Examining Retention and Transfer in Relation to Incoming Student Data for First-Generation and Continuing-Generation Students: What Differences Exist?

Abstract

First-generation (FG) students are generally less likely than their continuing-generation (CG) peers to persist and complete a degree. Using student data available at initial enrollment, this multi-institutional study examines retention and transfer at the second year in relation to academic readiness, financial resources, college intentions, enrollment attributes, and other demographic characteristics to determine whether the predictors and their effects differ between FG and CG students beginning at 4-year institutions. Students’ college intentions and enrollment attributes are included as possible barriers to academic and social integration at the initial institution. The study finds that parental education gaps in outcomes persist even after statistically controlling for incoming student information, and the effects of some predictors differ by parental education. The implications of the findings for early identification of students at risk of leaving their initial institution and for informing retention and transfer strategies that are aimed at equipping FG students for success are discussed.

Keywords

first-generation retention transfer attrition incoming student information academic readiness

First-generation (FG) students, or those whose parents have no college experience, comprise about one third of the undergraduate population (Skomsvold, 2015), and compared with their continuing-generation (CG) peers, they are generally less likely to persist in college and complete a degree (e.g., Ishitani, 2006, 2016; Redford & Hoyer, 2017). They also tend to be more likely to borrow and take out larger loans (Furquim, Glasener, Oster, McCall, & DesJardins, 2017), which can result in them accumulating debt without receiving the benefit of completing a degree. For these reasons, it is imperative for institutions and state systems to identify solutions and strategies that not only help students from these backgrounds enroll in college but also help them to persist to degree completion.

An initial step might include conducting local studies using incoming student information such as that to be illustrated in the current study. Incoming student information can be incorporated into early alert systems (Beck & Davidson, 2001; Faulconer, Geissler, Majewski, & Trifilo, 2014; Tampke, 2013) to identify students who may be at risk of leaving the institution and connect them with institutional services and supports when they begin college. Information from such local studies can help to (a) provide insights on gaps in outcomes among parental education groups, (b) identify unique barriers to success for FG students, and (c) inform retention and transfer strategies intended to help these students achieve their educational goals.

Background

A recent study by Cataldi, Bennett, and Chen (2018) suggests that only 56% of FG students had earned a credential or were still enrolled 6 years after initially entering college. In comparison, this percentage was considerably lower than that for CG students who had one or more parents with at least a bachelor’s degree (74%) and slightly lower than that for CG students whose parents had some college experience but neither had earned a bachelor’s degree (63%). Gaps also exist in earlier outcomes such as first-to-second-year persistence and attrition rates among parental education groups (e.g., Lohfink & Paulsen, 2005). Moreover, results from a study by Ishitani (2006) suggest that the greatest relative risk of dropping out occurs in the second year for FG students when compared with CG students whose parents earned a bachelor’s degree.

Given that research has consistently shown a strong positive relationship between students’ precollege academic readiness levels and their likelihood of persisting and completing a degree (Adelman, 2006; Kopp & Shaw, 2016; Schmitt et al., 2009), the gaps in college success rates by parental education are often partially attributed to FG students not entering the college environment as well prepared and equipped academically as their CG peers. For instance, proportionally fewer FG students take rigorous coursework in high school and earn Advanced Placement or International Baccalaureate credits (Cataldi et al., 2018), while more FG students take remedial coursework in college (Chen, 2016). FG students also tend to earn lower grades in their high school courses (Saenz, Hurtado, Barrera, Wolf, & Yeung, 2007) and lower scores on college admissions tests (ACT, 2015; College Board, 2015).

FG students have also been found to begin college with lower academic self-efficacy than their CG peers with comparable achievement levels (Cruce, Kinzie, Williams, Morelon, & Xingming, 2005; Ramos-Sánchez & Nichols, 2007) and are more apt to indicate that they feel less prepared for college (Bui, 2002). Academic self-efficacy, academic discipline, and other social and emotional learning (SEL) skills have been found to be positively related to college grades and retention even after controlling for academic achievement (Robbins, Allen, Casillas, Peterson, & Le, 2006; Robbins et al., 2004), even among FG students (Majer, 2009).

In addition to entering college less academically prepared, FG students generally have more financial needs and concerns than their CG peers that can reduce their chances of persisting and succeeding in college (Attewell, Heil, & Reisel, 2011; Pratt, Harwood, Cavazos, & Ditzfeld, 2017; Wilbur & Roscigno, 2016). For instance, FG students are generally more likely to come from lower income families (Redford & Hoyer, 2017), to be financially independent from their parents (Engle & Tinto, 2008), and to lack financial aid knowledge (Lee & Mueller, 2014). To overcome their financial challenges, FG students often work while attending college at a higher rate and for more hours than their CG peers (Engle & Tinto, 2008). Having to work many hours, especially off campus, can limit the amount of time students have to focus on their studies and can prevent them from academically and socially integrating into the college environment (Engle, 2007; Kuh, Kinzie, Buckley, Bridges, & Hayek, 2006).

According to Tinto (1975, 1993), academic and social integration into the college environment can positively influence students’ chances of returning to an institution. Besides working fewer hours while going to college, there are other related college-attending behaviors that can help foster academic and social integration into the college environment. These include enrolling full time, living on campus, and being involved in student organizations and campus clubs; these college-attending behaviors are ones that FG students are generally less likely than their CG peers to do (Engle & Tinto, 2008; Lohfink & Paulsen, 2005; Wilbur & Roscigno, 2016). While research suggests that students who enroll full time generally have higher persistence and degree completion rates than part-time students (Shapiro et al., 2016), findings on whether campus residency is associated with these outcomes have been mixed (e.g., Lohfink & Paulsen, 2005; Schudde, 2011). Yet, results from other studies suggest that campus residency helps to facilitate student engagement and a sense of belonging (Kuh, Kinzie, Schuh, Whitt, & Associates, 2010; Pascarella & Terenzini, 2005).

Extending beyond examining retention rates by generational status, there have been a few studies that have explored and found differences in the determinants of first-to-second-year retention by parental education group. For example, a single-institution study (D’Amico & Dika, 2013) that was limited to the inclusion of six predictors found that race/ethnicity and in-state residency played a more prominent role among FG students than among CG students. A second study (Latino et al., 2018) based on data from a predominantly Hispanic-serving postsecondary institution found that receiving need-based financial aid was positively associated with first-to-second-year retention among FG students but not among CG students. A third study (Lohfink & Paulsen, 2005) based on a national cohort of students beginning at 4-year institutions in 1995–1996 found that several demographic characteristics (e.g., being male, married, Hispanic); in-college experiences (e.g., academic integration index, total grant aid received); and institutional variables (e.g., enrollment size and control) played a greater role in reenrollment behaviors for FG students. In light of how the demographic composition of the college-going population has changed over the past two decades since the Lohfink and Paulsen study, the current study builds on these prior studies by examining these topics across multiple institutions for more recent cohorts of students using variables available at the time of college entry and known to be related to student retention. These variables are from the general areas of precollege academic readiness levels, financial resources, demographic characteristics, and initial enrollment attributes believed to help foster academic and social integration at the initial institution attended. This study extends beyond these studies by differentiating between two types of attrition (dropout and transfer) and examining whether there are group differences in the type of transfer (reverse vs. lateral) among those transferring to another institution.

Current Study

Students can find themselves in academic jeopardy or encountering problems assimilating into the college environment during the first year, making it more likely that they do not return in their second year (Kopp & Shaw, 2016). Many institutions set up early alert or warning systems as a mechanism for identifying students most likely to struggle within the first year (e.g., Faulconer et al., 2014; Tampke, 2013). The goal of these systems is early identification so that institutional supports and services can be offered when they might be most beneficial, such as upon college arrival. The current study takes place within the context of early alert systems and illustrates how institutions might use student information available at the time of initial enrollment to learn more about their incoming FG students and gain additional insights about how they might tailor their resources and supports to better meet their unique needs.

Building on prior research and using data available at initial enrollment related to students’ academic readiness, college intentions and goals, initial enrollment attributes, and demographic characteristics, a study was conducted to answer the following research questions: (a) Are there differences in student retention, transfer, and dropout rates at the beginning of the second year among parental education groups? (b) Are there differences in the relevant predictors and their effects on retention and attrition by parental education? and (c) Among students who transferred to another institution at the beginning of their second year, are there differences in the type of institution transferred to among parental education groups?

Method

Sample

Data were available for 111,177 ACT-tested students entering college for the first time in fall 2012, 2013, or 2014 at one of 23 4-year institutions from two state systems. The ACT-tested sample represented 68% of the initial sample of students who began at these institutions.¹ The 4-year institutions were somewhat diverse on their admissions policies (17% highly selective/selective, 48% traditional, and 35% liberal/open) and enrollment size (26% less than 5,000; 44% 5,000 to 19,999; 30% 20,000 or higher).² The two state systems provided students’ first-year outcomes that included credit hours attempted and earned and grade point average (GPA), as well as reenrollment status for fall of the second year. Subsequent enrollment for the second year was supplemented with data from the National Student Clearinghouse.

Study Outcomes

The primary outcome was whether a student returned during the fall of the second year to the initial institution. The outcome comprises the following three categories to allow for the examination of two types of attrition: returned to initial institution, transferred to another institution, or dropped out. These are point-in-time definitions of transfer and dropout. The secondary outcome was a binary outcome for the type of institution transferred to in the second year. Transferring to a 2-year institution (reverse transfer) was compared with transferring to another 4-year institution (lateral transfer).

Predictors

Many of the student-level predictors were obtained from the ACT Student Profile Section and Course and Grade Information Section that students complete when registering for the ACT.

Parental education

Students indicated the highest level of education attained by their mother/guardian 1 and father/guardian 2. This information was categorized into three groups: neither parent attended a higher education institution (labeled first-generation or FG), at least one parent had some college experience but neither completed a bachelor’s degree (labeled continuing generation—some college or CG-SC), or at least one parent earned a bachelor’s degree (labeled continuing generation—bachelor’s degree or higher or CG-BD). The definition of FG students in this study is consistent with that used in recent National Center for Education Statistics studies (Cataldi et al., 2018; Redford & Hoyer, 2017). In comparison, the Higher Education Act defines FG students as those for whom no parent or guardian has completed a bachelor’s degree (Higher Education Act of 1965, 1998 Higher Education Act Amendments, 1998); that is, the CG-SC group is combined with the FG group. CG-BD students served as the reference group.

Demographic characteristics

The demographic characteristics included the following: gender, race/ethnicity, annual family income, and median household income associated with student’s residential zip code. Race/ethnicity was categorized as African American, Asian, Hispanic, Other, White, and missing. The Other category comprises racial/ethnic groups with smaller sample sizes that included American Indian, Native Hawaiian/Pacific Islander, and Multiracial.

For annual family income, students were asked to estimate the approximate total combined annual income of their parents by selecting one of nine possible range options. These options were classified into the following three categories: less than $36,000 (low), $36,000 to $80,000 (medium), and more than $80,000 (high). Another measure of socioeconomic status included median household income according to student’s residential zip code; it was classified into the following three categories: $43,315 or less, $43,316 to $61,580, and more than $61,580.³

Academic readiness

Institutions often develop admission models using data from prior cohorts to estimate incoming students’ first-year grade point average (FYGPA) from their standardized test scores, high school coursework and grades, class rank, and other information (Clinedinst & Koranteng, 2018; Rigol, 2003) and use these predicted values as a single variable or index that encompasses students’ precollege readiness levels. Examples of institutions using this index or predicted FYGPA values to help identify students who may benefit from institutional supports and services or may be at risk of leaving the institution have been noted in the literature (e.g., Beaudoin & Kumar, 2012; D’Amico & Dika, 2013).

In a similar way, the academic readiness measure that was used in this study was a student’s institution-specific predicted FYGPA estimated from students’ ACT Composite scores and high school GPA (HSGPA). The ACT Composite score is the rounded average of the four subject area scores in English, mathematics, reading, and science that was obtained from a student’s latest test record prior to enrolling in college. HSGPA was based on students’ self-reports of their coursework taken in up to 23 specific courses in English, mathematics, social studies, and science, and the grades earned in those courses. Prior studies have shown that students report high school coursework and grades accurately relative to information provided in their official high school transcripts (Sanchez & Buddin, 2016; Shaw & Mattern, 2009). More details about how FYGPA was estimated are discussed in the Development of Academic Readiness Index section.

College intentions and educational goals

Students provided information about their intentions of living on campus (categorized as yes or no) and the number of hours they planned to work per week during their first year of college (options included none, 1 to 10, 11 to 20, 21 to 30, and more than 30 hours).⁴ Students were also asked about the highest level of education that they expected to complete, which was categorized as associate’s degree or vocational/technical program (labeled as associate’s degree or below), bachelor’s degree, beyond a bachelor’s degree, or other.

Enrollment characteristics

Enrollment characteristics included enrollment status (full time vs. part time) and distance from home. A student was considered to be a full-time student if they attempted 12 or more credit hours during their first fall term. Distance from home was calculated as the distance in miles between a student’s home address and college address.⁵ Due to the heavily right-skewed distribution of the distance values, it was categorized as 0 to 24 miles, 25 to 89 miles, and 90 or more miles from home. Distance from home has been shown to be positively related to the likelihood of transferring to another institution that is generally closer to home (Mattern, Wyatt, & Shaw, 2013).

Multiple Imputation

Some students did not provide responses to all the questionnaire items included in this study. The missing rate was 10% or below for most predictors; the rate ranged from <1% for median household income to 10% for parents’ education level. The one exception to this was annual family income that had a missing rate of 18%.

Multiple imputation was used to estimate missing values with plausible values based on nonmissing data (Rubin, 1987; Schafer & Graham, 2002). Five data sets were imputed. Models were developed for all five imputed data sets. The reported models were based on the average parameter estimates across the five imputed data sets.

Given that parental education was the primary variable of interest, follow-up analyses were conducted on the sample that excluded those who did not provide their parents’ education levels, as well as those that were missing race/ethnicity. In these sensitivity analyses, the estimates and significance levels for the individual predictors and interactions were similar to those reported here based on the imputed data sets.

Development of Academic Readiness Index

Hierarchical linear regression was used to estimate institution-specific predicted FYGPAs for students from their ACT Composite scores and HSGPAs. The outcome in these models was students’ actual FYGPA. The model included the two predictors as well as their interaction term. The intercept and the slopes for ACT Composite score and HSGPA were allowed to vary across institutions to develop institution-specific predictions. Estimates of the fixed effects from the prediction models and the variance estimates for the random effects averaged across the five imputed data sets are shown in Table 1.

Table 1.

Estimates of FYGPA Models.

Variable	Estimate	SE	Min	Max	p value
Fixed-effect parameter estimates
Intercept	2.645	0.028	2.644	2.645	<.001
HSGPA	0.383	0.011	0.382	0.385	<.001
ACT Composite	0.262	0.018	0.261	0.263	<.001
HSGPA × ACT Composite	0.050	0.003	0.049	0.051	<.001
Variance estimates of random effects
Intercept	0.017	0.006	0.017	0.018	.002
HSGPA	0.003	0.001	0.002	0.003	.006
ACT Composite	0.007	0.002	0.007	0.007	.004

Note. Estimates shown are the averages combined across the five imputed data sets. HSGPA and ACT Composite score were standardized to have a mean of 0 and standard deviation of 1. The interaction term did not randomly vary across institutions. SE = standard error; FYGPA = first-year grade point average; HSGPA = high school GPA.

The predicted FYGPAs at a typical institution as a function of the two predictors are shown in Figure 1. As illustrated in the figure, students with higher HSGPAs and ACT Composite scores had higher predicted FYGPAs. The typical correlation between predicted and actual FYGPA was .50 and ranged from .34 to .61 across institutions. These correlations are consistent with those reported in another study (D’Amico & Dika, 2013).

Figure 1.

Predicted FYGPAs at a typical institution as a function of ACT Composite score and HSGPA.

Analytic Techniques

Due to the nested structure of the data (i.e., students clustered within institutions), hierarchical regression models were developed to predict retention from the student characteristics. A hierarchical multinomial regression model was used for the three-category retention outcome, where those who returned to their initial institution in the second year were used as the base category. For the binary transfer type outcome, a hierarchical logistic regression model was used. In these models, intercepts were allowed to vary randomly across institutions. Institution-level variables of admission selectivity and enrollment size were also included in the models, as retention rates have been shown to vary by these characteristics (Kopp & Shaw, 2016; Lohfink & Paulsen, 2005). Interaction terms with parental education were examined to answer the second question; they were included when statistically significant at the .01 level.

For each variable, the odds ratio (OR) was reported as a means to compare the strength of the predictor-outcome relationships among student characteristics. Two ORs of attrition compared with the base category were estimated in the primary analyses: the OR of dropping out versus returning to the initial institution and the OR of transferring to another institution versus returning to the initial institution. In comparison with members of the referent group, an OR greater than 1.0 indicates that members of the subgroup of interest are generally more likely to experience the outcome, whereas an OR less than 1.0 indicates that they are less likely to do so. For each student characteristic that interacted with parental education on retention and attrition, adjusted ORs for the predictor were calculated within each parental education group.

Results

Description of Study Samples by Parental Education

The sample comprises 15% FG students, 33% CG-SC students, and 52% CG-BD students. Descriptive statistics on student demographics, college intentions, enrollment characteristics, and institution characteristics by parental education are provided in Table 2. Briefly, FG students and CG-SC students tended to be more likely than their peers to be female, Hispanic or African American, from a less affluent neighborhood, from a family with a lower annual income, to have plans of working more hours during their first year of college, and to attend a college that was closer to home. Compared with CG-BD students, FG students and CG-SC students were less likely to have intentions of living on campus, to enroll full time, and to attend a larger or more selective institution. In addition, predicted FYGPAs, actual FYGPAs, ACT Composite scores, and HSGPAs tended to be the lowest on average for FG students and the highest for CG-BD students (Table 3).

Table 2.

Description of Student Characteristics by Parental Education.

Student characteristic	FG		CG-SC		CG-BD
Student characteristic	n	Percent	n	Percent	n	Percent
Gender
Female	9,391	57.2	20,332	55.5	28,922	49.8
Male	7,029	42.8	16,296	44.5	29,207	50.2
Race/ethnicity
African American	3,130	19.1	5,582	15.2	4,686	8.1
Asian	499	3.0	406	1.1	1,102	1.9
Hispanic	1,312	8.0	1,320	3.6	1,661	2.9
Other	742	4.5	1,654	4.5	1,864	3.2
Missing	430	2.6	1,312	3.6	2,055	3.5
White	10,307	62.8	26,354	72.0	46,761	80.4
Residential median household income^a
Less than $43,316	8,034	48.9	14,251	38.9	13,387	23.0
$43,316 to $61,580	6,411	39.0	15,393	42.0	21,454	36.9
More than $61,580	1,975	12.0	6,984	19.1	23,288	40.1
Annual family income
Less than $36,000	8,486	51.7	11,241	30.7	5,841	10.0
$36,000 to $80,000	6,162	37.5	16,147	44.1	17,675	30.4
More than $80,000	1,772	10.8	9,240	25.2	34,613	59.5
Hours planned to work per week
None	1,560	9.5	4,902	13.4	13,615	23.4
1 to 10	2,955	18.0	7,857	21.5	17,643	30.4
11 to 20	7,559	46.0	17,208	47.0	21,758	37.4
21 to 30	3,508	21.4	5,744	15.7	4,496	7.7
31 or more	838	5.1	917	2.5	617	1.1
Intend to live on campus
No	6,797	41.4	12,268	33.5	11,140	19.2
Yes	9,623	58.6	24,360	66.5	46,989	80.8
Degree plans
Associate’s or below	837	5.1	1,196	3.3	549	0.9
Bachelor’s degree	10,513	64.0	22,021	60.1	28,769	49.5
Beyond bachelor’s	4,755	29.0	13,071	35.7	28,526	49.1
Other	315	1.9	340	0.9	285	0.5
Distance from home
Less than 25 miles	6,790	41.4	13,802	37.7	19,376	33.3
25 to 89 miles	6,145	37.4	13,202	36.0	15,930	27.4
90 or more miles	3,485	21.2	9,624	26.3	22,823	39.3
Enrollment status
Part time	2,200	13.4	3,226	8.8	2,313	4.0
Full time	14,220	86.6	33,402	91.2	55,816	96.0
Selectivity of institution attended
Selective	3,859	23.5	9,795	26.7	22,422	38.6
Traditional	8,931	54.4	19,977	54.5	29,214	50.3
Open/Liberal	3,630	22.1	6,856	18.7	6,493	11.2
Size of institution
Less than 5,000	2,205	13.4	3,747	10.2	3,028	5.2
5,000 to 19,999	6,834	41.6	14,590	39.8	18,979	32.6
20,000 or below	7,381	45.0	18,291	49.9	36,122	62.1

Note. Descriptive statistics based on first imputed data set. Similar statistics were seen for the other four imputed data sets. FG = first-generation; CG-SC = continuing generation—some college; CG-BD = continuing generation—bachelor’s degree.

^aMedian household income is based on students’ residential zip code.

Table 3.

Average Predicted FYGPA, ACT Composite Score, and HSGPA by Parental Education.

	Mean (SD)
Student characteristic	FG	CG-SC	CG-BD
ACT Composite score	20.34 (4.18)	21.55 (4.30)	23.76 (4.48)
HSGPA	3.19 (0.55)	3.27 (0.54)	3.44 (0.49)
Predicted FYGPA	2.42 (0.55)	2.54 (0.55)	2.77 (0.53)
Actual FYGPA	2.31 (1.11)	2.48 (1.07)	2.84 (0.93)

Note. Means and standard deviations based on first imputed data set. Similar statistics were seen for the other four imputed data sets. SD = standard deviation; FYGPA = first-year grade point average; HSGPA = high school GPA; FG = first-generation; CG-SC = continuing generation—some college; CG-BD = continuing generation—bachelor’s degree.

The lowest typical retention rate was seen for FG students (62%), and the highest rate was seen for CG-BD students (78%; unadjusted analyses; Figure 2). These gaps existed primarily because FG students and CG-SC students were more likely than CG-BD students to dropout (OR = 2.77 and 2.06, respectively) when compared with returning in the second year.

Figure 2.

Modeled retention and attrition rates by parental education.⁶

Multiple-Predictor Models of Attrition

The results of the multiple-predictor model are presented in Table 4. All of the student-level characteristics included in the models were found to be significantly related to student attrition. The institution-level characteristics were not significant predictors of either type of attrition. Briefly, the following characteristics were found to be associated with a reduced risk of dropping out when compared with returning in the second year: entering college better prepared academically (adjusted OR = 0.49 associated with a one standardized unit increase in predicted FYGPA), being female (adjusted OR = 0.77), being Hispanic or Asian (adjusted OR = 0.81 and 0.40, respectively, compared with White), coming from a more affluent neighborhood (adjusted OR = 0.86 and 0.68, compared with < $43,316), planning to work fewer hours while in college (adjusted OR = 0.43 to 0.87, compared with more than 30 hours), intending to live on campus (adjusted OR = 0.88), having educational goals of a bachelor’s degree or higher (adjusted OR = 0.72 to 0.73, compared with associate’s degree or below), enrolling full time (adjusted OR = 0.54), and being from a family with a higher annual income (adjusted OR = 0.84 and 0.67, compared with < $30,000).

Table 4.

Multiple-Predictor Results for First-to-Second Year Attrition.^a

Variable	Category	Dropped out vs. returned				Transferred vs. returned
Variable	Category	Estimate	SE	Adj-OR	p value	Estimate	SE	Adj-OR	p value
Intercept		–0.276	0.116		.017	–2.345	0.198		<.001
Parental education	FG	0.367	0.028	1.443	<.001	0.086	0.035	1.089	.014
Parental education	CG-SC	0.297	0.022	1.346	<.001	0.120	0.025	1.128	<.001
Gender	Female	–0.266	0.018	0.767	<.001	0.110	0.021	1.117	<.001
Race/ethnicity	African American	–0.075	0.029	0.928	.011	–0.063	0.036	0.939	.078
	Hispanic	–0.216	0.047	0.806	<.001	–0.099	0.055	0.906	.072
	Asian	–0.912	0.097	0.402	<.001	–0.401	0.097	0.670	<.001
	Other race	0.344	0.043	1.411	<.001	0.114	0.054	1.121	.034
	Missing	0.215	0.048	1.240	<.001	–0.364	0.068	0.695	<.001
Median household income	< $43,316	0.382	0.029	1.465	<.001	–0.066	0.032	0.936	.036
Median household income	$43,316 to $61,580	0.232	0.026	1.261	<.001	–0.022	0.027	0.978	.417
Hours planned to work per week	None	–0.835	0.059	0.434	<.001	–0.242	0.083	0.785	.004
	1 to 10	–0.711	0.057	0.491	<.001	–0.174	0.082	0.840	.035
	11 to 20	–0.439	0.053	0.645	<.001	–0.089	0.078	0.915	.257
	21 to 30	–0.139	0.054	0.870	.010	0.012	0.083	1.012	.882
Intend to live on campus	Yes	–0.131	0.021	0.878	<.001	0.152	0.028	1.164	<.001
Distance from home	25 to 89 miles	0.032	0.023	1.032	.162	0.546	0.029	1.726	<.001
Distance from home	90 or more miles	–0.288	0.027	0.750	<.001	0.645	0.030	1.906	<.001
Educational plans	Beyond bachelor’s	–0.325	0.053	0.723	<.001	0.144	0.080	1.155	.072
	Bachelor’s degree	–0.312	0.052	0.732	<.001	0.130	0.078	1.139	.097
	Other	–0.163	0.096	0.850	.089	0.067	0.148	1.070	.649
Enrollment status	Full time	–0.610	0.031	0.543	<.001	–0.281	0.043	0.755	<.001
Predicted FYGPA^b		–0.708	0.012	0.492	<.001	–0.436	0.013	0.646	<.001
Annual family income	$36,000 to $80,000	–0.178	0.025	0.837	<.001	0.016	0.034	1.016	.646
Annual family income	> $80,000	–0.409	0.028	0.665	<.001	–0.038	0.035	0.962	.280
Admission selectivity	Selective	–0.029	0.088	0.971	.741	–0.168	0.179	0.845	.347
Admission selectivity	Traditional	0.077	0.063	1.080	.219	0.067	0.125	1.069	.591
Institution size	< 5,000	0.043	0.082	1.044	.604	0.004	0.164	1.004	.980
Institution size	5,000 to 19,999	–0.094	0.064	0.910	.141	–0.007	0.131	0.993	.958

Note. SE = standard error; adj-OR = adjusted odds ratio; FYGPA = first-year grade point average; FG = first-generation; CG-SC = continuing generation—some college.

aEstimates shown are the averages computed across the five imputed data sets. The variability estimates (and SE) for the random intercepts were 0.011 (0.004) for dropped out versus returned and 0.052 (0.017) for transferred versus returned; both estimates were significantly different from zero; p = .004 and .002, respectively.

^bPredictor was standardized to have a mean of 0 and a standard deviation of 1.

The characteristics associated with a reduced risk of transferring to another institution when compared with returning to the initial institution in the second year included entering college better prepared academically (adjusted OR = 0.65 for predicted FYGPA), being male (adjusted OR = 0.90), being Asian (adjusted OR = 0.67, compared with White), not intending to live on campus (adjusted OR = 0.86), attending an institution closer to home (adjusted OR = 0.52 and 0.91, compared with 90 or more miles from home), and enrolling full time (adjusted OR = 0.76).

For model fit, the McFadden’s (1974) pseudo R² was 10.8%. This pseudo R² estimate is consistent with those reported in other studies on student retention (D’Amico & Dika, 2013; Kopp & Shaw, 2016). Pseudo R² values for binary or multinomial outcomes are typically smaller in magnitude than R² values for continuous outcomes.

First question

After statistically controlling for student and institution characteristics, the gaps in retention and dropout rates were reduced, but they were not completely eliminated. The adjusted odds of dropping out for FG students and CG-SC students were 1.44 and 1.35 times that of CG-BD students, when compared with returning in the second year (Table 4). The corresponding adjusted odds of transferring to another institution were 1.09 and 1.13, respectively.

The reductions in the gaps in attrition rates by parental education are further illustrated in Figure 2 (see adjusted analyses), where the other predictors in the model are set to the sample means. For this example, the difference in the dropout rate is 4 percentage points between FG and CG-BD students, when compared with 16 percentage points in unadjusted analyses.

Second question

The following predictors interacted with parental education on student attrition: predicted FYGPA (p < .001), gender (p < .001), race/ethnicity (p < .001), enrollment status (p < .01), and intentions to live on campus (p < .01). Table 5 compares the parameter estimates and adjusted ORs by parental education for the significant predictors. The effects for the other predictors such as the financial-related variables did not differ by parental education; their estimates would be the same as or similar to those shown in Table 4.

Table 5.

Effects of the Student Attrition Predictors That Differed by Parental Education.

Variable	FG			CG-SC			CG-BD
Variable	Estimate	SE	Adj-OR	Estimate	SE	Adj-OR	Estimate	SE	Adj-OR
Dropped out vs. returned
Predicted FYGPA^a	–0.615***	0.024	0.541	–0.650***	0.018	0.522	–0.809	0.018	0.445
Gender—female	–0.151***	0.040	0.860	–0.245*	0.029	0.783	–0.348	0.032	0.706
Race/ethnicity
African American	–0.100	0.055	0.905^b	–0.042	0.042	0.959^b	–0.102	0.050	0.903^b
Asian	–1.285***	0.187	0.277	–0.942	0.225	0.390	–0.588	0.150	0.556
Hispanic	–0.579***	0.080	0.561	–0.125*	0.081	0.883^b	0.127	0.085	1.135^b
Other	0.345	0.090	1.412	0.393	0.066	1.481	0.241	0.081	1.272
Missing	–0.344***	0.150	0.709^b	0.330	0.080	1.391	0.308	0.078	1.361
Full-time status	–0.499*	0.058	0.607	–0.632	0.053	0.532	–0.711	0.063	0.491
Plan to live on campus	–0.049*	0.041	0.953^b	–0.125	0.032	0.883	–0.179	0.035	0.836
Transferred vs. returned
Predicted FYGPA^a	–0.324***	0.034	0.724	–0.391***	0.022	0.677	–0.497	0.018	0.608
Gender—female	0.174	0.056	1.190	0.166*	0.037	1.180	0.061	0.030	1.063^b
Race/ethnicity
African American	0.102	0.079	1.108^b	–0.136	0.054	0.873^b	–0.077	0.056	0.926^b
Asian	–0.644	0.201	0.525	–0.607	0.233	0.545	–0.210	0.131	0.810^b
Hispanic	–0.296*	0.114	0.744	–0.057	0.099	0.945^b	0.007	0.088	1.007^b
Other	0.041	0.138	1.042^b	0.036	0.087	1.037^b	0.197	0.080	1.218^b
Missing	–0.550	0.215	0.577	–0.416	0.127	0.660	–0.293	0.096	0.746
Full-time status	–0.042**	0.091	0.959^b	–0.338	0.070	0.713	–0.416	0.074	0.659
Plan to live on campus	0.238*	0.065	1.269	0.214*	0.043	1.239	0.053	0.044	1.054^b

Note. SE = standard error; adj-OR = adjusted odds ratio; FYGPA = first-year grade point average; FG = first-generation; CG-SC = continuing generation—some college; CG-BD = continuing generation—bachelor’s degree.

p value comparing whether parameter estimate for FG students and those whose parents have some college experience is different from the estimate for students whose parents have earned a bachelor’s degree or higher. *p < .05. **p < .01. ***p < .001.

^aPredictor was standardized to have a mean of 0 and a standard deviation of 1.

^bIndicates that the 99% confidence interval for adjusted OR includes 1; that is, the associated parameter estimate is not significantly different from zero.

According to the results in Table 5, entering college better prepared academically as measured by having a higher predicted FYGPA played a greater role in reducing the likelihood of dropping out and transferring to another institution (when compared with returning in the second year) among CG-BD students (adjusted OR = 0.45 and 0.61, respectively) than it did among FG and CG-SC students (adjusted OR = 0.54 and 0.72 for FG students).⁶ ⁷ ⁸ Enrolling as a full-time student reduced the likelihood of dropping out for each parental education group and reduced the chances of transferring to another institution for CG-BD and CG-SC students but not for FG students. Having intentions of living on campus was associated with a reduced odds of dropping out for both CG groups. This was not seen among FG students. Instead, planning to live on campus was associated with greater odds of transferring to another institution for FG students, as well as for CG-SC students but not for CG-BD students.

In terms of student demographics, the odds of dropping out were smaller for females than for males though this effect was less pronounced among FG students than among CG-BD students. In comparison, the odds of transferring were greater for females than for males among FG and CG-SC students but not among CG-BD students. For race/ethnicity, Asian students were less likely than White students to dropout and to transfer to another institution, with a more pronounced effect among FG and CG-SC students than among CG-BD students (e.g., adjusted OR = 0.28 and 0.53 for FG and 0.56 and 0.81 for CG-BD, respectively). Hispanic students were less likely than White students to dropout and to transfer to another institution (adjusted OR = 0.56 and 0.74) among FG students but not among CG-SC and CG-BD students (e.g., adjusted OR = 1.14 and 1.01 for CG-BD).

To further illustrate the interaction with students’ incoming readiness levels, Figure 3 provides retention and attrition rates by parental education and predicted FYGPA, holding all other predictors constant at their sample means. In this example, we see that for all three parental education groups that as a student’s predicted FYGPA increases, their chances of returning to the initial institution increase while their chances of dropping out and their chances of transferring decrease. Another observation from this example is that there appears to be larger gaps in retention rates between FG and CG-BD students and between CG-SC and CG-BD students among those who are entering college better prepared academically.

Figure 3.

Retention and attrition rates by parental education and predicted FYGPA holding all other predictors constant at sample means.

Figure 4 provides retention and attrition rates by parental education and enrollment status, holding all other predictors constant at their sample means. According to this example, there is a slightly larger difference in retention rates between full- and part-time students among CG-BD students than there is among FG students. This difference is primarily due to full-time students being less likely than part-time students to transfer among CG-BD students (adjusted OR = 0.66) but not among FG students (adjusted OR = 0.96).

Figure 4.

Retention and attrition rates by parental education and enrollment status holding all other predictors constant at sample means.⁸

Third question—transfer type

Given that more than 90% of students from each parental education group indicated that they had educational aspirations of obtaining at least a bachelor’s degree, we examined where students who transferred to another institution were going. Among those who transferred in the second year, the odds of reverse transferring to a 2-year institution for FG students and CG-SC students were 1.90 and 1.44 times that of CG-BD students, after statistically controlling for the initial institution attended. This translated to 16 and 9 percentage point differences in reverse transfer rates when comparing FG (56%) and CG-SC (49%) students to CG-BD (40%) students, respectively. Even after statistically controlling for the other institution and student characteristics included in this study, FG and CG-SC students were significantly more likely than CG-BD students to reverse transfer to a 2-year institution (adjusted OR = 1.39 and 1.19, respectively).

Discussion

This study examined retention and attrition at the second year in relation to student information available at the time of college enrollment that could be used in early alert systems. The student attributes included precollege academic readiness levels, financial resources, demographic characteristics, and other variables thought to serve as proxies for barriers to academic and social integration at the initial institution attended. Student characteristics were examined in relation to two types of attrition—dropping out of college and transferring to another institution—in comparison with returning in the second year.

A Reduction in the Gaps

Compared with their CG-BD peers, FG and CG-SC students tended to be at greater risk of dropping out or transferring to another institution in the second year even after statistically controlling for other student attributes. A comparison between the unadjusted and adjusted analyses highlights that these other student characteristics help to explain but do not completely eliminate the gaps in retention and attrition rates among parental education groups. These findings are consistent with those reported by others on persistence (Ishitani, 2016; Kopp & Shaw, 2016) and degree completion (Wilbur & Roscigno, 2016). They support the need for institutional programs—such as early high school outreach programs, summer bridge programs, academic supports and enrichment programs, mentoring and advising programs, and skills learning support programs—designed to help FG and CG-SC students succeed and persist in college. The findings also illustrate how institutions could use incoming student information to identify those who might benefit early on from these types of institutional services and supports before it is too late and students find themselves in academic jeopardy or no longer enrolled at the institution (Beck & Davidson, 2001).

Differences in Predictors

Another finding of the study was that the effects of some but not all of the predictors related to student attrition differed across the parental education groups. These included academic readiness, enrollment status, gender, race/ethnicity, and intentions of living on campus. The findings from such analyses have implications for early alert systems in the sense that the parameter estimates for these predictors are allowed to vary across parental education groups when estimating students’ chances of returning in the second year.

Academic readiness

The measure of academic readiness used in this study was students’ predicted FYGPA based on their ACT Composite score and HSGPA, an index that institutions might develop for use in their admissions process and have available for use in their early alert warning systems (e.g., D’Amico & Dika, 2013). Academic readiness was negatively related to dropout and transfer for all three parental education groups, suggesting that students who entered better prepared academically were more likely to be retained than those entering less academically prepared. However, the strength of these associations was significantly smaller for FG and CG-SC students than for CG-BD students. These findings suggest that if institutions are only offering or recommending their institutional supports and services designed for FG students to those entering less prepared academically, they might want to open these opportunities to all FG students.

To better understand the supports and services that might be most beneficial to FG students, future research should explore whether the group differences in the academic readiness–student retention findings are due to differences in other factors that were not available in this study. For example, given that FG students often lack early exposure to and knowledge about the college environment (Engle, 2007) and the guidance at home that can help contribute to student success in college (Saenz et al., 2007; Westbrook & Scott, 2012), FG students can often experience a greater cultural shift upon matriculating to college and have a more difficult time making the transition and mastering the role of the college student. This can lead to them having poorer outcomes than their CG peers even after prior achievement levels are taken into account (Collier & Morgan, 2008). Relatedly, students’ academic self-efficacy or other SEL skills might be a missing factor because FG students have been found to have significantly lower academic self-efficacy (Ramos-Sánchez & Nichols, 2007) even among students with comparable achievement levels (Cruce et al., 2005).

Enrollment status

The negative effects of full-time enrollment on both types of attrition were estimated to be the largest among CG-BD students and the smallest among FG students. Given that full-time students generally spend more time on campus, they are more likely than part-time students to get involved in campus activities and clubs, interact with faculty, connect with peers, and take advantage of support services that can lead to greater academic engagement and social integration into the campus environment (Center for Community College Student Engagement, 2017). Unfortunately, FG students tend to be less likely to get involved in these types of activities (Wilbur & Roscigno, 2016), which may help to explain why full-time enrollment had a smaller effect on attrition for FG students. Yet, other research suggests that FG students may actually benefit more from participating in academic-related activities such as interacting with faculty (Lohfink & Paulsen, 2005). As such, institutions and their personnel may need to be more proactive in assisting and advising FG students on topics related to enrollment status and activity involvement, including helping them to overcome any unique challenges they may be experiencing that limits their opportunity to participate in these types of activities.

Intentions of living on campus

Having intentions of living on campus was found to be negatively related to dropout when compared with returning in the second year and positively related to transferring to another institution. Analyses by parental education revealed that the negative association with dropout was seen among CG students only, while the positive association with transfer was seen among FG and CG-SC students only. In contrast, the Lohfink and Paulsen (2005) study found that living on campus was not significantly related to first-to-second-year retention for FG students and CG students. Future research should explore this relationship using students’ actual campus residency status on a more recent cohort of students. In light of the reality that FG students are generally less likely than their CG peers to enroll full time and live on campus, institutions need to implement strategies and solutions that create ways for these students to have similar enriching campus experiences and interactions that may be more readily available to those enrolled full time and living on campus.

Demographic characteristics

Parental education was found to interact with gender and race/ethnicity on student attrition. For gender, there were smaller differences in dropout rates between female and male students among FG students. For race/ethnicity, Asian and Hispanic students were found to be less likely than White students to dropout and transfer in the second year with this finding being more pronounced among FG students than among CG students. Given that nearly one-half of Hispanic students entering college are FG (Skomsvold, 2015), further research is warranted on the Hispanic FG population. It could be that compared with the White FG students in this study, Hispanic FG students used institutional supports and services to a greater extent, leading to lower attrition. This possible explanation could not be validated as data on resource utilization was not available.

None of the financial-related variables—annual income, neighborhood median household income, or number of hours planned to work—interacted with parental education on student attrition, meaning that their effects were similar across the parental education groups. In contrast, Lohfink and Paulsen (2005) found that some of their financial-related variables positively influenced students’ likelihood of returning in the second year for FG students but not for CG students. These included total annual income and the grant aid received. Unfortunately, information about the financial aid received was not available in this study and highlights another area that could be explored in greater detail regarding its utility in early identification of students at risk of leaving their initial institution.

Transfer Type

While a majority of FG students indicated they had educational aspirations of obtaining a bachelor’s degree, they were more likely than their CG-BD peers to reverse transfer than to make a lateral transfer in the second year. This finding also held to a lesser degree for CG-SC students. Considering that reverse transfer is generally associated with lower rates of bachelor’s degree completion than lateral transfer (Hossler et al., 2012), these findings suggest that the bachelor’s degree aspirations for many FG students may go unfulfilled.

In a review of the literature on student transfer, Taylor and Jain (2017) noted that there are relatively few studies on the factors associated with reverse transfer and suggested that more research is needed on this topic to help inform transfer policies and practices and ensure the opportunity for upward social mobility for all, including FG and CG-SC students. One study on reverse transfer that was conducted by Hillman, Lum, and Hossler (2008) found that the two strongest predictors were high school preparation and major choice. They unfortunately did not have data available on generational status to compare reverse transfer rates among parental education groups. The findings from the current study not only help to fill this gap, but they also highlight the need for institutions and state systems to better understand why FG students are reverse transferring at higher rates, even after statistically controlling for academic readiness levels, financial measures, and other student characteristics. This type of information could help shape transfer policies and guidance programs that are aimed at helping FG students achieve their educational goals.

Limitations

Data for the study came from two state systems and were not a nationally representative sample. Therefore, some of the findings may be specific to these states and may not generalize. Moreover, this study focused on FG students initially enrolling at a 4-year institution. Given that FG students tend to be more likely than CG-BD students to initially enroll at a 2-year institution (Redford & Hoyer, 2017), future research could explore a similar topic for community colleges. Another limitation of the study is the lack of additional variables and information (e.g., financial aid received and SEL measures) that could contribute to identifying early on those who may be at greater risk of leaving the institution. Institutions know whether financial aid was received by students, and they could collect SEL measures on students at the time of freshman orientation or early on in a first-year college experience course. Future studies should consider incorporating some of these additional measures to better understand their role and whether they help to explain some of the results from the current study.

Despite these limitations, this multi-institutional study illustrates how institutions and state systems might conduct local studies of this nature to develop initial prediction models based on incoming student information that could be incorporated into early alert systems to (a) identify those who are at a greater risk of leaving the institution so that timely guidance and supports can be provided prior to or upon arrival to the institution and (b) learn more about the reenrollment and transfer behaviors of their incoming FG students and how they compare with their CG peers. Incoming student information was primarily collected from students at the time they were going through the college application process (from the ACT record). This included information on students’ college intentions and preferences about living on campus and the number of hours planned to work that were used as proxies of actual campus residency and outside work commitments. Students’ intentions about living on campus and the number of hours planned to work were both found to be related to student attrition in this study. Similar data could be collected by institutions from students at the time of the college application process. Another possible benefit of having students’ college intentions and preferences available is that it could be used to identify initial areas for discussion for faculty advisors and student services administrators to have with their advisees, especially in cases where actual enrollment behaviors differ from intentions and preferences. Given that students’ likelihood of returning in the second year is determined by multiple factors besides precollege attributes, early alert systems and predictions should be regularly updated and augmented with additional data and metrics from the first term and year of college as it becomes available to better gauge FG students’ chances of returning.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Justine Radunzel, PhD, is a principal research scientist in the Validity & Efficacy Research Department at ACT, Inc.

References

ACT. (2015). The condition of college & career readiness 2015: First-generation students. Iowa City, IA: ACT. Retrieved from https://equityinlearning.act.org/wp-content/uploads/2016/06/2015-first-generation.pdf

ACT. (2017). National collegiate retention and persistence-to-degree rates. Iowa City, IA: ACT. Retrieved from https://www.act.org/content/dam/act/unsecured/documents/Retention-Persistence-Tables-2017.pdf

Adelman

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

Attewell

Heil

Reisel

(2011). Competing explanations of undergraduate noncompletion. American Educational Research Journal, 48(3), 536–559. doi:10.3102/0002831210392018

Beaudoin

Kumar

(2012). Using data to identify at-risk students and develop retention strategies (University Leadership Council Custom Research Brief). Washington, DC: Education Advisory Board. Retrieved from https://www.csueastbay.edu/oaa/files/docs/acinfo/idatriskstud.pdf

Beck

H. P.

Davidson

W. D.

(2001). Establishing an early warning system: Predicting low grades in college students from survey of academic orientations scores. Research in Higher Education, 2(6), 709–723.

Bui

K. V. T.

(2002). First-generation college students at a four-year university: Background characteristics, reasons for pursuing higher education, and first-year experiences. College Student Journal, 36(1), 3–11.

Cataldi

E. F.

Bennett

C. T.

Chen

(2018). First-generation students: College access, persistence, and postbachelor’s outcomes (NCES 2018-421). U.S. Department of Education, Washington, DC: National Center for Education Statistics Retrieved from https://nces.ed.gov/pubs2018/2018421.pdf

Center for Community College Student Engagement. (2017). Even one semester: Full-time enrollment and student success. Austin, TX: The University of Texas at Austin. Retrieved from https://www.ccsse.org/docs/Even_One_Semester.pdf

10.

Chen

(2016). Remedial coursetaking at U.S. public 2- and 4-year institutions: Scope, experiences, and outcomes (NCES 2016–405). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2016/2016405.pdf

11.

Clinedinst

Koranteng

(2018). 2017 state of college admission. Alexandria, VA: National Association for College Admissions Counseling. Retrieved from https://www.nacacnet.org/globalassets/documents/publications/research/soca17final.pdf

12.

College Board. (2015). 2015 college-bound seniors total group profile report. New York, NY: The College Board. Retrieved from https://securemedia.collegeboard.org/digitalServices/pdf/sat/total-group-2015.pdf

13.

Collier

P. J.

Morgan

D. L.

(2008). “ Is that paper really due today?”: Differences in first-generation and traditional college students’ understandings of faculty expectations. Higher Education, 55(4), 425–446. doi:10.1007/s10734-007-9065-5

14.

Cruce

T. M.

Kinzie

J. L.

Williams

J. M.

Morelon

C. L.

Xingming

(2005, November). The relationship between first-generation status and academic self-efficacy among entering college students. Paper presented at the 30th Annual Meeting of the Association for the Study of Higher Education (ASHE), Philadelphia, PA. Retrieved from http://nsse.indiana.edu/pdf/conference_presentations/2005/First-Generation%20Academic%20Self-Efficacy%20-%20ASHE.pdf

15.

D’Amico

M. M.

Dika

S. L.

(2013). Using data known at the time of admission to predict first-generation college student success. Journal of College Student Retention: Research, Theory & Practice, 15(2), 173–192. doi:10.2190/CS.15.2.c

16.

Engle

(2007). Postsecondary access and success for first-generation college students. American Academic, 3(1), 25–48.

17.

Engle

Tinto

(2008). Moving beyond access: College success for low-income, first-generation students. Washington, DC: The Pell Institute. Retrieved from https://files.eric.ed.gov/fulltext/ED504448.pdf

18.

Faulconer

Geissler

Majewski

Trifilo

(2014). Adoption of an early-alert system to support university student success. Delta Kappa Gamma Bulletin, 80(2), 45–48.

19.

Furquim

Glasener

K. M.

Oster

McCall

B. P.

DesJardins

S. L.

(2017). Navigating the financial aid process: Borrowing outcomes among first-generation and non-first-generation students. ANNALS of the American Academy of Political and Social Science, 671(1), 69–91. doi:10.1177/0002716217698119

20.

Higher Education Act of 1965, 1998 Higher Education Act Amendments, Subpart 2—Federal Early Outreach and Student Services Programs, Chapter 1—Federal TRIO Programs SEC. 402A, 20 U.S.C. 1070a–11 (1998).

21.

Hillman

Lum

Hossler

(2008). Understanding Indiana’s reverse transfer students: A case study in institutional research. Community College Journal of Research and Practice, 32, 113–134. doi:10.1080/10668920701707797

22.

Hossler

Shapiro

Dundar

Chen

Zerquera

Ziskin

Torres

(2012). Reverse transfer: A national view of student mobility from four-year to two-year institutions (Signature Report No. 3). Herndon, VA: National Student Clearinghouse Research Center. Retrieved from https://nscresearchcenter.org/wp-content/uploads/NSC_Signature_Report_3.pdf

23.

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.

24.

Ishitani

T. T.

(2016). Time-varying effects of academic and social integration on student persistence for first and second years in college: National data approach. Journal of College Student Retention: Research, Theory & Practice, 18(3), 263–286. doi:10.1177/1521025115622781

25.

Kopp

J. P.

Shaw

E. J.

(2016). How final is leaving college while in academic jeopardy? Examining the utility of differentiating college leavers by academic standing. Journal of College Student Retention: Research, Theory, & Practice, 18(1), 2–30. doi:10.1177/1521025115579670

26.

Kuh

G. D.

Kinzie

Buckley

J. A.

Bridges

B. K.

Hayek

J. C.

(2006). What matters to student success: A review of the literature (Report commissioned for the National Symposium on Postsecondary Student Success: Spearheading a Dialog on Student Success). Washington, DC: National Postsecondary Education Cooperative. Retrieved from https://nces.ed.gov/npec/pdf/kuh_team_report.pdf

27.

Kuh

G. D.

Kinzie

Schuh

J. H.

Whitt

E. J.

, & Associates. (2010). Student success in college: Creating conditions that matter. San Francisco, CA: John Wiley & Sons.

28.

Latino

C. A.

Stegmann

Radunzel

Way

J. D.

Sanchez

Casillas

(2018). Reducing gaps in first-year outcomes between Hispanic first-generation college students and their peers: The role of accelerated learning and financial aid. Journal of College Student Retention: Research, Theory & Practice, 22, 441–463.

29.

Lee

Mueller

J. A.

(2014). Student loan debt literacy: A comparison of first-generation and continuing-generation college students. Journal of College Student Development, 55(7), 714–719. doi:10.1353/csd.2014.0074

30.

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.

31.

Majer

J. M.

(2009). Self-efficacy and academic success among ethnically diverse first-generation community college students. Journal of Diversity in Higher Education, 2(4), 243–250. doi:10.1037/a0017852

32.

Mattern

K. D.

Wyatt

J. N.

Shaw

E. J.

(2013). College distance from home: Implications for student transfer behavior. Journal of the First-Year Experience & Students in Transition, 25(1), 77–92.

33.

McFadden

(1974). Conditional logit analysis of qualitative choice behavior. In P. Zarembka (Ed.), Frontiers in econometrics (pp. 105–142). New York, NY: Academic Press.

34.

Pascarella

E. T.

Terenzini

P. T.

(2005). How college affects students, volume 2: A third decade of research. San Francisco, CA: Jossey-Bass.

35.

Pratt

I. S.

Harwood

H. B.

Cavazos

J. T.

Ditzfeld

C. P.

(2017). Should I stay or should I go? Retention in first-generation college students. Journal of College Student Retention: Research, Theory, & Practice, 36, 1–14. doi:10.1177/1521025117690868

36.

Ramos-Sánchez

Nichols

(2007). Self-efficacy of first-generation and non-first-generation college students: The relationship with academic performance and college adjustment. Journal of College Counseling, 10(1), 6–18. doi:10.1002/j.2161-1882.2007.tb00002.x

37.

Redford

Hoyer

K. M.

(2017). First-generation and continuing-generation college students: A comparison of high school and postsecondary experiences (NCES 2018-009). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2018/2018009.pdf

38.

Rigol

G. W.

(2003). Admissions decision making models: How U.S. institutions of higher education select undergraduate students. New York, NY: The College Board. Retrieved from https://files.eric.ed.gov/fulltext/ED562589.pdf

39.

Robbins

S. B.

Allen

Casillas

Peterson

C. H.

(2006). Unraveling the differential effects of motivational and skills, social, and self-management measures from traditional predictors of college outcomes. Journal of Educational Psychology, 98(3), 598–616. doi:10.1037/0022-0663.98.3.598

40.

Robbins

S. B.

Lauver

Davis

Langley

Carlstrom

(2004). Do psychosocial and study skill factors predict college outcomes? A meta-analysis. Psychological Bulletin, 130(2), 261–288. doi:10.1037/0033-2909.130.2.261

41.

Rubin

D. B.

(1987). Multiple imputation for nonresponse in surveys. New York, NY: Wiley.

42.

Saenz

V. B.

Hurtado

Barrera

Wolf

Yeung

(2007). First in my family: A profile of first-generation college students at four-year institutions since 1971. Los Angeles, CA: Higher Education Research Institute, UCLA. Retrieved from https://www.heri.ucla.edu/PDFs/pubs/TFS/Special/Monographs/FirstInMyFamily.pdf

43.

Sanchez

Buddin

(2016). How accurate are self-reported high school courses, course grades, and grade point average? Iowa City, IA: ACT. Retrieved from https://www.act.org/content/dam/act/unsecured/documents/5269-research-report-how-accurate-are-self-reported-hs-courses.pdf

44.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177. doi:10.1037//1082-989X.7.2.147

45.

Schmitt

Keeney

Oswald

F. L.

Pleskac

T. J.

Billington

A. Q.

Sinha

Zorzie

(2009). Prediction of 4-year college student performance using cognitive and noncognitive predictors and the impact on demographic status of admitted students. Journal of Applied Psychology, 94(6), 1479–1497. doi:10.1037/a0016810

46.

Schudde

L. T.

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

47.

Shapiro

Dundar

Wakhungu

P. K.

Yuan

Nathan

Hwang

(2016, November). Completing college: A national view of student attainment rates – Fall 2010 cohort (Signature Report No. 12). Herndon, VA: National Student Clearinghouse Research Center. Retrieved from https://nscresearchcenter.org/wp-content/uploads/SignatureReport12.pdf

48.

Shaw

E. J.

Mattern

K. D.

(2009). Examining the accuracy of self-reported high school grade point average (College Board Research Report 2009–5). New York, NY: The College Board. Retrieved from https://files.eric.ed.gov/fulltext/ED562616.pdf

49.

Skomsvold

(2015). Web tables—Profile of undergraduate students: 2011–12 (NCES 2015-167). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2015/2015167.pdf

50.

Tampke

D. R.

(2013). Developing, implementing, and assessing an early alert system. Journal of College Student Retention: Research, Theory, and Practice, 14(4), 523–532. doi:10.2190/CS.14.4.e

51.

Taylor

J. L.

Jain

(2017). The multiple dimensions of transfer: Examining the transfer function in American higher education. Community College Review, 45(4), 273–293. doi:10.1177/0091552117725177

52.

Tinto

(1975). Dropout from higher education: A theoretical synthesis of recent research. Review of Educational Research, 45(1), 89–125. doi:10.3102/00346543045001089

53.

Tinto

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

54.

Westbrook

S. B.

Scott

J. A.

(2012). The influence of parents on the persistence decisions of first-generation college students. Focus on Colleges, Universities, and Schools, 6(1), 1–9.

55.

Wilbur

T. G.

Roscigno

V. J.

(2016). First-generation disadvantage and college enrollment/completion. Socius: Sociological Research for a Dynamic World, 2, 1–11. doi:10.1177/2378023116664351