Social Mobility and Reproduction for Whom? College Readiness and First-Year Retention

Abstract

Completing college is now the minimum threshold for entry into the middle class. This has pushed college readiness issues to the forefront in efforts to increase educational attainment. Little is known about how college readiness improves outcomes for students traditionally marginalized in educational settings or if social background factors continue to impact students in the same way during college regardless of readiness. Examining first-year college retention using a nationally representative data set, this study asks if social background factors and financial resources for college differentially impact students based on readiness. Findings indicate that academic readiness matters and that parental income and college generation status differentially affect first-year college retention for less-ready students but not college-ready students. Students who begin college less prepared academically are also more disadvantaged than college-ready students by the funding sources they have for college. Implications for policy and practice are discussed.

Keywords

achievement gap college readiness effectively maintained inequality higher education retention social stratification

Attending college and earning a degree have never been more important for social mobility. In fact, earning a degree is now the minimum threshold for entry into the middle class (Carnevale, Smith, & Strohl, 2010; Rothwell, 2012). In the higher education literature and on campuses across the United States, reform efforts and research have focused on increasing access to higher education for previously marginalized groups. This attention has contributed to dramatic increases in higher education attendance among underrepresented students (Astin & Osegura, 2004; Posselt, Jaquette, Bielby, & Bastedo, 2012). Despite these gains and a college-for-all ethos (Rosenbaum, 2001), degree attainment has not significantly increased (Alon, Domina, & Tienda, 2010; Roksa, 2010), and social inequality has actually grown (Dwyer, 2013; Grusky, Western, & Wimer, 2011; McCall, 2001). Attrition among students who begin college at a four-year institution is nearly 30% (American College Testing [ACT], 2013a) and has been increasing since 2009 (National Student Clearinghouse, 2013). These realities make additional research related to retention and educational attainment increasingly urgent.

The current study examines first-year retention focusing on the role of social status background factors among students of differing college readiness. College students do not arrive on campus equally prepared for academic success. Low-income and underrepresented students are less likely to begin college academically prepared for success and are the most likely to leave college prior to earning a degree (Adelman, 2006). While it is possible that differences in academic preparation explain higher rates of attrition in the first year of college and studies consistently find a relationship between prior academic achievement and retention (see Cabrera, Miner, & Milem, 2013), the interrelationships between academic preparation, social status background factors, and retention remain unclear. Prior research has yet to address these relationships directly.

We address this gap in the literature by explicitly investigating how the relationship between social status background and first-year retention may be different based on readiness. To do this, we examine influences of social status background on retention separately for college-ready students and students who begin higher education with less academic preparation, a departure from past studies that have examined readiness as one or more factors among many. With this modeling approach, we investigate the ways in which college readiness might interrupt the mechanisms that transmit the effects of social status across generations. Understanding the continuing and differential role of social status background factors in the status transmission process at this key transition along the educational attainment pipeline has important consequences for designing policy and practices that are consistent with increasing educational success.

Literature Review

This review begins with a discussion of the theoretical grounding that serves as the foundation for this study. A discussion of college readiness and the nascent literature emerging on the relationship of college readiness to retention and degree attainment follows. This section concludes with a review of recent retention and degree completion literature examining the role of social background factors and the financial resources students use to fund college. In doing so, we frame this study with six decades of theory and related research.

Theoretical Grounding

Two competing theoretical narratives have emerged in the educational and social attainment literature that influence the approach we take in this study. One narrative, status attainment theory, focuses on academic achievement and expectations for achievement as the primary mechanism for social mobility and success over the life course. The second encompasses critiques of that notion and centers on the continuing and predominant role of social status of origin and material circumstances despite educational achievement.

Status attainment theory emerged as a frame to examine social reproduction in the 1960s. Blau and Duncan (1967) developed it to examine how social status positions perpetuate across generations, and Sewell and colleagues (Sewell, Haller, & Ohlendorf, 1970; Sewell, Haller, & Portes, 1969) extended it to focus on the socialization processes through which family status background and relationships with significant others influence social status attainment in adulthood. The framework states that expectations for educational attainment and subsequent success and achievement in the educational arena are the central factors through which individuals achieve both social status (mobility) and maintain it (reproduction). The status attainment framework posits that interactions with family and significant others as well as self-assessment of one's potential based on academic performance form expectations for educational attainment (Knotterus, 1987). Signals from these sources combine and work together as students form expectations about their educational attainment.

In their critiques of status attainment theory, Kerckhoff (1984) and Bozick, Alexander, Entwisle, Dauber, and Kerr (2010) represent the alternative understanding of the status attainment process. Kerckhoff's critique focuses on the social allocation aspects of reproduction, whereas Bozick and colleagues focus on the explanatory power of status attainment theory to explain differences in attainment between individuals from higher and lower status families. Specifically, Kerckhoff argued that status attainment theory fails to acknowledge the role of material circumstances in molding outcomes. Thus, educational continuation decisions, despite expectations, could be constrained for lower status individuals by what an individual assesses can reasonably be achieved given both known and perceived constraints. For instance, the experience of attending college may give lower income students information about the achievability of a degree that they did not have previously, such as the difficulty of performing at a college level while working at a full-time job or the lack of support their particular institution provides, whether monetarily or otherwise. Thus, students who begin college expecting to earn a degree may find that they can reasonably project they will not complete their degree on the basis of information they gain in the first year and elect not to continue.

Bozick et al. (2010) argue that students receive different signals about their potential for college success depending on their social stratum of origin. In a study examining the socialization messages regarding college that students received from their families, these scholars find that students from high status families receive consistent messages that they are college material early on and throughout their schooling experiences, whereas lower status families provide mixed messages about prospects for college. They conclude that status attainment theory underestimates the force through which family structural location influences attainment over the life course through the allocation of these messages, arguing that mixed messages will influence some students to be easily deterred from achieving a college degree. Taken together, Kerckhoff’s (1984) and Bozick et al.’s critiques address how signals in the environment about one's potential and limits on the realization of that potential take shape differently depending on social stratum of origin.

Focusing on how structural inequalities manifest themselves rather than status attainment among individuals, Lucas’s (2001) theory of effectively maintained inequality posits that families and individuals with status will secure for themselves and their children some degree of advantage wherever the educational system affords that possibility. He argues that the site of conflict in securing advantage is determined at each level of education by the degree to which that level is universally obtained. When a particular level of education is near universally attained (in the United States earning a high school diploma), families of higher status will work to secure qualitatively better education at that level for their children, and when different levels of education are attained (in the United States earning a college degree), families will work to secure quantitatively more education for their children. Thus, higher status students will secure both better and more education based on their status position. In an effectively maintained inequality framework, students who are not fully college ready will nonetheless continue past their first year of college—and probably to obtaining a degree—if they have a higher social background, while students from lower social status backgrounds will not.

College Readiness

The large majority of college students today begin college not fully prepared for success (ACT, 2013b; College Board, 2013). Assessments of college readiness most often use three factors that have both overlapping and distinctive qualities: high school course-taking patterns, high school GPA, and standardized test scores. As Roderick, Nagaoka, and Coca (2009) explain, high school course-taking patterns provide a measure of exposure to the content knowledge needed in introductory college coursework, high school GPA measures the development of core academic skills and content knowledge, and test scores provide a standard measure of ability and core academic skills. Thus, high school course-taking patterns and high school GPA both assess content knowledge, and test scores and high school GPA both assess core academic skills. High school GPA also distinctively assesses noncognitive skills such as effort and study skills that students need to succeed in college. Gaps in readiness are more apparent using high school GPA rather than test scores as a measure of readiness, particularly in examining differences by sex and race/ethnicity (Roderick, Nagaoka, & Allensworth, 2006). In addition, high school GPA is a stronger predictor of academic achievement during the first college year (Cabrera et al., 2013; Roderick et al., 2009; Wolniak & Engberg, 2010), first-year retention (Cabrera et al., 2013), and degree completion (Geiser & Santelices, 2007; Roderick et al., 2006) than standardized test scores.

Current policy in the United States focuses on increasing college readiness through more rigorous course-taking patterns in high school (Roderick et al., 2009). Results from NCES's National Association of Educational Progress (NAEP) and from the Higher Education Research Institute (HERI) as well as data from ACT and SAT test takers document persistent and continuing gaps in course-taking patterns by race/ethnicity (ACT, 2009; College Board, 2012; Planty, Bozick, & Ingels, 2006; Pryor, Hurtado, DeAngelo, Blake, & Tran, 2009; Roderick et al., 2009), socioeconomic status (College Board, 2012; Planty et al., 2006), and first-generation status (Lohfink & Paulsen, 2005). These differences contribute to measured gaps in achievement on college entrance exams (ACT, 2009; College Board, 2012).

The connection between high school course-taking patterns and first-year retention has also begun to emerge as a focus of retention and degree completion studies. Warburton, Bugarin, Nunez, and Carroll's (2001) study of degree completion using NCES data found that students who took rigorous courses in high school were more likely to be retained to degree. Additionally, in examining their descriptive data they found that first-generation students who had completed a rigorous course sequence in high school were just as likely to complete a degree as continuing generation students. Studying the role of high school course-taking patterns separately for first-generation and continuing generation students with regression, Lohfink and Paulsen (2005), using Beginning Postsecondary Student (BPS) data, found that level of rigor in course-taking made no difference to retention within either group. These findings suggest that readiness for college might help to mitigate social background factors that are thought to put students at risk for attrition.

The Role of Socioeconomic Status and Financial Resources for College on Retention

The positive connection between socioeconomic status and retention and degree completion is one of the most consistent findings in the literature (Astin & Oseguera, 2005; Bozick, 2007; Cabrera, Burkum, & La Nasa, 2005; DeAngelo, Frank, Hurtado, Pryor, & Tran, 2011; Franke, 2012; Herzog, 2005; Hu & St. John, 2001; Ishitani, 2006; Leppel, 2002; Lohfink & Paulsen, 2005; Paulsen & St. John, 2002). Among these, Bozick’s (2007) work using BPS data to study first-year retention is particularly relevant to the current study. He found that family income and family wealth had a direct and strong relationship with first-year retention and that family income and wealth mediate the experience students have on campus, which directly connects to retention. Thus, he identified a cumulative disadvantage for low-income students. Low-income students in the study were much more likely to have high intensity work schedules (more than 20 hours per week) than their high income peers and more likely to live with their parents, both of which connected directly to first-year attrition.

Other studies have confirmed the higher intensity work patterns in college of low-income students (Terenzini, Cabrera, & Bernal 2001; Walpole, 2003) and the effect of high intensity work on retention in the first college year (DeAngelo, 2014; Gilardi & Guglielmetti, 2011; Roksa, 2010). Intent to work full-time in college has also been negatively connected to degree completion (Astin & Oseguera, 2005; DeAngelo et al., 2011). On the other hand, Lohfink and Paulsen (2005) did not find a significant link between living on campus and first-year retention for either their first-generation or continuing generation student sample groups. In fact, first-generation students who chose the college they attended at least in part because it allowed them to live at home were more frequently retained. These findings challenge Bozick’s (2007) results associating living with parents and college attrition.

Herzog’s (2005) study addresses low-income students through a focus on financial aid. Using almost a decade's worth of cohort data from a single institution, Herzog found that when low-income students have a statewide scholarship paying full tuition in place, their likelihood of retention was higher than in years in which the scholarship was not in place and in which students had to use a combination of scholarships and loans to fund college. Thus, it seems these students benefited in terms of retention from having a single stable source of funding for college. At the same time, large amounts of grant aid did not entirely mitigate the disadvantage low-income students faced. They had lower odds of being retained at the end of the first year of college than high-income students regardless of funding.

Related directly to loans, Herzog’s (2005) study found that loans in any amount were detrimental to first-year retention for low-income students. Other studies show the benefits of grants and detriments of loans to first-year retention for first-generation students (Ishitani, 2006; Lohfink & Paulsen, 2005) and the benefit of grants to first-year retention overall (Chen & DesJardins, 2010; DesJardins, Ahlburg, & McCall, 2002). However, Cabrera and colleagues (2005) found that loans facilitated degree completion, even among low-income students. More generally, studies demonstrate a negative connection between unmet financial need and concern about financing college and retention (Alon, 2007; DeAngelo, 2014; Herzog, 2005; Paulsen & St. John, 2002) as well as degree completion (Astin & Oseguera, 2005; DeAngelo et al., 2011).

In considering these findings as they relate to financial aid and educational attainment, it is important to recognize that the current policy context in the United States increasingly favors merit-based grants over need-based grants as well as a much heavier reliance on loans as an aid mechanism (Perna & Finney, 2014). The movement from need-based to merit-based grant aid has resulted in an increase in grant aid for high-performing, college-ready students regardless of need (Doyle, 2006; Perna & Finney, 2014). This direction is part of an overall policy context in which the large burden of funding the cost of college has shifted to the individual and away from the public in the form of state support for institutions and other public mechanisms of providing support for educational attainment.

Summary and Gaps in the Literature

Factors related to socioeconomic status clearly exert an influence on students during college, producing a substantial cumulative effect on the likelihood of degree completion. Further, research shows that socioeconomic status operates in part by influencing several specific decisions and conditions that affect degree attainment, including the number of hours students work, their place of residence, and the amount and type of financial aid they receive, all of which influence educational attainment decisions. The college readiness literature also reveals that socioeconomic status and other background factors influence who begins college prepared to succeed and that readiness factors may have the potential to level out the playing field for students once in college. Overall, the literature suggests that the processes through which readiness influences educational attainment may operate differently based on social status background and that both competing theoretical notions of social mobility and reproduction may be at play. College readiness may be a mediator in relation to first-year retention, interrupting mechanisms that transmit the effects of social status background across generations.

Although research to date suggests that social status background factors may predict retention differently based on readiness and that the transmission of social status across generations might be different for students who begin college academically prepared and those who begin with less preparation, research examining this link directly is limited. To address this gap from a methodological standpoint, studies need to take into account that readiness shares covariance in predicting retention with social status background factors. One way to do this is to disaggregate the data by college readiness. We disaggregate our data into two distinct readiness groups to focus on how the pattern of relationships predicting retention might be different for students whose preparation is above or below a critical readiness threshold as established in the literature (Planty et al., 2006; Pryor et al., 2008; Roderick et al., 2009) and as a means to study the social processes of status transmission related to first-year retention for these two groups.

By isolating and investigating the relationship between social status factors at a specific educational transition, this study supplements life course studies of educational attainment by reducing the likelihood of confounding the effects of socioeconomic status at a particular educational transition with earlier attainment points (Ewert, 2010). The theoretical and conceptual frames we bring to this study offer a rich analysis. As Lucas and Beresford (2010) assert, bringing theory-driven research to studies of educational transitions and success is the key to making progress to increase overall educational attainment.

Research Questions

Studying first-year retention separately for college-ready and less-ready students and focusing on how the processes related to educational attainment and social status transmission at this juncture differ based on social status background, we investigated the following research questions:

Research Question 1: How do college-ready and less-ready first-time, full-time students differ with respect to socioeconomic background, financial resources for college, and demographic characteristics?

Research Question 2: To what extent do socioeconomic background, financial resources for college, and demographic characteristics contribute to first-year retention differently for college-ready and less-ready students?

Methodology

Data Source and Sample

The Higher Education Research Institute (HERI) at the University of California, Los Angeles provided the data for this study. The data set draws from two main sources, the 2004 Freshman Survey (TFS) from the Cooperative Institutional Research Program (CIRP) at HERI and the National Student Clearinghouse (NSC). The TFS is a national survey of incoming college students designed to gather information about student background, high school experiences, financial resources for college, the college choice process, and expectations for experiences in college among other topics. NSC, which has been tracking enrollment and degree completion for participating institutions since 1993, provides data on student persistence. Through merging these data sources, HERI created a unique data set that allows for the study of the movement, persistence, and degree attainment of students at higher education institutions in the United States. To compensate for missing values, HERI carried out a multiple imputation method based on a multivariate normal approach in order to preserve the data set in its near entirety. The final, combined 2004 TFS/NSC data set that this study used encompasses 210,056 full-time, first-time students at 356 four-year colleges and universities (see DeAngelo et al., 2011, for details on data and imputation methodology). For analyses, the sample was weighted to be nationally representative for entering first-year students in 2004 following the procedures used at HERI for over 40 years (see DeAngelo et al., 2011, for details on weighting).

Based on our review of the literature, we define college readiness using a combination of high school course-taking patterns and high school GPA. We defined course-taking readiness based on readiness definitions adopted by HERI (Pryor et al., 2008) and NAEP (Horn & Kojaku, 2001). College-ready students have a B+ or better high school GPA and have completed four years of English, three years of math, two years of a foreign language, one year each of biological and physical sciences, plus an additional year of one or the other (in total three years of science), one year of history/government, and one year of arts. In our data set, 113,167 students (53.8%, unweighted) met our course-taking patterns requirements, and 151,937 students (72.3%, unweighted) met our GPA requirements. Taking both readiness factors together, 86,863 students (41.4%, unweighted) were college-ready. The remaining 123,193 students (58.6%, unweighted) did not meet at least one of the two necessary criteria and were classified as less-ready.

Measures

For the purposes of our study, we coded the dependent variable 1 for individuals who continued at their initial institution of enrollment after one year and 0 for those who did not. Although some students who leave their initial institution do successfully enter and graduate at other institutions, attrition from one's initial institution and student movement generally dampens one's chances of degree attainment and is especially troublesome for underrepresented populations (Kalogrides & Grodsky, 2011). Thus, we focus on persistence at the initial college or university into the second college year, the traditional dependent variable used to study retention in the higher education literature. See Appendix A for a full listing of the variables and coding schemes.

To address our research questions as they relate to relationships between socioeconomic status, financial resources for college, demographics, and first-year retention, we included a set of 15 variables in our model. To represent student socioeconomic status, we included parental income and first-generation status. We recoded the categorical parental income variable in the TFS into quartiles and inserted them as dummy variables. We determined first-generation status by parental educational attainment and inserted it as a dummy variable as well. If neither parent had college experience, we considered a student first generation. To examine the influence of the financial resources students use to fund college during the first year, we included measures for various sources of financial assistance, including family resources used to finance one's education, grants received, and loans. We recoded the categorical aid measures in the TFS into five dummy variables for each of these three financial aid sources. We incorporated a measure for students’ financial concern, which we recoded into three dummy variables, each representing a different level of concern regarding having enough funds to complete college.

We incorporated students’ SAT composite scores, which research shows correlate with family income (Bowen, Chingos, & McPherson, 2009) and are predictive of retention and degree completion. From the senior year of high school, we include hours per week spent studying and on household duties, both of which are known to be associated with family income (Fuligni & Pederson, 2002). Furthermore, hours per week spent studying is thought to indicate, in part, noncognitive habits of mind indicative of college readiness and predictive of college success (Conley, 2005, 2010). We also include measures of how students make the choice of which college they will attend; specifically, we include the degree to which a student chooses their institution because of cost and the degree to which they choose based on wanting to live near home. The distance the college attended is from home, the living arrangements students have made for their first college year, and intent to work full-time while attending college are also included. Lastly, we included an aggregated parental income measure at the institutional level to test if the overall socioeconomic status climate on campus had an effect on retention over and above any student level effects. Palardy’s (2013) research on the significant role of aggregate socioeconomic status at high schools on college going supports this as an important variable for consideration.

To measure the relationship between student demographics and retention, we include student sex and race/ethnicity. Race/ethnicity is included as group of seven dummy variables in the analyses: African American, Asian American, American Indian, Latino/a, White (reference group), multiracial, and other.

To fully fill out our model based on the background literature on retention and degree completion in the higher education literature, we use Nora, Barlow, and Crisp's (2005) student/institution engagement model as a guide, focusing on the pre-college and intentions for college aspects of the model. By testing a fuller model, we can assess if factors of primary concern to this study are still significant even when other factors with a known connection to retention and degree completion are taken into account. At the student level based on this model, we include students’ level of academic and social self-concept prior to the start of college, the hours per week in the senior year of high school devoted to student clubs/groups and volunteering, advanced degree aspirations, the intention to transfer to another college before graduating, having a declared major at the start of college, willingness to change one's major after starting college, and planning to be involved in campus life (for empirical validation of the inclusion of these variables, see Astin & Oseguera, 2005; DeAngelo et al., 2011). At the institutional level based on the research literature, we include selectivity, private or public institutional control, and aggregate transfer likelihood among the student population (Astin & Oseguera, 2005; DeAngelo, et al., 2011; Oseguera, 2005; Oseguera & Rhee, 2009). Lastly, we incorporated an indicator for Historically Black Colleges and Universities (HBCU control), due to the strong representation of African American students at these institutions in the data set.

Data Analysis

For our analyses, we evaluate models of first-year retention separately for the two readiness groups (college-ready, less-ready) using hierarchical generalized linear modeling (HGLM). This analytic strategy is consistent with the purpose of this study and recognizes that college readiness is derived from a discrete number of readily identifiable factors and that students are identified and understood by policymakers, practitioners, and administrators through a dichotomy as it relates to readiness. Modeling in a manner that is consistent with how students are understood helps to make our findings actionable by these external groups. Methodologically, although it might be argued that college readiness has elements of a continuous latent trait at the theoretical level and should be modeled as such with interaction terms used to identify significant differences between readiness and other key factors in the model, this strategy is problematic because it assumes that the potential mediating effect of college readiness in the relationship between social background factors and retention are similar and linear at different levels of readiness (Steinberg & Fletcher, 1998). Instead, the retention process may be substantially different above or below a critical readiness threshold. In addition, the complexity of interacting readiness with other key variables to identify significant differences can make interpreting main effects difficult (Warner, 2012) and is not the most efficient way to analyze the data for our purposes.¹

The multilevel approach we carried out in this study has advantages over conventional analyses, such as ordinary least squares (OLS) or logistic regression, as it accounts for the dependence among students within colleges. It provides more efficient and reliable estimates in cases of unbalanced, nested data structures (Raudenbush, Bryk, Cheong, & Congdon, 2004). In building models within HGLM, analysts must ensure that the outcome significantly varies across institutions (Niehaus, Campbell, & Inkelas, 2014; Raudenbush & Bryk, 2004). To do this, we analyzed the random variance component from a fully unconditional model to determine whether the likelihood of retention at the end of the first year for both college readiness student groups varies significantly across institutions. The fully unconditional model suggested that institutions significantly differed in the average propensity of retention. Using the between-institution variance component to calculate the intraclass correlation (ICC), the results showed that 12.3% of the variability for less college-ready students is between institutions. For the college-ready students, the between-institution variability is only slightly lower at 10.2%, which indicates sufficient Level 2 variability to justify the use of a multilevel method. Considering the ICC results, we proceeded with building the full analytic models. We carried out all analyses using HLM 7.

For our analyses, we recoded categorical predictors in the data set into dummy variables before they were entered in the HGLM models. A series of diagnostic tests was carried out before the final analytic model was run, using procedures from Hosmer and Lemeshow (2004) and Tabachnick and Fidell (2007). All diagnostics were run on the full data set and separately for the two readiness groups under study. We also ran diagnostics separately for the original and all imputed data sets. First, we inspected all categorical variables for empty and low frequency cells but found no zero cells and no cells with expected frequency lower than five. Second, we examined the variance inflation factor (VIF) for each of the predictor variables, as a test of multicollinearity within the model, after visually inspecting coefficients and standard errors for exceedingly large values. All variables had a VIF below 2.0; thus, they were far below a critical threshold, and therefore all variables were retained in the analytical model. Third, we examined the linearity of the logit of the dependent variable for continuous predictors. For this, all continuous variables were transformed using the natural log and included together with the original predictors in the analysis. We inspected chi-square statistics in the likelihood ratio test but found no problematic values (using p = .001). Lastly, we plotted predicted probabilities against Cook's distance, a measure for multivariate outliers. For the group of college-ready students, we detected two data points that seemed distant from the rest of the distribution. However, when we excluded these points from the analysis, results did not change, thus we retained them in the data set.

To perform our main analyses, we specified the following within-college model. Due to the dichotomous nature of the outcome variable, the Level 1 (student-level) HGLM model uses a Bernoulli sampling distribution and logit link function:

Prob (Y_{i j} = | β_{i j}) = Φ_{i j} .

The following equation characterizes the Level 1 (student-level or within-institution) model:

\begin{matrix} Log [\frac{Φ_{ij}}{1 - Φ_{ij}}] = β_{0 j} + β_{1 j} \times {(Demographics)}_{ij} + β_{2 j} \times {(Socioeconomic Status)}_{ij} \\ + β_{3 j} \times {(Financial Resources & Concern)}_{ij} \\ + β_{4 j} \times {(Controls)}_{ij} + e_{ij}, \end{matrix},

where i denotes the individual student and j represents the institution. The variable vectors included in the model represent characteristics and experiences we believe influence the outcome variable—retention at the end of the first year.

The Level 2 (institution-level or between-institution) model is represented by:

\begin{matrix} β_{0 j} = γ_{00} + γ_{01} \times {(Aggregate Parental Income)}_{j} \\ + γ_{02} \times {(Institutional Characteristics)}_{j} + u_{0 j}, \end{matrix},

where j denotes the institution, γ₀₀ represents the average likelihood of retention across all institutions, and $u_{oj}$ the random variance component for institution j. The variables included at the institution level describe how the context at four-year colleges and universities affect the student's average likelihood of first-year retention.

Multilevel modeling techniques require a consideration of how variables are centered (Raudenbush et al., 2004). For this study, all dichotomous predictors at the student and institutional levels have been inserted uncentered into the model. Continuous variables at the student level are group-mean centered, centered on their unadjusted institutional mean. Group-mean centering allows the intercept to represent the log-odds of being retained in the first college year for students who have average characteristics for the Level 1 variables at the same institution and is the appropriate centering technique given the purposes of this study and Level 2 controls. At the institutional level, continuous measures are grand-mean centered, the standard choice at Level 2.

Limitations

The use of TFS/NSC data presents a limitation to this study. Participation in the NSC has been increasing since its inception in 1993; more than 3,600 postsecondary education institutions submitted student enrollment and degree completion data for cohort 2004. Though coverage is widespread and continuously increasing, the quality of the data used in our study is affected by the proportion of data that can be matched between CIRP's 2004 Freshmen Survey and the NSC. Not all students that participated in 2004’s administration of the TFS could be matched to NSC data, which could result in biases. However, from the 424,808 students that participated in the TFS, almost three-fifths (57.1% or 243,676) could be matched to NSC data (for a recent discussion of Clearinghouse data limitations, see Dynarski, Hemelt, & Hyman, 2013). Another limitation arises from the fact that the TFS survey relies on self-reported data for all its measures, which may also insert bias into the estimation.

Results

Descriptive Results

In the nationally weighted data set, 38% of students met the criteria for placement in the college-ready group. The remaining 62% did not meet at least one of the two necessary requirements and were thus classified into the less-ready group. These figures are on par with readiness figures from the College Board (2013), which classified 43% of incoming college students as ready for college. In our nationally weighted data set, 83% of the overall student population were retained into the second year. Among the college-ready group, the retention rate is higher at 88%, whereas among the less-ready group, the retention rate is lower at 78%; this difference is statistically significant (p < .000). This amounts to a loss of 58,041 students in the college-ready group and 178,468 students in the less-ready group in the first year. This means that students who begin college less-ready account for 75% of the attrition in the first year.

Table 1 provides the socioeconomic status, financial resources for college, and demographic characteristics of the college-ready and less-ready student groups in the weighted dataset. Female students have a higher representation in the college-ready group (60%) than in the less-ready group (52%). By race/ethnicity, more than 7 in 10 students (74%) are White in the college-ready group, whereas White students make up 66% of the population in the less-ready group. Asian American students likewise have a higher representation in the college-ready group (9%) and a lower representation in the less-ready group (6%). Group representation is equal among multiracial students (6% in each group), American Indian students (<1% in each group) and students of other race/ethnicity (2% in each group). On the other hand, African American and Latino/a students have higher representation in the less-ready group than in the college-ready group. African American students in particular are overrepresented in the less-ready group, making up 14% of the student population but only 5% of the student population in the college-ready group. First-generation students also make up a larger percentage of the less-ready (23%) than the college-ready (15%) group.

Table 1

Selected Descriptive Statistics (Weighted) by College Readiness Groups

	Less-Ready (N = 131,985)	College-Ready (N = 78,071)
Dependent variable: Retained into second year, %	78	88
Sex, %
Female	52	60
Male	48	40
Race/ethnicity, %
American Indian	1	1
Asian American	6	9
African American	14	5
Latino/a	6	4
White	66	74
Multirace/ethnicity	6	6
Other	2	2
First-generation status, %
First generation	23	15
Not first generation	77	85
Financial resources (median in dollars)^a
Family income	66,270	77,680
Aid for first year from family	4,300	6,630
Grants for first year	2,720	3,680
Loans for first year	1,840	1,750

Note. Data from 2004 Freshmen Survey (TFS) and Cooperative Institutional Research Program (CIRP) and National Student Clearinghouse. Sample weighted to represent incoming first-time, full-time cohort in 2004 at all four-year institutions in the United States. Weighted sample includes N = 210,056 students attending n = 356 colleges and universities.

Interpolated median, due to categorical nature of TFS survey data.

The two groups’ differences in socioeconomic status and financial resources for college reveal a noticeable, though not unexpected, association with academic preparation for college. Parental income is considerably higher in the college-ready group with a median income of $77,680, compared to $66,270 in the less-ready group. Students in the college-ready group also report more financial support from their families to pay for the first year of college ($6,630 median) than students who begin college less-ready for success ($4,300 median). College-ready students report larger median grant amounts ($3,680) and lower median loan amounts ($1,750) than students in the less-ready group for grants ($2,720) and loans ($1,840). Overall, the median amount college-ready students report they have available to fund their first year of college is almost one-third larger ($12,060) than the median amount among the less-ready group ($8,860).

Results From the HGLM

Table 2 contains results based on the full model but limited to the variables of interest in this study. Appendix B contains the complete results for all variables in the full model.

Table 2

Parameter Estimates From Hierarchical Generalized Linear Modeling Analysis Predicting First-Year Retention, by College Readiness Groups (Study Variables)

	Less-Ready (N = 131,985)				College-Ready (N = 78,071)
	Coefficient	SE	Significance	Odds Ratio	Coefficient	SE	Significance	Odds Ratio
Study variables
Parental income (low-income reference group)
Lower-middle income ($40,000–$74,999)	.123	.027	.000***	1.131	.042	.041	.306	1.043
Upper-middle income ($75,000–$149,999)	.146	.031	.000***	1.157	.108	.046	.019	1.114
High income (above $150,000)	.108	.038	.005**	1.114	−.011	.054	.834	0.989
First-generation status	−.117	.028	.000***	0.890	−.091	.042	.033	0.913
Financial resources for first year of college
Aid for first year from family (no family resources reference group)
Less than $2,999	.064	.041	.116	1.066	.042	.053	.426	1.043
$3,000–$5,999	.105	.044	.017	1.111	.076	.065	.247	1.078
$6,000–$9,999	.101	.042	.016	1.106	.132	.075	.079	1.142
More than $10,000	.091	.042	.033	1.095	.063	.064	.324	1.065
Grants for first year (no grants reference group)
Less than $2,999	.105	.036	.011	1.110	.143	.049	.004**	1.153
$3,000–$5,999	.081	.041	.051	1.085	.139	.050	.006**	1.149
$6,000–$9,999	.095	.034	.006**	1.100	.152	.055	.006**	1.134
More than $10,000	.151	.036	.000***	1.162	.177	.057	.003**	1.194
Loans for first year (no loans reference group)
Less than $2,999	−.115	.037	.004**	0.892	−.086	.042	.042	0.918
$3,000–$5,999	−.202	.044	.000***	0.817	−.111	.051	.034	0.895
$6,000–$9,999	−.202	.038	.000***	0.817	−.254	.059	.000***	0.776
More than $10,000	−.267	.041	.000***	0.766	−.299	.051	.000***	0.742
Some financial concern (none reference group)	.026	.025	.297	1.027	−.043	.045	.345	0.958
Major financial concern (none reference group)	−.187	.038	.000***	0.829	−.305	.054	.000***	0.737
SAT composite score	.067	.012	.000***	1.069	.113	.015	.000***	1.119
Hours per week in high school senior spent studying/homework	.101	.008	.000***	1.106	.081	.013	.000***	1.084
Hours per week in high school senior spent on household duties	−.033	.009	.000***	0.968	−.029	.010	.006**	0.971
Choose institution because of costs	.109	.014	.000***	1.115	.151	.027	.000***	1.163
Choose institution wanted to live near home	.077	.018	.000***	1.080	.025	.025	.317	1.026
Distance college away from home	−.058	.013	.000***	0.944	−.146	.017	.000***	.864
Place of residence in the fall (on-campus reference group)
Live with family	−.057	.049	.241	0.944	−.069	.070	.327	0.933
Other living arrangement	−.401	.067	.000***	0.670	−.501	.090	.000***	0.606
Plan to get full-time job to pay for college	−.111	.014	.000***	.895	−.087	.020	.000***	.916
Sex (female)	.134	.032	.000***	1.143	.106	.055	.054	1.111
Race (White reference group)
American Indian	−.260	.115	.023	0.771	−.493	.222	.027	0.611
Asian	.348	.063	.000***	1.416	.394	.086	.000***	1.483
African American	.098	.065	.132	1.103	.117	.094	.213	1.124
Latino/a	.045	.071	.528	1.046	.058	.082	.481	1.059
Other race/ethnicity	.280	.118	.020	1.323	.371	.126	.004**	1.449
Multi-ethnicity	−.161	.045	.001**	0.851	−.163	.054	.003**	0.850
Institutional context factor
Institutional income/socioeconomic status (aggregate)	.116	.052	.026*	1.123	.071	.053	.184	1.074
Intercept	.819	.458	.074	2.268	1.681	.620	.007**	5.369

Note. Due to use of multiple imputation, parameter estimates are based on population-average models with robust standard errors. Weighted samples using N = 131,985 (Group 1) and N = 78,071 (Group 2) at n = 356 four-year institutions.

p < .05. **p < .01. ***p < .000 (significance level at p < .05 only marked for Level 2 predictors).

Socioeconomic and Financial Resources for College

Results demonstrate that social status factors have differential impact on first-year retention based on college readiness. Thus, the gaps in socioeconomic status and financial resources for college found in the descriptive results for the groups further compound into (dis)advantage in terms of retention to the second college year. Specifically, for less-ready students, as family income rises, so does retention. Lower-middle (odds ratio [OR] = 1.131), upper-middle (OR = 1.157), and high-income (OR = 1.114) students continue at a higher rate than low-income students. First-generation students (OR = 0.890) are also less likely to be retained among the less-ready group. Importantly, social status factors do not affect retention for students in the college-ready group. Thus, these results seem to indicate that if the playing field in terms of readiness can be leveled prior to students beginning college, the social status background factors that put students at risk of attrition once they begin college might be largely ameliorated. These results also indicate that the college environment itself appears to provide incomplete support for students who begin college less-ready.

Consistent with past research (Herzog, 2005; Lohfink & Paulsen, 2005), the results with regard to financial resources demonstrate a differential impact of these resources on retention by readiness. Using loans and higher amounts of loans in the first year puts students at an attrition risk during their first college year but to differing degrees based on readiness. Among the less-ready students, a loan to fund college of any amount is an attrition risk compared to no loans (OR = .892 <$2,999, OR = .817 $3,000–$5,999 and $6,000–$9,999, OR = .766 $10,000 or more), whereas college-ready students only incur a significant attrition risk with loans in amounts over $6,000 in the first year (OR = .776 $6,000–$9,999, OR = .742 $10,000 or more).

Findings related to the supportive nature of grants to first-year retention also accord with past research (Chen & DesJardins, 2010; DesJardins et al., 2002) but again demonstrate differential impacts on retention by readiness (Herzog, 2005; Lohfink & Paulsen, 2005). Among college-ready students, students who received grant funding had increased odds of retention compared to students without grants, starting at 15% (OR = 1.153) for grants less than $2,999 and ending at 19% (OR = 1.194) for grants of more than $10,000. Thus, small grants, even those of $3,000 or less, serve to keep college-ready students matriculated at their institution of origin, while larger grants appear to have less marginal value. Students in the less-ready and less financially resourced group receive smaller amounts of grant funding than college-ready students (see Table 1). Moreover, grants of the size typically received by less-ready students have little impact on retention; a positive effect emerges only for grants in excess of $6,000 (OR = 1.100 $6,000-$9,999, OR = 1.162 $10,000 or more).

Unlike aid from loan and grant sources, financial support from the family was not a statistically significant predictor of retention for either readiness group, regardless of how much money the family provided toward the cost of college. This result should not be surprising since regardless of readiness, students without family support will be awarded Pell Grants and potentially other types of state or institutional grant aid, as well as packages that include loans, as necessary to fund college. Lastly, in examining financial resources for college, we found that students in both readiness groups who have major concerns about having enough funds to finance their college education are at a large attrition risk compared to those who report no concerns (OR = .829 for less-ready, OR = .737 college-ready). The known and perceived material circumstances that generate these concerns may serve as a signal that earning a degree is not a possibility (Kerckhoff, 1984). For the college-ready group, this factor has the largest effect on first-year retention of any we examined. For the less-ready group, among the financial supports and resources factors we studied, only a loan above $3,000 taken to fund the first year outpaces the deleterious effects of major financial concerns on retention. However, the difference between having no financial concerns about financing college and some concerns was not significant for either group.

In examining the relationship between first-year retention and student-level factors associated with socioeconomic status and financial resources for college, we find more similarity between readiness groups than difference in predictors of retention. For both readiness groups, higher SAT scores increase the likelihood of retention (OR = 1.069 less-ready, OR = 1.119 college-ready for each 100-point incremental score increase). Hours per week in senior year of high school spent on homework (OR = 1.106 less-ready, OR = 1.084 college-ready) was also positive for both readiness groups. On the other hand, time spent on household duties (OR = .968 less-ready, OR = .971 college-ready) shows a negative relationship for both groups. These results suggest a compounded impact on lower income students in both readiness groups since test scores (Bowen et al., 2009) and allocation of time to studying and household duties (Fuligni & Pederson, 2002) correlate with family income. In addition, students in both groups who placed greater importance in the college choice process on the cost of the college they selected (OR = 1.115 less-ready, OR = 1.163 college-ready) had higher retention rates, whereas students who planned to get a full-time job to pay for college (OR = .895 less-ready, OR = .916 college-ready) had lower rates.

In comparing students who were going to live on campus to those who were going live at home with family for the first year of college, we found no difference in terms of first-year retention. This adds to evidence from Lohfink and Paulsen (2005) that controlling for other factors—in this study readiness and in their study student generation status—there are no differences in first-year retention between living at home and living at campus. Conversely, students with living arrangements that were neither on campus nor with family were less likely to be retained (OR = 0.670 less-ready, OR = 0.606 college-ready). Distance between college and home (OR = 0.944 less-ready, OR = 0.864 college-ready) was also negative for both groups. Lastly, among less-ready students, those who placed importance on living near home in their college choice process (OR = 1.080) were more likely to be retained.

At the institutional level, we examined whether the overall socioeconomic status (SES) of the students on campus has an effect on individual retention. For college-ready students, institution-mean socioeconomic status was not significant, but less-ready students benefit from a campus environment with higher aggregate parental income levels among their students (OR = 1.123). Although this benefit in terms of retention from the institutional climate set on campuses with aggregate higher parental income levels accrues equally to less-ready students, this amounts to a kind of double SES advantage for higher income less-ready students that they receive on top of the benefit to retention they have based on their own family status and income backgrounds.

Sex and Race/Ethnicity

Among college-ready students, men are just as likely to be retained as women, but they are less likely to be retained in the less-ready group (OR = 1.143 for women). Race/ethnicity yields a more complex picture. Underrepresented racial minority students have a much higher representation in the less-ready group than in the college-ready group and therefore lower retention rates. But within each readiness group, American Indians, African Americans, and Latino/a students are just as likely to be retained as White students. Asian American students are more likely (OR = 1.416 less-ready, OR = 1.483 college-ready) and multiracial students less likely (OR = 0.851 less-ready, OR = 0.850 college-ready) to be retained than White students in both groups. College-ready students of other race/ethnicity are more likely to continue than their similarly ready White peers (OR = 1.449) but just as likely to be retained in the less-ready group.

Discussion and Implications

The key finding in this study is that not all lower income and first-generation students are equally at risk of attrition as they begin college. The transmission of social status across generations differs substantially depending on college readiness. Low-income and first-generation students who begin college academically prepared for success have as strong a chance of continuing past their first year as their equally prepared higher income and continuing generation peers. On the other hand, results show higher income and continuing generation students who are less ready for college have an advantage over similarly less-ready, lower income, and first-generation students. It is clear from these data that college readiness moderates retention for low-income and first-generation students. These results move our understanding of what contributes to educational attainment differences considerably beyond past studies that have found more generally that socioeconomic status (Astin & Oseguera, 2005; Bozick, 2007; Cabrera et al., 2005; DeAngelo et al., 2011; Herzog, 2005; Hu & St. John, 2001; Ishitani, 2006; Lohfink & Paulsen, 2005; Leppel, 2002; Paulsen & St. John, 2002) and first-generation status (Ishitani, 2006) significantly predict retention and degree completion. Specifically, results in this study, combined with prior research by Lofhink and Paulsen (2005) and Warburton, Bugarin, Nuñez, and Carroll (2001), solidify our understanding of the degree attainment process for students of differing readiness and social status background. In targeting efforts at a policy and practice level to improve first-year retention outcomes, the first important step is identifying not only which groups are at risk but why.

The main implication from this study, based on the findings related to family income and generation status, is that in the absence of academic readiness for college, current systems of higher education favor students from higher social status backgrounds. Higher income and continuing generation student characteristics thus amount to a distinct advantage in the college environment, which in the absence of beginning college ready for success, act as tacit markers of academic potential that are recognized, reinforced, and rewarded—and result in higher rates of first-year retention. For lower income and first-generation students who begin college-ready, readiness itself acts as a tacit marker of academic potential and a marker that they share equally with their higher income and continuing generation peers. All of this means that among the less-ready student group at this status transmission juncture, there are fewer possibilities for status mobility and more status maintenance or reproduction. Meanwhile, for all students who come to college prepared for academic success, both status mobility and status maintenance are possible.

Findings in this study also reveal that less-ready, first-generation, and lower income students benefit from a campus environment with higher aggregate parental income levels among students. This benefit likely stems from the higher expectations for success among the student peer group (Palardy, 2013), the higher expectations for student success overall on these campuses, and the climate for success that manifests itself through faculty, staff, and administrators’ interactions with students as well as the services aimed at developing student potential (Stuber, 2011). Significantly, less-ready, higher income students benefit as much as their lower income peers from this climate. In totality, this means that advantages accumulate for less-ready, higher income students.

To promote equity, institutions must develop specific supports aimed at recognizing and developing the academic potential of less-ready, lower income, and first-generation students, part of what Rendn (1994) terms validation. Validation may be particularly important to less-ready, lower income, and first-generation students since the customs, traditions, and cultural values at college are more familiar to students with higher social status backgrounds (Reid & Moore, 2008).

Results in this study indicate that 75% of the attrition in the first college year is among students who begin college less-ready and that African American, Latino/a, lower income, and first-generation students have a much higher representation in the less-ready group than in the college-ready group. Thus, the very students who are most likely to have been disenfranchised at all earlier points in the educational pipeline (Adelman, 2006; Roderick et al., 2009) are also the most likely to leave college in the first year. This means that higher education institutions act as a lever of reproduction during the first college year, sorting out students who lack the desired markers of academic potential recognized by college communities. Given current policies and practices at the higher education level, ensuring lower income and first-generation students begin college academically ready for success has vast potential to raise retention. For elementary and secondary education reform, this finding suggests that efforts to increase readiness (Conley, 2005, 2010) can make a difference.

There are also important implications in this study for policymakers with respect to the differential role of readiness in retention based on the resources students have available to fund college. We found differences in retention between college-ready and less-ready students in the effect of using loans (for less-ready students, loans in any amount pose a retention risk; for college-ready, only loans over $6,000 pose a retention risk) and grants (less-ready students require $6,000 or more to see a retention benefit; college-ready students benefit from all amounts of grant aid). This means that current financial aid policies, which increasingly favor merit over need in grant-based aid, heavier reliance on loans, and tax credits that amount to regressive taxation (Perna & Finney, 2014), only increase differential retention rates for less-ready students. Our results buttress findings from past studies that current financial aid policies at all levels—institutional, state, and federal—are failing to support our national goal of increasing degree attainment. During the first college year, these aid policies also act as a mechanism of social reproduction or exceptionalism (rewarding the “few” academically ready, lower income, and first-generation students), both of which effectively maintain social inequality. Indeed, these policies very often reward students who would remain in the degree attainment pipeline regardless of aid or aid type received (Alon, 2011; Perna & Finney, 2014).

Other findings in this study related to the financial resources students have to fund college add to our implications for policymakers regarding financial aid. Specifically, the finding that a student's family financial contribution to funding the first year of college does not predict retention for either readiness group (controlling for other financial aid and perception of concerns about financing college), when coupled with our findings related to the overall positive relationship of grants and overall negative relationship of loans to retention, likely means that retention is supported by a stable source of non–loan based funds in general rather than a particular source. Herzog’s (2005) findings for low-income students and retention confirm the importance of having a stable source of enough non–loan based aid to improving retention outcomes. The findings in this study that major concerns about the ability to finance college has a negative effect on retention for both groups further supports our conclusion about the need for enough available stable non–loan based aid and may be a strong additional signal to less-ready, lower income, and first-generation students about their actual potential for realizing a degree. Past research linking major financial concerns to attrition (Astin & Oseguera, 2005; DeAngelo, 2014; DeAngelo et al., 2011; Paulsen & St. John, 2002) mirror these findings. Additionally, findings in this study, with respect to college choice and the importance of various factors in the process of choosing which college to attend, also support the importance of amply funding marginalized groups in college. The more importance students place in the choice process on attending a college based on cost— which we believe encompasses having the appropriate type and amount of funds to cover costs—the higher the rates of first-year retention. Refocusing aid policies on need and providing enough aid to cover college costs irrespective of college readiness is an essential component of increasing degree attainment nationally.

How might the totality of results from this study inform basic theories of educational attainment? Findings from this study lend support both to the narrative that academic achievement and expectations for achievement are the primary mechanism for social mobility and success over the life course as well as narratives emphasizing the continuing and predominate role of social status of origin and material circumstances to social mobility and success over the life course. For students who begin college with less academic readiness, the weight of the evidence suggests that status begets status, with social status of origin and material circumstances prevailing (Kerckhoff, 1984). For students who begin college academically prepared, the evidence in this study suggests that social status of origin and material circumstances matter less to educational attainment at the first-year retention juncture. Academic readiness acts as a mediator to retention for lower income and first-generation students, interrupting mechanisms that transmit social status across generations.

Our findings may have further implications regarding the role of institutional signals in the status attainment process. We add to Bozick et al.’s (2010) push toward greater theoretical complexity in understanding the process through which individuals realize limits on the development of their potential. While the large-scale data employed here do not allow us to closely interpret students’ experiences, we hypothesize that less-ready, lower income, and first-generation students receive signals that do not support their success from institutional actors, policies, and practices, some of which are related to financial aid. Our results suggest that these factors strongly relate to attrition for lower income and first-generation students who are not college-ready. It appears that these students choose not to continue to invest in a degree, perhaps taking them as a signal that they are unlikely to earn a degree (Beekhoven, De Jong, & Van Hout, 2002). For less-ready, lower income, and first-generation students, these messages become toxic to retention.

Extending Lucas’s (2001) theory of effectively maintained inequality to higher education settings, our findings suggest the ways in which institutions recognize and support academic potential affect less-ready students in ways that effectively maintain societal inequality. Higher income and continuing generation less-ready students exhibit the markers of academic potential valued in higher education environments beyond readiness, and institutions invest in the retention of these students (Alon, 2011). Despite their lack of academic preparation, this gives higher income and continuing generation less-ready students a significant advantage that supports maintenance of their stratum of origin as adults. This advantage effectively maintains inequality in much the same way as tracking does at the primary and secondary levels. As Lucas describes it, tracking works as a system to maintain inequality earlier in the educational pipeline, with students from higher status backgrounds disproportionately dominating higher tracks.

In working to create more equitable practice once students begin college, we must recognize that the curricular and co-curricular reforms in higher education of the past 30 years aimed at improving outcomes in the first year of college (see Upcraft, Gardner, Barefoot, & Associates, 2005) have not been retaining less-ready, lower income, and first-generation students. For instance, many campuses have expanded on-campus housing and even required students to live on campus at least during their first college year as one measure of their efforts to increase retention (Zeller, 2005). Our finding that neither readiness group showed a difference in first-year retention if living on campus or living at home challenges the effectiveness of this effort. Lohfink and Palusen (2005) had similar results in their examination of retention by generation groups. While living on campus provides access to valuable resources and experiences that may be related to retention (see DeAngelo, 2014; Pike, Schroeder, & Berry, 1997; Schudde, 2011), our findings suggest that just living on campus makes little retention difference. In fact, given the additional cost of living on campus versus living at home, it is important to recognize that living at home with family does not itself put students at a disadvantage.

The challenge our study poses is for institutions to provide the on-campus experiences and access to resources that support retention to all students regardless of living arrangement. Currently, most institutions are organized to deliver the large majority of support and services to first-year students through mechanisms that work primarily for students who live on campus (Jacoby, 2014). Changing this will take a reimagining of co-curriculum to serve all students. This will only occur through a commitment to interrupting structures that reinforce the delivery of the benefits of the co-curriculum to students who have traditionally succeeded in higher education (Kezar, 2011). Certainly, results related to housing call on higher education professionals to consider more thoughtfully the purposes of housing investments and continued investment in increasingly more luxurious housing and amenities (Jacob, McCall, & Stange, 2013; Zeller, 2005). Specifically, we need to consider if these investments are about improving outcomes for all college students or wooing certain students to campus.

In developing practices to retain less-ready, lower income, and first-generation students, we also need to be mindful and supportive of family attachment and the importance and strength closeness with family confers (Guiffrida, 2005; Palmer, Davis, & Maramba, 2011; Stuber, 2011). Results in this study indicate that placing importance on living near home in their college choice process supports retention of less-ready students. Additionally, for both readiness groups, the further college is from home, the less likely a college is to retain the student. Guidance during the college choice process should reflect these realities.

Conclusion

One of the key strengths of this study is its use of a theory-rich analysis to examine the factors that contribute to retention in the first college year based on college readiness. As Lucas and Beresford (2010) discuss, studying significant educational transitions requires rich theoretical analyses. In this study, two theoretical narratives regarding the status attainment framework were employed to examine first-year retention separately for students who begin college academically ready for success and those who begin less ready. Although college degrees are increasingly necessary for entry into the middle class (Carnevale et al., 2010; Rothwell, 2012), degree attainment has not increased significantly among marginalized populations (Alon et al., 2010; Roksa, 2010). Increasing degree attainment in the United States depends on succeeding with students who begin college less academically ready and who are more vulnerable to attrition.

Findings from this study have revealed that academic readiness matters and that readiness mediates the relationships between social background factors and retention in the first college year. Specifically, our finding that academic readiness eliminates differences in retention for lower income and first-generation students while the relationship between family income and generation status remains significant for less-ready students adds significantly to our understanding of how status is transmitted across generations at this educational juncture. Evidence suggests that higher education environments contribute to social reproduction during the first college year, allocating students who begin college with less academic readiness toward different adult statuses based on social background factors. This means that the various policies and practices within and outside of the systems of higher education affect lower income and first-generation less-ready students in ways that their higher income and continuing generation peers do not experience. These processes effectively maintain inequality (Lucas, 2001). Mitigating these inequities will take a strong commitment at all educational levels. For policymakers and educators at the elementary and secondary level, the challenge from this study is to focus on readiness and practices that support success. Change within higher education will require a realignment and a reimagining of priorities, policies, and practices by all involved in the higher education effort. We must approach such reform holistically and creatively in order to fashion higher education environments conducive to broader student participation and success.

Footnotes

Appendix

Appendix B

Parameter Estimates Hierarchical Generalized Linear Modeling Analysis Predicting First Year Retention, by College Readiness Groups (Full Model)

	Less-Ready (N = 131,985)				College-Ready (N = 78,071)
	Coefficient	SE	Significance	Odds Ratio	Coefficient	SE	Significance	Odds Ratio
Study variable, Level 1
Parental income (low income reference group)
Lower-middle income ($40,000–$74,999)	.123	.027	.000***	1.131	.042	.041	.306	1.043
Upper-middle income ($75,000–$149,999)	.146	.031	.000***	1.157	.108	.046	.019	1.114
High income (above $150,000)	.108	.038	.005**	1.114	−.011	.054	.834	0.989
First-generation status	−.117	.028	.000***	0.890	−.091	.042	.033	0.913
Financial resources for first year of college
Aid for first year from family (no family resources reference group)
Less than $2,999	.064	.041	.116	1.066	.042	.053	.426	1.043
$3,000–$5,999	.105	.044	.017	1.111	.076	.065	.247	1.078
$6,000–$9,999	.101	.042	.016	1.106	.132	.075	.079	1.142
More than $10,000	.091	.042	.033	1.095	.063	.064	.324	1.065
Grants for first year (no grants reference group)
Less than $2,999	.105	.036	.011	1.110	.143	.049	.004**	1.153
$3,000–$5,999	.081	.041	.051	1.085	.139	.050	.006**	1.149
$6,000–$9,999	.095	.034	.006**	1.100	.152	.055	.006**	1.134
More than $10,000	.151	.036	.000***	1.162	.177	.057	.003**	1.194
Loans for first year (no loans reference group)
Less than $2,999	−.115	.037	.004**	0.892	−.086	.042	.042	0.918
$3,000–$5,999	−.202	.044	.000***	0.817	−.111	.051	.034	0.895
$6,000–$9,999	−.202	.038	.000***	0.817	−.254	.059	.000***	0.776
More than $10,000	−.267	.041	.000***	0.766	−.299	.051	.000***	0.742
Some financial concern (none reference group)	.026	.025	.297	1.027	−.043	.045	.345	0.958
Major financial concern (none reference group)	−.187	.038	.000***	0.829	−.305	.054	.000***	0.737
SAT composite score	.067	.012	.000***	1.069	.113	.015	.000***	1.119
Hours per week in high school senior spent studying/homework	.101	.008	.000***	1.106	.081	.013	.000***	1.084
Hours per week in high school senior spent on household duties	−.033	.009	.000***	0.968	−.029	.010	.006**	0.971
Choose institution because of costs	.109	.014	.000***	1.115	.151	.027	.000***	1.163
Choose institution wanted to live near home	.077	.018	.000***	1.080	.025	.025	.317	1.026
Distance college away from home	−.058	.013	.000***	0.944	−.146	.017	.000***	0.864
Place of residence in the fall (on-campus reference group)
Live with family	−.057	.049	.241	0.944	−.069	.070	.327	0.933
Other living arrangement	−.401	.067	.000***	0.670	−.501	.090	.000***	0.606
Plan to get a full-time job to pay for college	−.111	.014	.000***	0.895	−.087	.020	.000***	0.916
Sex (female)	.134	.032	.000***	1.143	.106	.055	.054	1.111
Race (White reference group)
American Indian	−.260	.115	.023	0.771	−.493	.222	.027	0.611
Asian	.348	.063	.000***	1.416	.394	.086	.000***	1.483
African American	.098	.065	.132	1.103	.117	.094	.213	1.124
Latino/a	.045	.071	.528	1.046	.058	.082	.481	1.059
Other race/ethnicity	.280	.118	.020	1.323	.371	.126	.004**	1.449
Multi-ethnicity	−.161	.045	.001**	0.851	−.163	.054	.003**	0.850
Control variables, Level 1
Construct: academic self-concept	.013	.002	.000***	1.013	.006	.002	.017	1.006
Construct: social self-concept	−.007	.002	.000***	0.993	−.005	.002	.007**	0.995
Hours per week spent on student clubs/groups	.058	.011	.000***	1.060	.043	.016	.006**	1.044
Performed volunteer work	.066	.017	.000***	1.068	.098	.027	.000***	1.103
Highest degree aspirations	.015	.022	.494	1.015	−.019	.039	.636	0.982
Intention to transfer to another college	−.247	.017	.000***	0.781	−.419	.027	.000***	0.658
Major declared	.005	.047	.924	1.005	−.005	.050	.915	0.995
Construct: college involvement	.012	.002	.000***	1.013	.011	.003	.000***	1.011
Plan to change major	.033	.012	.005**	1.034	.058	.022	.008**	1.060
Study variables, Level 2
Institutional income/socioeconomic status (aggregate)	.116	.052	.026*	1.123	.071	.053	.184	1.074
Control variables, Level 2
Selectivity	.030	.004	.000***	1.031	.027	.004	.000***	1.027
Control (private)	−.202	.060	.001**	0.817	−.139	.067	.038*	0.870
HBCU control	.682	.149	.000***	1.978	.740	.191	.000***	2.096
Institutional transfer climate (aggregate)	−.588	.148	.000***	0.556	−1.042	.200	.000***	0.353
Intercept	.819	.458	.074	2.268	1.681	.620	.007**	5.369

p < .05. **p < .01. ***p < .000 (significance level at p < .05 only marked for Level 2 predictors).

Acknowledgements

The authors thank Dr. Sean Kelly for his support and helpful feedback on this manuscript and the Higher Education Research Institute (HERI) at UCLA for allowing us data access.

Notes

Linda DeAngelo is assistant professor of higher education and Center for Urban Education Fellow at the University of Pittsburgh, 5908 Posvar Hall, 230 S. Bouquet St., Pittsburgh, PA 15260, USA; e-mail: deangelo@pitt.edu . She studies stratification, equity, and diversity issues, investigating how social inequalities are produced within higher education. Her work examines the differential effect of institutions on students, pipeline and educational transitions, and outcomes for first-generation, low-income, and underrepresented students. Currently her scholarship focuses on retention, degree completion, and access to and engagement in faculty mentorship.

Ray Franke is an assistant professor of higher education in the College of Education and Human Development at the University of Massachusetts Boston. His research foci include higher education finance and financial aid, education policy, and organizational change in the United States and internationally. His recent work has examined how federal, state, and institutional financial aid programs and policy affect access to and success in higher education for students from different socioeconomic and racial/ethnic backgrounds.

References

American College Testing Service. (2009). The condition of college readiness 2009. Iowa City, IA: ACT, Inc.

American College Testing Service. (2013a). 2013 retention/completion summary tables. Iowa City, IA: ACT, Inc.

American College Testing Service. (2013b). The reality of college readiness 2013. Iowa City, IA: ACT, Inc.

Adelman

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

Alon

(2007). The influence of financial aid in leveling group differences in graduating from elite institutions. Economics of Education Review, 26(3), 296–311. doi: 10.1016/j.econedurev.2006.01.003

Alon

(2011). Who benefits most from financial aid? The heterogeneous effect of need-based grants on students’ college persistence. Social Science Quarterly, 92(3), 807–829. doi: 10.1111/j.1540-6237.2011.00793.x

Alon

Domina

Tienda

(2010). Stymied mobility or temporary lull? The puzzle of lagging Hispanic college degree attainment. Social Forces, 88(4), 1807–1832.

Astin

A. W.

Oseguera

(2004). The declining “equity” of American higher education. The Review of Higher Education, 27(3), 321–341. doi: 10.1353/rhe.2004.0001

Astin

A. W.

Oseguera

(2005). Pre-college and institutional influences on degree attainment. In Seidman

(Ed.), College student retention: Formula for student success (pp. 245–276). Westport, CT: Greenwood Publishing Group.

10.

Beekhoven

De Jong

Van Hout

(2002). Explaining academic progress via combining concepts of integration theory and rational choice theory. Research in Higher Education, 43(5), 577–600. doi: 10.1023/A:1020166215457

11.

Blau

P. M.

Duncan

O. D.

(1967). The American occupational structure. New York, NY: John Wiley & Sons, Inc.

12.

Bowen

W. G.

Chingos

M. M.

McPherson

M. S.

(2009). Crossing the finish line: Completing college at America's public universities. Princeton, NJ: Princeton University Press.

13.

Bozick

(2007). Making it through the first year of college: The role of students’ economic resources, employment, and living arrangements. Sociology of Education, 80(3), 261–285.

14.

Bozick

Alexander

Entwisle

Dauber

Kerr

(2010). Framing the future: Revisiting the place of educational expectations in status attainment. Social Forces, 88(5), 2027–2052.

15.

Cabrera

A. F.

Burkum

K. R.

La Nasa

S. M.

(2005). Pathways to a four-year degree: Determinants of transfer and degree completion among socioeconomically disadvantaged students. In Seidman

(Ed.), College retention: A formula for success (pp. 167–210). Westport, CT: Praeger Publishers.

16.

Cabrera

N. L.

Miner

D. D.

Milem

J. F.

(2013). Can a summer bridge program impact first-year persistence and performance?: A case study of the New Start Summer Program. Research in Higher Education, 54(5), 481–498. doi: 10.1007/s11162-013-9286-7

17.

Carnevale

A. P.

Smith

Strohl

(2010). Help wanted. Washington, DC: Georgetown University Center on Education and the Workforce.

18.

Chen

DesJardins

S. L.

(2010). Investigating the impact of financial aid on student dropout risks: Racial and ethnic differences. The Journal of Higher Education, 81(2), 179–208. doi: 10.1353/jhe.0.0085

19.

College Board. (2012). The SAT report on college and career readiness: 2012. New York, NY: Author.

20.

College Board. (2013). 2013 SAT report on college & career readiness. New York, NY: Author.

21.

Conley

D. T.

(2005). College knowledge: What it really takes for students to succeed and what we can do to get them ready. San Francisco, CA: Jossey-Bass.

22.

Conley

D. T.

(2010). College and career ready: Helping all students succeed beyond high school. San Francisco, CA: Jossey-Bass.

23.

DeAngelo

(2014). Programs and practices that retain students from the first to second year: Results from a national study. New Directions for Institutional Research, 2013(160), 53–75.

24.

DeAngelo

Franke

Hurtado

Pryor

J. H.

Tran

(2011). Completing college: Assessing graduation rates at four-year institutions. Los Angeles, CA: Higher Education Research Institute.

25.

DesJardins

S. L.

Ahlburg

D. A.

McCall

B. P.

(2002). A temporal investigation of factors related to timely degree completion. Journal of Higher Education, 73, 555–581.

26.

Doyle

W. R.

(2006). Adoption of merit-based student grant programs: An event history analysis. Educational Evaluation and Policy Analysis, 28(3), 259–285.

27.

Dwyer

R. E.

(2013). The care economy? Gender, economic restructuring, and job polarization in the U.S. labor market. American Sociological Review, 78(3), 390–416.

28.

Dynarski

S. M.

Hemelt

S. W.

Hyman

J. M.

(2013). The missing manual: Using National Student Clearinghouse data to track postsecondary outcomes (No. w19552). Cambridge, MA: National Bureau of Economic Research.

29.

Ewert

(2010). Male and female pathways through four-year colleges: disruption and sex stratification in higher education. American Educational Research Journal, 47(4), 744–773. doi: 10.3102/0002831210374351

30.

Franke

(2012). Towards the education nation: Revisiting the impact of financial aid, college experience, and institutional context on baccalaureate degree attainment using a propensity score matching, multilevel modeling approach (Doctoral Dissertation). University of California Los Angeles, Los Angeles, CA.

31.

Fuligni

A. J.

Pedersen

(2002). Family obligation and the transition to young adulthood. Developmental Psychology, 38(5), 856.

32.

Geiser

Santelices

M. V.

(2007). Validity of high-school grades in predicting student success beyond the freshman year: High-school record vs. standardized tests as indicators of four-year college outcomes. Berkeley, CA: Center for Studies in Higher Education.

33.

Gilardi

Guglielmetti

(2011). University life of non-traditional students: Engagement styles and impact on attrition. The Journal of Higher Education, 82(1), 33–53.

34.

Grusky

D. B.

Western

Wimer

C. C.

(Eds.). (2011). The great recession. New York, NY: Russell Sage Foundation.

35.

Guiffrida

(2005). To break away or strengthen ties to home: A complex issue forAfrican American college students attending a predominantly White institution. Equity & Excellence in Education, 38(1), 49–60. doi: 10.1080/10665680590907864

36.

Herzog

(2005). Measuring determinants of student return vs. dropout/stopout vs. transfer: A first-to-second year analysis of new freshmen. Research in Higher Education, 46(8), 883–928. doi: 10.1007/sl1162-005-6933

37.

Horn

Kojaku

L. K.

(2001). High school academic curriculum and the persistence path through college. Washington, DC: National Center for Education Statistics.

38.

Hosmer

D. W.

Lemeshow

(2004). Applied logistic regression. New York, NY: John Wiley & Sons.

39.

St. John

E. P.

(2001). Student persistence in a public higher education system: Understanding racial and ethnic differences. Journal of Higher Education, 72(3), 265–286.

40.

Ishitani

T. T.

(2006). Studying attrition and degree completion behavior among first-generation college students in the United States. Journal of Higher Education, 77(5), 861–885. doi: 10.1353/jhe.2006.0042

41.

Jacob

McCall

Stange

K. M.

(2013). College as country club: Do colleges cater to students’ preferences for consumption? (NBER Working Paper Series). Cambridge, MA: National Bureau of Economic Research.

42.

Jacoby

(2014). Engaging commuter and part-time students. In Harper

S. R.

Quaye

S. J.

(Eds.), Student engagement in higher education: Theoretical perspectives and practical approaches for diverse populations (pp. 289–305). New York, NY: Taylor & Francis.

43.

Kalogrides

Grodsky

(2011). Something to fall back on: Community colleges as a safety net. Social Forces, 89(3), 853–878.

44.

Kerckhoff

A. C.

(1984). The current state of social mobility research. The Sociological Quarterly, 25, 139–153.

45.

Kezar

(2011). Rethinking postsecondary institutions for low-income student success: The power of post-structural theory. In Kezar

(Ed.), Recognizing and serving low income students in higher education: An examination of the institutional policies, practices, and culture (pp. 3–26). New York, NY: Routledge.

46.

Knotterus

J. D.

(1987). Status attainment research and its image of society. American Sociological Review, 52, 113–121.

47.

Leppel

(2002). Similarities and differences in the college persistence of men and women. The Review of Higher Education, 25(4), 433–450. doi: 10.1353/rhe.2002.0021

48.

Lohfink

M. M.

Paulsen

M. B.

(2005). Comparing the determinants of persistence for first-generation and continuing-generation students. Journal of College Student Development, 46(4), 409–428.

49.

Lucas

S. R.

(2001). Effectively maintained inequality: Education transitions, track mobility, and social background effects. American Journal of Sociology, 106(6), 1642–1690.

50.

Lucas

S. R.

Beresford

(2010). Naming and classifying: Theory, evidence, and equity in education. Review of Research in Education, 34(1), 25–84.

51.

McCall

(2001). Complex inequality. New York, NY: Routledge.

52.

National Student Clearinghouse. (2013). Snapshot report: Persistence-retention. College Station, TX: National Student Clearinghouse Research Center.

53.

Niehaus

Campbell

C. M.

Inkelas

K. K.

(2014). HLM behind the curtain: Unveiling decisions behind the use and interpretation of HLM in higher education research. Research in Higher Education, 55(1), 101–122. doi: 10.1007/s11162-013-9306-7

54.

Nora

Barlow

Crisp

(2005). Student persistence and degree attainment beyond the first year in college. In Seidman

(Ed.), College student retention: Formula for success (pp. 129–153). Westport, CT: Praeger.

55.

Oseguera

(2005). Four and six-year baccalaureate degree completion by institutional characteristics and racial/ethnic groups. Journal of College Student Retention: Research, Theory and Practice, 7(1), 19–59.

56.

Oseguera

Rhee

B. S.

(2009). The influence of institutional retention climates on student persistence to degree completion: A multilevel approach. Research in Higher Education, 50(6), 546–569. doi: 10.i007/s11162-009-9134-y

57.

Palardy

G. J.

(2013). High school socioeconomic segregation and student attainment. American Educational Research Journal, 50(4), 714–754. doi: 10.3102/0002831213481240

58.

Palmer

R. T.

Davis

R. J.

Maramba

D. C.

(2011). The impact of family support on the success of black men at an historically black university: Affirming the revision of Tinto's theory. Journal of College Student Development, 52(5), 577–597.

59.

Paulsen

M. B.

St John

E. P.

(2002). Social class and college costs: Examining the financial nexus between college choice and persistence. The Journal of Higher Education, 73(2), 189–236.

60.

Perna

L. W.

Finney

J. E.

(2014). The attainment agenda: State policy leadership in higher education. Baltimore, MD: Johns Hopkins University Press.

61.

Pike

G. R.

Schroeder

C. C.

Berry

T. R.

(1997). Enhancing the educational impact of residence halls: The relationship between residential learning communities and first-year college experiences and persistence. Journal of College Student Development, 38(6), 609–621.

62.

Planty

Bozick

Ingels

S. J.

(2006). Academic pathways, preparation, and performance: A descriptive overview of the transcripts from the high school graduating class of 2003. Washington, DC: National Center for Education Statistics.

63.

Pryor

J. H.

Hurtado

DeAngelo

Blake

L. P.

Tran

(2009). The American freshman: National norms, fall 2009. Los Angeles, CA: Higher Education Research Institute.

64.

Pryor

J. H.

Hurtado

DeAngelo

Sharkness

Romero

L. C.

Korn

W. S.

Tran

(2008). The American freshman: National norms, fall 2008. Los Angeles, CA: Higher Education Research Institute.

65.

Posselt

J. R.

Jaquette

Bielby

Bastedo

M. N.

(2012). Access without equity: Longitudinal analyses of institutional stratification by race and ethnicity, 1972–2004. American Educational Research Journal, 49(6), 1074–1111. doi: 10.3102/0002831212439456

66.

Raudenbush

S. W.

Bryk

A. S.

(2004). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage.

67.

Raudenbush

Bryk

Cheong

Y. F.

Congdon

(2004). HLM 6: Hierarchical and nonlinear modeling [Computer software]. Lincolnwood, IL: Scientific Software International.

68.

Reid

M. J.

Moore

J. L.

(2008). College readiness and academic preparation for postsecondary education oral histories of first-generation urban college students. Urban Education, 43(2), 240–261. doi: 10.1177/0042085907312346

69.

Rendn

L. I.

(1994). Validating culturally diverse students: Toward a new model of learning and student development. Innovative Higher Education, 19(1), 33–51.

70.

Roderick

Nagaoka

Allensworth

(2006). From high school to the future: A first look at Chicago Public School graduates’ college enrollment. College preparation and graduation from four year colleges. Chicago, IL: Consortium on Chicago School Research, University of Chicago.

71.

Roderick

Nagaoka

Coca

(2009). College readiness for all: The challenge for urban high schools. The Future of Children, 19(1), 185–210.

72.

Roksa

(2010). Differentiation and work: Inequality in degree attainment in U.S. higher education. Higher Education, 61(3), 293–308. doi: 10.1007/s10734-010-9378-7

73.

Rosenbaum

J. E.

(2001). Beyond college for all: Career paths for the forgotten half. London: Russell Sage Foundation.

74.

Rothwell

(2012). Education, job openings, and unemployment in metropolitan America. Washington, DC: Brookings.

75.

Schudde

L. T.

(2011). The causal effect of campus residency on college student retention. The Review of Higher Education, 34(4), 581–610.

76.

Sewell

W. H.

Haller

A. O.

Ohlendorf

G. W.

(1970). The educational and early occupational status attainment process: Replication and revision. American Sociological Review, 35(6), 1014–1027.

77.

Sewell

W. H.

Haller

A. O.

Portes

(1969). The educational and early occupational attainment process. American Sociological Review, 34(1), 82–92.

78.

Steinberg

Fletcher

A. C.

(1998). Data analytic strategies in research on ethnic minority youth. In McLoyd

V. C.

Steinberg

(Eds.), Studying minority adolescents: Conceptual, methodological, and theoretical issues (pp. 279–294). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

79.

Stuber

J. M.

(2011). Inside the college gates: How class and culture matter in higher education. Lanham, MD: Lexington Books.

80.

Tabachnick

B. G.

Fidell

L. S.

(2007). Experimental designs using ANOVA. Independence, KY: Thomson/Brooks/Cole.

81.

Terenzini

P. T.

Cabrera

A. F.

Bernal

E. M.

(2001). Swimming against the tide: The poor in American higher education. New York, NY: The College Board.

82.

Upcraft

M. L.

Gardner

J. N.

Barefoot

B. O.

, & Associates. (2005). Challenging and supporting the first-year student: A handbook for improving the first year of college. San Francisco, CA: Jossey-Bass.

83.

Walpole

(2003). Socioeconomic status and college: How SES affects college experiences and outcomes. The Review of Higher Education, 27(1), 45–73.

84.

Warburton

E. C.

Bugarin

Nuñez

A. M.

Carroll

C. D.

(2001). Bridging the gap: Academic preparation and postsecondary success of first-generation college students (NCES 2001–153). Washington, DC: U.S. Department of Education, National Center for Education Statistics.

85.

Warner

R. M.

(2012). Applied statistics: From bivariate through multivariate techniques. New York, NY: Sage.

86.

Wolniak

G. C.

Engberg

M. E.

(2010). Academic achievement in the first year of college: Evidence of the pervasive effects of the high school context. Research in Higher Education, 51(5), 451–467. doi: 10.1007/s11162-010-9165-4

87.

Zeller

W. J.

(2005). First-year student living environments. In Upcraft

M. L.

Gardner

J. N.

Barefoot

B. O.

& Associates (Eds.), Challenging and supporting the first-year student: A handbook for improving the first year of college (pp. 410–427). San Francisco, CA: Jossey-Bass.