Qualitive and Quantitive Matrix

https://doi.org/10.1177/0895904818823753

Educational Policy 2021, Vol. 35(3) 481 –521

© The Author(s) 2019 Article reuse guidelines:

sagepub.com/journals-permissions DOI: 10.1177/0895904818823753

journals.sagepub.com/home/epx

Article

From “First in Family” to “First to Finish”: Does College Graduation Vary by How First-Generation College Status Is Defined?

Robert K. Toutkoushian1, Jennifer A. May-Trifiletti1, and Ashley B. Clayton2

Abstract The purpose of our study was to examine the relationship between alternative definitions of first-generation college students (FGCSs) and graduation from 2- and 4-year colleges. Using Education Longitudinal Study of 2002 data, we constructed eight definitions of FGCSs based on parents’ highest level of education and the number of parents at that level. We identified a series of regression models to explain whether the student earned a 2- or 4-year degree, and focused on the association between different definitions of FGCSs and student success. We estimated both unconditional models for all 10th graders in the sample as well as conditional models for only those who enrolled in college, to see if FGCS status mattered even for those who overcame the access hurdle. Furthermore, we examined the relationship between FGCSs and the pathways to college completion for students who have initially enrolled in a 2- versus 4-year colleges.

1University of Georgia, Athens, USA 2Louisiana State University, Baton Rouge, USA

Corresponding Author: Robert K. Toutkoushian, Professor, Institute of Higher Education, Meigs Hall 114, University of Georgia, Athens, GA 30602, USA. Email: [email protected]

823753EPXXXX10.1177/0895904818823753Educational PolicyToutkoushian et al. research-article2019

482 Educational Policy 35(3)

Keywords higher education, educational policy, parent education, first-generation college student

Introduction

Higher education researchers, administrators, and policymakers are increas- ingly interested in understanding why some students succeed in college and others do not. Recently, policy conversations have shifted from expanding access to increasing completion among students who have already enrolled (Ma, Pender, & Welch, 2016). Several national organizations, such as Complete College America (CCA) and the Lumina Foundation, have made college completion a priority, with Lumina establishing a specific goal of 60% of adults with a degree or certificate by 2025 (Lumina Foundation, 2017). This college completion agenda recognizes that although access to higher education is important, so is increasing degree attainment rates, especially for underserved student populations (Lumina Foundation, 2017).

One underserved group that has received significant attention as part of the completion agenda is first-generation college students (FGCSs). Data have consistently shown that, on average, students whose parents did not attend college are not only less likely to enroll in college (Choy, 2001; Ward, Siegel, & Davenport, 2012) but also less likely to graduate even if they do enroll (Choy, 2001; Engle & Tinto, 2008). However, FGCSs often have other char- acteristics associated with lower rates of college enrollment and graduation, such as coming from lower income homes and beginning college with less academic preparation (D’Amico & Dika, 2013; Engle, 2007; Nguyen & Nguyen, 2018; Nunez & Cuccaro-Alamin, 1998; Terenzini, Springer, Yaeger, Pascarella, & Nora, 1996; Warburton, Bugarin, & Nunez, 2001). FGCSs also differ from their non-FGCS peers in terms of on-campus experiences, with FGCSs more likely to enroll part-time and less likely to participate in high- impact practices associated with college success (Billson & Terry, 1982; Chen & Carroll, 2005; Pascarella, Pierson, Wolniak, & Terenzini, 2004). Viewed differently, researchers have called attention to the ways in which many col- leges’ structures, policies, and practices presume students arrive with the social skills and “college knowledge” necessary to navigate the college envi- ronment, presumptions that disadvantage FGCSs (Nguyen & Nguyen, 2018; Rosenbaum, Deil-Amen, & Person, 2006). Thus, it remains difficult to distin- guish the effects of parental education from the effects of other characteristics on outcomes for FGCS and non-FGCS students.

Complicating our understanding of how and why FGCSs tend to be less successful than their peers is a lack of consensus on how to define FGCSs.

Toutkoushian et al. 483

First-generation college status can vary depending on who is counted as a parent, how many of their parents did not attend college, whether parents started or completed college, and the type of institution attended. Some researchers have defined first-generation college status as neither parent having earned a bachelor’s degree, which is consistent with the language in the U.S. government’s Higher Education Act (Auclair et al., 2008; Spiegler & Bednarek, 2013; U.S. Department of Education, 1998). Other studies have categorized FGCSs as only students whose parents never attended col- lege (Auclair et al., 2008; Spiegler & Bednarek, 2013). Furthermore, there are inconsistencies in how first generation is defined on campuses with pro- grams for this population of students (Ward et al., 2012). A recent national landscape report on FGCS programs in the country revealed more than six different definitions of first-generation used in practice today (Whitley, Benson, & Wesaw, 2018).

Inconsistencies in definitions of FGCSs and reliance on simple binary con- structions are problematic for several reasons. It is difficult to generalize across research using vastly different samples, and we know less about certain key groups, such as students enrolled in 2-year colleges whose parents have associ- ate’s degrees. This also limits our understanding of the exact nature of the rela- tionship between parental education and college outcomes. For example, if social and cultural capital are the mechanisms through which FGCS status affects college success, as often argued (Lohfink & Paulsen, 2005; Padgett, Johnson, & Pascarella, 2012; Pascarella et al., 2004; Peralta & Klonowski, 2017), then simple binary definitions might overlook the effects of differences in the amounts of these resources. In policy and practice, definitions of FGCSs can affect who receives and does not receive services and interventions designed to improve their success in college. Students with only one non-college-edu- cated parent are often not considered to be FGCSs, and yet they may be at a similar disadvantage as students with two parents without a college education. Clearly, how we define FGCSs can have significant implications, especially if we do not understand the differences between alternative definitions.

Prior research also indicates the need for a granular analysis of student outcomes under alternative definitions of FGCSs that consider a variety of differences in parental education and in student pathways. Recent work sug- gests that definition matters for understanding FGCS’ access to higher educa- tion (Toutkoushian, Stollberg, & Slaton, 2018). The few studies that have used multiple levels of parental education or considered the number of col- lege-educated parents found differences in graduation depending on whether parents had no college, some college, or a bachelor’s degree (e.g., Redford & Hover, 2017), and between those with one versus two parents with bachelor’s degrees (e.g., Ishitani, 2003, 2006). Yet only a few such studies exist; among

484 Educational Policy 35(3)

these, most considered only a limited set of options for parental education and limited their analysis to students enrolling in 4-year colleges. These find- ings suggest the importance of jointly analyzing a variety of definitions and pathways with a single dataset.

Purpose of the Study and Research Questions

The purpose of our study was to examine the relationship between alternative definitions of FGCSs and graduation from 2- and 4-year colleges. Drawing on findings from prior research employing complex definitions of FGCSs and theories of human, cultural, and social capital, we hypothesized that both the highest level of parental education in a student’s household and the num- ber of parents at that level might influence graduation. We thus used both dimensions to construct eight alternative definitions of FGCSs. Using these definitions and data from the nationally representative Education Longitudinal Study of 2002 (ELS:02), we addressed three research questions:

Research Question 1: What is the relationship between FGCS status and students’ graduation from college? Does this relationship differ under alternative definitions of FGCSs? Research Question 2: Do results differ for students initially enrolling in 2- versus 4-year colleges? Research Question 3: Do results differ for the subsample of students who successfully enrolled in college versus a sample of all 10th graders?

This study adds to the literature on FGCSs by comparing results of differ- ent definitions of FGCSs. Many prior studies have used only one definition, and these definitions have varied substantially. A study focusing on outcomes for FGCSs under different definitions provides context for interpreting the results of studies using each definition, insight into diversity within the FGCS population, and guidance for future research on FGCS. This study also has implications for the ways in which we define FGCSs in policies and pro- grams aimed at improving these students’ college success.

Literature Review

Defining a “First-Generation College Student”

Policymakers and researchers have employed a variety of definitions of “first-generation college student” (Auclair et al., 2008; Nguyen & Nguyen, 2018; Peralta & Klonowski, 2017; Spiegler & Bednarek, 2013; Toutkoushian et al., 2018). According to the Higher Education Act of 1965, a FGCS is

Toutkoushian et al. 485

(a) An individual both of whose parents did not complete a baccalaureate degree; or (b) In the case of any individual who regularly resided with and received support from only one parent, an individual whose only such parent did not complete a baccalaureate degree. (U.S. Department of Education, 1998, para. f1)

Through the Higher Education Act, several college access programs were developed specifically to help FGCSs, and this definition has had significant policy implications. Some examples of federal programs that serve FGCSs include Upward Bound, Educational Talent Search, and Student Support Services. The majority of university programs for first-generation students use this federal definition of FGCS (Whitley et al., 2018). Some university programs for FGCSs also use the federal definition, while others only con- sider a student to be FGCS if neither parent attended college at all (Ward et al., 2012). Likewise, community colleges may define students as FGCSs if their parents have less than an associate’s degree (Ward et al., 2012).

Researchers vary in how they have defined FGCSs and, more often than not, they have differed from the federal definition. A majority of studies focus- ing on FGCS persistence and graduation employed the stricter definition, which considered a student to be FGCS only when neither parent attended any postsecondary institution (e.g., Billson & Terry, 1982; DeAngelo & Franke, 2016; Dumais & Ward, 2010; Ishitani, 2006; Lohfink & Paulsen, 2005; Redford & Hover, 2017; Terenzini et al., 1996). Fewer studies followed the federal government’s lead and defined students as FGCSs when neither parent has a bachelor’s degree but either or both might have some college experience (e.g., D’Allegro & Kerns, 2011; D’Amico & Dika, 2013; Ishitani, 2003, 2016; Vuong, Brown-Welty, & Tracz, 2010; Wolniak, Mayhew, & Engberg, 2012). Policymakers and researchers thus seem to differ in their understanding of how much parental education is necessary to provide students with an advan- tage in accessing and progressing through higher education.

Another problem with the most common definitions of FGCSs is that they overlook the great deal of variation possible in the total “amount” of parental education in a student’s home. Only a few studies have explicitly analyzed outcomes of students whose parents have “some college,” and none of these distinguished between students whose parents earned only a few credits, those whose parents attained an associate’s degree, and those whose parents completed most of a bachelor’s degree (Cragg, 2009; D’Allegro & Kerns, 2011; Ishitani, 2006; Nunez & Cuccaro-Alamin, 1998; Redford & Hover, 2017; Warburton et al., 2001). In addition, few studies have considered two parents’ levels of education (in two-parent households), most often using only the highest level of education of either parent (e.g., Cragg, 2009; Fike & Fike, 2008; Hahs-Vaughn, 2004; Ishitani, 2003, 2006, 2016; Wolniak et al.,

486 Educational Policy 35(3)

2012). Most studies employing these more complex definitions found differ- ences in student outcomes. This suggests the need for a systematic analysis of these differences that jointly explores the effects of both level of parental education and number of college-educated parents. Given that a large number of FGCSs first enroll in 2-year colleges (Arbona & Nora, 2007; Redford & Hover, 2017), it may be especially important to understand whether having parent(s) with an associate’s degree confers an advantage in this context.

Recent research has called attention to these inconsistencies and their implications. Auclair et al. (2008) and Spiegler and Bednarek (2013) both summarized the range of definitions used in literature on college access, experiences, and outcomes for FGCSs, and discussed the importance of a consistent definition. Nguyen and Nguyen (2018) argued that, by treating studies using different definitions as if they applied to the same population, researchers may have made incorrect generalizations and comparisons as well as overlooking potentially important differences among students with different parental education backgrounds. Studies employing different defi- nitions sometimes even yielded conflicting results, especially in terms of col- lege experiences associated with positive outcomes. Ward et al. (2012) and Peralta and Klonowski (2017) not only described variations in definitions but also advocated to define FGCSs as students for whom neither parent has any college experience. They argued that any college experience increases the social and cultural capital parents can share with their children, and thus advantages them in the college environment. However, this fails to consider whether and to what extent parents with more education have more capital and the implications of this extra capital. Although these studies called for a better understanding of the FGCS construct and an appropriate, consistent definition, they did not relate definitions to differences in outcomes. We thus have little information with which to make an informed decision about who “should” be considered FGCSs and whether these decisions “should” vary in 2- versus 4-year contexts.

FGCSs and College Success

Most prior research on FGCS’ persistence and graduation has found that FGCSs are less likely than their non-FGCS peers to persist and graduate from college. For example, a recent descriptive study analyzing nationally repre- sentative data found that both students whose parents did not attend college and those whose parents had some college earned bachelor’s degrees at lower rates than those with at least one parent with a bachelor’s degree (Redford & Hover, 2017). Other studies using national-level data and controlling for other characteristics correlated to college outcomes yielded similar findings

Toutkoushian et al. 487

for persistence (e.g., Dowd, 2004; Ishitani, 2006, 2016; Lohfink & Paulsen, 2005; Warburton et al., 2001) and bachelor’s degree attainment (e.g., Arbona & Nora, 2007; Cragg, 2009; Dumais & Ward, 2010; Hahs-Vaughn, 2004; Ishitani, 2006, 2016; Nunez & Cuccaro-Alamin, 1998; Warburton et al., 2001), across a range of FGCS definitions.

In contrast to the wealth of literature on FGCS’ bachelor’s degree attain- ment, only a few studies have examined FGCSs and associate’s degree attain- ment (Auclair et al., 2008). Defining FGCSs as those whose parents did not earn bachelor’s degrees, Engle and Tinto (2008) found that, compared with non-FGCSs, FGCSs are less likely to earn a bachelor’s degree, but more likely to earn an associate’s degrees. Other studies found students to be equally likely to earn an associate’s degree regardless of parental education (Nunez & Cuccaro-Alamin, 1998; Roksa, 2011). Note that none of these studies exam- ined whether there was a relationship between having a parent with an associ- ate’s degree and student associate’s degree attainment comparable to the relationship between having a parent with a bachelor’s degree and student bachelor’s degree attainment. The minimal research, mixed findings, and lack of complexity in operationalization of parental education for students attend- ing 2-year colleges suggest the need for research that considers both 2- and 4-year pathways. This is especially important given the large proportion of FGCSs who begin their education in 2-year colleges (Redford & Hover, 2017).

Similarly, only a few studies have employed more complex variables for parental education. Ishitani (2003) examined persistence of students with nei- ther parent with bachelor’s degrees, one parent with a bachelor’s degree, and two parents with bachelor’s degrees. Students with two parents with bache- lor’s degrees were significantly more likely to persist, and differences in per- sistence increased over time, suggesting that the number of parents who attended college may be important. Chen and Carroll (2005) and Ishitani (2006) both found significant differences in graduation between students whose parents never attended college, attended college but did not graduate, and graduated with a bachelor’s degree. While these studies provide initial evidence for differences based on several levels of parental education and the number of parents at a given level, further research is necessary to explore both dimensions at the same time.

Clearly, despite substantial prior research on the relationship between parental education and student outcomes, our knowledge is incomplete. Few studies have operationalized parental education in appropriately complex ways, and those that have done so have at most examined either three levels of parental education or one versus two parents have bachelor’s degrees. We thus do not know whether the advantages associated with college-educated parents accrue incrementally or at a threshold, which has implications for

488 Educational Policy 35(3)

how we operationalize parental education in our research and where pro- grams and policies set thresholds for service. We also understand little about FGCSs in 2-year colleges, including both FGCSs who attend 2-year colleges and children of parents with 2-year degrees. Until we examine these issues simultaneously, we cannot begin to resolve discrepancies in definitions or resolve disagreements about which definition might be most appropriate.

Moreover, many studies of FGCSs have relied on single-institution data with small numbers of students of color and students from low-income back- grounds. In their review of studies of FGCSs, Nguyen and Nguyen (2018) noted that such data are especially problematic for studying FGCSs given the strong correlations between FGCS status and these other characteristics. Among studies employing nationally representative data with larger numbers of students of color and those from low-income backgrounds, only one descriptive study (Redford & Hover, 2017) has employed the most recent complete national dataset. Most studies using the national longitudinal data- sets also fail to take full advantage of these datasets’ abilities to track students from high school into and through postsecondary education. Examining path- ways beginning in high school can help us better understand whether parental education is more significant for access, persistence, or both.

Conceptual Frameworks

The complexity of students’ pathways to graduation suggests the need to examine them through the lenses of several different conceptual frameworks. Those especially relevant to understanding FGCS’ outcomes include human capital, social and cultural capital, and college persistence theories. All three forms of capital relate to the notion of social reproduction in education, in that students who come from families with higher levels of capital are them- selves more successful in education (Bourdieu, 1977). College persistence theories view persistence as a longitudinal process and emphasize the effects of academic and social interactions on campus (Tinto, 1993). They provide a framework for understanding the ways in which various forms of capital not only directly affect student persistence but may also indirectly affect persis- tence through their effects on student-institution interactions.

Human capital theory views the pursuit of a college degree as an invest- ment in oneself, and suggests that individuals will make the investment when the expected benefits outweigh the costs (e.g., Becker, 1993). There are sev- eral reasons why students with college-educated parents may be more likely to perceive the benefits of completing college as outweighing the costs. Because of their own educational attainment, college-educated parents are likely to have greater financial resources to support their children’s college

Toutkoushian et al. 489

education, and thus reduce the cost of a degree. Such financial resources may also enable college-educated parents to invest in their children’s precollege academic preparation and abilities, which in turn may make them more likely to persist in college. In addition, college-educated parents are in a better posi- tion to communicate to their children the economic and social benefits of a college degree. Therefore, first-generation students might have lower levels of academic preparation and fewer financial resources when they enter col- lege, compared to their peers with college-educated parents.

Two related but distinct concepts, cultural and social capital, offer additional explanations for differences in FGCS’ and non-FGCS’ college outcomes. Cultural capital includes knowledge and skills for navigating the college expe- rience from application to graduation (Berger, 2000; Bourdieu, 1977, 1986; Jaeger & Karlson, 2015; Mollegaard & Jaeger, 2015; Tan, 2017). Parents who have experience in college settings may be better able to share this knowledge with their children, improving their ability to navigate the college environment. Social capital refers to the way in which college-educated parents and their networks of friends and associates are more likely to realize and transfer infor- mation about the importance of a college degree to their children (Bourdieu, 1977, 1986; Coleman, 1988; Field, 2017; Lin, 2002; Mollegaard & Jaeger, 2015; Moschetti & Hudley, 2015). As with cultural capital, this social capital provides students of college-educated parents with important resources to sup- port them in graduating from college. Likewise, a FGCS may not be able to go to their parents or their parents’ networks for help in navigating college.

Students enter college with variation in these three forms of capital, or what Tinto (1993) describes as pre-entry attributes, which play a role in post- secondary persistence to completion. Tinto’s (1993) Theory of Student Departure focuses on student integration into their institutions, and has been one of the most widely cited explanations of college student persistence (Braxton et al., 2013; Mayhew et al., 2016). Tinto suggests that student back- ground characteristics, including parental education, influence students’ edu- cational goals and commitment to these goals. Parents who went to college might plausibly place a higher value on their child going to college and ulti- mately earning a degree; this may be especially important when students encounter setbacks or obstacles to persistence. All three forms of capital also shape students’ academic and social integration into college and likelihood to persist. For example, a FGCS with lower levels of cultural and social capital might have trouble navigating and integrating into a college campus that was likely not designed with their needs in mind. Alternatively, FGCSs may be less likely to participate in high-impact practices, especially extracurricular activities, associated with persistence due to limited financial capital and a lack of information on the value of such activities, which is often transferred

490 Educational Policy 35(3)

from parent to child as a form of cultural capital as well as through social networks (Billson & Terry, 1982; Chen & Carroll, 2005; Lohfink & Paulsen, 2005; Longwell-Grice & Longwell-Grice, 2008; Pascarella et al., 2004).

Some scholars have argued that Tinto’s (1993) emphasis on integration into college inherently disadvantages underserved students, including FGCSs (Attinasi, 1989; Braxton et al., 2013; Cabrera, Nora, & Castaneda, 1993; Hurtado & Carter, 1997; Melguizo, 2011; Museus, 2014; Tierney, 1991, 1992, 1993). According to these arguments, requiring students to separate from their home environments to integrate into their college environments may be problematic for students from some cultural and family backgrounds, including many FGCSs (Museus, 2014; Tierney, 1992). Recent critiques have also challenged the way in which Tinto’s model asks students to adapt to institutional culture rather than asking institutions to modify their environ- ments and structures (Museus, 2014). Still, alternative models explaining persistence for FGCSs and other underserved students often modify rather than reject Tinto’s work (Berger, 2000; Cabrera et al., 1993; Stuart, Rios- Aguilar, & Deil-Amen, 2014). Recent FGCS studies have often incorporated constructs from Tinto’s model as well as from other models to account for the variety of factors that may affect this population’s persistence decisions (e.g., Lohfink & Paulsen, 2005; Martinez, Sher, Krull, & Wood, 2009). Therefore, this study still acknowledges the tenets of Tinto’s (1993) model, while focus- ing on the variation in capital that is passed on to students based on parent education level.

Data and Method

Data

For the purpose of this study, we relied on the ELS:02. Conducted by the National Center for Education Statistics (NCES), ELS:02 is a national study of 10th graders who were first surveyed during their 10th grade year in 2002 and completed follow-up surveys in 2004, 2006, and 2012. We restricted our sample to students who indicated in 2002 that they were living full-time with two parents (whether biological, step, foster, or adoptive) who each reported their highest level of educational attainment which is roughly half of the ELS sample. This restriction enabled us to examine the relationship between the number of college-educated parents and whether a student graduates from college. After deleting cases with missing data on the dependent variable, our final weighted sample consisted of approximately 7,800 students (all sample sizes are rounded per NCES requirements).

Toutkoushian et al. 491

We selected ELS:02 for this study for several reasons. First, ELS:02 is a nationally representative dataset, and thus allows us to generalize findings to the broader U.S. student population. Second, the survey collected informa- tion from students during high school and throughout college, enabling us to examine their pathways through both K-12 and postsecondary education. Third, ELS:02 included both student and parent surveys with the latter pro- viding a more reliable measure of parental educational attainment and their relationship to the student.

Dependent variables. The primary dependent variable used in this study was whether or not a student had graduated from college as of 2012. The ELS:02 data further allowed us to determine the highest level of educational attain- ment completed by each student as of 2012. Accordingly, the dependent vari- able was categorical with three possible values: (a) highest degree earned was at least a bachelor’s degree, (b) highest degree earned was an associate’s degree, and (c) did not earn either a bachelor’s or an associate’s degree.

Explanatory variables. We created several groups of explanatory variables that theory and prior literature suggest may be associated with college comple- tion. The first group (E) was the main focus of our study, namely, the parental education level of each student’s parents. We created 12 different measures of parental education, where each variable depended on the level of education needed for a parent to be labeled college educated and the number of parents meeting the education criteria, as depicted in Figure 1. We used four levels to determine whether a parent was college educated: (a) enrolled in any postsec- ondary institution, (b) earned an associate’s degree or more, (c) attended a 4-year institution or more, or (d) earned a bachelor’s degree or more. These education levels were constructed so that they would be nested within each other. We used three values for the number of college-educated parents: zero, one, or two. When combined, these criteria gave rise to 4 × 3 = 12 different variables. Finally, we labeled a student being a FGCS when either zero or at most one parent met the designated criteria for being college educated, result- ing in eight possible definitions of FGCS status.

The next group of explanatory variables (P) were the student’s personal characteristics, including the student’s gender, race/ethnicity, ninth-grade GPA (grade point average), and PISA:2003 concordance test scores in read- ing and mathematics. The decision to include these variables was based on prior research identifying significant differences in persistence by gender and race/ethnicity, and especially strong effects of prior academic preparation (Adelman, 2006; Mayhew et al., 2016; Perna & Thomas, 2008). Most models

492 Educational Policy 35(3)

of student persistence, including Tinto’s (1993) model, likewise emphasize the importance of these attributes (Perna & Thomas, 2008).

The third group of explanatory variables (F) represents family measures such as whether the student lived with their biological parents, their number of siblings, number of dependents, and family income (four categories). The fourth group of explanatory variables (S) captures the following high school measures for each student: urbanicity of high school, percentage of students on free lunch, percentage of students who took at least one Advanced Placement (AP) course, public/private status, and Census region where the school is located. Perna and Thomas (2008) summarized research on the role of both family and high school characteristics not only on enroll- ment in college but also on longer term college outcomes. Controlling for family income is an especially important control given the high correlations between family income and FGCS status (e.g., Engle & Tinto, 2008) and the ways in which financial resources can inform the cost/benefit analysis resulting in persistence decisions (Long, 2007; Perna & Thomas, 2008).

Finally, the college-level factors for the first institution attended (C) include whether the student enrolled in college full-time, the type of college attended (public, private not-for-profit, private for-profit), the student’s

“What is the highest level of education you and your spouse/partner have reached?”

Education Category Father Mother

A High School or Less High School or Less

B Attend 2-Year, No Degree Attend 2-Year, No Degree

C Associate’s Degree Associate’s Degree

D Attend 4-Year, No Degree Attend 4-Year, No Degree

E Bachelor’s Degree or More Bachelor’s Degree or More

Exactly one parent is college-educated: E11 = 1 if Father or Mother (but not both) in A, 0 otherwise E12 = 1 if Father or Mother (but not both) in A or B, 0 otherwise E13 = 1 if Father or Mother (but not both) in A or B or C, 0 otherwise E14 = 1 if Father or Mother (but not both) in A or B or C or D, 0 otherwise Neither parent is college-educated: E21 = 1 if both Father and Mother in A, 0 otherwise E22 = 1 if both Father and Mother in A or B, 0 otherwise E23 = 1 if both Father and Mother in A or B or C, 0 otherwise E24 = 1 if both Father and Mother in A or B or C or D, 0 otherwise

Figure 1. Construction of parental education status variables from ELS. Note. Data obtained from Question 34 on Parent Questionnaire Base Year. ELS = Education Longitudinal Study.

Toutkoushian et al. 493

planned major (science, technology, engineering and mathematics [STEM], business, other), and whether the student relied on grants or loans to help pay for college. Full-time enrollment has been found to be an especially strong predictor of college completion, perhaps by increasing integration in ways consistent with Tinto’s (1993) model (Braxton et al., 2013). Although the mechanism by which financial aid influences persistence remains some- what unclear, research has at least established a clear, significant connec- tion between these variables (Long, 2007; Mayhew et al., 2016).

Method

To examine college completion, we relied on multinomial logistic regression. The dependent variable of interest in our study had three possible values: (a) highest degree earned was a bachelor’s degree or more, (b) highest degree earned was an associate’s degree, and (c) did not earn either an associate’s or bachelor’s degree. In each of the following models, we converted the coeffi- cients into marginal effects so that they could be interpreted as the effect of each variable on the probability of the given educational attainment levels. In all models, we clustered the standard errors at the high school level to account for possible nonindependence of the errors for students within schools. We also weighted the data to account for the complex sampling design NCES used in the ELS:02 survey.

In the first set of models, we focused on all 10th graders in our sample (N ~ 7,800). The multinomial logistic regression model was of the general form:

Y E E P F S N= + + + + + + =α β β γ δ θ ε1 1 2 2 , all 10 graders,th (1)

where Y = college outcome, P = personal characteristics, F = family char- acteristics, and S = high school characteristics. We included two dummy variables in the equation to capture parental education, where E1 = 1 if exactly one parent was college educated and E2 = 1 if neither parent was col- lege educated. Accordingly, the coefficients β1 and β2 measure the differ- ence in the rates at which these students earned college credentials relative to other 10th-grade students with two college-educated parents, holding con- stant their personal, family, and high school characteristics. We estimated Equation 1 four times for the sample of 10th graders, where each time we varied the education level needed for parents to be labeled college educated. In addition, we repeated the analysis after replacing the two dummy variables E1 and E2 with a single dummy variable for whether at least one parent was not college educated as this corresponds to our second definition of FGCSs.

494 Educational Policy 35(3)

Taken together, this gave us 12 different multinomial logistic regression models for the 10th-grade sample.

The effects of FGCSs on college graduation for 10th graders, however, entail two hurdles: first a student must enroll in college and then graduate from college conditional on enrollment. Accordingly, we also estimated the model for only those students who initially enrolled in college (n ~ 6,300) to determine if FGCSs were still at a greater risk of failure even after getting to college. These models were of the form:

Y E E P F S C n= + + + + + + + =α β β γ δ θ τ ε1 1 2 2 , college enrollees, (2)

where all variables are defined as before, and C = college-specific vari- ables defined earlier. The coefficients for the variables E1 and E2 now capture whether those FGCSs who enrolled in college were less likely than their peers to earn a degree. Taken together, Equations 1 and 2 allow us to determine whether the lower college graduation rate for FGCSs is due to their being less likely to go to college or graduate once they have enrolled or both.

Finally, we examined whether FGCS graduation rates varied depending on whether they initially enrolled at a 2- or 4-year institution. To do this, we partitioned the sample into those who reported initially enrolling at a 2-year institution (n ~ 1,600) or a 4-year institution (n ~ 3,900) and esti- mated separate conditional multinomial logistic regression models for each group.

Results

In Figure 2, we provide a breakdown of 10th-grade students by the paren- tal education categories used in our study. Each column represents a dif- ferent definition of a college-educated parent, ranging from “attend any college” to “earned a bachelor’s degree.” Within each column, we report the percentage of students with zero, one, or two college-educated parents according to the designated criteria. As expected, the percentage of stu- dents with no college-educated parents increased as we raised the level of education required to be labeled college educated. The bottom two por- tions of each column represent FGCSs. If we adhere to a strict definition where FGCSs are only those for whom neither parent is college educated, then these are shown as the bottom section of each column. Alternatively, if we loosen the criteria to also include students with exactly one non- college-educated parent as FGCSs, then the sum of the bottom two sec- tions of each column represents their proportions. Accordingly, the share

Toutkoushian et al. 495

of FGCSs in our 10th-grade sample could range from as low as 22% to a high of 76%.

Table 1 contains the descriptive statistics for the various explanatory vari- ables used in our statistical models. More specific details on how we con- structed the dependent and other explanatory variables are provided in the appendix. In the interest of space, we only present the descriptive statistics for the full sample of 10th graders.

In Table 2, we present breakdowns of college completion rates by the vari- ous definitions of parental education used in our study. The first column denotes whether the student graduated from a 4-year, 2-year, or either type of college. The second column describes the level of education needed for a par- ent to be categorized as college educated. Columns 3 through 5 show the percentages of students achieving each outcome broken down by the number of college-educated parents. The last three columns show differences in col- lege completion rates between students with two college-educated parents (non-FGCSs) and either one or no college-educated parents.

Figure 2. Breakdown of parental education—10th graders. Note. Each column represents a different measure of college education. Bottom bar = neither parent is college educated. Middle bar = only one out of two parents is college-educated. Top bar = both parents are college educated. Any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree.

496 Educational Policy 35(3)

Table 1. Descriptive Statistics—Full Sample of 10th Graders.

Variable M SD Minimum Maximum

Gender: Female 0.51 0.50 0 1 Gender: Male [REF] 0.49 0.50 0 1 Race: White [REF] 0.70 0.46 0 1 Race: Black 0.08 0.27 0 1 Race: Asian 0.09 0.29 0 1 Race: All other 0.07 0.25 0 1 Ethnicity: Hispanic 0.13 0.34 0 1 Ethnicity: Non-Hispanic [REF] 0.86 0.35 0 1 High School GPA 2.90 0.71 0.30 4.00 Score: Math 5.17 0.84 3.08 7.32 Score: Reading 5.49 0.86 3.44 7.09 Live w/biological parents 0.82 0.39 0 1 Number of siblings 2.07 1.50 0 6 Number of dependents 2.59 1.41 0 8 Income: Below US$20K 0.06 0.24 0 1 Income: US$20K to US$50K 0.28 0.45 0 1 Income: US$50K to US$100K [REF] 0.37 0.48 0 1 Income: Above US$100K 0.18 0.38 0 1 Enrollments: Grade 10 (in 100s) 3.02 2.24 0.03 12.75 School: Urban 0.31 0.46 0 1 School: Suburban 0.50 0.50 0 1 School: Rural [REF] 0.19 0.39 0 1 School: Pct free lunch 20.25 20.97 3 88 School: Pct AP 15.44 13.78 0 81 School: Public 0.74 0.44 0 1 School: Nonpublic [REF] 0.26 0.44 0 1 School: New England 0.04 0.20 0 1 School: Mid Atlantic 0.13 0.34 0 1 School: E North Central 0.19 0.40 0 1 School: W North Central 0.08 0.26 0 1 School: South Atlantic 0.18 0.38 0 1 School: E South Atlantic 0.08 0.27 0 1 School: W South Atlantic 0.10 0.31 0 1 School: Mountain 0.05 0.22 0 1 School: West [REF] 0.14 0.35 0 1

Note. GPA = grade point average; Pct = percentage; AP = Advanced Placement.

Toutkoushian et al. 497

Our results document that there are large gaps in 4-year and overall college completion rates by parental education. For example, Table 2 shows that the 4-year college completion gap between students with two college-educated parents and no college-educated parents ranges from 34% to 42% depending on how we defined college-educated parents. However, the 4-year college completion gaps were also quite large between students with one and two college-educated parents (18%-23%). As we increased the level of education required for a parent to be labeled college educated, the gaps in 4-year and overall college completion rates for students with one versus two college-educated parents fell. We found the opposite pattern, however, when comparing students with no versus two college-educated parents. Overall, the 4-year college completion gaps increased along with the difference in the number of parents meeting

Table 2. Comparison of College Graduation Rates by First-Generation College Student Definition.

Student graduated from:

Definition of college- educated parent:

Number of college- educated parents:

None vs. one

(%) One vs. two (%)

None vs. two

(%) Two (%)

One (%)

None (%)

Four-year college

Any college 54.3 30.9 20.2 −10.7*** −23.4*** −34.1*** AA degree 58.7 36.5 21.9 −14.6*** −22.2*** −36.8*** Some BA 62.7 42.1 24.2 −17.9*** −20.6*** −38.5*** BA degree 67.9 49.7 25.9 −23.8*** −18.2*** −42.0***

Two-year college

Any college 5.6 8.4 8.5 0.1 2.8 2.9 AA degree 4.7 8.3 8.7 0.4 3.6 4.0 Some BA 3.7 7.6 9.0 1.4 3.9 5.3 BA degree 2.7 7.2 8.6 1.4 4.5 5.9

Any college

Any college 59.9 39.4 28.7 −10.7*** −20.5*** −31.2*** AA degree 63.4 44.8 30.6 −14.2*** −18.6*** −32.8*** Some BA 66.4 49.7 33.2 −16.5*** −16.7*** −33.2*** BA degree 70.6 56.9 34.5 −22.4*** −13.7*** −36.1***

Note. Sample includes all 10th graders who live with two parents and have complete data on the variables used in subsequent statistical models (n ~ 7,800). Data are weighted using survey weights for participation in 10th grade. Parents include biological, step, adopted, and foster parents. Education level of a college-educated parent: any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree. ***p < .001.

498 Educational Policy 35(3)

the definition of college educated. Finally, we noted no statistically sig- nificant difference in 2-year college completion rates for students with different numbers of college-educated parents, and this result held across definitions of college-educated parent.

The results from the multinomial logistic regression model for the full (10th-grade) sample are shown in Table 3. Recall that the dependent variable has three possible values: (a) highest degree earned was at least a bachelor’s degree, (b) highest degree earned was an associate’s degree, and (c) did not earn either a bachelor’s or an associate’s degree. As noted earlier, the values in the table represent the marginal effect of each variable on the college com- pletion alternatives. Collectively, the independent variables accounted for approximately 25% of the variation in college completion outcomes.

Beginning with parental education, we found that 10th-grade students with at least one parent who did not have more than a high school education were significantly less likely than students with two college-educated parents to have earned a bachelor’s degree, even after controlling for a range of per- sonal, family, and school characteristics. The differentials were larger for stu- dents who had two non-college-educated parents than they were for those with exactly one non-college-educated parent. However, even those with one college-educated parent completed college at lower rates than their peers with two college-educated parents.

Although not the focus of our study, the findings among several non- parental education variables are also worth noting. Female students were more likely than comparable male students to earn a college degree. Black students were 6.6% more likely than similar White students to earn a 4-year degree. Not surprisingly, academic performance and aptitude in high school was positively and strongly associated with a student’s chance of earning a bachelor’s degree. Family characteristics also seemed to mat- ter for 10th graders, in that those who lived with both biological parents and/or resided in higher income families were more likely to complete 4-year degrees.

In Table 4, we present the results for the 12 alternative parental education variables for the full 10th-grade sample. Recall that the first four rows (zero college-educated parents) and the last four rows (zero or one college-edu- cated parents) correspond to definitions that could be used for “first-genera- tion college students.” In every instance, we found that students without two college-educated parents were significantly less likely than their counter- parts to earn a 4-year degree. The estimated marginal effects, however, var- ied considerably by definition from a low of 5.4% to a high of 11.4%. This also means that students with only one college-educated parent were at greater risk of not graduating from college relative to their peers with two

Toutkoushian et al. 499

Table 3. Multinomial Logit Models for College Graduation—All 10th Graders.

Variable

(1) (2) (3)

No graduation Two-year grad Four-year grad

Both parents: HS 0.082*** 0.009 −0.091*** (0.016) (0.011) (0.015)

One parent: HS 0.055*** 0.009 −0.064*** (0.015) (0.010) (0.014)

Female −0.039** 0.019* 0.020†

(0.012) (0.008) (0.011) Black −0.070** 0.004 0.066**

(0.022) (0.014) (0.021) Asian −0.036 −0.018 0.055*

(0.031) (0.019) (0.026) All other races −0.022 0.016 0.006

(0.021) (0.015) (0.021) Hispanic 0.021 0.000 −0.021

(0.024) (0.015) (0.022) High school GPA −0.261*** −0.015* 0.276***

(0.010) (0.007) (0.010) Score: Math −0.047*** 0.005 0.042***

(0.012) (0.008) (0.010) Score: Reading −0.012 −0.008 0.020*

(0.010) (0.007) (0.010) Live w/biological parents −0.065*** 0.002 0.063***

(0.015) (0.010) (0.014) Number of siblings 0.003 0.002 −0.006

(0.005) (0.003) (0.005) Number of dependents 0.009† −0.005 −0.005

(0.005) (0.003) (0.005) Income below US$20K 0.043 −0.013 −0.030

(0.027) (0.018) (0.026) Income US$20K to US$50K 0.056*** −0.008 −0.048***

(0.014) (0.009) (0.013) Income above US$100K −0.015 −0.036* 0.051**

(0.021) (0.015) (0.017) Enrollments: Grade 10 −0.008* −0.000 0.008**

(0.004) (0.003) (0.003) School: Urban −0.023 −0.027† 0.051**

(0.021) (0.015) (0.018)

(continued)

500 Educational Policy 35(3)

college-educated parents. The results were mixed with regard to the connec- tion between parental education and completion of a 2-year degree, with roughly half of the models showing a positive and significant association

Variable

(1) (2) (3)

No graduation Two-year grad Four-year grad

School: Suburban 0.004 0.001 −0.005 (0.017) (0.011) (0.015)

School: Pct free lunch 0.001† −0.000† −0.000 (0.000) (0.000) (0.000)

School: Pct AP −0.000 −0.000 0.001†

(0.001) (0.000) (0.000) School: Public 0.062** −0.001 −0.061***

(0.020) (0.014) (0.017) School: New England −0.069† −0.041 0.111***

(0.037) (0.029) (0.026) School: Mid Atlantic −0.119*** 0.014 0.105***

(0.025) (0.017) (0.023) School: E North Central −0.045† −0.005 0.050*

(0.023) (0.017) (0.020) School: W North Central −0.030 0.019 0.011

(0.027) (0.018) (0.024) School: South Atlantic −0.057* 0.009 0.048*

(0.024) (0.018) (0.021) School: E South Central −0.027 −0.007 0.033

(0.028) (0.020) (0.026) School: W South Central −0.009 0.003 0.006

(0.024) (0.018) (0.023) School: Mountain 0.037 0.014 −0.052†

(0.031) (0.019) (0.027) Pseudo R2 = .25 Chi-square = 1,755.16***

Note. Sample size ~ 7,800. Coefficients are reported as average marginal effects. Standard errors are shown in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Reference category for race is White. Reference category for income is US$50K to US$100K. Models also include variables for missing income, ethnicity, siblings, dependents, family income, school free lunch percentage, school AP, and race. HS = high school; GPA = grade point average; Pct = percentage; AP = Advanced Placement. †p < .10. *p < .05. **p < .01. ***p < .001.

Table 3. (continued)

Toutkoushian et al. 501

Table 4. Effects of First-Generation College Status on College Graduation—10th Graders.

No. of college- educated parentsa

Definition of college-educated

parentb

Outcome:

No college

Two-year grad

Four-year grad

Zero Any college 0.082*** (0.016)

0.009 (0.011)

−0.091*** (0.015)

AA degree 0.070*** (0.015)

0.026* (0.011)

−0.097*** (0.014)

Some BA 0.057*** (0.016)

0.041*** (0.012)

−0.098*** (0.014)

BA degree 0.073*** (0.019)

0.042** (0.016)

−0.114*** (0.016)

One Any college 0.055*** (0.015)

0.009 (0.010)

−0.064*** (0.014)

AA degree 0.031* (0.016)

0.024* (0.010)

−0.054*** (0.014)

Some BA 0.028 (0.018)

0.035** (0.013)

−0.063*** (0.015)

BA degree 0.016 (0.021)

0.042* (0.016)

−0.058*** (0.016)

Zero or one Any college 0.065*** (0.013)

0.009 (0.009)

−0.074*** (0.012)

AA degree 0.049*** (0.013)

0.025** (0.009)

−0.073*** (0.012)

Some BA 0.043** (0.015)

0.038*** (0.011)

−0.081*** (0.013)

BA degree 0.048** (0.018)

0.041** (0.015)

−0.089*** (0.015)

Note. Marginal effects for the alternative first-generation college student variables. Standard errors are in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Each model also controls for gender, race, GPA, standardized test score, number of siblings and dependents, family status and income, Grade 10 enrollments, and school attributes. GPA = grade point average. aParents include biological, step, adopted, and foster parents (n ~ 7,300). bEducation level of a college-educated parent: any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree. †p < .10. *p < .05. **p < .01. ***p < .001.

502 Educational Policy 35(3)

and the rest having insignificant marginal effects. As a result, in most mod- els FGCSs were more likely to not earn any type of postsecondary credential.

Our next step was to reestimate the multinomial logistic regression models for only those students who enrolled in college. Accordingly, the marginal effects show the relationship between each factor and college completion conditional on the student enrolling in college. College enroll- ees represent approximately 80% of the sample of 10th graders in our dataset. The complete results are shown in Table 5 for the case where we defined a non-college-educated parent as someone who never enrolled in college at any level. We found that students without two college-educated parents were still significantly less likely than their peers to earn a bache- lor’s degree. The marginal effects for students with exactly one college- educated parent in Table 5 were virtually the same as for the full sample (Table 3). This finding means that the disadvantage these students face in terms of completion is driven mostly by their being less likely to graduate once they enrolled in college. In contrast, the marginal effects for students with two non-college-educated parents were negative and significant for only college enrollees, but these effects were several percentage points smaller than for the full sample of 10th graders. For them, the disadvan- tage in terms of college completion is thus driven by both access and com- pletion conditional on enrollment.

Table 6 is structured in the same way as Table 4, where we present only the results for the 12 parental education variables across multiple ways of defin- ing this construct. Overall, the results were very similar to what we observed for the full sample, except that the marginal effects in the conditional models in Table 5 were about one percentage point smaller than for the full sample. As before, the marginal effects for FGCSs (first four and last four rows) var- ied substantially across definitions, further emphasizing the importance of being specific about how one defines FGCSs in empirical studies. Regardless, the results show that overall FGCSs were less likely than non-FGCSs to earn a 4-year degree, even among those who had successfully navigated the col- lege search process and enrolled in college. Across many definitions, stu- dents with exactly one college-educated parent were less likely to graduate than were students with two college-educated parents. It is also worth noting that out of those students who enrolled in college, FGCSs were often found to be more likely than non-FGCS students to earn a 2-year degree depending on how we defined FGCSs.

In Table 7, we repeated the analysis in Table 6 after further limiting the sample to only those students who reported that their first postsecondary institution attended was a 4-year college. Overall, the results in Table 7 paral- lel what we found in Table 6, which in large part reflects the fact that the

Toutkoushian et al. 503

Table 5. Multinomial Logit Models for College Graduation—All College Enrollees.

Variable

(1) (2) (3)

No graduation Two-year grad Four-year grad

Both parents: HS 0.050* 0.018 −0.067*** (0.020) (0.012) (0.019)

One parent: HS 0.055** 0.012 −0.067*** (0.018) (0.012) (0.017)

Female −0.036* 0.021* 0.016 (0.014) (0.010) (0.014)

Black −0.028 0.003 0.025 (0.027) (0.017) (0.025)

Asian −0.004 −0.028 0.032 (0.033) (0.022) (0.030)

All other races −0.037 0.028 0.010 (0.028) (0.018) (0.027)

Hispanic 0.046 −0.017 −0.029 (0.030) (0.017) (0.027)

GPA high school −0.236*** −0.030*** 0.266*** (0.013) (0.008) (0.013)

Score: Math −0.031* −0.001 0.033* (0.015) (0.008) (0.013)

Score: Reading 0.001 −0.012 0.012 (0.012) (0.008) (0.012)

Live w/biological parents −0.071*** 0.005 0.066*** (0.017) (0.011) (0.017)

Number of siblings −0.003 0.003 −0.000 (0.006) (0.004) (0.006)

Number of dependents 0.017* −0.005 −0.011†

(0.007) (0.004) (0.006) Income below US$20K 0.003 0.005 −0.008

(0.034) (0.022) (0.032) Income US$20K to US$50K 0.045** −0.002 −0.043*

(0.017) (0.011) (0.017) Income above US$100K −0.016 −0.042* 0.058**

(0.023) (0.018) (0.020) Enrollments: Grade 10 −0.008† −0.002 0.010**

(0.005) (0.003) (0.004) School: Urban −0.026 −0.029† 0.055**

(0.025) (0.017) (0.021) School: Suburban 0.005 −0.001 −0.004

(0.020) (0.013) (0.017) School: Pct free lunch 0.001 −0.000† −0.000

(0.000) (0.000) (0.000) School: Pct AP −0.000 −0.001 0.001†

(0.001) (0.000) (0.001) School: Public 0.054* 0.004 −0.058**

(0.022) (0.016) (0.020)

(continued)

504 Educational Policy 35(3)

Variable

(1) (2) (3)

No graduation Two-year grad Four-year grad

School: New England −0.061 −0.030 0.091** (0.045) (0.032) (0.031)

School: Mid Atlantic −0.104*** 0.013 0.091*** (0.029) (0.021) (0.027)

School: E North Central −0.028 −0.012 0.040†

(0.027) (0.021) (0.023) School: W North Central −0.008 0.020 −0.012

(0.031) (0.022) (0.026) School: South Atlantic −0.044† 0.004 0.040†

(0.027) (0.020) (0.023) School: E South Central −0.010 −0.008 0.019

(0.032) (0.024) (0.030) School: W South Central −0.004 0.007 −0.004

(0.029) (0.020) (0.027) School: Mountain 0.037 0.017 −0.054†

(0.037) (0.022) (0.032) Enrolled in College FT −0.183*** 0.003 0.179***

(0.022) (0.012) (0.024) First attend public college −0.089* −0.043** 0.132***

(0.040) (0.016) (0.040) First attend private NP college −0.070 −0.095*** 0.165***

(0.044) (0.022) (0.042) Planned STEM major −0.033† 0.015 0.018

(0.019) (0.013) (0.018) Planned business major −0.055** −0.003 0.058**

(0.020) (0.013) (0.018) Pay for college w/grants −0.033* −0.006 0.039**

(0.015) (0.010) (0.014) Pay for college w/loans −0.021 −0.028** 0.049***

(0.015) (0.010) (0.013) Pseudo R2 = .20 Chi-square = 1,370.43***

Note. Sample size ~ 6,300. Coefficients are reported as average marginal effects. Standard errors are shown in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Reference category for race is White. Reference category for income is US$50K to US$100K. Models also include variables for missing income, ethnicity, siblings, dependents, family income, school free lunch percentage, school AP, and race. HS = high school; GPA = grade point average; Pct = percentage; AP = Advanced Placement; FT = full-time; NP = not-for-profit; STEM = science, technology, engineering and mathematics. †p < .10. *p < .05. **p < .01. ***p < .001.

Table 5. (continued)

Toutkoushian et al. 505

Table 6. Effects of First-Generation College Status on College Graduation—All College Enroll.

No. of college- educated parentsa

Definition of college-educated

parentb

Outcome:

No college

Two-year grad

Four-year grad

Zero Any college 0.050* (0.020)

0.018 (0.012)

−0.067*** (0.019)

AA degree 0.047** (0.018)

0.033** (0.012)

−0.080*** (0.017)

Some BA 0.039* (0.018)

0.049*** (0.014)

−0.087*** (0.017)

BA degree 0.062** (0.021)

0.047** (0.018)

−0.110*** (0.019)

One Any college 0.055** (0.018)

0.012 (0.012)

−0.067*** (0.017)

AA degree 0.034† (0.018)

0.020† (0.011)

−0.054** (0.017)

Some BA 0.035† (0.020)

0.033* (0.014)

−0.067*** (0.018)

BA degree 0.019 (0.023)

0.045* (0.019)

−0.064*** (0.019)

Zero or one Any college 0.053*** (0.015)

0.014 (0.010)

−0.067*** (0.015)

AA degree 0.039** (0.015)

0.026* (0.010)

−0.065*** (0.015)

Some BA 0.036* (0.017)

0.041** (0.013)

−0.077*** (0.016)

BA degree 0.042* (0.020)

0.046** (0.017)

−0.088*** (0.017)

Note. Sample only includes students who enrolled in college (~6,300). Marginal effects for the alternative first-generation college student variables. Standard errors are in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Each model also controls for gender, race, GPA, standardized test score, number of siblings and dependents, family status and income, Grade 10 enrollments, school attributes, and college attributes. GPA = grade point average. aParents include biological, step, adopted, and foster parents. bEducation level of a college-educated parent: any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree. †p < .10. *p < .05. **p < .01. ***p < .001.

506 Educational Policy 35(3)

Table 7. Effects of First-Generation College Status on College Graduation—4- Year Enroll.

No. of college- educated parentsa

Definition of college-educated

parentb

Outcome:

No college

Two-year grad

Four-year grad

Zero Any college 0.055* (0.025)

0.010 (0.014)

−0.066** (0.025)

AA degree 0.031 (0.023)

0.027* (0.012)

−0.058* (0.024)

Some BA 0.030 (0.021)

0.039*** (0.012)

−0.069** (0.022)

BA degree 0.049* (0.023)

0.052*** (0.015)

−0.101*** (0.024)

One Any college 0.056** (0.020)

0.012 (0.011)

−0.068** (0.021)

AA degree 0.023 (0.021)

0.025* (0.012)

−0.048* (0.021)

Some BA 0.040† (0.022)

0.025† (0.013)

−0.065** (0.022)

BA degree 0.041† (0.023)

0.040* (0.016)

−0.081*** (0.022)

Zero or one Any college 0.056** (0.018)

0.011 (0.010)

−0.067*** (0.019)

AA degree 0.026 (0.018)

0.026** (0.010)

−0.052** (0.019)

Some BA 0.035† (0.018)

0.032** (0.011)

−0.067*** (0.019)

BA degree 0.045* (0.020)

0.046** (0.014)

−0.091*** (0.020)

Note. Sample includes students who initially enrolled in a 4-year institution (~3,900). Marginal effects for the alternative first-generation college student variables. Standard errors are in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Each model also controls for gender, race, GPA, standardized test score, number of siblings and dependents, family status and income, Grade 10 enrollments, school attributes, and college attributes. GPA = grade point average. aParents include biological, step, adopted, and foster parents. bEducation level of a college-educated parent: any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree. †p < .10. *p < .05. **p < .01. ***p < .001.

Toutkoushian et al. 507

majority of students in this dataset who enrolled in college chose a 4-year institution. Regardless of how we defined parental education, students with- out two college-educated parents were less likely than other students to earn a bachelor’s degree.

Finally, we estimated the multinomial logistic regression model for only those students who initially enrolled at a 2-year institution (Table 8). Unlike the 4-year sector, for this subset of students we found relatively few differences between FGCSs and non-FGCSs that were statistically signifi- cant. FGCSs were equally likely to not earn any degree or to earn a 2-year degree. Depending on how FGCSs was defined, however, we did note sev- eral instances where FGCSs who began at a 2-year institution were less likely than their non-FGCS counterparts to earn a 4-year degree.

Discussion

Researchers have long been interested in the role of parental education in their children’s postsecondary outcomes, especially for FGCSs. However, definitions of FGCSs have varied across studies, limiting our understand- ing of how and why outcomes vary for different students. The primary goal of this study was to examine whether the relationship between FGCSs and college graduation differed with different definitions of FGCSs. We created a series of alternative definitions of FGCSs based on the level of parental education and the number of parents with this level of education. In addition, we were interested in whether results differed for FGCSs and non-FGCSs beginning in the 2-year versus 4-year sector and whether differences persisted even among those who successfully enrolled in college.

We found that FGCSs were less likely to graduate from a 4-year col- lege than were non-FGCSs across most definitions; however, the magni- tude of the effect varied with each definition. In general, the likelihood of 4-year college graduation increased along with the level of parent educa- tion and the number of parents at that level. These results are consistent with prior research on FGCSs, but provide additional insight into the out- comes of students who fall somewhere in the middle of the spectrum of zero parents attending college and two parents with bachelor’s degrees. It is important to note that students with one college-educated parent also face challenges in completing a college education. They are often over- looked, however, in terms of support and attention because they are not typically included in definitions of FGCSs.

The lower likelihood of graduation for students whose parents have less than bachelor’s degrees indicates an important difference between

508 Educational Policy 35(3)

Table 8. Effects of First-Generation College Status on College Graduation—2- Year Enroll.

No. of college- educated parentsa

Definition of college-educated

parentb

Outcome:

No college

Two- year grad

Four- year grad

Zero Any college 0.025 (0.032)

0.024 (0.029)

−0.050† (0.030)

AA degree 0.024 (0.032)

0.040 (0.031)

−0.064* (0.028)

Some BA 0.020 (0.037)

0.054 (0.036)

−0.073* (0.032)

BA degree 0.087† (0.047)

0.000 (0.046)

−0.087* (0.038)

One Any college 0.042 (0.033)

0.001 (0.029)

−0.043 (0.029)

AA degree 0.036 (0.035)

0.007 (0.031)

−0.043 (0.031)

Some BA 0.026 (0.040)

0.033 (0.039)

−0.060† (0.034)

BA degree 0.000 (0.051)

0.029 (0.049)

−0.029 (0.041)

Zero or one Any college 0.035 (0.027)

0.011 (0.025)

−0.046† (0.024)

AA degree 0.029 (0.030)

0.025 (0.028)

−0.054* (0.025)

Some BA 0.022 (0.035)

0.046 (0.035)

−0.068* (0.030)

BA degree 0.058 (0.046)

0.009 (0.045)

−0.067† (0.037)

Note. Sample includes students who initially enrolled in 2-year institutions (~1,600). Marginal effects for the alternative first-generation college student variables. Standard errors are in parentheses and are clustered at the school level. Data are weighted using survey weights for participation in 10th grade. Each model also controls for gender, race, GPA, standardized test score, number of siblings and dependents, family status and income, Grade 10 enrollments, school attributes, and college attributes. GPA = grade point average. aParents include biological, step, adopted, and foster parents. bEducation level of a college-educated parent: any college = parent(s) attended any postsecondary institution; AA degree = parent(s) earned an associate’s degree or attended a 4-year institution; some BA = parent(s) enrolled in a 4-year institution (does not include enrolling in or completing an associate’s degree); BA degree = parent(s) earned a bachelor’s degree. †p < .10. *p < .05. **p < .01. ***p < .001.

Toutkoushian et al. 509

students whose parents attended but did not complete college and those whose parents completed college. Definitions of FGCSs often use college attendance as the threshold, which our results suggest may be problematic. Perhaps parents who complete bachelor’s degrees better understand the full range of benefits of a college degree, and can pass this on to their chil- dren. Their children may also benefit from parents’ greater social, cultural, and human capital that can translate to resources to support their children in attending and completing college. Their understanding of the benefits of completing college and access to the resources necessary to support col- lege completion may also encourage and enable their children to devote more time and effort to college, which researchers have identified as an important factor in persistence (Mayhew et al., 2016; Tinto, 1993). On the contrary, FGCSs, who may lack this social and cultural capital even though many campuses assume students possess it, might be less academically and socially integrated into college, which according to theory can lead to higher student attrition (Rosenbaum et al., 2006; Tinto, 1993).

There was relatively little prior research on the outcomes of FGCSs who initially enrolled in 2-year colleges, which was surprising given that FGCSs enroll in 2-year colleges at higher rates (Redford & Hover, 2017). We found that, under some definitions, FGCSs were slightly more likely to graduate from 2-year colleges; under other definitions, there are no dif- ferences in 2-year graduation rates for FGCSs and non-FGCSs. Many students who enroll at community colleges, including FGCSs, may view the community college as their only viable option for higher education due to location, family and community obligations and expectations, lower levels of academic preparation, or concerns about cost (Cohen, Brawer, & Kisker, 2014; Rosenbaum et al., 2006). Such students may necessarily stop at the associate’s degree, thus accounting for the spike in associate’s degree attainment among some FGCS students relative to non- FGCS students. As our definitions shifted to include more highly edu- cated parents, students may have faced fewer of these constraints and been more likely to pursue a bachelor’s degree, eliminating the FGCS edge in associate’s degree attainment.

When limiting the sample to only students who initially enrolled in 2-year college, there was no difference between FGCSs and non-FGCSs in the likelihood of graduating from a 2-year college. We also noted that hav- ing a parent with a 2-year degree did not increase the likelihood of a stu- dent completing a 2-year degree even though having a parent with a 4-year degree increases the likelihood of completing a 4-year degree. These dif- ferences between 2-year enrollment and attainment and 4-year enrollment and attainment may reflect the differences between 2- and 4-year students.

510 Educational Policy 35(3)

While we controlled for many key differences, such as family income and race/ethnicity, we were not able to control for all factors. For example, 2-year college students tend to have a more diverse set of educational goals, are more likely to be unsure of their educational goals, and are more likely to have chosen between a 2-year college and no higher education (Cohen et al., 2014; Rosenbaum et al., 2006). These differences may explain why non-FGCSs have no advantage in graduation among students who initially enroll in 2-year college. Additional research is necessary to more fully understand the experiences of FGCSs on community college campuses. However, regardless of the reason, it is clear that employing different definitions of FGCSs can lead to significantly different results for students attending and graduating from 2-year colleges in particular.

FGCS effects were largely similar for both a sample of all 10th graders and a sample of only those who successfully enrolled in college. FGCSs who enrolled in college were less likely than non-FGCSs to graduate from a 4-year college, suggesting that FGCSs remain at a disadvantage even after surpassing the access to college hurdle. Again, perhaps college-edu- cated parents are more likely or better able to help their children accli- mate to the college environment or take advantage of campus resources, and thus persist to completion. They are also likely in a better position to communicate the benefits of completing a college degree, which may also encourage persistence.

Implications

These findings also have implications for policy and practice. Colleges, government agencies, and nonprofit organizations sponsor many programs to support college enrollment and completion of historically underserved students, including FGCS students. When participation is conditional upon FGCSs, these organizations must carefully consider an appropriate defini- tion. Limiting services to only students whose parents had no postsecond- ary experience would overlook students who, according to our findings, may experience difficulties and could benefit from such services. Based on the findings of this study, we suggest that more policies and programs use the federal definition of FGCSs, which TRIO programs have always uti- lized. This more inclusive definition will broaden our conception of an FGCSs and expand their access to key services and programs.

If institutions and policymakers have varying definitions, it is also challenging for students to know if they should personally identify as a FGCS. For example, a student might not apply for TRIO programs, such as Student Support Services or McNair Scholars, because their parent(s)

Toutkoushian et al. 511

attended some college (did not graduate) and thus they do not perceive themselves to be a FGCS. However, the TRIO programs use the federal definition and serve students whose parents have not earned a bachelor’s. Definitions of FGCSs are also especially important in community college contexts given our finding that students of one or two parents with associ- ate’s degrees were still less likely to complete either an associate’s or bachelor’s degree. When the definition of first-generation status is not clearly defined, it is plausible that students might miss out on services that are dependent on FGCS status.

In addition to streamlining the definition of FGCSs, institutions of higher education need to recognize their role in FGCS success. Our find- ings indicate that this population is at a unique disadvantage net of the highly correlated financial disadvantage, and higher education profes- sionals need to provide the necessary support to help FGCS persist and graduate. Tinto’s (2012) Model of Institutional Action acknowledges the role of pre-entry attributes (e.g., parent education level), but focuses on what the institution can do to improve student success outcomes. A uni- versity with a strong institutional commitment to student success “sets the tone for the exceptional climate for success that students encounter their everyday interactions with the institution, its policies, practices, and vari- ous members (faculty, staff, administrators, and other students)” (Tinto, 2012, p. 259). Institutions should make a strong commitment to provide services and support for students with lower levels of parental education and invest in their success in college.

This study also has several implications for future research. Currently, researchers are using a variety of different definitions of FGCSs. This variation highlights the importance of carefully considering the most appropriate definition of FGCSs for a given study, and clearly articulating this definition. Researchers must consider who is counted as a parent, the threshold at which a parent is considered “college-educated,” and the number of parents who are “college-educated.” When possible, research- ers should determine whether their findings are robust with regard to dif- ferent definitions. They might also consider a multilevel construct or distinguishing by the number of parents who attained a particular level of education. Furthermore, qualitative research can examine the experiences of students with various levels of parental education, to better understand the social and cultural capital that is passed down through parents, and how this shapes postsecondary experiences and outcomes.

One potential limitation of this study is that our sample is limited to students with two parents. More research is needed to examine students, including FGCSs, from single-parent households, as this could form an

512 Educational Policy 35(3)

additional barrier to successful educational outcomes. Furthermore, future studies should examine postsecondary outcomes for students from nontraditional family structures. While our study included biological, step, adopted, and foster parents, we did not have sufficient numbers of observations in these groups to examine whether differences exist based on who is counted as a parent. For example, do FGCSs living with foster parents have different outcomes than FGCSs living with biological par- ents? Due to the ever-evolving nature of families, this line of inquiry will be important in the future.

Given that many FGCSs are also from low-income backgrounds or students of color, future research should employ data enabling examina- tion of the combined effects of these multiple characteristics. Our study found that the group of students commonly considered “FGCSs” is much more diverse than previously thought and that such diversity has implica- tions for their outcomes. It would not be surprising to also find differ- ences in outcomes among FGCSs from different racial/ethnic and family income backgrounds. Other researchers have also called for an increasing understanding of the effects of multiple intersecting identities (Deil- Amen, 2015; Nguyen & Nguyen, 2018). Although our study was primar- ily interested in isolating as best as possible the effect of FGCS status net of these other characteristics and data limitations precluded analysis of interaction variables, studying these intersections can further enhance our understanding of FGCS experiences and outcomes.

It is important that researchers continue this line of inquiry to deter- mine why parental education matters and whether FGCS status affects longer term outcomes. While quantitative methods allow researchers to rule out certain variables, future research can help us further clarify why FGCS matters to college completion. However, this study and other quan- titative work will not be able to fully understand why this group is at a disadvantage. Qualitative work might be helpful for better understanding the experiences of college students with varying levels of parental educa- tion to better inform practice.

Still, our study provides important evidence about the relationship between parental education level and increased college completion rates. Students with slightly more parental education have slightly more capital in navigating the college completion process. Future research should examine the relationship between parental education on outcomes beyond a bacca- laureate, such as graduate school enrollment and persistence, and labor market outcomes. It would also be worthwhile to examine career choices of FGCSs and whether they select different types of professions. Through these future studies, we can better unpack the ways in which parental edu- cation shapes the education and subsequent careers of their children.

513

V ar

ia bl

e D

es cr

ip ti o ns

.

V ar

ia bl

e na

m e

EL S:

02 v

ar ia

bl e

D es

cr ip

ti o n

G ra

d 4-

ye ar

F3 IC

R ED

1 if

hi gh

es t

de gr

ee e

ar ne

d ac

ro ss

a ll

re po

rt ed

h ig

he r

ed uc

at io

n cr

ed en

ti al

s w

as a

4 -y

ea r

de gr

ee b

y 20

12 G

ra d

2- ye

ar F3

IC R

ED 1

if hi

gh es

t de

gr ee

e ar

ne d

ac ro

ss a

ll re

po rt

ed h

ig he

r ed

uc at

io n

cr ed

en ti al

s w

as a

2 -y

ea r

de gr

ee b

y 20

12 G

ra d

an y

F3 IC

R ED

1 if

G ra

d 4-

ye ar

= 1

o r

G ra

d 2-

ye ar

= 1

En ro

ll F2

B 07

1 if

st ud

en t

en ro

lle d

in a

p o st

se co

nd ar

y in

st it ut

io n

as o

f se

co nd

f o llo

w -u

p su

rv ey

, 0 o

th er

w is

e En

ro ll2

Y r

F2 PS

1S EC

1 if

st ud

en t

fir st

e nr

o lle

d in

a 2

-y ea

r o r

le ss

t ha

n 2-

ye ar

p o st

se co

nd ar

y in

st it ut

io n,

0 o

th er

w is

e En

ro ll4

Y r

F2 PS

1S EC

1 if

st ud

en t

fir st

e nr

o lle

d in

a 4

-y ea

r po

st se

co nd

ar y

in st

it ut

io n,

0 o

th er

w is

e N

ei th

er p

ar en

t: A

ny c

o lle

ge B

Y P3

4A , B

Y P3

4B 1

if ed

uc at

io na

l a tt

ai nm

en t

fo r

bo th

p ar

en ts

w as

a t

m o st

a h

ig h

sc ho

o l d

ip lo

m a

o r

G ED

, 0 o

th er

w is

e N

ei th

er p

ar en

t: A

A d

eg re

e B

Y P3

4A , B

Y P3

4B 1

if ne

it he

r pa

re nt

e ar

ne d

an a

ss o ci

at e’

s de

gr ee

n o r

en ro

lle d

at a

4 -y

ea r

in st

it ut

io n,

0 o

th er

w is

e N

ei th

er p

ar en

t: S

o m

e B

A B

Y P3

4A , B

Y P3

4B 1

if ed

uc at

io na

l a tt

ai nm

en t

fo r

bo th

p ar

en ts

w as

a t

m o st

a n

as so

ci at

e’ s

de gr

ee , 0

o th

er w

is e

N ei

th er

p ar

en t: B

A d

eg re

e B

Y P3

4A , B

Y P3

4B 1

if ne

it he

r pa

re nt

e ar

ne d

a ba

ch el

o r’

s de

gr ee

, 0 o

th er

w is

e O

ne p

ar en

t: A

ny c

o lle

ge B

Y P3

4A , B

Y P3

4B 1

if ed

uc at

io na

l a tt

ai nm

en t

fo r

o nl

y o ne

p ar

en t

w as

a t

m o st

a h

ig h

sc ho

o l d

ip lo

m a

o r

G ED

, 0 o

th er

w is

e O

ne p

ar en

t: A

A d

eg re

e B

Y P3

4A , B

Y P3

4B 1

if o nl

y o ne

p ar

en t

di d

no t

ea rn

a n

as so

ci at

e’ s

de gr

ee n

o r

en ro

ll at

a 4

-y ea

r in

st it ut

io n,

0 o

th er

w is

e O

ne p

ar en

t: S

o m

e B

A B

Y P3

4A , B

Y P3

4B 1

if ed

uc at

io na

l a tt

ai nm

en t

fo r

o nl

y o ne

p ar

en t

w as

a t

m o st

a n

as so

ci at

e’ s

de gr

ee , 0

o th

er w

is e

O ne

p ar

en t: B

A d

eg re

e B

Y P3

4A , B

Y P3

4B 1

if o nl

y o ne

p ar

en t

di d

no t

ea rn

a b

ac he

lo r’

s de

gr ee

, 0 o

th er

w is

e Fe

m al

e B

Y S1

4 1

if se

x =

f em

al e,

0 o

th er

w is

e M

al e

B Y

S1 4

1 if

se x =

m al

e, 0

o th

er w

is e

[R ef

er en

ce ]

M is

si ng

: G en

de r

B Y

S1 4

1 if

se x =

m is

si ng

, 0 o

th er

w is

e W

hi te

B Y

S1 7A

1 if

ra ce

= o

nl y

W hi

te , 0

o th

er w

is e

[R ef

er en

ce ]

B la

ck B

Y S1

7B 1

if ra

ce =

o nl

y B

la ck

, 0 o

th er

w is

e A

si an

B Y

S1 7C

1 if

ra ce

= o

nl y

A si

an , 0

o th

er w

is e

H is

pa ni

c B

Y S1

5 1

if et

hn ic

it y =

H is

pa ni

c o r

La ti no

/a , 0

o th

er w

is e

A ll

o th

er r

ac es

B Y

S1 7D

, B Y

S1 7E

1 if

ra ce

= a

ll o th

er c

at eg

o ri

es (

in cl

ud es

m ul

ti pl

e ra

ce s)

, 0 o

th er

w is

e M

is si

ng : R

ac e

B Y

S1 7A

-E 1

if ra

ce =

m is

si ng

, 0 o

th er

w is

e

A p

p e n

d ix

( co

nt in

ue d)

514

V ar

ia bl

e na

m e

EL S:

02 v

ar ia

bl e

D es

cr ip

ti o n

M is

si ng

: E th

ni ci

ty B

Y S1

5 1

if et

hn ic

it y =

m is

si ng

, 0 o

th er

w is

e G

PA G

ra de

9 F1

R G

P9 Fr

es hm

an y

ea r

G PA

( m

is si

ng c

as es

d ro

pp ed

) Sc

o re

: M at

h B

Y PI

SA M

E Sc

o re

o n

m at

he m

at ic

s te

st a

dm in

is te

re d

by t

he P

IS A

in 2

00 3

(m is

si ng

c as

es d

ro pp

ed )

Sc o re

: R ea

di ng

B Y

PI SA

R E

Sc o re

o n

re ad

in g

te st

a dm

in is

te re

d by

t he

P IS

A in

2 00

3 (m

is si

ng c

as es

d ro

pp ed

) N

um be

r o f si

bl in

gs B

Y P0

8 N

um be

r o f si

bl in

gs a

s o f G

ra de

1 0.

I nc

lu de

s ad

o pt

iv e,

h al

f- a

nd s

te p-

br o th

er s

an d

si st

er s

M is

si ng

: N um

be r

o f si

bl in

gs B

Y P0

8 1

if nu

m be

r o f si

bl in

gs is

n o t

re po

rt ed

, 0 o

th er

w is

e N

um be

r o f de

pe nd

en ts

B Y

P0 6

N um

be r

o f pe

o pl

e w

ho a

re d

ep en

de nt

o n

th e

pa re

nt o

r sp

o us

e/ pa

rt ne

r (n

o t

co un

ti ng

p ar

en ts

) M

is si

ng : N

um be

r o f de

pe nd

en ts

B Y

P0 6

1 if

nu m

be r

o f de

pe nd

en ts

is n

o t

re po

rt ed

, 0 o

th er

w is

e In

co m

e be

lo w

U S$

20 K

B Y

P8 5

1 if

to ta

l f am

ily in

co m

e in

2 00

1 fr

o m

a ll

so ur

ce s

is U

S$ 20

,0 00

o r

le ss

, 0 o

th er

w is

e In

co m

e U

S$ 20

K t

o U

S$ 50

K B

Y P8

5 1

if to

ta l f

am ily

in co

m e

in 2

00 1

fr o m

a ll

so ur

ce s

is b

et w

ee n

U S$

20 ,0

01 a

nd U

S$ 50

,0 00

, 0 o

th er

w is

e In

co m

e U

S$ 50

K t

o U

S$ 10

0K B

Y P8

5 1

if to

ta l f

am ily

in co

m e

in 2

00 1

fr o m

a ll

so ur

ce s

is b

et w

ee n

U S$

50 ,0

01 a

nd U

S$ 10

0, 00

0, 0

o th

er w

is e

[R ef

er en

ce ]

In co

m e

ab o ve

U S$

10 0K

B Y

P8 5

1 if

to ta

l f am

ily in

co m

e in

2 00

1 fr

o m

a ll

so ur

ce s

is a

t le

as t

U S$

10 0,

00 1,

0 o

th er

w is

e M

is si

ng : I

nc o m

e B

Y P8

5 1

if to

ta l f

am ily

in co

m e

in 2

00 1

fr o m

a ll

so ur

ce s

is n

o t

re po

rt ed

, 0 o

th er

w is

e En

ro llm

en ts

: G ra

de 1

0 B

Y G

10 ER

N um

be r

o f st

ud en

ts in

t he

r es

po nd

en t’ s

10 th

g ra

de le

ve l (

in 1

00 s)

Sc ho

o l:

U rb

an B

Y U

R B

A N

1 if

sc ho

o l i

s lo

ca te

d in

a n

ur ba

n ar

ea , 0

o th

er w

is e

Sc ho

o l:

Su bu

rb an

B Y

U R

B A

N 1

if sc

ho o l i

s lo

ca te

d in

a s

ub ur

ba n

ar ea

, 0 o

th er

w is

e Sc

ho o l:

R ur

al B

Y U

R B

A N

1 if

sc ho

o l i

s lo

ca te

d in

a r

ur al

a re

a, 0

o th

er w

is e

[R ef

er en

ce ]

Sc ho

o l:

Pc t

fr ee

lu nc

h B

Y 10

FL P

Pe rc

en ta

ge o

f st

ud en

ts in

t he

h ig

h sc

ho o l w

ho r

ec ei

ve d

fr ee

-l un

ch s

er vi

ce s

(m id

po in

ts f o r

ca te

go ri

es u

se d

as

va lu

es )

M is

si ng

: P ct

f re

e lu

nc h

B Y

10 FL

P 1

if sc

ho o l i

s m

is si

ng d

at a

o n

pe rc

en t

o f st

ud en

ts r

ec ei

vi ng

f re

e lu

nc h,

0 o

th er

w is

e Sc

ho o l:

Pc t

A P

F1 A

22 F

Pe rc

en ta

ge o

f hi

gh s

ch o o l s

tu de

nt b

o dy

in A

P cl

as se

s M

is si

ng : P

ct A

P F1

A 22

F 1

if sc

ho o l i

s m

is si

ng d

at a

o n

pe rc

en t

o f st

ud en

ts t

ak in

g an

A P

ex am

, 0 o

th er

w is

e Sc

ho o l:

Pu bl

ic B

Y SC

T R

L 1

if sc

ho o l i

s pu

bl ic

, 0 o

th er

w is

e Sc

ho o l:

N ew

E ng

la nd

B Y

C EN

D IV

1 if

C en

su s

di vi

si o n

is N

ew E

ng la

nd (

C T

, M E,

M A

, N H

, R I,

V T

), 0

o th

er w

is e

Sc ho

o l:

M id

A tl an

ti c

B Y

C EN

D IV

1 if

C en

su s

di vi

si o n

is M

id dl

e A

tl an

ti c

(N J,

N Y

, P A

), 0

o th

er w

is e

Sc ho

o l:

E N

o rt

h C

en tr

al B

Y C

EN D

IV 1

if C

en su

s di

vi si

o n

is E

as t

N o rt

h C

en tr

al (

IL , I

N , M

I, O

H , W

I) , 0

o th

er w

is e

A p

p e n

d ix

( co

n ti

n u

e d

)

( co

nt in

ue d)

515

V ar

ia bl

e na

m e

EL S:

02 v

ar ia

bl e

D es

cr ip

ti o n

Sc ho

o l:

W N

o rt

h C

en tr

al B

Y C

EN D

IV 1

if C

en su

s di

vi si

o n

is W

es t

N o rt

h C

en tr

al (

IA , K

S, M

N , M

O , N

E, N

D , S

D ), 0

o th

er w

is e

Sc ho

o l:

So ut

h A

tl an

ti c

B Y

C EN

D IV

1 if

C en

su s

di vi

si o n

is S

o ut

h A

tl an

ti c

(D E,

D C

, F L,

G A

, M D

, N C

, S C

, V I,

W V

), 0

o th

er w

is e

Sc ho

o l:

E So

ut h

C en

tr al

B Y

C EN

D IV

1 if

C en

su s

di vi

si o n

is E

as t

So ut

h C

en tr

al (

A L,

K Y

, M S,

T N

), 0

o th

er w

is e

Sc ho

o l:

W S

o ut

h C

en tr

al B

Y C

EN D

IV 1

if C

en su

s di

vi si

o n

is W

es t

So ut

h C

en tr

al (

A R

, L A

, O K

, T X

), 0

o th

er w

is e

Sc ho

o l:

M o un

ta in

B Y

C EN

D IV

1 if

C en

su s

di vi

si o n

is M

o un

ta in

( A

Z , C

O , I

D , M

T , N

V , N

M , U

T , W

Y ), 0

o th

er w

is e

Sc ho

o l:

Pa ci

fic B

Y C

EN D

IV 1

if C

en su

s di

vi si

o n

is P

ac ifi

c (A

K , C

A , H

A , O

R , W

A ), 0

o th

er w

is e

[R ef

er en

ce ]

En ro

ll in

c o lle

ge F

T F2

PS 1F

T P

1 if

en ro

lle d

fu ll-

ti m

e at

f ir

st p

o st

se co

nd ar

y in

st it ut

io n

Fi rs

t at

te nd

p ub

lic c

o lle

ge F2

PS 1C

T R

1 if

fir st

in st

it ut

io n

at te

nd ed

w as

p ub

lic Fi

rs t

at te

nd p

ri va

te N

P co

lle ge

F2 PS

1C T

R 1

if fir

st in

st it ut

io n

at te

nd ed

w as

p ri

va te

n o t-

fo r-

pr o fit

Pl an

ne d

ST EM

m aj

o r

F2 B

15 1

if pl

an ne

d m

aj o r

w as

in a

S T

EM f ie

ld (

R es

po ns

es 4

, 5 , 6

, 7 )

Pl an

ne d

bu si

ne ss

m aj

o r

F2 B

15 1

if pl

an ne

d m

aj o r

w as

b us

in es

s (R

es po

ns e

1) Pa

y fo

r co

lle ge

w /g

ra nt

s F2

B 25

A 1

if st

ud en

t us

ed g

ra nt

s to

h el

p pa

y fo

r co

lle ge

Pa y

fo r

co lle

ge w

/lo an

s F2

B 25

B 1

if st

ud en

t us

ed lo

an s

to h

el p

pa y

fo r

co lle

ge

N ot

e. E

LS =

E du

ca ti

o n

Lo ng

it ud

in al

S tu

dy ; a

ny c

o lle

ge =

p ar

en t(

s) a

tt en

de d

an y

po st

se co

nd ar

y in

st it

ut io

n; A

A d

eg re

e =

p ar

en t(

s) e

ar ne

d an

a ss

o ci

at e’

s de

gr ee

o r

at te

nd ed

a 4

-y ea

r in

st it

ut io

n; s

o m

e B

A =

p ar

en t(

s) e

nr o

lle d

in a

4 -y

ea r

in st

it ut

io n

(d o

es n

o t

in cl

ud e

en ro

lli ng

in o

r co

m pl

et in

g an

a ss

o ci

at e’

s de

gr ee

); B

A d

eg re

e =

p ar

en t(

s) e

ar ne

d a

ba ch

el o

r’ s

de gr

ee ; G

PA =

g ra

de p

o in

t av

er ag

e; P

IS A

= P

ro gr

am f o

r In

te rn

at io

na l S

tu de

nt A

ss es

sm en

t; P

ct =

p er

ce nt

ag e;

A P =

A dv

an ce

d Pl

ac em

en t;

F T

= f ul

l- ti

m e;

N P =

n o

t- fo

r- pr

o fit

; S T

EM =

s ci

en ce

, t ec

hn o

lo gy

, e ng

in ee

ri ng

a nd

m at

he m

at ic

s.

A p

p e n

d ix

( co

n ti

n u

e d

)

516 Educational Policy 35(3)

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors received funding for this study from the National Science Foundation and the Association for Institutional Research grant no. RG14-5499.

ORCID iD

Ashley B. Clayton https://orcid.org/0000-0003-1800-4504

References

Adelman, C. (2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education.

Arbona, C., & Nora, A. (2007). The influence of academic and environmental factors on Hispanic college degree attainment. Review of Higher Education, 30, 247- 269.

Attinasi, L. C., Jr. (1989). Getting in: Mexican Americans’ perceptions of university attendance and the implications for freshman year persistence. The Journal of Higher Education, 60, 247-277.

Auclair, R., Bélanger, P., Doray, P., Gallien, M., Groleau, A., Mason, L., & Mercie, P. (2008). Transitions: Research Paper 2—First-generation students: A prom- ising concept? Montreal, Québec: Canada Millennium Scholarship Foundation. Retrieved from https://www.researchgate.net/publication/236592424_First- Generation_Students_A_Promising_Concept

Becker, G. (1993). Human capital: A theoretical and empirical analysis, with spe- cial reference to education (3rd ed.). Chicago, IL: The University of Chicago Press.

Berger, J. (2000). Optimizing capital, social reproduction, and undergraduate persis- tence. In J. M. Braxton (Ed.), Reworking the student departure puzzle (pp. 195- 224). Nashville, TN: Vanderbilt University Press.

Billson, J., & Terry, M. (1982). In search of the silken purse: Factors in attrition among first-generation students. College and University, 58, 57-75.

Bourdieu, P. (1977). Cultural reproduction and social reproduction. In J. Karabel & A. H. Halsey (Eds.), Power and ideology in education (pp. 487-510). New York, NY: Oxford University Press.

Bourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.), Handbook of the- ory and research for the sociology of education (pp. 241-258). New York, NY: Greenwood.

Braxton, J., Doyle, W., Hartley, H., III, Hirschy, A., Jones, A., & McLendon, M. (2013). Rethinking college student retention. San Francisco, CA: Jossey-Bass.

Toutkoushian et al. 517

Cabrera, A., Nora, A., & Castaneda, M. (1993). College persistence: Structural equa- tions modeling test of an integrated model of student retention. The Journal of Higher Education, 64, 123-139.

Chen, X., & Carroll, C. (2005). First-generation students in postsecondary education: A look at their college transcripts (NCES 2005-171). Washington, DC: National Center for Education Statistics.

Choy, S. (2001). Students whose parents did not go to college: Postsecondary access, persistence, and attainment (The Condition of Education 2001, pp. 18-43). Washington, DC: National Center for Education Statistics.

Cohen, A., Brawer, F., & Kisker, C. (2014). The American community college (6th ed.). San Francisco, CA: Jossey-Bass.

Coleman, J. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95-S120.

Cragg, K. (2009). Influencing the probability for graduation at four-year institutions: A multi-model analysis. Research in Higher Education, 50, 394-413.

D’Allegro, M., & Kerns, S. (2011). Is there such a thing as too much of a good thing when it comes to education? Reexamining first generation student success. The Journal of College Student Retention: Research, Theory & Practice, 12, 293-317.

D’Amico, M., & Dika, S. (2013). Using data known at the time of admission to predict first-generation college student success. Journal of College Student Retention: Research, Theory and Practice, 15, 173-192.

DeAngelo, L., & Franke, R. (2016). Social mobility and reproduction for whom? College readiness and first-year retention. American Educational Research Journal, 53, 1588-1625.

Deil-Amen, R. (2015). The traditional college student: A smaller and smaller minor- ity and its implications for diversity and access. In M. Stevens & M. Kirst (Eds.), Remaking college: The changing ecology of higher education (pp. 134-165). Palo Alto, CA: Stanford University Press.

Dowd, A. (2004). Income and financial aid effects on persistence and degree attain- ment in public colleges. Education Policy Analysis Archives, 12, 1-33.

Dumais, S., & Ward, A. (2010). Cultural capital and first-generation college success. Poetics, 38, 245-265.

Engle, J. (2007). Postsecondary access and success for first-generation college stu- dents. American Academic, 3, 25-48.

Engle, J., & Tinto, V. (2008). Moving beyond access: College success for low-income, first-generation students. Pell Institute for the Study of Opportunity in Higher Education. Retrieved from http://www.pellinstitute.org/publications-Moving_ Beyond_Access_2008.shtml

Field, J. (2017). Social capital. London, England: Routledge. Fike, D., & Fike, R. (2008). Predictors of first-year student retention in the commu-

nity college. Community College Review, 36, 68-88. Hahs-Vaughn, D. (2004). The impact of parents’ education level on college students:

An analysis using the beginning postsecondary students longitudinal study 1990- 92/94. Journal of College Student Development, 45, 483-500.

518 Educational Policy 35(3)

Hurtado, S., & Carter, D. (1997). Effects of college transition and perceptions of the campus racial climate on Latino students’ sense of belonging. Sociology of Education, 70, 324-345.

Ishitani, T. (2003). A longitudinal approach to assessing attrition behavior among first-generation students: Time-varying effects of pre-college characteristics. Research in Higher Education, 44, 433-449.

Ishitani, T. (2006). Studying attrition and degree completion behavior among first-generation college students in the United States. The Journal of Higher Education, 77, 861-885.

Ishitani, T. (2016). First-generation students’ persistence at four-year institutions. College and University, 91, 22-34.

Jaeger, M., & Karlson, K. (2015). Parental cultural capital investments and the production of social class differentials in educational attainment: A coun- terfactual analysis. In Abstract from ISA RC28 Spring Meeting, Tilburg, Netherlands, May.

Lin, N. (2002). Social capital: A theory of social structure and action (Vol. 19). Cambridge, UK: Cambridge University Press.

Lohfink, M., & Paulsen, M. (2005). Comparing the determinants of persistence for first-generation and continuing-generation students. Journal of College Student Development, 46, 409-428.

Long, B. (2007). The contributions of economics to the study of college access and success. Teachers College Record, 109, 2367-2443.

Longwell-Grice, R., & Longwell-Grice, H. (2008). Testing Tinto: How do reten- tion theories work for first-generation, working-class students? The Journal of College Student Retention: Research, Theory & Practice, 9, 407-420.

Lumina Foundation. (2017). Strategic plan for 2017 to 2020. Retrieved from https:// www.luminafoundation.org/files/resources/strategic-plan-2017-to-2020-apr17. pdf

Ma, J., Pender, M., & Welch, M. (2016). Education pays 2016: The benefits of higher education for individuals and society. The College Board. Retrieved from https:// trends.collegeboard.org/sites/default/files/education-pays-2016-full-report.pdf

Martinez, J., Sher, K., Krull, J., & Wood, P. (2009). Blue-collar scholars? Mediators and moderators of university attrition in first-generation college students. Journal of College Student Development, 50, 87-103.

Mayhew, M., Rockenbach, A., Bowman, N., Seifert, T., Wolniak, G., Pascarella, E., & Terenzini, P. (2016). How college affects students: Twenty-first century evi- dence that higher education works (Vol. 3). San Francisco, CA: Jossey-Bass.

Melguizo, T. (2011). A review of the theories developed to describe the process of persistence and attainment. In J. C. Smart (Ed.), Higher education: Handbook of theory and research (Vol. 24, pp. 321-358). Dordrecht, The Netherlands: Springer.

Mollegaard, S., & Jaeger, M. (2015). The effect of grandparents’ economic, cultural, and social capital on grandchildren’s educational success. Research in Social Stratification and Mobility, 42, 11-19.

Toutkoushian et al. 519

Moschetti, R., & Hudley, C. (2015). Social capital and academic motivation among first-generation community college students. Community College Journal of Research and Practice, 39, 235-251.

Museus, S. (2014). The Culturally Engaging Campus Environments (CECE) Model: A new theory of college success among racially diverse student populations. In M. B. Paulsen (Ed.), Higher education: Handbook of theory and research (Vol. 29, pp. 189-228). New York, NY: Springer.

Nguyen, T., & Nguyen, B. (2018). Is the “first-generation student” term useful for understanding inequality? The role of intersectionality in illuminating the implications of an accepted—yet unchallenged—term. Review of Research in Education, 42, 146-176.

Nunez, A., & Cuccaro-Alamin, S. (1998). First-generation students: Undergraduates whose parents never enrolled in postsecondary education (NCES 98-082). Washington, DC: National Center for Educational Statistics.

Padgett, R., Johnson, M., & Pascarella, E. (2012). First-generation undergraduate stu- dents and the impact of the first year of college: Additional evidence. Journal of College Student Development, 53, 243-266.

Pascarella, E., Pierson, C., Wolniak, G., & Terenzini, P. (2004). First-generation col- lege students: Additional evidence on college experiences and outcomes. The Journal of Higher Education, 75, 249-284.

Peralta, K. J., & Klonowski, M. (2017). Examining conceptual and operational defi- nitions of “first-generation college student” in research on retention. Journal of College Student Development, 58, 630-636.

Perna, L., & Thomas, S. (2008). Theoretical perspectives on student success: Understanding the contributions of the disciplines (ASHE Higher Education Report, Vol. 34, No. 1). Hoboken, NJ: Wiley.

Redford, J., & Hover, K. M. (2017). First-generation and continuing-generation college students: A comparison of high school and postsecondary experiences (NCES 2018-009). Washington, DC: National Center for Education Statistics, U.S. Department of Education.

Roksa, J. (2011). Differentiation and work: Inequality in degree attainment in us higher education. Higher Education, 61, 293-308.

Rosenbaum, J., Deil-Amen, R., & Person, A. (2006). After admission: From college access to college success. New York, NY: Russell Sage Foundation.

Spiegler, T., & Bednarek, A. (2013). First-generation students: What we ask, what we know and what it means: An international review of the state of research. International Studies in Sociology of Education, 23, 318-337.

Stuart, G., Rios-Aguilar, C., & Deil-Amen, R. (2014). “How much economic value does my credential have?” Reformulating Tinto’s model to study students’ per- sistence in community colleges. Community College Review, 42, 327-341.

Tan, C. (2017). Conceptual diversity, moderators, and the theoretical issues in quanti- tative studies of cultural capital theory. Educational Review, 69, 600-619.

Terenzini, P., Springer, L., Yaeger, P., Pascarella, E., & Nora, A. (1996). First- generation college students: Characteristics, experiences, and cognitive develop- ment. Research in Higher Education, 37, 1-22.

520 Educational Policy 35(3)

Tierney, W. (1991). Culture and ideology in higher education: Advancing a critical agenda. New York, NY: Praeger.

Tierney, W. (1992). Official encouragement, institutional discouragement: Minorities in academe—The Native American experience. Norwood, NJ: Ablex.

Tierney, W. (1993). The college experience of Native Americans: A critical anal- ysis. In L. Weis & M. Fine (Eds.), Beyond silenced voices: Class, race, and gender in United States schools (pp. 309-324). Ithaca: State University of New York Press.

Tinto, V. (1993). Leaving college: Rethinking the causes and cures of student attrition (2nd ed.). Chicago, IL: The University of Chicago Press.

Tinto, V. (2012). Moving from theory to action: A model for institutional action for student success. In A. Seidman (Ed.), College student retention: Formula for student success (pp. 251-266). Lanham, MD: Rowan & Littlefield.

Toutkoushian, R., Stollberg, R., & Slaton, K. (2018). Talking ’bout my genera- tion: Defining “first-generation college students” in higher education research. Teachers College Record, 120, 1-38.

U.S. Department of Education. (1998). Higher Education Act of 1965, 1998 Higher Education Act Amendments, Subpart 2—Federal Early Outreach and Student Services Programs (Chapter 1—Federal Trio Programs, SEC. 402A. 20 U.S.C. 1070a–11). Retrieved from https://www2.ed.gov/policy/highered/leg/hea98/ sec402.html

Vuong, M., Brown-Welty, S., & Tracz, S. (2010). The effects of self-efficacy on aca- demic success of first-generation college sophomore students. Journal of College Student Development, 51, 50-64.

Warburton, E., Bugarin, R., & Nunez, A. (2001). Bridging the gap: Academic prepa- ration and postsecondary success of first-generation students (NCES 2001-153). Washington, DC: National Center for Education Statistics.

Ward, L., Siegel, M., & Davenport, Z. (2012). First-generation college students: Understanding and improving the experience from recruitment to commence- ment. San Francisco, CA: John Wiley.

Whitley, S., Benson, G., & Wesaw, A. (2018). First-generation student success: A landscape of analysis of programs and services at four-year institutions. NASPA—Student Affairs Administrators in Higher Education. Retrieved from https://firstgen.naspa.org/2018-landscape-analysis.

Wolniak, G., Mayhew, M., & Engberg, M. (2012). Learning’s weak link to persis- tence. The Journal of Higher Education, 83, 795-823.

Author Biographies

Robert K. Toutkoushian is a Professor of higher education in the Institute of Higher Education at the University of Georgia. His research focuses on the application of economic models and theories to issues in higher education. He is also the coauthor of the recent book Economics of Higher Education: Background, Concepts, and Applications (Springer, 2016).

Toutkoushian et al. 521

Jennifer A. May-Trifiletti is a PhD Student in the Institute of Higher Education at the University of Georgia. Her research examines policies and practices affecting student access, equity, and success with a focus on community college students.

Ashley B. Clayton is an Assistant Professor of higher education in the School of Education at Louisiana State University. Her research examines the barriers, systems, and policies that aid or inhibit postsecondary access and success for underserved stu- dent populations.