing values.
Multiple imputation was selected because—unlike mean value replacement—it provides an
102
accurate estimate of missing data and, like maximum likelihood, multiple imputation estimates
are consistent and asymptotically normal (Allison, 2009). Moreover, multiple imputation can be
applied to any type of data or model, an advantage not offered by maximum likelihood (Allison,
2009). Using IBM SPSS 20, the fully conditional specification approach of multiple imputation
was used to determine missing values. However, missing values for the variable parents’
educational levels were not imputed and were deleted prior to imputation. As recommended,
during the imputation process, five copies of the data were created, each with missing values
imputed; five data sets are enough to yield parameter estimates that are close to being fully
efficient (Allison, 2009). The final results showed the original data set and the pooled results
across the five imputed data sets.
Multicollinearity. To screen for multicollinearity, two linear regressions per outcome
were run. Menard (1995) suggests tolerance values less than 0.1 indicate a collinearity problem.
Similarly, Myers (1990) suggests a VIF value greater than 10 is cause for concern. The
collinearity diagnostics for all linear regressions revealed no major collinearity between the
independent variables. All linear regressions showed tolerance values greater than 0.1 and VIF
values less than 10. Hence, there were no major multicollinearity issues.
Logistic regression. Logistic regression was selected because it allows the prediction of
a discrete outcome—for this study, “do not plan to attend graduate/professional school or plan to
attend graduate/professional school” and “do not plan to work full-time or plan to work full-
time”—and predicts the category of outcome for individual cases using the most parsimonious
model (Field, 2009; Tabachnick & Fidell, 2007). Because two continuous variables (i.e.,
undergraduate loan amount and institutional selectivity, as shown in Figure 9 and Figure 10)
were included, it was necessary to check that each continuous variable was linearly related to the
103
log of each outcome variable (i.e., “plan to attend graduate/professional school” and “plan to
work full-time”). To test for linearity, logistic regressions were run with the interactions between
each continuous variable and the log of that variable (e.g., Selectivity x LnSelectivity, where Ln
is the natural log transformation). The results of all regressions indicated the interaction terms
were not significant and the assumption of linearity of the logit was met for the two continuous
variables.
Four binary logistic regressions were conducted—two regressions for the outcome “plan
to attend graduate/professional school” and two regressions for the outcome “plan to work full-
time.” For the outcome “plan to attend graduate/profession school” one regression was for REM
women with S&E majors and another regression was for REM women with non-S&E majors.
Similarly, for the outcome “plan to work full-time,” one regression was for REM women with
S&E majors and another regression was for REM women with non-S&E majors. Variables were
entered simultaneously for all four regressions.
For the logistic regressions of the outcome “plan to attend graduate/professional school,”
the following functional form expressed the relationship between the dependent and the
independent variables:
REM undergraduate women with S&E majors and REM undergraduate women with non-S&E
majors grad
gi
= f ([HD
i
OR
i
SB
i
UG
i
PG
i
UA
i
PE
i
PI
i
CS
i
IS
i
], u
i
)
Where students’ grad
gi
= 0 do not plan to attend graduate/professional school, 1 plan to
attend graduate/professional school. HD
i
= helping others who are in difficulty; OR
i
=
obtaining recognition from my colleagues for contributions to my special field; SB
i
=
becoming successful in a business of my own; UG
i
= undergraduate GPA; PG
i
=
preparedness for graduate or advanced education; UA
i
= undergraduate loan amount; PE
i

104
= parents’ level of education; PI
i
= participated in internship program; CS
i
= college
social capital; IS
i
= institutional selectivity; and u
i
= a stochastic error term.
Hence, the following is a logit model where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
HD
i
+ b
3
OR
i
+...+ b
7
UA
i
… +b
10
CS
i
+b
11
IS
i
+ u
i

For the logistic regressions of “plan to work full-time,” the following functional form
expressed the relationship between the dependent and the independent variables:
REM undergraduate women with S&E majors and REM undergraduate women with non-S&E
majors work
gi
= f ([HD
i
RF
i
BF
i
UG
i
EC
i
UA
i
PE
i
FG
i
FL
i
FR
i
IS
i
], u
i
)
Where students’ work
gi
= 0 do not plan to work full-time, 1 plan to work full-time. HD
i
=
helping others who are in difficulty; RF
i
= raising a family; BF
i
= being very well-off
financially; UG
i
= undergraduate GPA; EC
i
= preparedness for employment after college;
UA
i
= undergraduate loan amount; PE
i
= parents’ level of education; FG
i
= faculty
provided advice and guidance about educational program; FL
i
= faculty provided letter of
recommendation; FR
i
= faculty provided opportunity to work on a research project; IS
i
=
institutional selectivity; and u
i
= a stochastic error term.
Hence, the following is a logit model where L represents the log of the odds ratio:
L
i
= ln (P
i
/ 1 –P
i
) = b
i
+ b
2
HD
i
+ b
3
RF
i
+...+ b
7
UA
i
…+b
11
FR
i
+ b
12
IS
i
+ u
i

Limitations
As with any study, there are limitations to address. First: data from the 2003 Freshmen
Survey and 2007 College Senior Survey were self-reported. Similarly, the dependent variables
(“plan to attend graduate school” and “plan to work full-time”) only provide information about a
student’s “intended plans” to choose or not to choose graduate school or to choose or not to
choose to work full-time. They do not confirm whether a student actually chose or did not
105
choose graduate school or whether a student chose or did not choose to work full-time.
Second: REM women were treated as a homogeneous group. That is, because of the
sample sizes, it was not possible to conduct separate regression analyses—for example, a
regression analysis for African American women or for Latina women.
Third: the samples did not include women of non-traditional ages or women who had
transferred from two-year institutions. Finally, despite its advantages, multiple imputation
produces different results every time it is used (Allison, 2009). This is because the “imputed
values are random draws rather than deterministic quantities,” (Allison 2009, p. 81).
Findings
“Plan to Attend Graduate/Professional School”
38

REM undergraduate women with S&E majors. The results of the logistic regression
for REM women with S&E majors indicated the final block was statistically reliable in
distinguishing between “do not plan to attend graduate/professional school” and “plan to attend
graduate/professional school.” The inferential goodness-of-fit test—the Hosmer–Lemeshow (H–
L)—yielded a X
2
(8) of 5.99 and was insignificant (p >.05), indicating the model fit the data well;
the null hypothesis of a good model fit to data was plausible. The Cox and Snell R
2
and the
Nagelkerke R
2
were .11 and .16, respectively (see Table 13).
Wald statistics of the variables included to represent human capital indicated “helping
others who are in difficulty” (X
2
(1) 10.02, p<.01) and “obtaining recognition from my colleagues
for contributions to my special field” (X
2
(1) 3.71, p<.05) significantly predicted “plan to attend
graduate/professional school.” Respectively, odds ratios demonstrated women who reported
“obtaining recognition from my colleagues for contributions to my special field” was “very

38
The findings reported are the pooled results across the five imputed data sets. For the non-
imputed results, see Table A-9 and Table A-10 in Appendix A.
106
important/essential” were about 2 times more likely to report they planned to attend
graduate/professional school. Conversely, the odds ratio for “helping others who are in
difficulty” was below 1, hence, women who reported “helping others who are in difficulty” was
“very important/essential” were about .3 times less likely to report they planned to attend
graduate/professional school.
For the habitus component (i.e., layer one), Wald statistics specified parents’ levels of
education (X
2
(1) 3.79, p<.05), and “college social capital” (X
2
(1) 5.32, p<.05), were significant.
The odds ratios showed REM women who reported frequent college social capital interactions
with faculty were 2 times more likely to report they planned to attend graduate/professional
school. However, women who reported parents with a college degree or higher were about .6
times less likely to report they planned to attend graduate/professional school (the odds ratio was
below 1.)
Finally, for the higher education context (i.e., layer three), the Wald statistic for
institutional selectivity (X
2
(1) 9.0, p<.01) was significant. The odds ratio for institutional
selectivity (.997) indicated that, as institutional selectivity decreased, the odds of planning to
attend graduate/professional school decreased (see Table 13).
REM undergraduate women with Non-S&E majors. The results of the logistic
regression for REM women with non-S&E majors indicated the model was statistically reliable
in distinguishing between “do not plan to attend graduate/professional school” and “plan to
attend graduate/professional school.” The inferential goodness-of-fit test—the Hosmer–
Lemeshow (H–L)—yielded a X
2
(8) of 5.45 and was insignificant (p >.05), indicating the model
fit the data well. The Cox and Snell R
2
and the Nagelkerke R
2
were .05 and .08, respectively (see
Table 14). However, Wald statistics indicated only one variable was significant. This variable
107
Table 13

Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Women with S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who are in Difficulty -1.32 .417 10.02 .002** .267 .118 .605

Obtaining Recognition From My Colleagues For
Contributions to My Special Field
.578 .300 3.71 .054* 1.78 .989 3.21

Becoming Successful in a Business of My Own -.245 .284 8.12 .389 .783 .449 1.37

Undergraduate GPA .155 .310 0.25 .617 1.17 .636 2.14

Preparedness for Graduate/Advanced Education .906 .543 2.78 .096 2.47 .852 7.18

Undergraduate Loan Amount -.102 .135 .571 .452 .903 .690 1.18

Parents’ Levels of Education -.576 .296 3.79 .051* .562 .315 1.00

Participated in Internship Program .134 .303 0.19 .658 1.14 .631 2.07

College Social Capital .766 .332 5.32 .021* 2.15 1.12 4.12

Institutional Selectivity -.003 .001 9.00 .014** .997 .995 .999

Constant 3.56 2.00 3.17 .077

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.99 8 .648

Omnibus Test of Model 35.14 10 .000***

-2 Log Likelihood 336.17 +

Cox & Snell R
2
.113

Nagelkerke R
2
.158

* p<.05; **p<.01; ***p<.000 + 371.36 with only constant in model

108
Table 14

Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Women with Non-S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who are in Difficulty .183 .393 .217 .641 1.20 .556 2.59

Obtaining Recognition From My Colleagues For
Contributions to My Special Field
-.164 .285 .331 .564 .848 .485 1.48

Becoming Successful in a Business of My Own -.002 .292 .047 .994 .998 .563 1.77

Undergraduate GPA .400 .311 1.65 .198 1.49 .811 2.74

Preparedness for Graduate/Advanced Education 1.46 .622 5.51 .019* 4.32 1.28 14.63

Undergraduate Loan Amount -.128 .199 .414 .527 .880 .581 1.33

Parents’ Levels of Education -.188 .292 .414 .520 .829 .467 1.47

Participated in Internship Program -.199 .282 .498 .480 .819 .471 1.42

College Social Capital .632 .337 3.52 .061 1.88 .971 3.65

Institutional Selectivity .000 .002 0 .812 1.00 .996 1.00

Constant -1.60 3.47 .213 .651

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.45 8 .708

Omnibus Test of Model 17.97 10 .05*

-2 Log Likelihood 336.79 +

Cox & Snell R
2
.048

Nagelkerke R
2
.077

* p<.05; **p<.01; ***p<.000 + 354.76 with only constant in model
109
was “preparedness for graduate/advanced education” (X
2
(1) 5.51, p<.05), which was included to
represent demand for higher education (i.e., human capital). The odds ratio revealed women who
reported their “preparedness for graduate/advanced education” was “stronger/much stronger”
than when they began college were 4.3 times more likely to report they planned to attend
graduate/professional school (see Table 14).
“Plan to Work-Full Time”
39

REM undergraduate women with S&E majors. The results of the logistic regression
for REM women with S&E majors indicated the final block was statistically reliable in
distinguishing between “do not plan to work full-time” and “plan to work full-time.” The
inferential goodness-of-fit test—the Hosmer–Lemeshow (H–L)—yielded a X
2
(8) of 6.64 and
was insignificant (p >.05), indicating the model fit the data well; the null hypothesis of a good
model fit to data was plausible. The Cox and Snell R
2
and the Nagelkerke R
2
were .10 and .14,
respectively (see Table 15).
Wald statistics indicated three variables were significant. Two of those variables were
“helping others who are in difficulty” (X
2
(1) 8.12, p<.01) and “raising a family” (X
2
(1) 4.23,
p<.05), which were included to represent expected benefits (i.e., human capital). Odds ratios
revealed women who reported “helping others who are in difficulty” or “raising a family” was
“very important/essential” were about 3 and 2 times, respectively, more likely to report they
planned to work full-time. The third significant variable was “faculty provided a letter of
recommendation” (X
2
(1) 12.50, p<.000), which was included as part of the habitus component
(i.e., layer one). However, the odds ratio was below 1; women who reported faculty frequently

39
The findings reported are the pooled results across the five imputed data sets. For the non-
imputed results, see Table A-11 and Table A-12 in Appendix A.
110
Table 15

Results for “Plan to Work Full-Time” for REM Undergraduate Women with S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who Are in Difficulty 1.10 .386 8.12 .004** 3.00 1.41 6.40

Raising a Family
.617 .300 4.23 .04* 1.85 1.03 3.33

Being Very Well-Off Financially -.104 .275 .143 .707 .902 .526 1.55

Undergraduate GPA -.250 .305 .672 .413 .779 .427 1.42

Preparedness for Employment After College .235 .389 .365 .545 1.26 .590 2.71

Undergraduate Loan Amount .125 .152 .676 .425 1.13 .820 1.56

Parents’ Levels of Education -.168 .313 .288 .591 .845 .456 1.57

Faculty Provided Advice and Guidance About
Educational Program
.022 .328 .004 .946 1.02 .537 1.95

Faculty Provided a Letter of Recommendation -1.05 .297 12.50 .000*** .349 .195 .625

Faculty Provided Opportunity to Work on a Research
Project
.401 .336 1.42 .233 1.49 .773 2.89

Institutional Selectivity .001 .001 1.00 .478 1.00 .998 1.04

Constant -2.84 1.79 2.52 .113

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 6.64 8 .576

Omnibus Test of Model 32.28 11 .000***

-2 Log Likelihood 360.10 +

Cox & Snell R
2
.105

Nagelkerke R
2
.142

* p<.05; **p<.01; ***p<.000 + 392.38 with only constant in model

111
Table 16

Results for “Plan to Work Full-Time” for REM Undergraduate Women with Non-S&E Majors

Variable

 

SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who Are in Difficulty -.235 .375 .393 .532 .791 .379 1.65

Raising a Family
-.213 .327 .424 .514 .808 .426 1.53

Being Very Well-Off Financially .159 .278 .327 .567 1.17 .680 2.02

Undergraduate GPA -.285 .288 .979 .322 .752 .427 1.32

Preparedness for Employment After College .047 .341 .019 .891 1.05 .537 2.05

Undergraduate Loan Amount .238 .152 2.45 .123 1.27 .936 1.72

Parents’ Levels of Education -.292 .268 1.19 .276 .746 .441 1.26

Faculty Provided Advice and Guidance About
Educational Program
.251 .312 .647 .421 1.29 .697 2.37

Faculty Provided a Letter of Recommendation -.846 .293 8.34 .004** .429 .241 .763

Faculty Provided Opportunity to Work on a Research
Project
-.455 .318 2.05 .153 .634 .340 1.18

Institutional Selectivity -.002 .001 4.00 .140 .998 .995 1.00

Constant 2.31 2.30 1.01 .317

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 6.94 8 .543

Omnibus Test of Model 35.20 11 .000***

-2 Log Likelihood 382.43 +

Cox & Snell R
2
.091

Nagelkerke R
2
.135

* p<.05; **p<.01; ***p<.000 + 417.63 with only constant in model
112
provided a letter of recommendation were about .3 times less likely to report they planned to
work full-time (see Table 15).
REM undergraduate women with Non-S&E majors. The results of the logistic
regression for REM women with non-S&E majors indicated the model was statistically reliable
in distinguishing between “do not plan to work full-time” and “plan to work full-time.” The
inferential goodness-of-fit test—the Hosmer–Lemeshow (H–L)—yielded a X
2
(8) of 6.94 and
was insignificant (p >.05), indicating the model fit the data well. The Cox and Snell R
2
and the
Nagelkerke R
2
were .09 and .13, respectively (see Table 16). However, Wald statistics indicated
only one variable was significant. This variable was “faculty provided a letter of
recommendation” (X
2
(1) 8.34, p<.01), which was included to represent the habitus component
(i.e., layer one). As illustrated in Table 16, the odds ratio showed women who reported faculty
frequently provided a letter of recommendation were about .4 times less likely to report they
planned to work full-time (the odds ratio was below 1) (see Table 16).
Discussion
In this study I focused on the “graduate education choice” and the “full-time employment
choice” of REM undergraduate women with S&E majors. I also included a comparison group of
REM undergraduate women with non-S&E majors. The following subsections provide an
overview of the key findings for each “choice.”
“Graduate Education Choice”
Four discussion points arise from the results of the analyses for “graduate education
choice.” The first concerns the human capital aspect of Perna’s model. In the economic model of
human capital, college choice is based on weighing the expected benefits of collegiate education
against its expected costs (Becker, 1993; Perna, 2006). Of the variables included to account for
113
expected benefits, two variables were significant: “helping others who are in difficulty” and
“obtaining recognition from my colleagues for contributions to my special field.” Specifically,
these variables were significant only for REM women with S&E majors. REM women with S&E
majors who reported “gaining recognition from my colleagues to my special field” was “very
important/essential” were 2 times more likely to report they planned to choose to attend
graduate/professional school. Zhao, Carini, and Kuh (2005) note that women in S&E fields
underestimate their collegiate educational accomplishments to a greater extent than do their male
counterparts. If such is the case, it is possible undergraduate REM women perceive gaining
recognition as a benefit of attending graduate/professional school.
While the variable “helping others who are in difficulty” was significant, this variable
had a negative effect. REM women with S&E majors who reported “helping others who are in
difficulty” was “very important/essential” were about .3 times less likely to report they planned
to choose to attend graduate/professional school. This finding extends previous research.
Although not specific to REM women with S&E majors, past research indicates undergraduate
women with S&E majors are less likely to pursue S&E graduate degrees if staying in physical
proximity to family (Villarejo et al., 2008) or raising a family is a high priority and if they are
committed to helping others or effecting social change (Sax, 2001). Within the context of my
study, it appears REM women with S&E majors did not perceive graduate studies as a means
through which they could help others.
Aligned with expected benefits is the demand for human capital. I hypothesized that
REM women with S&E majors who believed their undergraduate studies prepared them for a
graduate education would be more likely to choose to attend graduate school. While the results
supported this hypothesis, they also indicated the assumption was true for REM undergraduate
114
women with non-S&E majors. REM undergraduate women with non-S&E majors who reported
their “preparedness for graduate/advanced education” was “stronger/much stronger” than it was
when they had begun college, were about 4 times more likely to report they planned to choose to
attend graduate/professional school. This finding lends to the second discussion point:
“Preparedness for graduate/advance education” was the only statistically significant variable for
the sample of REM women with non-S&E majors. Although the model was statistically reliable
in distinguishing between “do not plan to attend graduate/professional school” and “plan to
attend graduate/professional school,” it appears the variables included to represent the human
capital aspect, layer one, and layer three of Perna’s model were more pertinent for REM women
with S&E majors than for REM women with non-S&E majors.
The third discussion point relates to the first layer of Perna’s model. The first layer, an
individual’s habitus, reflects demographic characteristics (including sex, race, and ethnicity) and
socioeconomic status, as well as his or her cultural capital and social capital (Perna, 2006). Of
the three variables included to account for habitus, two were significant: “parents’ levels of
education,” which was included to account for cultural capital, and the composite “college social
capital,” which was included as a proxy for social capital. The odds ratios showed REM women
with S&E majors who reported frequent “college social capital” interactions with faculty were 2
times more likely to report they planned to choose to attend graduate/professional school. The
composite “college social capital,” was composed of four variables: faculty provided emotional
support and encouragement, faculty provided advice and guidance about educational program,
faculty provided an opportunity to work on a research project, and faculty provided a letter of
recommendation. The variables that made up the composite highlight key interactions with
faculty members. Similar to prior research (Cole & Espinoza, 2009; Huss et al., 2002;
115
MacLachlan, 2006; Sax, 2001; Villarejo et al., 2008), this finding underlines the importance of
student-faculty interactions. Conversely, REM women with S&E majors who reported their
parents had college degrees or higher were about .6 times less likely to report they planned to
attend graduate/professional school. Given that cultural capital is primarily acquired through
family (Bourdieu, 1979/1984), I postulated the parents’ levels of education could bestow relevant
skills that could inform a REM woman’s graduate education choice. In this case, it is possible
that parents with college degrees or higher are able to provide their daughters with information
about the advantages and disadvantages of choosing to pursue a graduate education immediately
after completing a baccalaureate degree.
The fourth discussion point addresses the third layer of Perna’s model. This third layer,
the higher education context, represents the role post-secondary institutions play in shaping
college choice (Perna, 2006). For this study in particular, institutional type and control, as well as
institutional selectivity, were considered. Concerning institutional type and control, the data were
filtered to consider only four-year, private, predominately White institutions. Concerning
institutional selectivity, a scale variable of median SAT scores and/or ACT composite scores
was included to represent institutional selectivity. The results revealed institutional selectivity
was significant for REM women with S&E majors. The effect, however, was negative. The odds
ratio for institutional selectivity (.997) indicated that, as institutional selectivity decreased, the
odds of planning to make the choice to attend graduate/professional school decreased. This
finding is noteworthy because, while prior research indicates REM individuals who attend
selective institutions are more likely to complete bachelor’s degrees (Alon & Tienda, 2005;
Melguizo, 2008), this finding highlights that, for REM women with S&E majors, institutional
selectivity influences intended “graduate education choice.”
116
“Full-Time Employment Choice”
Three discussion points arise from the results of the analyses for “full-time employment
choice.” The first concerns the variables included to account for expected benefits (i.e., human
capital). Of the three variables included, two were significant: “helping others who are in
difficulty” and “raising a family.” REM women with S&E majors who reported “helping others
who are in difficulty” and “raising a family” were “very important/essential” were, respectively,
3 and 2 times more likely to report they planned to choose to work full-time. These findings are
somewhat similar to prior research. For example, research by Sax (1994) and Seymour and
Hewitt (1997) indicates that undergraduate women, more so than undergraduate men, switch out
of S&E majors, in part, because another major appears to offer greater intrinsic interest or a
better overall educational experience, or because the career options and/or lifestyles associated
with S&E majors are less appealing. The findings of my study seem to indicate that perhaps
REM women with S&E majors perceive “helping others who are in difficulty” and “raising a
family” as benefits associated with working full-time because these benefits possibly align with
their personal goals.
The second discussion point concerns habitus. Of the four variables included, only one
was significant. This variable was “faculty provided a letter of recommendation,” which was
included as a proxy for social capital. Interestingly, this variable had a negative effect for REM
women with S&E majors and for REM women with non-S&E majors. REM women with S&E
majors who reported faculty frequently provided a letter of recommendation were about .3 times
less likely to report they planned to make the choice to work full-time. Similarly, REM women
with non-S&E majors who reported faculty frequently provided a letter of recommendation were
about .2 times less likely to report they planned to make the choice to work full-time. It is
117
plausible, then, that receiving a letter of recommendation from a faculty member provides an
alternative post-baccalaureate choice—that is, a choice other than working full-time.
It is important to note that “faculty provided a letter of recommendation” was the only
statistically significant variable for REM women with non-S&E majors. This relates directly to
the final discussion point: Although the model was statistically reliable in distinguishing between
“do not plan to work full-time” and “plan to work full-time,” the variables included to account
for human capital, layer one, and layer three of Perna’s model were more pertinent for REM
women with S&E majors.

118
CHAPTER SEVEN: CONCLUSION
Since his first inauguration, President Obama has emphasized the importance of making
the U.S. the world leader in college attainment, positioning America to offer the “best educated,
most competitive workforce in the world” (President Obama, 2009, as cited by Kanter, 2011).
The President’s call for a college-educated workforce is perhaps most pertinent within the
country’s S&E sectors. By 2018, S&E occupations will account for about 8.6 million jobs in the
U.S. economy and will require about 1.2 million employees with bachelor’s degrees (Carnevale
et al., 2010). To ensure the domestic workforce will possess the academic credentials necessary
to satisfy future job openings, the National Science Board (2010) has stressed the need to
capitalize on untapped domestic talent by increasing the number of REM individuals pursuing
baccalaureate degrees in S&E-related fields.
While S&E bachelor’s degree attainment is important, procurement of advanced degrees
is equally valuable. Of the 8.6 million S&E jobs projected by 2018, about 779,000 jobs will
require employees with master’s degrees or higher (Carnevale et al., 2010). Yet, despite the need
for individuals to pursue advanced degrees in S&E, and despite the increased need for
individuals to enter the S&E workforce, the student college choice discussion does not often
consider students’ post-baccalaureate choices.
In this dissertation, I addressed two post-baccalaureate options—graduate education and
full-time employment—with a focus on the post-baccalaureate choices of REM undergraduate
students with S&E majors. Specifically, this dissertation was composed of three studies. The first
focused on the “graduate education choice” of REM undergraduate students, the second
addressed the “full-time employment choice” of REM undergraduate students, and the third
119
study focused on the “graduate education choice” and the “full-time employment choice” of
REM undergraduate women.
Because the focus of this dissertation was the post-baccalaureate choices of REM
students with S&E majors, all three studies were guided by Perna’s conceptual model of student
college choice
40
. In each of the three studies, I also included a comparison group of REM
undergraduate students with non-S&E majors. Moreover, in all three studies I utilized the 2003
Freshmen Survey and the 2007 College Senior Survey, used the same definition of REM
students, and employed the same classification of S&E majors
41
.
In this final chapter, I provide a summary of the key findings across the three studies—a
summary organized around the conceptual model utilized across all three studies. Thereafter, I
discuss the implications for practice and research.
Key Findings
At the center of Perna’s model of student college choice is the economic model of human
capital investment. The human capital model proposes college choice is based on weighing the
expected benefits of collegiate education against its expected costs (Becker, 1993; Perna, 2006).
Variables were included to account for expected benefits in all three studies. The variables
“helping others who are in difficulty” and “raising a family” were significant in more than one
study. In particular, “helping others who are in difficulty” had a negative and positive effect. On
the one hand, REM students with S&E majors and REM women with S&E majors who reported
“helping others who are in difficulty” was “very important/essential” were, respectively, .5 and
.3 times less likely to report they planned to choose to attend graduate/professional school. On
the other hand, REM women with S&E majors who reported “helping others who are in

40
For an overview of the model, see Chapter 3.
41
For more information about the similarities, see Chapter 1.
120
difficulty” was “very important/essential” were 3 times more likely to report they planned to
choose to work full-time. Similarly, REM women and REM students with S&E majors who
reported “raising a family” was “very important/essential” were about 2 times more likely to
report they planned to choose to work full-time. Given these findings, it seems neither male nor
female REM students with S&E majors perceives graduate school as a means through which
they can help others. It is also possible REM students do not want to delay contributing to
society and view pursuing a graduate education as delaying that contribution.
For REM women with S&E majors in particular, “helping others who are in difficulty”
and “raising a family” were benefits associated with planning to choose to work full-time.
Former research indicates women, more often than men, switch out of S&E majors because the
career options and/or lifestyles associated with S&E majors are less appealing (Sax, 1994;
Seymour & Hewitt, 1997). For this dissertation, the sample of REM women consisted of women
who had persisted toward their final year of undergraduate studies. It is plausible that, when
considering their post-baccalaureate choices, REM women choose the benefits that would align
with their personal goals—in this case, the benefits associated with choosing to work full-time.
In the economic model of human capital, expected costs and benefits are aligned with the
demand for human capital and the supply of resources to invest in human capital (Becker, 1993;
Perna, 2006). In regards to making the transition from high school to college, differences in the
demand for a post-secondary education are expected to reflect variations across individuals,
according to the respective academic preparation and achievement of those individuals (Paulsen,
2001). Accordingly, individuals who are more likely to choose to attend college are those who
are academically prepared (Paulsen 2001).
Within the context of this dissertation, one of the variables included to account for
121
“demand” was “preparedness for a graduate/advanced education.” I assumed REM students with
S&E majors who believed their undergraduate studies prepared them for a graduate education
would be more likely to choose to attend graduate/professional school than those students who
did not hold such a belief. This assumption proved to be true for REM students with S&E
majors, REM students with non-S&E majors, and REM women with S&E majors. The results
from the first study revealed REM students with S&E majors and REM students with non-S&E
majors who reported their “preparedness for a graduate/advanced education” was “stronger/much
stronger” during their last year of undergraduate studies than it had been when they began
college were, respectively, 3.2 and 4 times more likely to report they planned to choose to attend
graduate/professional school. Similarly, REM women with S&E majors who reported their
“preparedness for a graduate/advanced education” was “stronger/much stronger” during their last
year of undergraduate studies than it had been when they began college were 4.3 times more
likely to report they planned to choose to attend graduate/professional school. Whether a student
has an S&E major or a non-S&E major, being prepared for a graduate education informs a REM
student’s intended choice.
Moving beyond the economic model of human capital, Perna’s model contends that an
individual’s calculations of the expected costs and benefits are nested within four contextual
layers, each of which shapes an individual’s choice. These layers
42
are: 1) the individual’s
habitus, 2) the school and community contexts, 3) the higher education context, and 4) the
broader social, economic, and policy contexts. For all three studies in this dissertation, I
considered the first and third layers of Perna’s model.
The first layer, an individual’s habitus, reflects an individual’s demographic

42
For an overview of the layers, see Chapter 3.
122
characteristics (including sex, race, and ethnicity) and socioeconomic status, as well as his or her
cultural capital and social capital (Perna, 2006). To account for cultural capital, in all three
studies I used the variable “parents’ levels of education” as a proxy. I assumed parents’ levels of
education could inform REM students’ “graduate education choice” and “full-time employment
choice.”
The results of the second study revealed “parents’ levels of education” were significant,
but that this variable had a negative effect. REM students with S&E majors who reported their
parents had a college education or higher were .5 times less likely to report they planned to make
the choice to work full-time. It is possible that REM students with parents who have college
degrees or higher are exposed to information about their choices beyond full-time employment.
That is, parents with a college education are perhaps better equipped to provide advice about the
post-baccalaureate choices—besides full-time employment—associated with an S&E degree.
In Study 3, “parents’ level of education” again had a negative effect. REM women with
S&E majors who reported their parents had college degrees or higher were about .6 times less
likely to report they planned to choose to attend graduate/professional school. This finding seems
somewhat contradictory, as it would seem more plausible for “parents’ level of education” to
have a positive effect on students’ decisions to attend graduate/professional school. I propose
that parents with a college degree or higher are able to provide their daughters with information
about the advantages and disadvantages of making the choice to pursue a graduate education
immediately after completion of a baccalaureate degree.
The findings across all three studies highlighted the significance of social capital.
According to Coleman (1988, 1990), one form of social capital is “information channels,” a term
that refers to the gathering and sharing of information through social relations (Coleman, 1988,
123
1990). All three studies included variables of interactions between REM students and faculty
members. I assumed that, by interacting with faculty members, REM students would be exposed
to social relations through, which they could gather information about graduate school or full-
time employment. For the “graduate education choice,” student-faculty interactions were
statistically significant for REM students with S&E majors, REM students with non-S&E
majors, and for REM women with S&E majors. REM students with S&E majors and REM
students with non-S&E majors who reported frequent “social capital through student-faculty
interactions” were, respectively, 2.5 and 2 times more likely to report they planned to choose to
attend graduate/professional school. REM women with S&E majors who reported frequent
“college social capital” were 2 times more likely to report they planned to make the choice to
attend graduate/professional school.
The significance of student-faculty interactions was also apparent in relation to the “full-
time employment choice.” REM students with S&E majors who reported faculty frequently
provided an opportunity to work on a research project were 2 times more likely to report they
planned to make the choice to work full-time. These findings reinforce prior research (Cole &
Espinoza, 2009; Sax, 2001) that has stressed the importance of student-faculty interactions. In
particular, while prior research has established the importance of research experience for
students pursuing S&E baccalaureate degrees (Huss et al., 2002; Russell, 2006; Strayhorn,
2010), the findings of this dissertation suggest research experience is also essential to students
with non-S&E majors. Moreover, having the opportunity to work on a research project can
provide REM students with S&E majors with insight with which to inform their “full-time
employment choice.”
124
Concerning the third layer (i.e., higher education context) addressed in Perna’s model, all
three studies considered institutional type and control, as well as institutional selectivity. In
regards to institutional type and control, the data were filtered to consider only four-year, private,
predominately White institutions. In regards to institutional selectivity, a scale variable of
median SAT scores and/or ACT composite scores was included to represent institutional
selectivity.
The results of Study 1 and Study 3 revealed institutional selectivity mattered for the
“graduate education choice.” In Study 1, as institutional selectivity decreased, the odds ratio of
REM students with S&E majors planning to choose to attend graduate/professional school
decreased. In Study 3, as institutional selectivity decreased, the odds ratio of REM women with
S&E majors planning to choose to attend graduate/professional school decreased. As has been
illustrated in prior research (Eide et al., 1998; Mullen et al., 2003), these findings confirm
institutional selectivity influences graduate school attendance—or, within the context of this
dissertation, whether students will (or will not) plan to choose to attend graduate school.
Additionally, while prior research indicates REM individuals who attend selective institutions
are more likely to complete bachelor’s degrees (Alon & Tienda, 2005; Melguizo, 2008), these
findings reveal that attending a private, predominately White, selective institution also matters as
such attendance relates to the intended “graduate education choice” of REM students with S&E
majors.
Implications
While noteworthy, the findings of this dissertation should be just the beginning of
research into the post-baccalaureate choices of REM students with S&E majors. As noted
previously, because of limitations with the data, the dependent variables utilized in all three
125
studies only provided information about students’ “intended choices.” A challenge associated
with this line of research is the lack of longitudinal data (information about students’
undergraduate experiences, graduate studies, and employment) specific to REM individuals in
S&E fields. Tracking REM individuals as they move through their undergraduate studies and on
to their post-baccalaureate choices would provide valuable insights into how to attract and retain
REM individuals in S&E fields. The collection of data specific to REM individuals in S&E fields
would also make it possible to conduct studies that disaggregate by S&E field, sex, and race and
ethnicity. To better serve REM students, it is necessary to acknowledge that there are differences
associated with S&E field, sex, and race and ethnicity. As such, the collection of both
longitudinal data and of studies that disaggregate is needed.
One of the key findings across all three studies is the importance of student-faculty
interactions. For REM students, in particular, interactions with faculty are often facilitated
through STEM-specific opportunity and support programs
43
—programs that also provide REM
students with undergraduate research opportunities. Given the documented success of these
programs (Musesus et al., 2001), their support and funding should continue. Practitioners
heading STEM-specific opportunity and support programs should also continue to retain
longitudinal information
44
concerning their participants. Such information can lend itself to
research specific to REM students in S&E fields. Moreover, given the importance of student-
faculty interactions, future research should consider the educational and career paths of REM
individuals in S&E faculty positions. Current figures, for example, reveal the share of full-time,

43
For more information about these programs, see Chapter 2.
44
The Meyerhoff Program is an example of a program that keeps longitudinal information of its
participants.
126
full professorships held by REM individuals is lower, and has risen more slowly, than the share
of full-time, full professorships held by White women (NSF & NCSES, 2013).
In all three studies, a variable that was included and was not significant was
“undergraduate loan amount.” It could be that this variable was not significant because of the
statistical techniques utilized (Malcom & Dowd, 2012). Given students’ continued reliance on
loans to fund their undergraduate studies, future research should further explore how student debt
may (or may not) influence the post-baccalaureate choices of REM students with S&E majors.
Additionally, given the changes unfolding throughout financial aid policies, future studies should
consider how such adjustments may (or may not) influence the post-secondary choices of REM
students considering an S&E field as a major, as well as the post-baccalaureate choices of REM
students who have chosen an S&E field as a major.
As the discourse of increasing the number of college-educated individuals continues, it
will be necessary to continue research of student college choice—and, as I have argued in this
dissertation, it is necessary to consider post-baccalaureate choice. Overall, this dissertation
extends the college choice discussion by considering the “graduate education choice” and the
“full-time employment choice” of REM students with S&E majors. Specifically, as suggested by
Perna (2006), although college choice—or, for the purpose of this dissertation, “graduate
education choice” and “full-time employment choice”—is based on a comparison of the benefits
and costs, assessments of those benefits and costs are shaped by more than simply the demand
and supply of resources. As indicated by the findings of this dissertation, assessments are shaped,
in part, by an individual’s habitus and by the higher education context.

127
References
Alon, S., & Tienda, M. (2005). Assessing the “mismatch” hypothesis: Differences in college
graduation rates by institutional selectivity. Sociology of Education, 78, 294-315.
Retrieved from: http://www.jstor.org/stable/4150500
Allison, P.D. (2009). Missing data. In R.E. Millsap & Maydeu-Olivares, A. (Eds.), The SAGE
handbook of quantitative methods in psychology (pp. 72-89). Thousand Oaks, CA: Sage
Publications.
Astin, A. (1985). Minorities in American higher education: Recent trends, current prospects,
and recommendations. San Francisco, CA: Jossey-Bass.
Barlow, A.E.L., & Villarejo, M. (2004). Making a difference for minorities: Evaluation of an
educational enrichment program. Journal of Research in Science Teaching, 1-21. doi:
10.1002/tea.20029
Becker, G.S. (1962). Investment in human capital: A theoretical analysis. Journal of Political
Economy, 70(5), 9-49.
Bourdieu, P. (1977). Outline of a theory of practice. New York, NY: Cambridge University
Press.
Bourdieu, P. (1979/1984). Distinction: A social critique of the judgment of taste. (R. Nice,
Trans.). Cambridge, MA: Harvard University Press. (Original work published 1979)
Bourdieu, P. (1986). The forms of capital. In J.G. Richardson (Ed.), Handbook of theory
research for the sociology of education (pp. 241-258). New York, NY: Greenwood Press.
Brewer, D.J., Eide, E.R., & Ehrenberg, R.G. (1999). Does it pay to attend an elite private
college? Cross-cohort evidence of the effects of college type on earnings. The Journal of
Human Resources, 34(10), 104-123. Retrieved from http://www.jstor.org/stable/146304
128
Brazziel, M.E., & Brazziel, W.F. (2001). Factors in decisions of underrepresented minorities to
forego science and engineering doctoral study: A pilot study. Journal of Science
Education and Technology, 10(3), 273-281.
Bowen, W.G., & Bok, D. (1998). The shape of the river: Long-term consequences of considering
race in college and university admissions. Princeton, NJ: Princeton University.
Carnevale, A.P., Smith, N., & Strohl, J. (2010). Help wanted: Projections of jobs and education
requirements through 2018. Georgetown University: Center on Education and the
Workforce.
Chen, X. (2009). Stats in brief: Students who study science, technology, engineering, and
mathematics (STEM) in postsecondary education (NCES 2009-161). National Center for
Education Statistics, U.S. Department of Education. Washington, D.C.
Choy, S.P., & Gies, S. (1997). Early labor force experiences and debt burden (NCES 97-286).
National Center for Education Statistics, Institute of Education Sciences, U.S.
Department of Education. Washington, D.C.
Choy, S.P., & Bradburn, E.M. (2008). Ten years after college: Comparing the employment
experiences of 1992–93 bachelor’s degree recipients with academic and career-oriented
majors (NCES 2008-155). National Center for Education Statistics, Institute of Education
Sciences, U.S. Department of Education. Washington, DC.
Coleman, J.S. (1988). Social capital in the creation of human capital. Sociology of Education,
94(Issue supplement), S95-S120.
Cole, D., & Espinoza, A. (2009). When gender is considered: Racial ethnic minority students in
STEM. Journal of Women and Minorities in Science and Engineering, 15, 263-277. doi:
10.1615/JWomenMinorScienEng.v15.i3.50
129
Coleman, J.S. (1990). Foundations of social theory. Cambridge, MA: The Belknap Press of
Harvard University Press.
Dale, S., & Krueger, A. (2002). Estimating the payoff to attending a more selective college: An
application of the selection of the unobservables. Quarterly Journal of Economics,
117(4), 1491-1528. doi: 10.1162/003355302320935089
DesJardins, S.L., & Toutkoushian, R.K. (2005). Are students really rational? The development of
rational thought and its application to student choice. In J.C. Smart (Ed.), Higher
Education: Handbook of theory and research (Vol. 20, pp. 191-240). Dordrecht, The
Netherlands: Kluwer Academic Publishers.
Dika, S.L., & Singh, K. (2002). Applications of social capital in educational literature: A critical
synthesis. Review of Educational Research, 72(1), 31-60. doi: 10.3102/00346530720010
31
Douglass, G.K. (1997). Economic returns on investments in higher education. In H.R. Bowen
(Ed.), Investment in learning: The individual and social values of American higher
education (pp. 359-387). Baltimore, MD: The Johns Hopkins University Press.
Eide, E., Brewer, D.J., & Ehrenberg, R.G. (1998). Does it pay to attend an elite private college?
Evidence of the effects of undergraduate college quality on graduate school attendance.
Economics of Education Review, 17(4), 371-376. Retrieved from
http://www.jstor.org/stable/146304
Ellwood, D.T. & Kane, T.J. (2000). Who is getting a college education? Family background and
the growing gaps in enrollment. In S. Danziger & J. Waldfogel (Eds.), Securing the
future: Investing in children from birth to college (pp. 283-324). New York, NY: Russell
Sage Foundation.
130
Elton, C.F., & Smart, J.C. (1988). Extrinsic job satisfaction and person-environment congruence.
Journal of Vocational Behavior, 32, 226-238. doi: 10.1016/0001-8791(88)90016-4
Ethington, C.A., & Smart, J.C. (1986). Persistence to graduate education. Research in Higher
Education, 24(3), 287-303. doi: 10.1007/BF00992076
Field, A. (2009). Discovering statistics using SPSS (3
rd
ed.). Thousand Oaks, CA: Sage
Publications.
Frome, P.M., Alfeld, C.J., Eccles, J.S., & Barber, B.L. (2006). Why don’t they want a male-
dominated job? An investigation of young women who changed their occupation
aspirations. Educational Research and Evaluation, 12(4), 359-372. doi:
10.1080/13803610600765786
Gonzalez, J. (2010). In Texas speech, Obama renews his educational goals for the nation.
Chronicle of Higher Education. Retrieved from http://chronicle.com/article/In-Texas-
Speech-Obama-Rene/123861/
Graham, J.W. & Smith, S.A. (2005). Gender differences in employment and earnings in science
and engineering in the U.S. Economics of Education Review, 24, 341-354. doi:
10.1016/j.econedurev.2004.06.005
Hearn, J.C. (1987). Impacts of undergraduate experiences on aspirations and plans for graduate
and professional education. Research in Higher Education, 27(2), 119-141. doi:
10.1007/BF00992365
Hossler, D., Braxton, J., & Coopersmith, G. (1989). Understanding student college choice. In J.
Smart (ed.), Higher education: Handbook of theory and research (Vol. 5, pp. 231-288).
New York, NY: Agathon Press.

131
Hossler, D., Schmit, J., Vesper, N. (1999). Going to college: How social, economic, and
educational factors influence the decisions students make. Baltimore, MD: John Hopkins
University Press.
Huang, G., Taddese, N., & Walter, E. (2000). Entry and persistence of Women and Minorities in
college science and engineering education (NCES Rep. No. 2000-601). Washington, DC:
National Center for Education Statistics, U.S. Government Printing Office.
Huss, M.T., Randall, B.A., Patry, M., Davis, S.F., & Hansen, D.J. (2002). Factors influencing
self-rated preparedness for graduate school: A survey of graduate students. Teaching of
Psychology, 29(4), 275-281.
Kanter, M.J. (2011, May-June). American higher education: First in the world. Change the
Magazine of Higher Learning. Retrieved from http://www.changemag.org/Archives/
Back%20Issues/2011/May-June%202011/first-in-the-world-full.html
Kinkead, J. (2003). Valuing and supporting undergraduate research. New Directions for
Teaching and Learning, no. 93. San Francisco, CA: Jossey-Bass.
Leslie, L.L., & Brinkman, P.T. (1988). The economic value of higher education. New York, NY:
American Council on Education and Macmillan Publishing Company.
Lewis, B.F., & Collins, A. (2001). Interpretive investigation of the science-related career
decisions of three African American college students. Journal of Research in Science
Teaching, 38(5), 599-621.
Lopatto, D. (2004). Survey of undergraduate research experiences (SURE): First findings. Cell
Biology Education, 3(4), 270–277. doi: 10.1187/cbe.04-07-0045

132
MacLachlan, A.J. (2006). Developing graduate students of color for the professoriate in science,
technology, engineering, and mathematics (STEM). Center for Studies in Higher
Education University of California, Berkeley (CSHE 6.06).
Malcom, L.E., & Dowd, A.C. (2012). The impact of undergraduate debt on the graduate school
enrollment of STEM baccalaureates. The Review of Higher Education, 35(2), 265-305.
doi: 10.1353/rhe.2012.0007
Manksi, C.F., & Wise, D.A. (1983). College choice in America. Cambridge, MA: Harvard
University Press.
Mare, R.D. (1980). Social background and school continuation decisions. Journal of the
American Statistical Association, 75(370), 295-305.
Maton, K.I., Hrabowski, F.A. III, Schmitt, C.L. (2000). African American college students
excelling in the sciences: College and postcollege outcomes in the Meyerhoff Scholars
Program. Journal of Research in Science Teaching, 37(7), 629-654.
Maton, K.I., Sto Domingo, M.R., Stolle-McAllister, K.E., Zimmerman, J.L., & Hrabowski, F.A.
III (2009). Enhancing the number of African Americans who pursue STEM PhD:
Meyerhoff scholarship program outcomes, processes, and individual predictors. Journal
of Women and Minorities in Science and Engineering, 15, 15-37. doi:
10.1615/JWomenMinorScienEng.v15.i1.20
McDonough, P.M. (1997). Choosing colleges: How social class and schools structure
opportunity. Albany, NY: State University of New York Press.
Melguizo, T. (2008). Quality matters: Assessing the impact of attending more selective
institutions on college completion rates of minorities. Research in Higher Education, 49,
214-236. doi: 10.1007/s11162-007-9076-1.
133
Menard, S. (1995). Applied logistic regression analysis. Sage University paper series on
quantitative applications in the social sciences, 07-106. Thousand Oaks, CA: Sage.
Millet, C.M. (2003). How undergraduate loan debt affects application and enrollment in graduate
or first-professional school. Journal of Higher Education, 74(4), 386-427.
Moore, J.L., III. (2006). A qualitative investigation of African American male’s career
trajectories in engineering: Implication for teachers, school counselors, and parents.
Teachers College Record, 108(2), 246-266.
Mullen, A.L., & Goyette, K.A., & Soares, J.A. (2003). Who goes to graduate school? Social and
academic correlates of educational continuation after college. Sociology of Education, 76,
143-169.
Museus, S.D., & Kiang, P.N. (2009). Deconstructing the model minority myth and how it
contributes to the invisible minority reality in higher education. New Directions for
Institutional Research, 142, 5-15. doi: 10.1002/ir.292
Museus, S.D., Palmer, R.T., Davis, R.J., & Maramba, D.C. (2011). Racial and ethnic minority
students success in STEM education ASHE higher education report (Vol. 36, No.
6). San Francisco, CA: Jossey-Bass.
Myers, R. (1990). Classical and modern regression with applications (2
nd
ed.). Boston, MA:
Duxbury.
National Science Board (2010). Preparing the next generation of STEM innovators: Identifying
and developing our nation’s human capital. Arlington, VA: National Science Board
(NSB 10-33).
National Science Foundation, Division of Science Resources Studies (1999). SESTAT: A tool for
studying scientists and engineers in the United States. NSF 99-337. Arlington, VA.
134
National Science Foundation, & National Center for Science and Engineering Statistics.
(2013). Women, minorities, and persons with disabilities in science and engineering:
2013. Special Report NSF 13-304. Arlington, VA.
Nevill, S.C., & Chen, X. (2007). The path through graduate school: A longitudinal examination
10 years after bachelor’s degree (NCES 2007-162). U.S. Department of Education.
Washington, DC: National Center for Education Statistics.
Pascarella, E.T., & Terenzini, P.T. (2005). How college affects students: A third decade of
research (Vol. 2). San Francisco, CA: Jossey-Bass.
Pallant, J. (2011). SPSS survival manual (4
th
ed.). England: McGraw Hill.
Paulsen, M.B. (1990). College choice: Understanding student enrollment behavior. ASHE-ERIC
Higher Education Report (No. 6). Washington, DC: School of Education and Human
Development, The George Washington University.
Paulsen, M.B., & St. John, E.P. (2002). Social class and college costs: Examining the financial
nexus between college choice and persistence. The Journal of Higher Education, 73(2),
189-236. doi: 10.1353/jhe.2002.0023
Perna, L.W. (2004). Understanding the decision to enroll in graduate school: Sex and
racial/ethnic group differences. The Journal of Higher Education, 75(5), 487-527.
Perna, L.W., Steele, R., Woda, S., & Hibbert, T. (2005). State of public policies and the
racial/ethnic stratification of college access and choice in the state of Maryland. Review
of Higher Education, 28(2), 245-272. doi: 10.1353/rhe.2004.0037
Perna, L. W. (2006). Studying college access and choice: A proposed conceptual model. In J.C.
Smart (ed.), Higher Education: Handbook of Theory and Research (Vol. 21, pp.99-157).
Amsterdam, The Netherlands: Springer.
135
Portes, A., & Landolt, P. (1996). The downside of social capital. The American Prospect, 26,
18-21.
Portes, A. (1998). Social capital: Its origins and applications in modern sociology. Annual
Review of Sociology, 24, 1-24.
Robst, J. (2007). Education, college major, and job match: Gender differences in reasons for
mismatch. Education Economics, 15(2), 159-175. doi: 10.1080/0964529071263070
Russell, S. H. (2006). Evaluation of NSF support for undergraduate research opportunities:
Draft synthesis report (SRI Project P16346). Menlo Park, CA: SRI International.
Sax, L.J. (1994). Retaining tomorrow’s scientists: Exploring the factors that keep male and
female college students interested in science careers. Journal of Women and Minorities in
Science and Engineering, 1, 45-61.
Sax, L.J., (2001). Undergraduate science majors: Gender difference in who goes to graduate
school. The Review of Higher Education, 24(2), 153-172.
Schultz, T.W. (1961). Investment in human capital. The American Economic Review, 51(1),
1-17.
Seymour, E., & Hewitt, N.M. (1997). Talking about leaving: Why undergraduates leave the
sciences. Boulder, CO: Westview Press.
Stanton-Salazar, R.D. (1997). A social capital framework for understanding the socialization of
racial minority children and youth. Harvard Educational Review, 67, 1-40.
Stolzenberg, R.M. (1994). Educational continuation by college graduates. American Journal of
Sociology, 99, 1042-77.

136
Strayhorn, T.L. (2010). Undergraduate research participation and STEM graduate degree
aspirations among students of color. New Directions for Institutional Research, 148, 85-
93. doi: 10.1002/ir.364
Swartz, D. (1997). Culture and power: The sociology of Pierre Bourdieu. Chicago, IL:
University of Chicago Press.
Tabachnick, B.G., & Fidell, L.S. (2007). Using multivariate statistics (5
th
ed.). Boston, MA:
Pearson Education.
Tierney, W.G., & Venegas, K.M. (2006). Fictive kind and social capital: The role of peer groups
in applying and paying for college. American Behavioral Science, 49(12), 1687-1702.
doi: 10.1177/000276406289145
Tierney, W. G., & Venegas, K. M. (2007). The cultural ecology of financial aid decision-
making. In E. P. St. John (Ed.), Readings on Equal Education (Vol. 22, pp. 1–35). New
York, NY: AMS Press.
Villarejo, M., Barlow, A.E.L., Kogan, D., Veazey, B.D., & Sweeney, J.K. (2008). Encouraging
minority undergraduates to choose science careers: Career paths survey results. Life
Sciences Education, 7, 394-409. doi: 10.1187/cbe.08-04-0018
Walters, N.B. (1997, November). Retaining aspiring scholars: Recruitment and retention of
students of color in graduate and professional science degree programs. Paper presented
at the annual meeting of the Association for the Study of Higher Education, Albuquerque,
NM. (ERIC Document Reproduction Service, No. 415816)
Weiler, W.C. (1991). The effect of undergraduate student loans on the decision to pursue
postbaccalaureate study. Educational Evaluation and Policy Analysis, 13(3), 212-220.
doi:10.3102/01623737013003212
137
Winkle-Wagner, R. (2010). Cultural capital: The promises and pitfalls in educational research
ASHE higher education report (Vol. 36, No. 1). San Francisco, CA: Jossey-Bass.
Wolniak, G.C. & Pascarella, E.T. (2005). The effects of college major and job field congruence
on job satisfaction. Journal of Vocational Behavior, 67, 233-251. doi:
10.1016/j.jvb.2004.08.010
Zhao, C.M., Carini, R.M., & Kuh, G.D. (2005). Searching for the peach blossom Shangri-La:
Student engagement of men and women SMET majors. The Review of Higher Education,
28(4), 503-525. doi: 10.1353/rhe.2005.0054

138
Appendix A

Tables

Table A-1

Science and Engineering Majors for Study 1 and Study 2 (N=446)

Major

Frequency

Percent

Anthropology

15

3.4

Biochemistry or Biophysics

16

3.6

Biology (General)

46

10.3

Botany

1

.2

Chemistry

17

3.8

Chemical Engineering 1 .2

Civil Engineering

7

1.6

Computer Engineering

5

1.1

Computer Science

9

2.0

Earth Science

2

.4

Economics

55

12.3

Electrical or Electronic Engineering

2

.4

Environmental Science

6

1.3

Ethnic Studies

13

2.9

Geography

3

.7

Industrial Engineering

1

.2

Kinesiology

7

1.6

Mathematics

13

2.9

Mechanical Engineering

3

.7

Microbiology or Bacteriology

1

.2

Other Biological Science

13

2.9

Other Engineering

3

.7

Other Social Science

15

3.4

Physics

6

1.3

Political Science

73

16.4

Psychology

80

17.9

Sociology

32

7.2

Statistics

1

.2

139
Table A-2

Non-Science and Engineering Majors for Study 1 and Study 2 (N=518)

Major

Frequency

Percent

Accounting

31

6.0

Architecture or Urban Planning

1

.2

Art, Fine and Applied

37

7.1

Business Administration (General)

32

6.2

Business Education

1

.2

Communications

34

6.6

Data Processing or Computer Programming

1

.2

Elementary Education

20

3.9

English (Language and Literature)

55

10.6

Finance

15

2.9

History

45

8.7

Home Economics

1

.2

International Business

15

2.9

Journalism

3

.6

Language and Literature (except English)

36

6.9

Law

1

.2

Law Enforcement

2

.4

Management

16

3.1

Marketing

14

2.7

Medicine, Dentistry, Veterinarian

1

.2

Music

19

3.7

Music or Art Education

3

.6

Nursing

18

3.5

Other Arts and Humanities

38

7.3

Other Business

7

1.4

Other Education

8

1.5

Other Field

7

1.4

Other Professional

2

.4

Other Technical

1

.2

Pharmacy

1

.2

Philosophy

10

1.9

Physical Education or Recreation

5

1.0

Secondary Education

3

.6

140
Social Work

6

1.2

Speech

1

.2

Special Education

1

.2

Theater or Drama

12

2.3

Theology or Religion

9

1.7

Therapy (Occupational, Physical Speech)

3

.6

Women’s Studies

3

.6

141
Table A-3

Non-Imputed Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Students with S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others in Difficulty -1.63 .664 6.00 .014** .197 .054 .722

Being Very Well-off Financially 1.17 .481 5.92 .015* 3.23 1.26 8.29

Becoming an Authority in my Field .264 .457 .335 .563 1.30 .532 3.19

Becoming Successful in a Business of my Own -.550 .436 1.59 .207 .577 .246 1.35

Undergraduate GPA .644 .426 2.29 .130 1.90 .827 4.39

Preparedness for Graduate/Advanced Education 1.23 .722 2.91 .088 3.43 .833 14.10

Undergraduate Loan Amount -.047 .161 .084 .772 .954 .696 1.31

Student’s Sex .691 .442 2.44 .118 1.99 .839 4.75

Parents’ Levels of Education -.818 .415 3.89 .048* .441 .196 .995

Participated in Internship Program -.098 .430 .052 .820 .907 .391 2.10

Participated in a Racial/Ethnic Student Organization -.520 .416 1.57 .211 .594 .263 1.34

Social Capital through Student-Faculty Interactions 1.12 .444 6.33 .012** 3.05 1.28 7.28

Institutional Selectivity -.003 .002 2.23 .135 .998 .994 1.00

Constant 1.23 2.87 .184 .668

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 4.06 8 .852

Omnibus Test of Model 33.63 13 .001***

-2 Log Likelihood 169.51 +

Cox & Snell R
2
.18

Nagelkerke R
2
.26

* p<.05; **p<.01; ***p<.000 + 203.14 with only constant in model
142
Table A-4

Non-Imputed Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Students with Non-S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others in Difficulty .364 .523 .485 .486 1.44 .516 4.01

Being Very Well-off Financially .108 .396 .074 .785 1.11 .513 2.42

Becoming an Authority in my Field .500 .385 1.69 .194 1.65 .775 3.51

Becoming Successful in a Business of my Own -.708 .387 3.34 .068 .493 .231 1.05

Undergraduate GPA .340 .376 .821 .365 1.41 .673 2.94

Preparedness for Graduate/Advanced Education 1.27 .777 2.67 .102 3.56 .777 16.35

Undergraduate Loan Amount -.540 .195 7.70 .006** .583 .398 .853

Student’s Sex -.775 .369 4.41 .036* .461 .224 .949

Parents’ Levels of Education .058 .353 .027 .870 1.06 .531 2.11

Participated in Internship Program -.144 .359 .162 .687 .865 .428 1.75

Participated in a Racial/Ethnic Student Organization .169 .360 .220 .639 1.18 .585 2.39

Social Capital through Student-Faculty Interactions .386 .377 1.05 .306 1.47 .702 3.08

Institutional Selectivity -.001 .002 .552 .457 .999 .996 1.02

Constant 3.74 3.11 1.44 .229

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.95 8 .653

Omnibus Test of Model 24.09 13 .03*

-2 Log Likelihood 220.47 +

Cox & Snell R
2
.09

Nagelkerke R
2
.15

* p<.05; **p<.01; ***p<.000 + 244.56 with only constant in model

143
Table A-5

Non-Imputed Results for “Plan to Work Full-Time” for REM Undergraduate Students with S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Raising a Family .452 .409 1.220 .269 1.57 .705 3.51

Being Very Well-off Financially .364 .406 .805 .370 1.44 .650 3.19

Becoming Successful in a Business of my Own .485 .394 1.51 .218 1.62 .750 3.52

Undergraduate GPA .052 .394 .018 .894 1.05 .487 2.28

Preparedness for Employment After College .570 .482 1.40 .237 1.77 .688 4.54

Undergraduate Loan Amount .222 .137 2.65 .104 1.25 .956 1.63

Student’s Sex -.246 .399 .379 .538 .782 .358 1.71

Parents’ Levels of Education -.296 .371 .638 .425 .744 .360 1.54

Participated in Internship Program .067 .379 .031 .860 1.07 .509 2.25

Met with an Advisor/Counselor About Career Plans -.022 .476 .002 .963 .978 .384 2.49

Faculty Provided Help in Achieving Professional
Goals
-.061 .458 .018 .893 .940 .383 2.31

Faculty Provided Opportunity to Work on a Research
Project
.658 .452 2.13 .145 1.93 .797 4.68

Faculty Provided Letter of Recommendation -1.41 .426 10.94 .001*** .224 .106 .563

Institutional Selectivity .001 .002 .497 .481 1.00 .998 1.00

Constant -3.40 2.54 1.79 .181

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 10.98 8 .203

Omnibus Test of Model 21.73 14 .084

-2 Log Likelihood 195.19 +

Cox & Snell R
2
.112

Nagelkerke R
2
.168

* p<.05; **p<.01; ***p<.000 + 216.92 with only constant in model
144
Table A-6

Non-Imputed Results for “Plan to Work Full-Time” for REM Undergraduate Students with Non-S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Raising a Family -.425 .533 .636 .425 .654 .230 1.86

Being Very Well-off Financially .358 .382 .879 .348 1.43 .677 3.02

Becoming Successful in a Business of my Own .179 .372 .232 .630 1.20 .577 2.48

Undergraduate GPA -.142 .372 .146 .703 .868 .418 1.80

Preparedness for Employment After College -.103 .516 .040 .841 902 .328 2.48

Undergraduate Loan Amount .156 .194 .642 .423 1.17 .798 1.71

Student’s Sex .846 .372 5.18 .023* 2.33 1.12 4.83

Parents’ Levels of Education .005 .352 .000 .989 1.00 .504 2.00

Participated in Internship Program .623 .361 2.98 .084 1.86 .920 3.78

Met with an Advisor/Counselor About Career Plans .921 .497 3.44 .064 2.51 .949 6.65

Faculty Provided Help in Achieving Professional
Goals
-.592 .416 2.02 .155 .553 .245 1.25

Faculty Provided Opportunity to Work on a Research
Project
-.335 .397 .711 .399 .716 .329 1.56

Faculty Provided Letter of Recommendation -.572 .403 2.01 .156 .565 .256 1.24

Institutional Selectivity -.003 .002 2.81 .094 .997 .994 1.00

Constant 2.70 3.17 .727 .394

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 14.69 8 .065

Omnibus Test of Model 24.90 14 .036*

-2 Log Likelihood 218.79 +

Cox & Snell R
2
.096

Nagelkerke R
2
.153

* p<.05; **p<.01; ***p<.000 + 243.69 with only constant only in the model
145
Table A-7

Science and Engineering Majors for Study 3 (N=292)

Major

Frequency

Percent

Anthropology

12

4.1

Biochemistry or Biophysics

11

3.8

Biology (General)

35

12.0

Chemistry

11

3.8

Chemical Engineering 1 .3

Civil Engineering

2

.7

Computer Science

3

1.0

Earth Science

1

.3

Economics

23

7.9

Electrical or Electronic Engineering

1

.3

Environmental Science

5

1.7

Ethnic Studies

12

4.1

Geography

2

.7

Industrial Engineering

1

.3

Kinesiology

6

2.1

Mathematics

6

2.1

Mechanical Engineering

1

.3

Microbiology or Bacteriology

1

.3

Other Biological Science

10

3.4

Other Engineering

1

.3

Other Social Science

9

3.1

Physics

2

.7

Political Science

41

14.0

Psychology

70

24.0

Sociology

24

8.2

Statistics

1

.3

146
Table A-8

Non-Science and Engineering Majors for Study 3 (N=367)

Major

Frequency

Percent

Accounting

12

3.3

Art, Fine and Applied

33

9.0

Business Administration (General)

20

5.4

Communications

24

6.5

Data Processing or Computer Programming

1

.3

Elementary Education

19

5.2

English (Language and Literature)

44

12.0

Finance

6

1.6

History

28

7.6

Home Economics

1

.3

International Business

9

2.5

Journalism

3

.8

Language and Literature (except English)

25

6.8

Law

1

.3

Law Enforcement

1

.3

Management

9

2.5

Marketing

10

2.7

Medicine, Dentistry, Veterinarian

1

.3

Music

16

4.4

Music or Art Education

2

.5

Nursing

17

4.6

Other Arts and Humanities

32

8.7

Other Business

5

1.4

Other Education

6

1.6

Other Field

4

1.1

Other Technical

1

.3

Pharmacy

1

.3

Philosophy

4

1.1

Physical Education or Recreation

3

.8

Secondary Education

2

.5

Social Work

3

.8

Speech

1

.3

Special Education

1

.3

147
Theater or Drama

11

3.0

Theology or Religion

6

1.6

Therapy (Occupational, Physical Speech)

2

.5

Women’s Studies

3

.8

148
Table A-9

Non-Imputed Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Women with S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who are in Difficulty -1.70 .718 5.61 .018* .182 .045 .746

Obtaining Recognition From My Colleagues For
Contributions to My Special Field
.552 .475 1.35 .245 1.74 .684 4.41

Becoming Successful in a Business of My Own -.567 .478 1.41 .235 .567 .222 1.45

Undergraduate GPA .983 .507 3.76 .053* 2.67 .989 7.22

Preparedness for Graduate/Advanced Education .809 .796 1.03 .310 2.25 .472 10.69

Undergraduate Loan Amount -.155 .189 .668 .414 .857 .591 1.24

Parents’ Levels of Education -1.11 .475 5.45 .020* .330 .130 .837

Participated in Internship Program -.470 .476 .977 .323 .625 .246 1.59

College Social Capital .766 .542 .957 .328 1.70 .587 4.92

Institutional Selectivity -.002 .002 .834 .361 .998 .995 1.00

Constant 3.25 3.10 1.10 .295

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 4.96 8 .761

Omnibus Test of Model 18.80 10 .043*

-2 Log Likelihood 129.80 +

Cox & Snell R
2
.151

Nagelkerke R
2
.208

* p<.05; **p<.01; ***p<.000 + 148.60 with only constant in model
149
Table A-10

Non-Imputed Results for “Plan to Attend Graduate/Professional School” for REM Undergraduate Women with Non-S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who are in Difficulty .672 .678 .982 .322 1.96 .519 7.39

Obtaining Recognition From My Colleagues For
Contributions to My Special Field
.381 .444 .736 .391 1.46 .613 3.50

Becoming Successful in a Business of My Own -.420 .441 .909 .341 .657 .277 1.56

Undergraduate GPA -.256 .432 .353 .553 .774 .332 1.80

Preparedness for Graduate/Advanced Education 1.66 1.06 2.45 .118 5.25 .657 42.02

Undergraduate Loan Amount -.403 .225 3.21 .073 .668 .430 1.04

Parents’ Levels of Education -.100 .432 .053 .817 .905 .388 2.11

Participated in Internship Program .091 .426 .046 .830 1.09 .476 2.52

College Social Capital .173 .493 .122 .727 1.19 .452 3.13

Institutional Selectivity -.001 .002 .256 .613 .999 .995 1.00

Constant -1.60 3.87 .126 .722

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 4.45 8 .814

Omnibus Test of Model 10.31 10 .413

-2 Log Likelihood 152.61 +

Cox & Snell R
2
.055

Nagelkerke R
2
.093

* p<.05; **p<.01; ***p<.000 + 162.92 with only constant in model

150
Table A-11

Non-Imputed Results for “Plan to Work Full-Time” for REM Undergraduate Women with S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who Are in Difficulty 1.77 .702 6.34 .012** 5.86 1.48 23.21

Raising a Family
.310 .498 .388 .533 1.36 .514 3.62

Being Very Well-off Financially .309 .480 .414 .520 1.36 .532 3.49

Undergraduate GPA -.645 .493 1.71 .191 .525 .200 1.38

Preparedness for Employment After College -.677 .708 .914 .339 .508 .127 2.03

Undergraduate Loan Amount .389 .183 4.53 .033* 1.47 1.03 2.11

Parents’ Levels of Education .166 .467 .127 .722 1.18 .473 2.95

Faculty Provided Advice and Guidance About
Educational Program
-.363 .538 .457 .499 .695 .242 1.99

Faculty Provided a Letter of Recommendation -1.51 .517 8.53 .003** .221 .080 .609

Faculty Provided Opportunity to Work on a Research
Project
.565 .566 .998 .318 1.76 .580 5.34

Institutional Selectivity .001 .002 .163 .687 1.00 .997 1.00

Constant -4.60 3.02 2.31 .128

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.55 8 .698

Omnibus Test of Model 23.47 11 .015*

-2 Log Likelihood 128.40 +

Cox & Snell R
2
.183

Nagelkerke R
2
.251

* p<.05; **p<.01; ***p<.000 + 151.87 with only constant in model

151
Table A-12

Non-Imputed Results for “Plan to Work Full-Time” for REM Undergraduate Women with Non-S&E Majors

Variable

 


SE

Wald’s X
2

p

Odds Ratio

95% C.I.

Helping Others Who Are in Difficulty -.530 .718 .544 .461 .589 .144 2.41

Raising a Family
-.474 .649 .535 .465 .622 .174 2.22

Being Very Well-off Financially .672 .466 2.08 .149 1.96 .786 4.88

Undergraduate GPA .056 .462 .015 .904 1.06 .428 2.61

Preparedness for Employment After College -.520 .681 .583 .445 .595 .157 2.26

Undergraduate Loan Amount .094 .242 .153 .696 1.10 .685 1.76

Parents’ Levels of Education -.282 .441 .410 .522 .754 .318 1.79

Faculty Provided Advice and Guidance About
Educational Program
.193 .510 .142 .706 1.21 .446 3.30

Faculty Provided a Letter of Recommendation -1.07 .479 4.95 .026* .344 .135 .881

Faculty Provided Opportunity to Work on a Research
Project
-.468 .481 .946 .331 .626 .244 1.61

Institutional Selectivity -.003 .002 1.90 .168 .997 .993 1.00

Constant 5.57 4.11 1.83 .176

Model Fit

X
2

df

p
Intercept &
Covariates

Hosmer & Lemeshow 5.31 8 .724

Omnibus Test of Model 15.87 11 .146

-2 Log Likelihood 147.05 +

Cox & Snell R
2
.084

Nagelkerke R
2
.141

* p<.05; **p<.01; ***p<.000 + 162.92 with only constant in model
152
Appendix B
Figures
Figure B-1. Science and Engineering Fields and Occupational Categories in SESTAT
1

Main
Educational
Field
2

Minor Groups of Disciplines and
Sub-disciplines
Main
Occupational
Category
3

Minor Categories of Occupations and
Subcategories
SCIENCE AND ENGINEERING
Computer &
Mathematical
Sciences
Computer & Information Sciences
 Computer & Information Sciences
 Computer Science
 Computer Systems Analysis
 Information Services & Systems
 Other Computer & Information
Sciences

Mathematical Sciences
 Applied Mathematics
 Mathematics (General)
 Operations Research
 Statistics
 Other Mathematical Sciences
Computer &
Mathematical
Scientists
Computer & Information Scientists
 Computer Systems Analysts
 Computer Scientists (except systems
analysts)
 Information Systems Scientists & Analysts
 Other Computer & Information Science
Occupations
 Computer Engineers (Software)

Mathematical Scientists
 Mathematics
 Operations Research Analysts, Modeling
 Statisticians
 Other Mathematical Scientists

Postsecondary Teachers – Computer &
Mathematical Sciences
 Computer Science
 Mathematics
Life & Related
Sciences
Agricultural & Food Services
 Animal Sciences
 Food Sciences & Technology
Life Scientists Agricultural & Food Scientists
 Agricultural & Food Scientists

153
 Plant Services
 Plan Sciences
 Other Agricultural Sciences

Biological Sciences
 Biochemistry & Biophysics
 Biology
 Botany
 Cell & Molecular Biology
 Ecology
 Genetics (Plant & Animal)
 Microbiology
 Nutritional Science
 Pharmacology (Human & Animal)
 Physiology (Human & Animal)
 Zoology
 Other Biological Sciences

Health & Related (these fields are
included under the life sciences for
doctoral programs only)
 Audiology & Speech Pathology
 Health Services Administration
 Health & Medical Assistants
 Health & Medical Technologies
 Medical Preparatory Programs
 Medicines
 Nursing (4 years or longer programs)
 Pharmacy
 Physical Therapy & Other
Rehabilitation

Biological Scientists
 Biochemists & Biophysicists
 Biological Scientists

 Medical Scientists (except practitioners)
 Other Biological & Life Scientists

Environmental Scientists
 Forestry & Conservation Scientists

Postsecondary Teachers – Life & Related
Sciences
 Agriculture
 Biological Science
 Medical Science
 Natural Science
154
 Public Health (Including
Environment)
 Other Health & Medical Sciences

Environmental Life Sciences
Environmental Science Studies
Forestry Services

Physical &
Related Sciences
Chemistry
 Chemistry (except Biochemistry)

Earth Science, Geology &
Oceanography
 Atmospheric Sciences & Meteorology
 Earth Sciences
 Geology
 Other Geological Sciences
 Oceanography

Physics & Astronomy
 Physics
 Astronomy & Astrophysics

Other Physical Sciences
 Other Physical & Related Sciences
Physical Scientists

Chemists
 Chemists (except Biochemists)

Earth Scientists, Geologists &
Oceanographers
 Atmospheric & Space Scientists
 Geologists
 Oceanographers

Physicists & Astronomers
 Astronomer
 Physicists

Other Physical Scientists
 Other Physical & Related Scientists

Postsecondary Teachers –Physical &
Related Sciences
 Chemistry
 Physics
 Earth, Environmental, & Marine Science
Social & Related
Sciences

Economics
 Agricultural Economics
 Economics

Social Scientists

Economists
 Economists

Political & Related Scientists
 Political & Relates Scientists
155
Political & Related Sciences
 Public Policy Studies
 International Relations
 Political Science & Government

Psychology
 Educational Psychology
 Clinical Psychology
 Counseling Psychology
 Experimental Psychology
 Psychology (General)
 Industrial & Organizational
Psychology
 Social Psychology
 Other Psychology

Sociology & Anthropology
 Anthropology & Archaeology
 Criminology
 Sociology
 Area & Ethnic Studies
 Linguistics
 Philosophy of Science
 Geography
 History of Science

Other Social Sciences
 Other Social Sciences

Psychologists
 Psychologists

Sociologists & Anthropologists
 Anthropologists
 Sociologists

Other Social & Related Scientists
 Historians, Science & Technology
 Other Social Scientists

Postsecondary Teachers – Social Sciences
 Economics
 Political Science
 Psychology
 Sociology
 Other Social Sciences
Engineering Aerospace & Related Engineering
 Aerospace, Aeronautical, &
Astronautical

Engineers Aerospace & Related Engineers
 Aerospace & Related Engineers

156
Chemical Engineering
 Chemical Engineering

Civil & Architectural Engineering
 Architectural Engineering
 Civil Engineering

Electrical & Related Engineering
 Computer & Systems Engineering
 Electrical, Electronics, &
Communications Engineering

Industrial Engineering
 Industrial Engineering

Mechanical Engineering
 Mechanical Engineering

Other Engineering
 Agricultural Engineering
 Bioengineering & Biomedical
Engineering
 Engineering Sciences, Mechanics &
Physics
 Environmental Engineering
 Engineering (General)
 Geophysical Engineering
 Materials Engineering (Including
Ceramics & Textiles)
 Metallurgical Engineering
 Mining & Minerals Engineering
 Naval Architecture & Marine
Engineering
Chemical Engineers
 Chemical Engineers

Civil & Architectural Engineers
 Civil Engineers

Electrical & Related Engineers
 Computer Engineers (Hardware)
 Electrical & Electronics Engineers

Mechanical Engineers
 Mechanical Engineers

Other Engineers
 Agricultural Engineers
 Bioengineers & Biomedical Engineers
 Environmental Engineers
 Marine Engineers or Naval Architects
 Materials & Metallurgical Engineers
 Mining & Geological Engineers
 Nuclear Engineers
 Petroleum Engineers
 Sales Engineers
 Other Engineers

Postsecondary Teachers –Engineering
 Engineering
157
 Nuclear Engineering
 Petroleum Engineering
 Other Engineering

Main
Educational
Field

Minor Groups of Disciplines and
Sub-disciplines
Main
Occupational
Category

Minor Categories of Occupations and
Subcategories
NON-SCIENCE AND ENGINEERING
Non-S&E
Disciplines
Management & Administration
 Agricultural Business & Production
 Accounting
 Business Administration &
Management
 Business (General)
 Business & Management Economics
 Financial Management
 Other Business
Management/Administrative Services

Health & Related (these fields are
included in Non-S&E for bachelor’s and
master’s programs only)
 Audiology & Speech Pathology
 Health Services Administration
 Health & Medical Assistants
 Health & Medical Technologies
 Medical Preparatory Programs
 Medicine
 Nursing (4 year or longer programs)
 Pharmacy
 Physical Therapy & Other
Non-S&E
Occupations
Managers & Administrators
 Top & Mid-Level Managers, Executives,
Administrators
 Accountants, Auditors, & Other Financial
Specialists
 Personnel, Training, & Labor Relations
Specialists
 Other Management Related Occupations

Health Related Occupations
 Diagnosing & Treating Practitioners
 Registered Nurses, Pharmacists, Dieticians,
Therapists, etc.
 Health Technologists & Technicians
 Other Health Occupations

Teachers (except S&E postsecondary
teachers)
 Pre-Kindergarten & Kindergarten
 Elementary School
 Secondary – Computer, Math or Science
 Social Sciences
 Other Subjects
158
Rehabilitation
 Public Health (Including
Environment)
 Other Health & Medical Sciences

Teaching & Education
 Education Administration
 Computer Teacher Education
 Counselor Education & Guidance
 Elementary Teacher Education
 Mathematics Teacher Education
 Physical Education & Coaching
 Pre-Elementary Teacher Education
 Science Teacher Education
 Secondary Teacher Education
 Special Education
 Social Science Teacher Education
 Other Education

Social Services & Related
 Social Work
 Other Philosophy, Religion, Theology

Technology & Technical
 Computer Programming
 Data Processing Technology
 Electrical & Electronics Technologies
 Industrial Production Technologies
 Mechanical Engineering-Related
Technologies
 Other Engineering-Related
Technologies
 Special Education
 Other Pre-collegiate Education

Non-S&E Postsecondary Teachers
 Art, Drama, & Music

 Business, Commerce, & Marketing
 Education
 English
 Foreign Language
 History
 Home Economics
 Law
 Physical Education
 Social Work
 Theology
 Trade & Industrial
 Other Health Specialists
 Other, Non-S&E Not Listed Above

Social Services & Related Occupations
 Clergy & Other Religious Workers
 Counselors, Educational & Vocational
 Social Workers

Technologists & Technicians
 Technologists & Technicians in Biology &
Life Sciences
 Computer Programmers
 Electrical, Electronics, Industrial, &
Mechanical Engineering Technicians
 Drafting Occupations

159

Sales & Marketing
 Business Marketing/Marketing
Management
 Marketing Research
Arts, Humanities & Related
 English Language, Literature &
Letters
 Other Foreign Languages & Literature
 Liberal Arts & General Studies
 History
 Dramatic Arts
 Fine Arts
 Music
 Other Visual & Performing Arts

Other Non-S&E
 Architecture & Environmental Design
 Other Conservational, Renewable
Natural Resources
 Actuarial Science
 Communications
 Journalism
 Other Communications
 Criminal Justice & Protective Services
 Home Economics
 Law, Pre-Law, Legal Studies
 Library Science
 Parks, Recreation, Leisure, & Fitness
Studies
 Public Administration
 Surveying & Mapping Engineering
Technicians
 Other Engineering Technologists &
Technicians
 Surveyors
 Technologists & Technicians in
Mathematical Sciences
 Technologists & Technicians in Physical
Sciences

Sales & Marketing Occupations
 Sales/Marketing – Insurance, Securities,
Real Estate & Business Services
 Sales Occupations – Commodities
 Sales Occupations – Retail
 Other Marketing & Sales Occupations

Art, Humanities, & Related Occupations
 Artists, Editors, Entertainers, Public
Relations, Writers
 Historians (except Science & Technology)

Other Non-S&E Occupations
 Accounting clerks & Bookkeepers
 Secretaries, Receptionists & Typists
 Other Administrative
 Architects
 Farmers, Foresters & Fishermen
 Lawyers & Judges
 Librarians, Archivists & Curators
 Actuaries
 Food Preparation & Service Workers
160
 Other Public Affairs
 Other Fields Not Listed

 Protective Service Workers
 Other Service Occupations (except Health)
 Construction Trades, Miners & Well-
Drillers
 Mechanics & Repairers
 Precision Production Occupations
 Operators & Related Occupation
 Transportation & Material Moving
Occupations
 Other Occupations

1. This figure is adapted from National Science Foundation, Division of Science Resources Studies (1999). SESTAT: A tool for
studying scientists and engineers in the United States. NSF 99-337. Arlington, VA.
2. The National Science Foundation recognizes five main S&E educational fields. They are: computer and mathematical
sciences, life and related sciences, physical and related sciences, social and related sciences, and engineering. All other
educational fields are considered non-S&E.
3. The National Science Foundation defines S&E occupational categories as: computer and mathematical scientists, life
scientists, physical scientists, social scientists, and engineers. The National Science Foundation developed the major and minor
science and engineering groupings of occupations shown in this figure.
161
Figure B-2. Majors Classified based on SESTAT: A Tool for Studying Scientists and Engineers in
the United States

“College Major” Variable
1

Label

Accounting Non-S&E (Management & Administration)
Aeronautical or Astronautical Engineering S&E (Engineering-Aerospace & Related
Engineering)
Agricultural

S&E (Life & Related Science-Agricultural &
Food Sciences)
Anthropology

S&E (Social & Related Sciences-Sociology &
Anthropology)
Architecture or Urban Planning Non-S&E (Other Non-S&E)
Art, Fine, and Applied Non-S&E (Art, Humanities, & Related)
Astronomy S&E (Physical & Related Sciences-Physics &
Astronomy)
Atmospheric Science (Including Meteorology) S&E (Physical & Related Sciences-Earth
Science, Geology, & Oceanography)
Biochemistry or Biophysics

S&E (Life & Related Sciences-Biological
Sciences)
Biology (General)

S&E (Life & Related Sciences-Biological
Sciences)
Botany

S&E (Life & Related Sciences-Biological
Sciences)
Building Trades Non-S&E (Other Non-S&E)
Business Administration Non-S&E (Management & Administration)
Business Education Non-S&E (Teaching & Education)
Chemical Engineering S&E (Engineering-Chemical Engineering)
Chemistry S&E (Physical & Related Sciences-Chemistry
Civil Engineering

S&E (Engineering-Civil & Architectural
Engineering)
Communications Non-S&E (Other Non-S&E)
Computer Engineering S&E (Engineering-Electrical & Related
Engineering)
Data Processing or Computer Programming Non-S&E (Technology & Technical)
Drafting or Design Non-S&E (Other Non-S&E)
Earth Science

S&E (Physical & Related Sciences-Earth
Science, Geology, & Oceanography)
Economics S&E (Social & Related Sciences-Economics)

Electrical or Electronic Engineering

S&E (Engineering-Electrical & Related
Engineering)
Electronics Non-S&E (Technology & Technical)
Elementary Education Non-S&E (Teaching & Education)

English Non-S&E (Art, Humanities, & Related)
162
Environmental Science S&E (Life & Related Sciences-Environmental
Life Sciences)
Ethnic Studies S&E (Social & Related Sciences-Sociology &
Anthropology)
Finance Non-S&E (Management & Administration)
Forestry

S&E (Life & Related Sciences-Environmental
Life Sciences)
Geography

S&E (Social & Related Sciences-Sociology &
Anthropology)

Health Technology (Medical, Dental
Laboratory)
2

Non-S&E (Health & Related)

History Non-S&E (Art, Humanities, & Related)

Home Economics Non-S&E (Other Non-S&E)

International Business Non-S&E (Management & Administration)
Industrial Engineering

S&E (Engineering-Industrial Related
Engineering)
Journalism Non-S&E (Other Non-S&E)

Kinesiology S&E (Life & Related Sciences-Biological
Sciences)
Language & Literature Non-S&E (Art, Humanities, & Related)
Law Non-S&E (Other Non-S&E)
Law Enforcement Non-S&E (Other Non-S&E)
Library/Archival Science Non-S&E (Other Non-S&E)
Management Non-S&E (Management & Administration)
Marine (Life) Science

S&E (Life & Related Sciences-Biological
Sciences)
Marketing Non-S&E (Management & Administration)
Marine Science (Including Oceanography)

S&E (Physical & Related Sciences-Earth
Science, Geology, & Oceanography)
Mathematics S&E (Computer & Mathematical Sciences-
Mathematical Sciences)
Mechanical Engineering S&E (Engineering-Mechanical Engineering)
Mechanics Non-S&E (Technology & Technical)
Medicine, Dentistry, Veterinarian
2
Non-S&E (Health & Related)
Microbiology or Bacteriology S&E (Life & Related Sciences-Biological
Sciences)
Military Science Non-S&E (Other Non-S&E)
Music Non-S&E (Art, Humanities, & Related)
Music or Art Education Non-S&E (Teaching & Education)
Nursing
2
Non-S&E (Health & Related)
Other Arts & Humanities Non-S&E (Art, Humanities, & Related)
Other Biological Science

S&E (Life & Related Sciences-Biological
Sciences)
163
Other Business Non-S&E (Management & Administration)
Other Education Non-S&E (Teaching & Education)
Other Engineering S&E (Engineering-Other Engineering)
Other Physical Science

S&E (Physical & Related Sciences-Other
Physical Sciences)
Other Professional
2
Non-S&E (Health & Related)
Other Social Science

S&E (Social & Related Sciences-Other Social
Sciences)
Other Technical Non-S&E (Technology & Technical)
Pharmacy
2
Non-S&E
Philosophy Non-S&E (Art, Humanities, & Related)
Physical Education or Recreation Non-S&E (Teaching & Education)
Physics

S&E (Physical & Related Sciences-Physics &
Astronomy)
Political Science (Government & International
Relations)
S&E (Social & Related Sciences-Political &
Related Sciences)
Psychology S&E (Social & Related Sciences-Psychology)
Secondary Education Non-S&E (Teaching & Education)
Secretarial Studies Non-S&E (Management & Administration)
Social Work Non-S&E (Social Services & Related)
Sociology S&E (Social & Related Sciences-Sociology &
Anthropology)
Special Education Non-S&E (Teaching & Education)
Speech Non-S&E (Art, Humanities, & Related)

Statistics

S&E (Computer & Mathematical Sciences-
Mathematical Sciences)
Theater or Drama Non-S&E (Art, Humanities, & Related)
Theology or Religion Non-S&E (Social Service & Related)
Therapy (Occupational, Physical, Speech)
2
Non-S&E (Health & Related)
Women Studies Non-S&E (Other Non-S&E)
Zoology

S&E (Life & Related Sciences-Biological
Sciences)
1. This variable is from the 2007 College Senior Survey.
2. These majors are categorized as non-S&E fields for bachelor’s and master’s programs.
For more information see the report SESTAT: A Tool for Studying Scientists and
Engineers in the United States.

164
Figure B-3. Selectivity of Private Four-Year Institutions

Selectivity

Median SAT
1
Scores and/or ACT Composite Scores
2

Low 800-989
Medium 990-1069
High 1070-1189
Very High 1190-1600

1. Median SAT is based on a combination of verbal and math scores (i.e., verbal + math).
2. Median SAT scores and/or ACT composite scores of the entering class as reported to the
Integrated Postsecondary Educational Data System

Abstract (if available)

Abstract Despite the need for individuals to pursue advanced degrees in science and engineering (S&E), and despite the increased need for individuals to enter the S&E workforce, the student college choice discussion does not often consider students’ post-baccalaureate choices. This dissertation addresses two post-baccalaureate options—graduate education and full-time employment—with a specific focus on the post-baccalaureate choices of racial and ethnic (REM) students with S&E majors. The dissertation is composed of three studies. The first study focuses on the ""graduate education choice"" of REM students, the second addresses the ""full-time employment choice"" of REM students, and the third focuses on the ""graduate education choice"" and the ""full-time employment choice"" of REM women. Moreover, because the focus of this dissertation is the post-baccalaureate choices of REM students with S&E majors, its investigation is guided by Laura Perna’s conceptual model of student college choice.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The college labyrinth: the educational journey of first-generation Latino students in engineering

PDF

A mixed method examination of available supports for secondary school students’ college and military aspirations

PDF

How religious engagement shapes the college experience of African American Christian males at a predominantly White institution: a phenomenlogical approach

PDF

Nursing students' perceptions of formal faculty mentoring

PDF

Latino high school students: Self-efficacy and college choice

PDF

Equity for students of color through practitioner accountability

PDF

The impact of student-faculty interaction on undergraduate international students' academic outcome

PDF

Investigating the association of student choices of major on college student loan default: a propensity-scored hierarchical linear model

PDF

Cultural centers and students: are the diversity needs of students being met?

PDF

Conceptualizing college writing readiness for the 21st century: a tale of two classrooms

PDF

A study of the pedagogical strategies used in support of students with learning disabilities and attitudes held by engineering faculty

PDF

Whiteness: a narrative analysis on student affairs professionals, race, identity, and multicultural competency

PDF

Chinese students’ college choice: targeting the U.S as the top study destination

PDF

Impact of culturally responsive education on college choice

PDF

"Creaming" students in the charter school admission process: a case study of admission practices in charter schools

PDF

Self-efficacy beliefs and intentions to persist of Native Hawaiian and non-Hawaiian science, technology, engineering, and mathematics majors

PDF

Making meaning: using constructivism to understand students' relationship to engineering

PDF

Collaborative social networks in student affairs: an exploration of the outcomes and strategies associated with cross‐institutional collaboration

PDF

The influence of a Latina-based sorority on the academic experiences of Latina college students

PDF

Teachers’ perceptions of student behavior and relationship quality: an exploration of racial congruence and self-identity development

Asset Metadata

Creator Espinoza, Araceli A. (author)

Core Title The post-baccalaureate choices of racial and ethnic minority students with science and engineering majors

School Rossier School of Education

Degree Doctor of Philosophy

Degree Program Education

Publication Date 09/19/2013

Defense Date 08/22/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,post-baccalaureate choice,racial and ethnic minority,science and engineering

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Cole, Darnell G. (committee chair), Madni, Azad M. (committee member), Slaughter, John Brooks (committee member)

Creator Email aaespino@callutheran.edu,aaespino@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-328509

Unique identifier UC11293254

Identifier etd-EspinozaAr-2044.pdf (filename),usctheses-c3-328509 (legacy record id)

Legacy Identifier etd-EspinozaAr-2044.pdf

Dmrecord 328509

Document Type Dissertation

Format application/pdf (imt)

Rights Espinoza, Araceli A.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA