Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

The Influence of Co-Enrollment on the Success of Traditional Age Community College Students


by Gloria Crisp - 2013

Background/Context: Student utilization of complex enrollment patterns has been identified as a significant recent development in higher education. Nearly a dozen different multi-institutional attendance patterns have been identified, including co-enrollment. Co-enrollment is simultaneous enrollment at two or more colleges or universities during the course of a given term or semester. Although research exists to understand the factors promoting persistence and degree completion for community college students, work does not properly account for co-enrollment or other forms of multi-institutional attendance.

Purpose: The purpose of the study is to measure the impact of co-enrollment on success outcomes among a national sample of traditional age community college students. The following research questions are addressed: (a) In what ways are the characteristics of traditional aged community college students who co-enroll similar or different from students who do not co-enroll? (b) Does co-enrollment significantly increase or decrease the odds that a student will earn a degree and/or persist through the sixth year of college?

Subjects: The sample was drawn from participants in the second follow-up of the Beginning Postsecondary Students Longitudinal Study (BPS: 04/09). The pre-matched sample was limited to the 4,920 students who began their postsecondary education at a two-year institution and were traditional in terms of age. The matched sample included a sub-sample of 700 co-enrollees and 700 matched students.

Research Design: The study utilized a non-experimental quantitative design.

Data Collection and Analysis: Propensity score matching techniques were used on observational data as a means of reducing observed selection bias. After the sample was shown to be balanced on observed covariates, logistic regression analyses were used to measures the influence of co-enrollment on the success of community college students.

Results: Results demonstrate that, even after controlling for observable selection bias and variables previously shown to influence success outcomes for community college students, co-enrolling at more than one institution during a given semester significantly increases the odds that community college students will succeed.

Conclusions/Recommendations: Co-enrollment may promote access to four-year institutions for community college students desiring to transfer and earn a four-year degree. State and federal policy makers and institutional leaders need to be aware that non-linear attendance is a legitimate way for students to experience and be successful in college. Descriptive work is needed to understand how and at what point students are attending multiple institutions.


Improving key success outcomes such as persistence, degree completion, and transfer rates among community college students is critical to economic and educational welfare of the United States (Wells, 2008) as students attending two-year institutions comprise 40% of all postsecondary students. Community college students are much less likely to persist or earn a bachelor’s degree when compared to students who begin college at a four-year institution. Only about 50% of first-time community college students persist to the second year, compared with nearly 75% of students who begin college at a four-year institution (McIntosh & Rouse, 2009). Moreover, although roughly 80% of two-year students indicate that they intend to transfer to a four-year institution and earn a bachelor’s degree, only 23% of students successfully transfer within six academic years (United States Department of Education, 2005).


Although a good amount of research has been done to understand the characteristics, behaviors, and programmatic efforts that promote persistence and degree completion for community college students (e.g., Attewell, Lavin, Domina, & Levey, 2006; Calcagno, Crosta, Bailey, & Jenkins, 2006; Crisp & Nora, 2010; Dowd & Coury, 2006; Porchea, Allen, Robbins, & Phelps, 2010), existing work assumes attendance at a single institution or vertical transfer to a four-year institution after completing two years of coursework, failing to properly account for recent changes in how students are choosing to enroll at postsecondary institutions (McCormick, 2003). The following study considers the impact of one of many multi-institutional enrollment patterns in higher education, co-enrollment, on community college persistence and degree completion. For the present inquiry, co-enrollment is defined as simultaneous enrollment at two or more colleges or universities during the course of a given term or semester.


Results demonstrate that, even after controlling for observable selection bias and variables previously shown to influence success outcomes for community college students, co-enrolling at more than one institution during a given semester significantly increases the odds that community college students will succeed (i.e., complete a degree and/or persist in college). Results have important implications for research, demonstrating the importance of accounting for multi-institutional attendance patterns, including co-enrollment, in theoretical models to explain student success. Further, findings have direct implications for policy impacting students’ ability to seamlessly enroll at multiple institutions and institutional types as a system.


MULTI-INSTITUTIONAL ENROLLMENT PATTERNS


Supported by changes in the political, social, and economic environments (Longanecker & Blanco, 2003), student utilization of increasingly complex enrollment patterns has been identified as one of the most significant developments in higher education in recent decades (Adelman, 1999). It is estimated that roughly 60% of traditional-age undergraduates attend more than one institution before graduating college (Adelman, 2005), moving among independent institutions and multiple institutional types as a system (Kinnick, Ricks, Bach, Walleri, Stoering, & Tapang, 1998). Nearly a dozen different multi-institutional attendance patterns have been identified by researchers (e.g., Adelman, 2005; McCormick, 2003), thereby challenging the appropriateness of limiting measures of institutional effectiveness to earning a one- or two-year certificate or degree after attending a single community college or to vertical transfer after up to sixty hours of course work at a single community college (McCormick, 2003).  


The most typical multi-institutional attendance pattern is one-way transfer, in which a student completes no more than two years of course work at a community college and then permanently transfers to complete a degree at a four-year university (McCormick, 2003). However, it is becoming increasingly common for students (even those pursuing a one- or two-year degree or certificate) to move from institution to institution, vertically moving from a community college to a four-year university, floating laterally from a community college to another community college, or transferring from a four-year university to a community college (Townsend & Dever, 1999). Multi-institutional attendance patterns are becoming increasingly common for online students and those living in urban areas who can take courses from multiple local institutions (Borden, 2004; Crawley & LeGore, 2009; McCormick, 2003). These students are said to make up nearly one-third (28%) of students attending college today (Adelman, 2005). Many of these students co-enroll before and/or after transferring to another institution. Various names have been given to this type of multi-institutional attendance including “co-enrollment”, “swirling,” “concurrent enrollment,” “overlapping enrollment,” and “dual enrollment” (de los Santos & Wright, 1990; Peter, Cataldi, & Carroll, 2005). This type of enrollment behavior (hereinafter referred to as co-enrollment) may include simultaneously attending two or more community colleges, or enrolling at both a four-year institution and community college during a given semester.


Co-enrollment behavior may take many forms, both within and outside of cross-institutional agreements. In the simplest form, during a given semester a student may take three classes at a single community college and enroll in a fourth class at another two- or four-year institution. There are a growing number of formalized programs/agreements that promote co-enrollment at various institutions and institutional types. For example, California Senate Bill 1914 permits undergraduate students enrolled in any state community college to co-enroll, without formal admission, in one class per academic term at any California State or University of California institution. There are also a growing number of multi-institutional agreements that, presumably designed to promote community college transfer, allow co-enrollment, including El Paso Community College and The University of Texas at El Paso, and Oregon community colleges and Oregon State University.  


CO-ENROLLMENT LITERATURE


There has been a wealth of research conducted on students who utilize traditional enrollment and transfer patterns (Kinnick, et al., 1998). Yet, researchers have begun only recently to study multi-institutional enrollment and its impact on student outcomes (Peter et al., 2005). Studies to date conducted on co-enrollment have been predominantly descriptive, with national data revealing the majority of students who co-enroll are enrolled at both two- and four-year institutions, while a quarter of students concurrently enroll at two or more community colleges. Data also show that the large majority of students who co-enroll exclusively attend public institutions (National Student Clearinghouse Research Center, 2011).


Although there is much more work to be done, researchers have also begun to document student characteristics associated with co-enrollment behavior. For instance, Adelman (2005) found that more than half of students classified as co-enrollers earned 30 or more credits from a single community college. In addition, findings by Kinnick et al. (1998) revealed that older students (ages 22-30) were more likely to follow co-enrollment patterns, moving among institutions, while African American students were more likely to follow more traditional linear enrollment patterns when compared to other ethnic groups. Moreover, findings by Rab (2004) suggested that students who were from lower socioeconomic backgrounds were more likely to co-enroll. In contrast, findings by Peter et al. (2005) indicated that younger students classified as dependent, who did not delay enrollment in college, and attended college full-time, were more likely to attend multiple institutions (including but not limited to co-enrollment).


Somewhat mixed findings have been found by researchers investigating the impact of co-enrollment on student outcomes. For example, an analysis of the earlier Beginning Postsecondary Students Longitudinal Study (BPS: 96/2001) data revealed that students who co-enrolled had higher first-year grades than students who transferred without co-enrolling. Moreover, students who co-enrolled but who did not transfer were found to perform comparably in the first year to students who enrolled at a single institution. Further, the persistence rates were found to be higher among students who attended other institutions but who did not transfer when compared to students who attended only one institution (McCormick, 2003). Evidence to support the positive impact of co-enrollment on student outcomes has also been found by Herzog (2005), who used multinomial logistic regression analyses to measure variables that increased or decreased the odds of four-year students transferring and persisting using co-enrollment as an independent variable. Results indicated co-enrollment significantly reduced the odds of student attrition and transfer-out risk during the first year of college. Furthermore, Peter et al. (2005) used the BPS:96/01 data to examine the relationship between co-enrollment and persistence, degree attainment, and time to degree. Descriptive results indicated that students who co-enrolled were more likely than those who did not co-enroll to earn a degree or persist to the sixth year of college.


In contrast, Maggard (2009) used data from the National Student Clearinghouse to descriptively compare graduation rates between students who did and did not co-enroll in the University of Wyoming system. No substantial differences were found between the graduation rates of the two groups. Evidence also exists to suggest that co-enrollment may in fact be negatively related to degree completion among students who begin college at a four-year institution, including findings by Adelman (2005) who used data from the National Education Longitudinal Study of 1988 (NELS:88/2000) to descriptively compare the degree completion rates of students who did and did not co-enroll. Similarly, Rab (2004) used the NELS: 88/2000 dataset to examine the relationship between co-enrollment and timely degree completion among students who began college at a four-year institution. Regression results indicated that co-enrollment decreased the odds of earning a degree for students from lower socio-economic groups.


PURPOSE OF THE STUDY


In sum, research to date on co-enrollment has been mainly descriptive, and somewhat contradictory results have been found by researchers utilizing multivariate statistics to measure the impact of co-enrollment on student outcomes. Conflicting findings are thought to be due, in part, to methodological limitations, including selection bias, as it appears that students who co-enroll may be systematically from those who do not co-enroll (e.g., Kinnick et al., 1998; Rab, 2004). As a result, the covariate distributions are likely substantially different between groups, thereby causing imprecise coefficient estimates and model misspecification in regression models (Titus, 2007) that likely contribute to contradictory findings within this line of work. It should be noted that the majority of co-enrollment research has been predominately focused on four-year students, and with the exception of studies by Adelman (2005) and Rab (2004), the selection of covariates used in quantitative modeling has been largely atheoretical, failing to control for variables shown in the broader retention literature to influence outcomes.   


Therefore, the present study utilizes national data from the second follow-up of the Beginning Postsecondary Students Longitudinal Study (BPS: 04/09) to measure the impact of co-enrollment on success outcomes among a nationally representative sample of traditional age community college students. The following research questions are addressed:


1.

In what ways are the characteristics of traditional aged community college students who co-enroll similar or different from students who do not co-enroll?

2.

After accounting for the chance of co-enrollment on observed characteristics, does co-enrollment significantly increase or decrease the odds that a student will earn a degree and/or persist through the sixth year of college?


METHOD


DATASET AND SAMPLE


As suggested by McCormick (2003), data from the Beginning Postsecondary Students Longitudinal Study (BPS: 04/09) are well suited to examining multi-institutional enrollment behavior. The BPS Longitudinal Study tracks students longitudinally in an attempt to collect data specific to transfer patterns, co-enrollment, persistence, and degree attainment over six academic years from 2003-04 to 2008-09. Students sampled in the BPS Longitudinal Study (n = 23,090) were classified as first-time beginners (FTB’s) during the base-year survey of the National Postsecondary Student Aid Study (NPSAS:04). FTB’s were defined as students who first enrolled at a postsecondary institution during the 2003-04 academic year. Data sources included in the BPS Study were derived from institutional records, federal and Pell grant records, federal financial aid applications, National Student Clearinghouse enrollment records, college admissions test agencies, and student interviews. Furthermore, in an effort to provide additional descriptive information regarding enrollment behavior among students who co-enroll, several data elements were used from the recently released BPS Postsecondary Education Transcript study (PETS).  


The pre-matched analytic sample used in this study was limited to the 4,9201 students in the BPS dataset who began their postsecondary education at a two-year institution and were traditional in terms of age (i.e., 23 or younger). The decision to limit the sample to traditional aged students was based on limitations of the BPS dataset, which does not capture complete data for high school related data elements (e.g., GPA, mathematics courses) for older students. Note that although co-enrollees have been previously shown to be older, traditional age students represented 76% of the community college students and 85% of the students who co-enrolled in the national dataset.


ANALYTIC FRAMEWORK/TECHNIQUES


According to Goldrick-Rab (2010), selection bias is a statistical problem plaguing higher education research, as college outcomes can only be observed for students who participate and participants may differ in important ways from non-participants.  Research on community college students as well as co-enrollment more specifically has been predominately descriptive and existing multivariate work has failed to take steps to minimize selection bias. In turn, the current study utilized a propensity score matching technique on observational data as a means of reducing observed selection bias (Schneider, Carnoy, Kilptrick, Schmidt, & Shavelson, 2007). As demonstrated by Titus (2007), unlike linear and non-linear regression techniques, propensity score matching can be used by higher education researchers using national databases to produce more accurate parameter estimates to make stronger inferences of causality.


In short, the use of propensity score matching relies on Rosenbaum and Rubin’s (1983) counterfactual framework. For a student who co-enrolls, the counterfactual argument is the success outcome had the student not co-enrolled at more than one institution during a given semester. In contrast, the counterfactual for a student who does not co-enroll is the potential probability of an outcome if the student had chosen to co-enroll. Because the counterfactual cannot be tested, the use of this framework requires the estimation of propensity scores that allow for the comparison of students who concurrently enroll at two or more institutions during a given term with individuals with similar observed characteristics who do not co-enroll.  


Diagnostic analyses, including correlations, variance inflation factors (VIF), Mahalanobis distances, and stem and leaf plots were analyzed to identify outliers and multicollinearity issues. Data assumed to be missing at random (3.88%) were handled using multiple imputations (MI) (Enders, 2008). The matching analysis was preceded by bivariate tests using co-enrollment as a grouping variable and the covariates hypothesized to have theoretical relevance to co-enrollment. Significant differences between the treatment and control group verified the presence of observed selection bias and the need for propensity score matching (Guo & Fraser, 2010).


Propensity Score Matching Analysis


Following the recommendation of Rosenbaum and Rubin (1984), logistic regression was used to estimate propensity scores, defined as:

p(Χ) = Pr{T=1/ Χ} = E{T /Χ}                                                        (1)

where T=0, 1 indicates co-enrollment, X is a vector of pre-college characteristics, and E is the mathematical expectations operator, which refers to expectations in the overall population of individuals, conditional on X. The PSMATCH2 program in Stata 11 was used to conduct the propensity score matching analysis. Nearest neighbor matching within a caliper (a form of greedy matching) was chosen over optimal or fine balance matching due to its flexibility in allowing the researcher to conduct post-matching multivariate analyses. The size of the caliper was set as .25 of the standard deviation of the estimated propensity score of the sample (Guo & Fraser, 2010). Post-matching bivariate tests were run to assess the degree to which the propensity score analysis reduced the observed selection bias. Moreover, differences in the success outcome between students who co-enrolled and did not co-enroll were compared before and after matching as a form of sensitivity analysis as suggested by Graham and Kurlaender (2011).


Post-Matching Analysis


After the sample was shown to be balanced on observed covariates, logistic regression analyses were used to measures the influence of co-enrollment on the success of community college students. As detailed in the next section, the logistic models were structured using Crisp and Nunez’s (2011) community college success model. Following the recommendations of Peng, So, Stage and St. John (2002), the overall fit of the regression models were evaluated, followed by an examination and interpretation of the p-values, beta weights, standard errors, and odds ratios.


VARIABLES


Propensity Score Modeling


Co-enrollers were identified by an item in the BPS dataset that flagged students who had concurrently enrolled at more than one institution for at least a month over the period of six academic years. The choice of conditioning variables used to calculate propensity scores was largely based on prior research (Guo & Fraser, 2010) suggesting that co-enrollment could be predicted by students’ age, ethnicity, enrollment status, delayed enrollment into college, and socio-economic status (Adelman, 2005; Kinnick et al., 1998; Rab, 2004; Peter et al., 2005). However, the line of research on co-enrollment is not well developed, and initial descriptive analyses suggested that additional variables hypothesized to be related to student success might also have theoretical relevance to concurrent enrollment. Therefore, the decision was made to add gender, first-generation status, and educational expectations (as a rough proxy of student goal commitment) to the propensity score model.


Post-Matching Analysis


Crisp and Nunez’s (2011) community college model was used to guide the selection of covariates used in the post-matching analyses. The framework is an expansion of Nora’s student/institution engagement model (2004) and explains that success outcomes for community college students are influenced by a combination of socio-demographic variables, pre-college behaviors and motivations, environmental pull factors, students’ educational expectations, and several on-campus academic and social experiences. As detailed in Appendix A, the present study utilized several variables as measures of socio-demographic characteristics including age, gender, ethnicity, and first generation status. Several pre-college factors were also controlled for, including students’ high school grade point average (GPA) range, highest math course taken during high school (e.g., Algebra II, Calculus), whether a student earned college credit prior to high school in the form of Advanced Placement courses, and a dichotomous measure of delayed enrollment into college. Pull factors were also included in the post-matching analysis including whether a student worked more or less than 20 hours per week, the amount of financial aid received from all sources (e.g., grants, loans), and enrollment intensity measured whether a student enrolled exclusively full-time across their undergraduate academic career. Students’ educational expectations and various academic and social experiences during college were also used as covariates in the model. Academic and social experiences included students’ initial degree program type (i.e., technical, transfer program), academic integration (as measured by the BPS index), first year cumulative GPA, and whether or not students enrolled in distance education or developmental courses.


Given the diversity in academic goals among community college students and the prevalence of part-time students who are not able to complete a certificate or degree within a traditional timeframe (Cohen & Brawer, 2008), the dependent variable, “success,” was measured using a broad and inclusive longitudinal measure of academic accomplishments. Community college students were coded as being successful if they accomplished at least one of the following within six academic years: (a) earned a certificate, (b) earned an Associate’s degree, (c) earned a Bachelor’s degree, or (d) remained enrolled at a postsecondary institution.


LIMITATIONS


The results should be considered in the context of several limitations.  First, the propensity score matching analysis did not control for unobserved variables (i.e., hidden bias) that may impact students’ attendance patterns. It assumed that there are additional student characteristics that predict co-enrollment that were not accounted for due to limitations in theory and data. As such, it is important to note that there is an unknown amount of error in the specification of the conditioning model. Second, findings should not be assumed to generalize to non-traditional aged community college students, as only students 23 or younger were included in the present analysis (represents 76% of the students in the BPS data). Third, the BPS and PETS data do not much provide information about when or the degree to which students co-enrolled. Finally, due to the nature of co-enrollment involving student attendance at multiple institutions, the present study does not measure or control for the impact of institutional level variables on student outcomes.


RESULTS


DESCRIPTIVE FINDINGS


As shown in Appendix A, 14% of the sample of traditional aged students who began their postsecondary education at a community college co-enrolled at some point within six academic years (n=700). Eighty percent of the co-enrollees were coded as successful, with 29% earning a bachelor’s degree. This compares to only 57% of the students who did not co-enroll, only 13% of who earned a bachelor’s degree within six academic years. Students who co-enrolled were also shown to be descriptively different from students who did not co-enroll. For instance, 63% of students who co-enrolled were female, compared to only 55% of non co-enrollees. Students who co-enrolled were also less likely to be the first in their family to attend college (32% compared to 39% of non co-enrollees). Additionally, co-enrollees were shown to enroll in college immediately following high school (verses delaying enrollment between one to five years) and were more likely to exclusively attend college full-time (62% compared to only 47% of non co-enrollees). Students who co-enrolled were also found to have higher degree expectations, with 61% intending to earn a graduate level degree compared to only 42% of non co-enrollees. Students who co-enrolled were also more likely to be enrolled in a transfer degree program and on average earned higher first year grade point averages when compared to students who did not co-enroll.


As previously mentioned, additional enrollment data were drawn from the recently released PETS dataset in an effort to provide a more detailed description of students who co-enrolled. Although the exact timing of co-enrollment is not provided, the data do show that only 45% of the students who co-enrolled at some point in six academic years did so before transferring from their first institution. In other words, the majority of students who co-enrolled did so after transferring from their first institution. The PETS data also show that 75% of the students transferred to another institution across six years, with nearly half (48%) transferring from their first college to a four-year institution. Further, the average student who co-enrolled earned 51% of their college credit at a community college.  


PROPENSITY SCORE MATCHING


As shown in Table 1, bivariate chi square tests revealed that conducting multivariate analyses on the unmatched sample would generate biased results. Consistent with prior research on co-enrollment, bivariate analyses revealed a significant relationship between co-enrollment and various student characteristics including age (p<.001), gender (p<.001), ethnicity (p<.05), first generation college status (p<.01), delaying enrollment into college (p<.001), the amount of financial aid received (p<.01), and students’ educational expectations (p<.001). However, the most notable difference between students who did and did not co-enroll was their likelihood of earning a degree and/or persisting in college through the sixth year, as co-enrollers were significantly more likely to be successful (p<.001) when compared to students who did not co-enroll. The logistic regression model predicting co-enrollment was found to be significant χ2(8, n = 4,220) = 129.66,  p <.001. Predicted probabilities ranged from .029 to .416 (M=.14, SD=.06).


POST-MATCHING ANALYSES


Post-matching bivariate chi-square analyses revealed that the matching analysis had substantially reduced the amount of observed sampling bias, as no significant differences remained between students who did and did not co-enroll (see Table 1). The matched sample included 1,400 students including the sample of co-enrollees and 700 matched students who had similar observed pre-college characteristics but who did not co-enroll. The average propensity score for the matched co-enrollment sample was nearly identical to the average propensity for the matched sample of students who did not co-enroll (M=.165, SD=.062). Figure 1 illustrates a comparison of the distribution of the probability of co-enrolling before and after matching, providing additional evidence of the overlap between the propensity scores generated by the modeling for the matched sample.  


Table 1. Differences Between Students Who Co-enroll and Students Who Do Not Co-enroll for Conditioning Variables and Success Outcome

 

Before PSM

(Unmatched)

After PSM

(Matched)

 

Chi Square

Sig.

Chi Square

Sig.

Conditioning Variables

    

Age

35.21

***

7.97

ns

Gender

18.85

***

.05

ns

Ethnicity

16.76

*

5.50

ns

First generation status

16.34

**

.51

ns

Delayed enrollment into college

13.58

***

.16

ns

Amount of financial aid received

18.49

**

.87

ns

Educational expectations

73.04

***

4.96

ns

     

Outcome

    

Success

127.99

***

140.89

***

* p < .05, ** p < .01, *** p < .001

Note: Pre-matched sample included 4,920 students. Matched sample includes 1,400 students (all 700 who co-enrolled and 700 matched cases). Data are rounded to the nearest 10th per IES guidelines.

Source: BPS:04/09 survey data


As shown in Appendix B, the matched sample of students who did not co-enroll was found to be similar to the larger un-matched sample in terms of age, ethnicity, high school GPA, math courses taken during high school, earning college credit during high school, hours worked, degree program, and enrollment in distance education courses. Females and students expecting to earn a master’s or doctoral degree were slightly over-represented in the matched sample. Similarly, first generation students and students who delayed entry to college were slightly underrepresented among students who did not co-enroll in the matched sample. At the same time, with a few exceptions (e.g., degree program), the matched sample was found to include a robustly similar group of students who did and did not co-enroll.


Pre-matching regression findings χ2(33, n = 4,920) = 384.31,  p <.001 suggested that co-enrollment was significantly related to student success. As shown in Table 2, results of the post-matching logistic regression analysis were similar, demonstrating that co-enrollment significantly increased students’ odds of success, even after controlling for observed selection bias, socio-demographic variables, pre-college behaviors and motivations, environmental pull factors, students’ educational expectations, and several on-campus academic and social experiences χ2(33, n = 1,400) = 232.34,  p <.001). More specifically, the odds of successfully earning a degree and/or in persisting through the sixth year of college was found to be 3.73 times as large for students who co-enrolled at some point within six academic years (p<.001). Additionally, consistent with prior research on community college students, the odds of success were also shown to be uniquely positively influenced by having a parent with postsecondary experience, enrolling in college exclusively full-time, not working, having expectations to earn a doctoral degree, and students’ first year cumulative grade point average.


Table 2. Post-Matching Regression Analysis

 (reference categories in parentheses)

b

S.E.

Odds Ratio1

Socio-demographic Variables

   

Gender (Male)

.195

.130

--

Ethnicity (White)

   

African American

-.068

.277

--

Hispanic

.035

.303

--

Asian American

.218

.311

--

Other or more than one race

-.029

.381

--

First generation status (First Gen)

.321*

.135

1.378

Precollege Factors

   

High school GPA (Less than 2.0)

   

2.0 to 2.4

.501

.295

--

2.5 to 2.9

.426

.287

--

3.0 to 3.4

.341

.277

--

3.5 to 4.0

.592

.310

--

Highest math course taken (Other)

   

Algebra II

-.034

.158

--

Trig and Algebra II

-.041

.199

--

Precalculus

.058

.224

--

Calculus

.528

.309

--

Earned college credit (No credit)

.217

.208

--

Delayed enrollment (Delayed entry)

.024

.162

--

Environmental Pull Factors

   

Hours worked (More than 20 hours per week)

   

     Did not work

.440*

.174

1.552

1 to 19 hours per week

.144

.152

--

Dependent (Independent)

-.208

.223

--

Amount of financial aid (None)

   

Less than 2,500 dollars

.016

.177

--

Between 2,500 and 4,999 dollars

.178

.178

--

Between 5,000 and 9,999 dollars

.163

.203

--

More than 10,000

.141

.222

--

Enrolled full-time (Part or Mixed)

.348**

.127

1.417

Highest Degree Expected (less than a bachelor’s degree)

   

Bachelor’s degree

-.078

.222

--

Master’s degree

.282

.220

--

Doctoral or professional degree

.538*

.248

1.712

Academic and Social Experiences

   

Transfer degree program (Technical)

-.111

.472

--

Academic Integration

.001

.001

--

No developmental education (Took DE course)

-.059

.133

--

First year GPA

.003***

.001

1.003

No distance education (Took distance course)

-.106

.200

--

Co-Enrollment (Did not co-enroll)

1.317***

.128

3.733

    

Model evaluation

   

-2 Log likelihood for final model

1594.29

  

Chi Square

232.34***

  

Cox and Snell R2

.153

  

Nagelkerke R2

.210

  

* p < .05, ** p < .01, *** p < .001

Note: Matched sample includes 1,400 students (all 700 who co-enrolled and 700 matched cases). Data are rounded to the nearest 10th per IES guidelines.

1Odds ratios only presented for significant variables

Source: BPS:04/09 survey data


CONCLUSIONS/DISCUSSION


Findings from the current study contribute to both the community college and co-enrollment literature by providing better understanding of the characteristics of community college students who co-enroll as well as the relationship between co-enrollment and success outcomes using a national dataset. The results contribute to descriptive knowledge showing that traditional aged community college students who co-enroll at multiple two- and/or four-year institutions are systematically different in several observable ways when compared to students who do not concurrently enroll. Namely, findings suggest that co-enrollees are more likely to be female, African American or Asian American, and have parents who have higher levels of postsecondary education. Additionally, students who co-enroll are more likely to have enrolled in college immediately after high school, receive higher amounts of financial aid, and have higher degree expectations when compared to non co-enrollees. Further, the BPS and PETS transcript data add to the understanding of co-enrollment behavior among students who first enroll at a community college. Most notably, data show that students may be equally likely to co-enroll both before and after transferring to a second or third institution, as only 45% of students were found to co-enroll prior to transferring from their first institution.  Data also demonstrate that co-enrollees are likely to earn a substantial percentage of credit at a two-year institution (both before and after transferring to four-year institution) and on the whole, are likely to remain enrolled in college much longer than students who do not co-enroll.


As with most higher education research, the multivariate work to date specific to co-enrollment has failed to account for observed or unobserved self-selection bias. The current study therefore provides a first step at addressing the methodological weaknesses in the existing research specific to concurrent enrollment by providing a less biased estimation of the influence of co-enrollment in predicting success outcomes for a national sample of traditional aged community college students. The post-matching multivariate findings are consistent with prior research that included both two- and four-year students (i.e., McCormick, 2003; Peter et al., 2005), as studies that have failed to find a positive relationship between co-enrollment and student outcomes have utilized four-year student samples (i.e., Herzog, 2005; Maggard, 2009; Rab, 2004). As such, inconsistencies in prior research findings may be due to differences in the influence of co-enrollment among students attending different institutional types as well as methodological approaches used.  


Results support the argument by Sturtz (2006) that co-enrollment promotes access by providing multiple points of entry and educational options. More specifically, the results suggest that co-enrollment may promote access to four-year institutions for community college students desiring to transfer and earn a four-year degree, as descriptive data show that co-enrollees are almost twice as likely to transfer. Intuitively, it makes sense that co-enrollment would have a positive influence on success outcomes, as it provides community college students the flexibility they need to take courses in locations and at times that best accommodates their work and family schedules (Sturtz, 2006). Opportunities to select from the class schedules at multiple institutions should make it more likely that students will be able to take the classes that they need at the time they need them. Moreover, co-enrollment is consistent with related efforts such as distance education to accommodate the needs of students. What is more, concurrent enrollment can help facilitate and ease transfer to four-year institutions for community college students by allowing them to take university courses and connect with a four-year institution prior to actually transferring.


SUGGESTIONS FOR FUTURE RESEARCH


According to McCormick (2003), theoretical models that presume single-institutional attendance are inadequate. Findings of the present study support this idea, specifically suggesting that co-enrollment behavior is an important variable to include in theoretical models of student success for community college students. At the same time, as this was one of the first studies to examine the impact of co-enrollment on students who begin college at a two-year institution, much more work is needed to understand how and why co-enrollment is positively related to outcomes. Namely, additional work is needed to capture additional ways co-enrollers may be systematically different from students who do not co-enroll.


Also, although the findings of the present study demonstrate a positive impact of co-enrollment on success outcomes, the BPS and PETS data capture little information about the co-enrollment behavior. As such, descriptive work is needed to understand how and at what point students are attending multiple institutions. Possible questions include whether or not students are simultaneously attending more than one community college, whether students are using co-enrollment to ease the transition to a four-year institution and to what degree distance education is influencing co-enrollment behavior? Moreover, research is recommended to understand why students choose to co-enroll and to what degree this behavior may be related to what Hagedorn and colleagues refer to as “course shopping” (Hagedorn, Maxwell, Cypers & Moon, 2007).  


Research is also recommended to understand whether students co-enroll as part of formalized agreements with four-year institutions and whether student outcomes differed among students who co-enrolled outside of articulation or course sharing agreements. Moreover, qualitative research is suggested to understand how and why co-enrollment is effective and the particular characteristics of co-enrollment that make it so beneficial for community college students. Furthermore, although the focus of this study is on community college students, future work is also necessary to examine the influence of co-enrollment on students who begin at four-year institutions.


IMPLICATIONS FOR POLICY AND PRACTICE


Findings are relevant to the growing conversation around outcomes-based accountability for community colleges (Bailey, Calcagno, Jenkins, Leinbach, & Kienzl, 2006) as well as current or future policy that may impact students’ ability to seamlessly move among independent community colleges and two and four-year institutions as a system. It is expected that co-enrollment will continue to become more prevalent among both two- and four-year students given the rapid expansion of online distance education and other efforts to make it easier to enroll at more than one institution (Borden, 2004; Crawley & LeGore, 2009; McCormick, 2003). State and federal policy makers and institutional leaders therefore need to be aware that non-linear attendance is a legitimate way for students to experience and be successful in college (Sturtz, 2006) as current findings suggest that there is tremendous benefit to the student having opportunities to attend multiple institutions within a given semester.


Unfortunately, many administrators and policymakers are not comfortable with the reality of student flow, and traditional linear-matriculation among first time, full-time students remains the dominant model influencing policy formation and educational practice (Borden, 2004; Sturtz, 2006). Findings suggest that measures of institutional effectiveness should be redefined to accommodate and support student co-enrollment, recognizing many entry points as effective measures of institutional access and success (Sturtz, 2006). As a first step, this requires expanded state and national data systems that effectively track student enrollment across institutions and accurately document concurrent enrollment, such as the data collection systems supported by the Statewide Longitudinal Data Systems Grant Program. Secondly, this requires administrators and policy makers (e.g., Voluntary Framework of Accountability) to continue to think outside of the box in developing measures of institutional effectiveness and accountability. For instance, effectiveness measures could be revised to support institutions that utilize articulation agreements and/or effective course sharing agreements with other institutions and institutional types.  


Notes

1. All raw data rounded to nearest 10 per NCES security guidelines.


References


Adelman, C. (1999). Answers in the tool box: Academic intensity, attendance patterns, and bachelor’s degree attainment. Washington, DC: National Center for Education Statistics.


Adelman, C. (2005). Moving into Town – And Moving On: The Community College in the Lives of the Traditional-age Student. Washington, DC: US Department of Education.


Attewell, P., Lavin, D., Domina, T., & Levey, T. (2006). New evidence on college remediation. Journal of Higher Education, 77(5), 886-924.


Bailey, T., Calcagno, J. C., Jenkins, D., Leinbach, T., & Kienzl, G. (2006). Is student right to know all you should know? An analysis of community college graduation rates. Research in Higher Education, 47(5), 491-519. DOI: 10.1007/s11162-005-9005-0


Borden, V. (2004). Student swirl: When traditional students are no longer the tradition. Change, 36(2), 10-17.


Calcagno, J. C., Crosta, P., Bailey, T., & Jenkins, D. (2006). Stepping stones to a degree: The impact of enrollment pathways and milestones on community college student outcomes. New York: Columbia University, Teachers College, Community College Research Center.


Cohen, A. M., & Brawer, F. B. (2008). The American community college. San Francisco: Jossey-Bass.


Crawley, A., & LeGore, C. (2009). Supporting online students. In G. S. McClellan and J. Stringer (Eds.), The Handbook of Student Affairs Administration (3rd ed.). San Francisco, CA: John Wiley and Sons, Inc.


Crisp, G., & Nora, A. (2010). Hispanic student success: Factors influencing the persistence and transfer decisions of Latino community college students enrolled in developmental education. Research in Higher Education, 51(2), 175-194. DOI: 10.1007/s11162-009-9151-x


Crisp, G., & Nunez, A. (2011, November). Modeling transfer among minority and non-minority community college students who intend to transfer and earn a 4-year degree. Paper presented at the 2011 ASHE conference, Charlotte, NC.


de los Santos, A., & Wright, I. (1990). Maricopa’s swirling students: Earning one-third of Arizona State’s bachelor’s degrees. Community, Technical, and Junior College Journal, 60(6), 32-34.


Dowd, A., & Coury, T. (2006). The effect of loans on the persistence and attainment of community college students. Research in Higher Education, 47(1), 33-62.


Enders, C. K. (2008, March 25). Analysis of missing data. Paper presented at the annual meeting of the American Educational Research Association, New York, NY.


Goldrick-Rab., S. (2010). Challenges and opportunities for improving community college student success. Review of Educational Research, 80(3), 437-469.


Graham, S. E., & Kurlaender, M. (2011). Using propensity scores in educational research: General principles and practical applications. The Journal of Educational Research, 104(5), 340-353.


Guo, S., & Fraser, M. W. (2010). Propensity score analysis: Statistical Methods and Analysis. Los Angeles, CA: Sage Publisher.


Hagedorn, L. S., Maxwell, W. E., Cypers, S., & Moon, H. S. (2007). Course shopping in urban community colleges: An analysis of student drop and add activities. The Journal of Higher Education, 78(4), 464-485.


Herzog, S. (2005). Measuring determinants of student return vs. dropouts/stopouts vs. transfer: A first to second year analysis of new freshmen. Research in Higher Education, 46(8), 883-928.


Kinnick, M. K., Ricks, M. F., Bach, S., Walleri, R. D., Stoering, J., & Tapang, B. (1998). Student transfer between community colleges and a university in an urban environment. Journal of Applied Research in the Community College 5(2), 89–99.


Longanecker, D. A., & Blanco, C. D. (2003). Public policy implications of changing student attendance patterns. New Directions for Higher Education, 121, 51-68.


Maggard, E. (2009, May). A Research Analysis of Swirling at the University of Wyoming Outreach School. Presented at the Annual Forum of the Association for Institutional Research, May, 2009.


McCormick, A. C. (2003). Swirling and double-dipping: New patterns of student attendance and their implications for Higher Education. New Directions for Higher Education, 121, 13-24.


McIntosh, M. F., & Rouse, C. E. (2009). The other college: Retention and completion rates among two-year college students. Center for American Progress. Retrieved from www.americanprogress.org/issues/2009/02/pdf/two_year_colleges.pdf


National Student Clearinghouse Research Center. (2011). Snapshot report: Concurrent enrollment. Herndon, VA: National Student Clearinghouse Research Center. Retrieved from http://research.studentclearinghouse.org


Nora, A. (2004). The role of habitus and cultural capital in choosing a college, transitioning from high school to higher education, and persisting in college among minority and non-minority students. Journal of Hispanic Higher Education, 3(2), 180-208.


Peng, C. J., So, T. H., Stage, F. K., & St. John, E. P. (2002). The use and interpretation of logistic regression in higher education journals: 1988-1999. Research in Higher Education, 43(3), 259-293.


Peter, K., & Cataldi, E. F., & Carroll, C. D. (2005). The Road Less Traveled? Students Who Enroll in Multiple Institutions (NCES 2005–157). U.S. Department of Education, National Center for Education Statistics. Washington, DC: U.S. Government Printing Office.


Porchea, S. F., Allen, J., Robbins, S., & Phelps, R. P. (2010). Predictors of long-term enrollment and degree outcomes for community college students: Integrating academic, psychosocial, socio-demographic and situational factors. The Journal of Higher Education, 81(6), 750–778.


Rab, S. Y. (2004). Swirling students: Putting a new spin on college attrition. Unpublished doctoral dissertation. University of Pennsylvania, Philadelphia, PA.  


Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41-55.


Rosenbaum, P. R., & Rubin, D. B. (1984). Reducing bias in observational studies using sob-classification on the propensity score. Journal of the American Statistical Association, 79, 516-524.


Schneider, B., Carnoy, M., Kilpatrick, J. Schmidt, W. H., & Shavelson, R. J. (2007). Estimating causal effects using experimental and observational designs: A think tank white paper. Washington, D.C.: American Educational Research Association.


Sturtz, A. J. (2006). The multiple dimension of student swirl, The Journal of Applied Research in the Community College, 13(2), 151-158.


Titus, M. A. (2007). Detecting selection bias using propensity score matching and estimating treatment effects: An application to the private returns to a Master’s degree. Research in Higher Education, 48(4), 487-521. DOI: 10.1007/s11162-006-9034-3


Townsend, B. K., & Dever, J. T. (1999). What do we know about reverse transfer students? New Directions for Community Colleges, 106, 5-13.


United States Department of Education (2005). Beginning postsecondary students: Data analysis system (Technical report). Washington, DC: National Center for Education Statistics.


Wells, R. (2008). The effects of social and cultural capital on student persistence: Are community colleges more meritocratic? Community College Review, 36(1), 25–46.


Figure 1. Distribution of the Probability of Co-enrolling Before and After Matching


[39_17156.htm_g/00001.jpg]



Appendix A.

 

Variable Specifications for PSM and Regression Analyses

 Variable Name

Description and Coding

Student Level Variables

 

Socio-demographic Variables

 

Age

Continuous variable ranging from 15 to 23

Gender

Female = 0, Male = 1

Ethnicity

White=0*, African American=1, Hispanic=2, Asian=3, Other or more than one race=4

First generation status

Neither parent attended college=0, earned less than a 4-year degree=1, 4-year degree=2, more than a 4-year degree=3*

Precollege Factors

 

High school GPA

Less than 2.0=0, 2.0 to 2.4=1, 2.5 to 2.9=2, 3.0 to 3.4=3, 3.5 to 4.0=4*

Highest math course taken

Other=0, Algebra II=1, Trigonometry and Algebra II=2, Pre-calculus=3, Calculus=4*

Earned advanced placement (AP) credit

Did not earn Advanced Placement credits during high school=0, Earned AP credits during high school=1

Delayed enrollment into college

Delayed enrolling in college=0, Enrolled in college immediately following high school=1

Environmental Pull Factors

 

Hours worked

Worked more than 20 hours per week (excluding work study)=0, Worked less than 20 hours per week=1, Did not work during first year of college=2*

Dependency status

Student was classified as independent=0, Student was classified as a dependent in 2003-04=1

Amount of financial aid

No aid received from all sources in 2003-04=0, less than 2,500 dollars=1, between 2,500 and 4,999 dollars=2, between 5,000 and 9,9999 dollars=3, more than 10,000=4*

Enrollment intensity

Enrolled college part-time or a mix of part and full-time through 2009=0, Enrolled full-time for all semesters attended through 2009=1

Educational Expectations

Expected to earn a bachelor’s degree in 2003-04=0, Expected to earn a Master’s degree=1, Expected to earn a doctoral or professional degree=2*


Academic and Social Experiences

Degree program

Enrolled in a technical or vocational program in 2003=04=0, Enrolled in general education or transfer degree program=1

Academic integration

BPS academic integration index for 2003-04 year calculated from an average of students’ frequency in participating in study groups, social contact with faculty, meeting with an academic advisor, talking with faculty outside of class

First year GPA

Cumulative grade point average in 2003-04 academic year

Distance education

Did not enroll in distance education classes in 2003-04=0, Enrolled in distance education courses=1

Developmental course

Did not enroll in developmental coursework in 2003-04=0, Enrolled in one or more developmental courses=1

Co-enrollment

Did not co-enroll at more than one institution for one or more semesters across six academic years=0, Co-enrolled at more than one institution during a given semester across six academic years=1

Outcome

 

Student success

Student did not earn a degree and did not persist in college through 2009=0, Student earned one or more degrees (associates or bachelor’s) or certificates and/or persisted in college through 2009=1

Source: BPS:04/09 survey data

*Indicates reference group


Appendix B.

 Unmatched and Matched Samples- Comparing Students who did and did not Co-enroll (Percentages and Means/Standard Deviation)

 

Unmatched Sample

Matched Sample

 

Co-Enrolled

(n=700)

Did not Co-Enroll

 (n=4,220)

Co-Enrolled

(n=700)

Did not Co-Enroll

(n=700)

Socio-demographic Variables

    

Age

18.9(1.2)

18.6(1.4)

18.9(1.2)

18.6(1.0)

Female

63%

55%

63%

64%

Ethnicity

    

White

57%

61%

57%

61%

African American

18%

15%

18%

15%

Hispanic

12%

15%

12%

14%

Asian

6%

4%

6%

5%

Other or more than one race

7%

5%

7%

4%

First generation college student

32%

39%

32%

33%

Precollege Factors

    

High school GPA

    

Less than 2.0

4%

6%

4%

6%

2.0 to 2.4

15%

18%

15%

19%

2.5 to 2.9

21%

25%

21%

24%

3.0 to 3.4

41%

36%

41%

35%

3.5 to 4.0

19%

15%

19%

15%

Highest math course taken

    

Other

20%

25%

20%

27%

Algebra II

41%

39%

41%

40%

Trig and Algebra II

16%

17%

16%

17%

Precalculus

15%

13%

15%

11%

Calculus

8%

5%

8%

6%

Earned college credit during HS

14%

9%

14%

10%

Delayed enrollment into college

21%

27%

21%

20%

Environmental Pull Factors

    

Hours worked

    

More than 20 hours per week

55%

59%

55%

59%

1 to 19 hours per week

21%

16%

21%

16%

Did not work

24%

26%

24%

26%

Amount of financial aid

    

Did not receive financial aid

33%

35%

33%

34%

Less than 2,500 dollars

18%

23%

18%

19%

Between 2,500 and 4,999 dollars

22%

20%

22%

21%

Between 5,000 and 9,999 dollars

15%

14%

15%

15%

More than 10,000 dollars

12%

8%

12%

11%

Exclusive full time enrollment

62%

47%

62%

47%

Highest Degree Expected to Earn

    

Less than a bachelor’s degree

9%

17%

9%

10%

Bachelor’s degree

31%

39%

31%

32%

Master’s degree or certificate

43%

32%

43%

38%

Doctoral or professional degree

18%

12%

18%

21%

Academic and Social Experiences

    

Degree program

    

Technical or occupational degree

16%

25%

16%

24%

General ed. or transfer degree

84%

75%

84%

76%

Academic Integration

67.6(45.3)

61.0(43.9)

67.6(45.3)

64.0(45.3)

Enrolled in developmental course

32%

31%

32%

35%

First year GPA

2.93(.83)

2.79(.84)

2.93(.83)

2.70(.91)

Distance education courses

12%

10%

12%

10%

     

Enrollment Behavior

    

Co-enrolled prior to first transfer

45%

--

45%

--

Transfer type – first transfer

    

Never transferred

25%

56%

25%

55%

Vertical transfer (2 to a 4-year)

48%

28%

48%

25%

Lateral transfer (2 to a 2-year)

25%

15%

25%

16%

Downward transfer (2 to less than a 2-year)

2%

1%

2%

4%

Credit earned at two-year institution

51(36)

42(31)

51(36)

34(28)

     

Outcome

    

Success – Earned a degree and/or was still enrolled in college through the sixth year

79%

57%

79%

49%

Earned a Bachelor’s degree

29%

13%

29%

10%

Earned an Associate’s degree

20%

18%

20%

11%

Earned a Certificate

7%

8%

7%

9%

Still enrolled

23%

18%

23%

19%

No degree and not enrolled

21%

43%

21%

51%

     
     

Note: Pre-matched sample included 4,920 students. Matched sample includes 1,400 students (all 700 who co-enrolled and 700 matched cases).

Data are rounded to the nearest 10th per IES guidelines.

Note that some categories may not sum to 100 due to rounding.

Source: BPS:04/09 and PETS survey data






Cite This Article as: Teachers College Record Volume 115 Number 10, 2013, p. 1-25
https://www.tcrecord.org ID Number: 17156, Date Accessed: 12/2/2021 2:27:44 PM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Gloria Crisp
    University of Texas at San Antonio
    E-mail Author
    GLORIA CRISP is an associate professor of higher education at The University of Texas at San Antonio. The focus of her scholarship includes understanding the factors influencing the success of community college and/or populations traditionally underrepresented in college and the impact of institutional and state policy on student transfer and persistence. Recent publications include “The role of discriminatory experiences on Hispanic students’ college choice decisions” in the Hispanic Journal of Behavioral Sciences (with Amanda Taggart) and an article in The Review of Higher Education entitled “The impact of mentoring on community college students’ intent to persist.”
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS