An Examination of Student Retention at a 2-Year College Through Structural Equation Modeling

Abstract

The purpose of this study was to develop a retention model specific to the 2-year college environment using factors not typically combined with the study of student retention. The study operationalized factors representing the basic psychological needs of autonomy, competence, and relatedness combined with elements of Bean and Metzner’s nontraditional student attrition model. The data were analyzed using structural equation modeling techniques. The results indicated acceptable model fit and small to moderate effects were noted for competence and autonomy with grade point average (GPA). GPA exhibited the sole direct effect on student retention. The results further indicated the student characteristics of full-time enrollment and completion of 30 or more college-level credits combined with GPA explained a higher variance on student retention than did the factors of autonomy, competence, relatedness, external pull, or external support combined.

Keywords

community college structural equation modeling student retention 2-year college

Why Is the Retention of Community College Students Important?

Two-year colleges, commonly referred to as community colleges, play a significant role in higher education. They provide opportunities and services to students who often have no other avenue to engage in postsecondary education. Community college students are diverse in age, gender, race, academic ability, and goals concerning degree completion and transfer opportunities (American Association of Community Colleges [AACC], 2014; Community College Research Center, n.d.; Hagedorn, 2005; Marti, 2008). These differences in student characteristics, preparation, and goals create challenges in assessing student retention rates and developing strategies for improvement.

The issue of student retention becomes increasingly important as community colleges strive to meet federal and state completion goals, often with decreased state and federal funding. According to Schuetz (2005), approximately 90% of first-time community college students begin college with the intent of completing a credential, but only 36% will do so within 6 years. Nearly 50% of first-time community college students will not persist to the second year (Berger, Ramirez, & Lyons, 2012). Alarmingly, the majority of students who attempt completion of a credential at the community college level will not succeed (Shapiro et al., 2015).

There is a focus on higher education resulting from projects such as Complete College America and the Lumina Foundation as well as a challenge posed by former President of the United States, Barack Obama, in 2010 for 2-year colleges to produce an additional five million graduates by the year 2020 (The White House, n.d.). The increased scrutiny of performance measures, such as student retention and completion rates, affects funding and drives higher education leaders to seek measures that will predict and improve student retention. The current study aims to provide another avenue to examine student retention through the development of a retention model.

Current Study

The Community College Survey of Student Engagement (CCSSE) was used to operationalize the factors of competence, autonomy, relatedness, external pull, and external support comprising the retention model (CCSSE, 2003). The study explored the relationships between the factors and their influence on student grade point average (GPA) and retention at a 2-year college in the Southeast. Observed variables included in the model were academic success, as measured by GPA, student retention, and student characteristics.

I sought to answer three research questions in the current study:

Do the individual items selected from the Community College Survey of Student Engagement instrument measure the factors of relatedness, competence, autonomy, external pull, and external support?

To what extent does the proposed conceptual model of student retention based on relatedness, competence, autonomy, external pull, external support, and cumulative GPA fit the data?

If the proposed model fits the data, what are the direct and indirect effects among the variables?

If the proposed model does not fit, what combinations of variables produce the best fit to the data?

How do student characteristics identified as gender, age, race, number of credit hours completed, and enrollment status directly or indirectly influence retention in the final accepted structural model?

Theoretical Framework

This study utilized Deci and Ryan’s (2002) self-determination theory (SDT) as the guiding theoretical lens. According to Deci and Ryan, basic psychological needs such as competence, autonomy, and relatedness are essential for individuals to achieve integration. Integration, commonly used interchangeably with the term engagement, is directly linked to student retention (Astin, 1999; Pascarella & Terenzini, 2005; Tinto, 1993). I hypothesized students with a strong sense of relatedness, competence, and autonomy coupled with external support and academic success, as demonstrated by GPA, will persist at a higher rate than students with low levels of relatedness, competence, autonomy, external support, and academic success.

Literature Review

Community colleges are important access points for higher education for many traditional and nontraditional students. The Community College Research Center (n.d.) reported approximately 10.1 million undergraduate students enrolled in a 2-year college in the 2012 to 2013 school year. Many of these students entered the community college environment with academic deficiencies requiring remediation and demands on their time in the form of work and family obligations (Berkner, Horn, & Clune, 2000; Community College Research Center, n.d.; Complete College America, 2013a; Nakajima, Dembo, & Mossler, 2012). These challenges impact student retention and completion in ways not expected at 4-year institutions. In addition to academic challenges, a large percentage of students enrolled in community colleges face financial challenges. The American Association of Community Colleges (AACC; 2014) reported 22% of full-time and 41% of part-time enrolled students are employed full-time, working more than 35 hours per week. AACC reported part-time employment rates of 40% for full-time and 32% for part-time enrolled students. External commitments such as work and family obligations leave the typical community college student with little time to focus on the demands of academic work at the postsecondary level (AACC, 2014; Berkner et al., 2000; Nakajima et al., 2012).

Enrollment in colleges and universities has more than doubled in the United States between 1970 and 2009 (Complete College America, 2014a); however, the number of college graduates has increased very little during the same period (Bean, 1980; Berger et al., 2012). This fact is concerning because of the increase in the number of new jobs requiring a college education outpaces the number of new jobs requiring a high school diploma (Carnevale & Rose, 2011). Carnevale, Smith, and Strohl (2013) estimated by the year 2020, 65% of jobs will require an education beyond high school compared with 23% in 1973.

Community College Retention

The Complete College America (2013a) initiative requires colleges to increase their focus on persistence and completion through the evaluation of their enrollment and retention programs. It is imperative that institutions improve student retention as more states move to completion rate formulas to determine levels of public funding (National Conference of State Legislatures [NCSL], 2015). Expanding the knowledge and understanding of the variables that influence student retention through decades of research has not led to significantly higher completion rates (Burrus et al., 2013). Therefore, college leaders must critically evaluate student retention strategies to determine which factors may help to improve their institution’s completion rates (Burrus et al., 2013; McClenney & Waiwaiole, 2005; Wilds & Ebbers, 2002).

Community College Student Involvement and Engagement

One area of research positively linked to student retention is student involvement or engagement (Pascarella & Terenzini, 2005; Schuetz, 2008). According to Astin (1999), student involvement with the college environment is essential for student retention. Student involvement with the college environment is defined as meaningful interactions with faculty, staff, and peers (Kuh, 2001, 2003). Astin’s theory of student involvement is particularly important to the community college environment because it asserts that students’ time and effort may be the most important resource on campus. Community college students typically have a variety of demands on their time, such as family and work, as well as the academic barriers resulting from a lack of academic preparedness (AACC, 2014; Center for Community College Student Engagement, 2016; National Center for Education Statistics, n.d.). These demands can limit community college students’ availability to dedicate sufficient effort to tasks required to excel academically and to integrate into the college culture. Astin postulated the level of student involvement is proportional to the level of student learning and development and likely leads to increased persistence. This is particularly relevant to the current study’s focus on competence, autonomy, and relatedness because Astin’s theory of student involvement served as a foundational theory in the development of the survey instrument used in this study (Kuh, 2009) and helped to guide variable selection for the factors competence, autonomy, and relatedness.

Research focused on competence, autonomy, and relatedness has shown these areas directly influenced academic success and student retention (Beachwood, Beachwood, Li, & Adkinson, 2011; Elliott & DiPerna, 2002; Hurtado & Carter, 1997; Kusurkar, Cate, Vos, Westers, & Croiset, 2012; Maurer, Allen, Gatch, Shankor, & Sturges, 2012; M. P. Ryan & Glenn, 2003; Schuetz, 2007, 2008). Schuetz (2008) focused on the relationship between competence, autonomy, and relatedness with student engagement and advanced the idea of other relationships between these factors and outcomes related to student success. Examining the areas of competence, autonomy, and relatedness in a retention model may reveal more insight into the relationship between these basic psychological needs and their influence on academic performance and retention.

Competence is derived from academic integration and engagement inside and outside of the classroom (Reason, Terenzini, & Domingo, 2006). M. P. Ryan and Glenn (2003) found increasing academic competence through improving academic skills deemed necessary for success in college such as note-taking, goal-setting, and test preparation had a direct link to improved student retention. Autonomy can be derived when students freely explore career options or seek information about careers and academic programming (Schuetz, 2007). Feelings of autonomy are associated with environments that provide choices for students, respect their differences, and facilitate the development of personal goals and judgments (Deci & Ryan, 2002). Students who reported a strong sense of autonomy demonstrated increased study effort and better academic performance as measured through GPA, making autonomy an important facet of student retention (Kusurkar et al., 2012). In the college environment, relatedness refers to the student’s feeling that the faculty care about their success and have created a learning environment that is respectful and nurturing (R. M. Ryan & Deci, 2000). Studies have demonstrated that relatedness is associated with improved persistence, higher levels of engagement, and increased academic achievement (Beachwood et al., 2011; Bean, 1980; Bean & Metzner, 1985; Furrer & Skinner, 2003; Inkelas & Weisman, 2003; Tinto, 1993).

Method

Data Source and Participants

The study was conducted at a moderate-sized, public, 2-year college located in the southeastern United States with multiple campus locations throughout the state. The current study site was identified with the pseudonym of Team College. The institution reported student demographics at the time of data collection were 40.1% Black, 39.7% White, 0.5% Hispanic, 1.6% Asian, 3.7% other races, and 14.4% unknown race/ethnicity. The institution reported 56.1% of enrolled students were under the age of 24, and 61.3% were women.

The college administered the CCSSE in the winter term of 2013 to 1,768 students across all campuses. The 66.6% response rate yielded 1,178 surveys of which 937 surveys were usable for the current study because those participants provided their student identification numbers. The student identification number enabled the collection of retention data from the college’s student record system. The racial identification of the participants in the study was reported as 42.4% Black, 41.1% White, 3.6% Hispanic, 1.6% Asian, 5% other races, and 5.9% unknown race/ethnicity. The sample was 67.6% traditional aged and 55.7% women. Traditional-aged students were slightly overrepresented in the sample as were male students in comparison with the college population. Students were not compensated for participation and were provided a statement indicating participation was voluntary along with instructions for opting out of survey completion.

Data for this study were archival and collected by the college’s Office of Institutional Research through the administration of the CCSSE. Student retention, GPA, and demographic data from the college’s student record system supplemented the data. The survey was administered in adherence to CCSSE guidelines using a stratified random cluster sampling scheme based on classes grouped by class start times. Students completed the survey at the beginning of a normally scheduled class period. The stratified random cluster sampling procedures were biased to include more full-time enrolled students because full-time students had an increased chance of being in a class selected to participate. The data were weighted based on enrollment status using a numerical value provided by CCSSE developed for the college based on their most recent Integrated Postsecondary Education Data System enrollment data to address sample bias for full-time enrolled participants.

Analytical Procedures

Before beginning data analysis to answer the research questions, the data set was parsed into two files. The first data set identified as group one was used to confirm the factors and analyze the model. The second data set identified as group two was used for cross-validation of the final model.

Structural equation modeling (SEM) was the primary analytical procedure used for this study. SEM consisted of the development and evaluation of a measurement model followed by the testing of a structural model. SEM was used to test the relationships between the factors relatedness, competence, autonomy, external pull, external support, and the observed items of GPA and retained. SEM is a statistical technique applicable for use with nonexperimental data containing observed and unobserved variables (Schumaker & Lomax, 2010; Ullman, 2006). Confirmatory factor analysis (CFA) was used to test the reliability and validity of the factors included in the model and the assessment of measurement invariance. Analyses were conducted using Mplus with the weighted least squares means and variance adjusted (WLSMV) estimator for categorical items.

Individual item factor loadings, average factor scores, composite reliability (CR), and average variance explained were evaluated to assess the reliability and validity of the factors. Goodness-of-fit indices were reviewed as outlined by Kline (2011) and Hoyle (2012) to evaluate model fit. The fit indices reviewed included chi-square tests for independence (χ²), Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), and Root Mean Square Error of Approximation (RMSEA) goodness-of-fit indices. A nonsignificant chi-square value was desirable and indicated good model fit; however, this test is sensitive to sample sizes above 250. Acceptable model fit can be determined with a significant chi-square value by evaluating other fit indices. CFI and TLI values greater than .95 indicated good model fit as did RMSEA values below .04 (Keith, 2015). RMSEA values between .05 and .08 indicated adequate model fit (Keith, 2015). In addition to the goodness-of-fit indices, the model pathways illustrated between the factors and the observed items, GPA and retained, were evaluated for the direction of influence and strength. The chi-square difference testing procedure within Mplus was used to determine changes in the chi-square values between data sets to evaluate measurement invariance between the group one and group two data (Hoffman, 2014; Muthen & Muthen, 2015), along with an evaluation of the CFI value. A change of .01 or less in the CFI value was desirable during measurement invariance testing.

Variables Studied

The student characteristics included in the model were gender, age, race, credit hours completed, and enrollment status. Each of the student characteristics selected has been shown to relate to student retention (Berkner et al., 2000; Horn & Carroll, 1996). The variables are consistent with other retention models, such as Bean and Metzner’s (1985) model of nontraditional student attrition.

The factors of relatedness, competence, autonomy, external pull, and external support were operationalized with items from the CCSSE (Table 1). CFA was conducted to assess the reliability and validity of each factor. Individual factors with a standardized factor loading below .6 were reviewed for removal. CR values of .6 were considered acceptable while values greater than .7 suggested good reliability (Table 1; Hair, Black, Babin, & Anderson, 2015). In some cases, the CFA was repeated as items were removed based on unacceptable factor loading values. Items related to marital and dependent status along with time spent caring for dependents and working comprised the factor external pull. The items hours working and have children were removed during CFA analysis because of unacceptable factor loadings. The remaining items for the factor external pull had a CR score of .50, indicating poor reliability. The factor external pull was removed from the conceptual model before further analysis.

Table 1.

Standardized Factors Loadings Resulting From Final CFA Conducted for Each Model Factor.

Factors	Measured variables	Factor loadings	Composite reliability
Relatedness			.79
	Quality of relationships with students	.61
	Quality of relationships with instructors	.81
	Quality of relationships with administrative personnel	.82
Competence			.91
	Acquire work-related skills or knowledge	.61
	Write clearly and effectively	.85
	Speak clearly and effectively	.88
	Analytical and critical thinking	.84
	Numerical problem-solving	.68
	Use computer technology	.70
	Acquire a broad general education	.75
Autonomy			.91
	Develop a personal ethical code	.78
	Develop clearer career goals	.90
	Gain information about careers	.88
	Understanding yourself	.81
External pull^a			.50
	Hours spent caring for dependents	.58
	Martial status	.58
External support^a			.81
	Support from family	.82
	Support from friends	.82

Note. n = 463. CFA = confirmatory factor analysis.

^aEquality constraint added to factor loadings for model identification.

GPA in the study referred to the cumulative GPA of the participants at the time of survey administration. GPA demonstrated direct effects on retention but also mediated the effects of the factors competence, autonomy, and external support on retention. Student retention was the outcome variable of this study. Student participants who had graduated, or enrolled in the fall terms of 2013 or 2014 were considered retained. The retained variable was measured on a dichotomous scale.

Results and Findings

Research Question 1—The Building Blocks For Model Assessment

Each factor was evaluated individually in a one-factor model to eliminate possible influence from other factors. The analysis was conducted using CFA. The results of CFA revealed the items selected measured the factors of relatedness, competence, autonomy, and external support (Table 1). The average factor loadings for each factor ranged from .75 to .84 with equally acceptable reliability and validity values. The items selected did not measure the factor external pull. The items for external pull exhibited low factor loadings or invalid factor loadings resulting in poor reliability values. As a result, the factor external pull was removed from the model.

The external pull items were selected based on prior research identifying variables such as being married, having children, and hours spent working or caring for dependents as valid measures of external pull (Bean & Metzner, 1985; Berkner et al., 2000; Brooks-Leonard, 1991; Horn & Carroll, 1996; Nakajima et al., 2012; St. John, Paulsen, & Starkey, 1996). The sample for this study did not exhibit many of the characteristics associated with studies that have included external pull as a measure. The majority of the study participants were not married (82%) and did not have children (67%). The number of participants who worked part-time or full-time was very similar to the employment status reported by the American Association of Community Colleges (2014). However, 67% of the participants of the current study indicated it was not likely, or only somewhat likely, they would withdraw from classes because of working. While Team College participants share other risk factors associated with 2-year college students, such as remedial course placement, first-generational status, part-time enrollment status, and commuter status, they may not be typical in all aspects associated with two-year college student demographics.

The factor external support was originally included in the model to determine if it influenced the effects of external pull on student retention and GPA. Removal of the factor external pull meant the reconsideration of the placement of external support. Rather than remove this factor, external support was retained in the model and evaluated for its relationship with the other factors, GPA, and the outcome retained.

Research Question 2—Evaluating the Model

The measurement model was adjusted once to improve model fit after a review of modification indices revealed a residual covariance existed between two items in the autonomy factor. The second measurement model indicated a significant chi-square value, χ² (121) = 236.67, p < .001, but acceptable CFI (.98), TLI (.98), and RMSEA (.05) values. A review of the unstandardized and standardized parameters estimates revealed each loading was strongly related to the perspective factor and a review of the correlations indicated all factors correlated with each other but not all factors correlated with the outcome retained or the observed item GPA (Table 2). Ideally, all factors and observed items in the measurement model will be correlated. The factor relatedness did not correlate with the outcome retained, or the observed item GPA although the relationship with the factor relatedness and student retention and GPA are well documented (Astin, 1999; Bean & Metzner, 1985; Tinto, 1993). All factors were retained in the model to measure their effects in the structural model analysis because nonsignificant pathways do not have to be removed in the measurement model analysis when supported by theory (Kline, 2011).

Table 2.

Standardized Parameter Estimates for Factor Correlations Including Observed Items in the Second Measurement Model.

Factor/Item	1	2	3	4	5
1. Relate
2. Competence	.54*
3. Autonomy	.55*	.88*
4. Support	.47*	.42*	.35*
7. GPA	−.02	−.07	−.19*	.12*
8. Retained	.11	.26*	.25*	.14	.17*

Note. n = 463. Relate = relatedness; Support = external support; GPA = grade point average.

*p < .05.

The structural model appeared to fit the data well. However, despite having CFI and TLI values that approached 1.0 and an RMSEA value of .05, the model explained less than 12% of the variance in student retention and only approximately 10% of the variance in cumulative GPA. There were few significant pathways in the model, and the observed item GPA was the sole predictor of student retention.

The factor relatedness did not exhibit significant total, direct, or indirect effects on the outcome retained. Rather than remove the factor during the measurement phase, it was retained for analysis in the structural model phase because it moderately correlated with the factors competence and autonomy and was supported by the literature (Astin, 1999; Beachwood, Beachwood, Li, & Adkinson, 2011; Bean, 1980; Bean & Metzner, 1985; Furrer & Skinner, 2003; Inkelas & Weisman, 2003; Tinto, 1993). Perhaps there was not enough variance in the responses of the participants to measure the effects of relatedness on GPA or retained accurately. Relatedness, as measured in this study, referred to the participants’ quality of relationships with other students, faculty, and administrative staff. The study participants rated their relationships with others at the level of 5 or higher on a 7-point scale at the rate of 84% for relationships with other students, 88% for relationships with faculty, and 76% for relationships with administrative personnel. Furthermore, several studies have demonstrated the importance of students’ feelings of belonging as influencing their level of academic and social integration (engagement) and commitment to the institution (Berger & Milem, 1999; Pascarella & Terenzini, 2005; Tinto, 1993). If a factor for engagement or commitment had been included in the model, the effects of relatedness might have been observed indirectly on GPA and retained.

As stated earlier, the model demonstrated a good fit to the data as measured by model fit indices, but the lack of significant pathways in the model warranted a review of the model structure to determine if adjustments would improve the number of significant pathways without sacrificing model fit. Two adjustments were made to the model. The first adjustment tested if the factors relatedness, competence, and autonomy would have an indirect effect on GPA or retained through the factor external support. Analysis indicated model fit indices comparable with the original structural model and no improvement in the number of significant pathways.

The second adjustment was the removal of the factor relatedness. Relatedness was not correlated to the observed items GPA or retained in the measurement model phase and did not exhibit any significant association with the observed items in the structural model phase. The removal of the factor relatedness resulted in similar model fit indices to the original structural model; however, the number of significant pathways increased (Table 3). The adjusted model was identified as the alternative model and the remaining factors of competence, autonomy, and external support each demonstrated a significant effect on GPA. In the alternative model, competence exhibited a small to moderate positive total and direct effect on GPA.

Table 3.

Parameters Estimates for Alternative Model.

Parameter estimates	Unstd. est.	SE	Est./SE	p	Std. est.	SE	Est./SE	p
Competence → Support	0.47	0.19	2.49	.01	0.51	0.19	2.72	.01
Autonomy → Support	−0.09	0.16	−0.53	.59	−0.10	0.20	−0.52	.59
Competence → GPA	0.33	0.17	1.99	.05	0.33	0.17	2.00	.05
Autonomy → GPA	−0.47	0.14	−3.33	.001	−0.53	0.16	−3.41	.001
Support → GPA	0.17	0.07	2.35	.02	0.16	0.07	2.40	.02
Competence → Retained	.07	.33	0.20	.84	.05	.25	.20	.84
Autonomy → Retained	.29	.30	0.99	.32	.25	.25	.99	.32
Support → Retained	.01	.14	0.06	.95	.01	.10	.06	.95
GPA → Retained	.30	.08	3.62	<.001	.22	.06	3.59	<.001

Note. n = 463.Unstd. Est. = unstandardized estimate; SE = standard error; Est./SE = estimate divided by standard error; Std. Est. = standardized estimate; Support = external support; GPA = cumulative grade point average; Retained = retained or graduated.

The factor autonomy exhibited a moderately negative association with GPA and a small negative indirect effect on retained through GPA. The direction of influence was unexpected based on the findings of other researchers, which supports positive associations with autonomy and GPA (Black & Deci, 2000; Friedman & Mandel, 2011; O’Reilly, 2014). The results of the study revealed that as the participant’s level of autonomy increased, their GPA decreased. Further analysis revealed an association with the factor autonomy and Learning Support Services, as well as autonomy and non–traditional-aged participants. The factor autonomy was negatively associated with GPA for those participants enrolled in or who had completed Learning Support Services (remedial) courses. Gains in the area of autonomy may not equate to one’s ability to navigate the academic rigor of the college-level coursework in this sample successfully.

Autonomy was negatively associated with GPA for non–traditional-aged students. More mature students may gain in autonomy as they validate their degree program choice with their career aspirations, but this does not lead to an increased GPA. In addition, some nontraditional students enter college with college transfer credits and may be attending to complete prerequisite courses for a program at a different college or to test their aptitude for the college environment after a period of stop-out (Wilds & Ebbers, 2002). Therefore, they may gain in their level of autonomy but may not be enrolled long enough to translate that awareness into a higher GPA or longer retention.

The factor external support had a small positive direct influence on GPA but no total, direct, or indirect effects on retained. The factor was originally included in the model to determine its mediating effects on external pull with GPA. The study results support the literature that external support in the form of support from family and friends has a positive influence on GPA (Bean & Metzner, 1985; Dennis, Phinney, & Chuateco, 2005).

The observed item GPA was the only item with a total and direct effect on the outcome retained. The amount of variance explained by the alternative model on the observed item GPA and the outcome retained changed by a few tenths of a percentage point. The changes to the structural model did not increase the acceptable model fit indices values or the percentage of variance explained by the model (Table 3); however, the number of significant pathways increased leading to a more parsimonious model. Therefore, the alternative model (Figure 1) was retained to answer Research Question 3 and to conduct cross-validation analysis.

Figure 1.

Alternative model before testing influence of student characteristics. Dashed lines represent nonsignificant pathways. Doubled line represents untested influence. Values reported are standardized coefficients.

Research Question 3—Influence of Student Characteristics

Each student characteristic was assessed individually and collectively to determine the influence on the items GPA and retained. Individually, each student characteristic demonstrated some level of effect on GPA and retained except for the student characteristic race which did not exhibit a correlation with GPA or retained. The model with all student characteristics included demonstrated good model fit with CFI and TLI values approaching 1.0 and the lowest reported RMSEA value in the study. The alternative model with all student characteristics exhibited the highest model fit indices values of any other model. Unexpectedly, after student characteristics were added to the model, the factors autonomy, competence, and external support ceased to exhibit significant direct effects on GPA or retained, meaning autonomy, competence, and external support were not significant measures of GPA or retained after student characteristics were added to the model. The variance explained for GPA and retained in the model with student characteristics was attributed to student characteristics and not to the factors autonomy, competence, or external support.

The student characteristic of age, credits completed, and gender exhibited direct effects on GPA, indicating non–traditional-aged participants had higher GPAs than traditional-aged participants. Second-year participants had higher GPAs than first-year participants. Female participants had higher GPAs than male participants. Student characteristics accounted for 19.7% of the variance for GPA in the model.

The student characteristics of credits completed and full-time enrollment status, along with GPA had direct effects on retained while age had an indirect effect on retained through GPA, demonstrating that second-year participants were more likely to be retained or to graduate than first-year participants. Full-time enrolled participants were more likely to be retained or to graduate than part-time enrolled participants. Participants with higher GPAs were more likely to be retained or to graduate than participants with lower GPAs. Student characteristics accounted for 17% of the variance for retained in the model.

The results of Question 3 indicated the student characteristics of credits completed and full-time enrollment along with the item GPA had direct effects on the outcome retained. These characteristics provided a better measure for predicting student retention than did the factors autonomy, competence, and external support based on the significant pathways and amount of variance explained in the final model.

The current study revealed students who had completed 30 or more college-level credits were more likely to be retained or graduate than students who had completed less than 30 credit hours. This result was consistent with research that indicates students who complete 30 or more college-level credits in the first year of enrollment have increased retention and graduation rates (Complete College America, 2013b; Klempin, 2014).

This study confirmed prior research that students with higher GPAs are more likely to be retained and graduate (Nakajima et al., 2012). While there are likely many strategies to help students improve their GPA, most colleges need to focus on strategies that will result in increased completion rates to keep pace with Complete College America (2013a) goals and performance funding initiatives. The results of the current study indicated students under the age of 24, male students, and students who have completed less than 30 credit hours have lower GPAs.

Recommendations

Team College could improve their retention by increasing the number of students who are enrolled full time, increasing the number of credits completed by students during enrollment with Team College, and helping students increase their GPA.

The results of the current study indicated full-time enrollment was linked to increased student retention, which is supported by prior research (Berkner et al., 2000; Complete College America, 2013b). To increase retention, Team College should develop strategies to encourage students to enroll full time. Team College could implement a multiterm or annual registration option to enable students to register for multiple terms. This process would commit students to Team College for more than one term and allow long-term planning by the student. This approach was adopted by Cleveland State University in 2012 and has increased fall to spring and fall to fall retention rates (Grasgreen, 2014). Team College should consider a fee structure that offers a flat rate for 12 or more credit hours per term. Currently, Team College assesses tuition and mandatory fees on a per credit hour structure meaning the more credit hours a student enrolls the higher the tuition and mandatory fee charges. Offering a flat tuition and mandatory fee rate for students enrolled in 12 or more credit hours each term will encourage students to enroll full-time and has been recommended by Complete College America (2013a) as a strategy to increase retention and degree completion.

Team College should adopt strategies to help ensure students complete 30 or more college-level credits in the first two to three terms of enrollment ensuring students reach the 30 credit hour mark quickly. One strategy is a restructuring of the remedial (Learning Support Services) education program of the college. Remedial coursework does not count toward the number of college-level credits completed and serves as a barrier to college-level gateway courses. According to Complete College America (2013a), only 22.3% of students who complete remediation will go on to complete the associated college-level course within 2 years, and only 9.5% of those completers will go on to graduate within 3 years. Team College should explore corequisite remediation courses in place of prerequisite course completion for gateway courses. According to Complete College America (2014b), the corequisite strategy has proven successful at the Community College of Baltimore County in Maryland and Austin Peay University. Both colleges have reported students completing the corequisite remedial option, while enrolled in credit level gateway, had twice the success rate as students who followed the traditional remedial coursework pathway (Complete College America, 2014b).

The results of the current study indicated that students under the age of 24, male students, and students who have completed less than 30 credit hours have lower GPAs. These characteristics include thousands of students at Team College, and with the college’s current allocation of Success Coaches, it would not be feasible to monitor every student who fits into one of these categories. Team College should consider utilizing data analytics to determine which characteristics constitute an at-risk student at Team College based on prior student enrollment records (Taylor & McAleese, 2012). The information provided by internal data analytics would enable Team College to explore targeted strategies and programs designed to assist at-risk students early in the matriculation process. Using data analytics proved successful for Arizona State University who reported an 8% increase in freshman retention through the use of data analytics (Phillips, 2013). Phillips (2013) reported the data analytics used by Arizona State University included a financial aid report that outlined financial aid awarded, scholarships applied, and academic progress information. An academic status report provided details on individual course grades, including mid-term and other progress grades reported by faculty, and feedback from faculty regarding recommended strategies for the student. An internal index score representing college readiness was developed using admissions information such as high school GPA, class rank, and test scores. These data were coupled with additional student data to build a student profile. The full profile supplied information related to dropped courses, course withdrawal, GPA, academic standing, enrollment holds, transcripts from prior institutions, and the level of engagement the student had with the online student portal. Academic advisors used the student profile data to help guide students through the enrollment process of degree major selection, course selection, information concerning available campus resources, and the student’s most direct path to graduation (Phillips, 2013). Phillips reported data analysis revealed that students who did not frequently access the student portal were more likely to not be retained. The data revealed which courses could be used as success indicators for degree completion. This information helped advisors make recommendations for degree major selection based on student performance in those courses (Phillips, 2013).

Beyond Team College, the discussion of the recommendations may be generalizable to other colleges with similar student populations. The demographics of this study indicated the majority of the sample at Team College were unmarried, without children, and worked less than 20 hours per week, which is not typical of community college student populations (Skomsvold, 2014). Therefore, other colleges with similar student populations, including 4-year colleges, may find the results of this study useful. This study affirms that some variables typically included in the study of student retention at 4-year institutions have relevance to the 2-year college environment. Specifically, the student characteristics of full-time enrollment, credit hours completed, and GPA discovered to correlate with the outcome student retention in this study have been linked to increased student retention at 4-year and 2-year institutions (Bean & Metzner, 1985; Brooks-Leonard, 1991; Cofer & Somers, 2000; Nakajima et al., 2012).

Limitations

The first limitation noted for this study was the use of secondary source data collected with an instrument designed to measure student engagement. A survey instrument designed specifically for this study may have provided better results. The ability to operationalize the factors of the model was limited to the survey questions contained within the CCSSE. One example of this limitation was the lack of questions concerning external commitments to friends or community organizations in the conceptualization of the factor external pull. A second limitation is the generalizability of the results to other 2-year colleges. Other 2-year colleges may have a more homogeneous student population with the majority of their students attending what is considered their main campus as opposed to the institution in this study where the majority of the student population attend a campus or center location other than the main campus. The fact that a majority of the students attend a campus or center outside of the main campus may be significant because the allocation of physical and technological resources and staffing levels are different at the branch campuses and extension centers than the main campus, which could have influenced student responses to the survey. A third limitation was data were collected primarily from the students’ self-reported responses to the survey instrument. Although student participation was not required, and the survey items were not of a sensitive nature, there is a possibility that some students may not have answered all items accurately.

Suggestions for Future Research

Several suggestions for future research can be provided resulting from this study. First, the relationship between autonomy and Learning Support Services as well as student age could be explored further. It would be of interest to evaluate students from other 2-year colleges to examine if the negative association with autonomy and GPA for Learning Support Service or student age was comparable with the results obtained in this study.

Another avenue of research is to evaluate the influence of the factors relatedness, competence, and autonomy on outcomes such as student engagement, campus climate, satisfaction, or institutional commitment in connection with student retention. These additional measurements may have improved the model’s influence on student retention. The possible moderating or mediating effects of additional measures in the model may have improved the variance explained by the model.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Author Biographies

Leslie C. Hafer holds an EdD in Leadership from Valdosta State University and an MS in Biology from Georgia College and State University. Dr. Hafer serves as the executive director for the Valdosta and Albany campuses of Georgia Military College. Before assuming the duties of executive director, Dr. Hafer served as an academic dean, Department Chairperson of Natural Science, and an assistant professor of Biology at Georgia Military College.

Nicole M. Gibson serves as a research faculty advisor to doctoral students. She completed her PhD in Educational Psychology at the University of Georgia. Her research interests include assessment of students' depth of knowledge and the development of statistical models in educational research.

Travis T. York is the director of Student Success, Research & Policy at the Association of Public & Land-grant Universities. His research centers on issues of college student access, success, and educational equity. Dr. York's work has focused on examining pathways into and through postsecondary environments for low-income and first-generation students. Dr. York earned his BA and MA in Higher Education from Geneva College where he worked in Student Affairs and his doctorate in Higher Education Administration from The Pennsylvania State University.

Herbert R. Fiester holds a PhD in Instructional Systems Technology from Indiana University. He is currently working as an associate professor in the Department of Curriculum, Leadership, and Technology at Valdosta State University and serves as a program coordinator of the Curriculum and Instruction Doctoral Program.

Rudo Tsemunhu is an associate professor in the Department of Educational Leadership at Valdosta State University. She has international school site experience that includes serving as a school principal in Zimbabwe and a program coordinator for an educational initiative in Boston, MA, USA.

References

American Association of Community Colleges. (2014). Fast facts from our fact sheet. Retrieved from http://www.aacc.nche.edu/AboutCC/Documents/Facts14_Data_R3.pdf

Astin

A. W.

(1999). Student involvement: A developmental theory for higher education. Journal of College Student Development, 40(5), 518–529.

Beachwood

M. R.

Beachwood

J. C.

Adkinson

S. R.

(2011). Cohorts and relatedness: Self-determination theory as an explanation of how learning communities affect educational outcomes. Research in Higher Education, 52, 853–874. doi:10.1007/s11162-011-9221-8

Bean

J. P.

(1980). Dropouts and turnovers: The synthesis and test of a casual model of student attrition. Research in Higher Education, 12(2), 155–187.

Bean

J. P.

Metzner

B. S.

(1985). A conceptual model of nontraditional undergraduate student attrition. Review of Educational Research, 55(4), 485–540. Retrieved from https://www-jstor-org.web.bisu.edu.cn/stable/1170245

Berger

J. B.

Milem

J. F.

(1999). The role of student involvement and perceptions of integration in a causal model of student persistence. Research in Higher Education, 40(6), 641–664.

Berger

J. B.

Ramirez

G. B.

Lyons

(2012). Past to present: A historical look at retention. In Seidman

(Ed.), College student retention (2nd ed., pp. 7–34). Lanham, MD: Rowman & Littlefield Publishers.

Berkner

Horn

Clune

(2000). Descriptive summary of 1995–96 beginning postsecondary students: Three years later with an essay on students who started at less-than-4-year institutions (NCES 2000154). Retrieved from National Center for Education Statistics website: https://nces.ed.gov/pubs2000/2000154.pdf

Black

A. E.

Deci

E. L.

(2000). The effects of instructors’ autonomy support and students’ autonomous motivation on learning organic chemistry: A self-determination theory perspective. Science Education, 84(6), 740–756.

10.

Brooks-Leonard

(1991). Demographic and academic factors associated with first-to-second-term retention in a two-year college. Community/Junior College Quarterly of Research and Practice, 15(1), 57–69.

11.

Burrus

Elliott

Brenneman

Markle

Carney

Moore

Roberts

R. D.

(2013). Putting and keeping students on track: Toward a comprehensive model of college persistence and goal attainment (Research Report No. ETS RR-13-14). Retrieved from Educational Testing Service website: http://ww.ets.org/Media/Research/pdf/RR-13-14.pdf

12.

Carnevale

A. P.

Rose

S. J.

(2011). The undereducated American. Retrieved from Georgetown University, Center on Education and the Workforce website: https://cew.georgetown.edu/wp-content/uploads/2014/11/undereducatedamerican.pdf

13.

Carnevale

A. P.

Smith

Strohl

(2013). Recovery: Job growth and education requirements through 2020. Retrieved from http://www.luminafoundation.org/files/resources/recovery2020-fr-web.pdf

14.

Center for Community College Student Engagement. (2003). About the survey. Retrieved from http://www.ccsse.org/aboutsurvey/aboutsurvey.cfm

15.

Center for Community College Student Engagement. (2016). Expectations meet reality: The underprepared student and community colleges. Retrieved from http://www.ccsse.org/docs/Underprepared_Student.pdf

16.

Cofer

Somers

(2000). Within-year persistence of students at two-year colleges. Community College Journal of Research and Practice, 24(10), 785–807. doi:10.1080/10668920050179808

17.

Community College Research Center. (n.d). Community college FAQs. Retrieved from ccrc.tc.columbia.edu\Community-college-FAQs.html

18.

Complete College America. (2013a). The game changers: Are states implementing the best reform to get more college graduates? Retrieved from http://completecollege.org/pdfs/CCA%20Nat%20Report%20Oct18-FINAL-singles.pdf

19.

Complete College America. (2013b). The power of 15 credits: Enrollment intensity and postsecondary student achievement. Retrieved from http://www.completecollege.org/docs/CCA%20Intensity%20Brief-April3.pdf

20.

Complete College America. (2014a). About CCA: Our work. Retrieved from completecollege.org/about-cca/

21.

Complete College America. (2014b). Best practices: Corequisite remediation. Retrieved from http://completecollege.org/strategies/#stratHolderCoreqRemediation

22.

Deci

E. L.

Ryan

R. M.

(Eds.). (2002). Handbook of self-determination research. Rochester, NY: The University of Rochester Press.

23.

Dennis

J. M.

Phinney

J. S.

Chuateco

L. I.

(2005). The role of motivation, parental support, and peer support in the academic success of the ethnic minority first-generation college students. Journal of College Student Development, 46(3), 223–236.

24.

Elliott

S. N.

DiPerna

J. C.

(2002). Assessing the academic competence of college students: Validation of a self-report measure of skills and enablers. Journal of Postsecondary Education and Disability, 15(2), 87–100.

25.

Friedman

B. A.

Mandel

R. G.

(2011). Motivation predictors of college student academic performance and retention. Journal of College Student Retention, 13(1), 1–15.

26.

Furrer

Skinner

(2003). Sense of relatedness as a factor in children’s academic engagement and performance. Journal of Educational Psychology, 95(1), 148–162.

27.

Grasgreen

(2014). Registering toward completion. Retrieved from https://www.insidehighered.com/news/2014/04/11/cleveland-state-aims-boost-completion-through-multiterm-registration#sthash.4twRW96D.eUqEXKRO.dpbs

28.

Hagedorn

L. S.

(2005). How to define retention: A new look at an old problem. In Seidman

(Ed.), College student retention: Formula for student success. (pp. 89–105). Westport, CT: Praeger Publishers.

29.

Hair

J. F.

Black

W. C.

Babin

B. J.

Anderson

R. E.

(2015). Multivariate data analysis. (7th ed.). Noida, India: Pearson India Education Services.

30.

Hoffman

(2014). Testing measurement invariance across groups in item factor models in Mplus version 7.11. Retrieved from http://www.lesahoffman.com/948/948_Example9c_IFA_Multiple_Group_Invariance.pdf

31.

Horn

L. J.

Carroll

C. D.

(1996). Nontraditional undergraduates: Trends in enrollment from 1986 to 1992 and persistence and attainment among 1989–1990 beginning postsecondary students (NCES 97-578). Retrieved from National Center for Education Statistics website: https://nces.ed.gov/pubs/97578.pdf

32.

Hoyle

R. H.

(Ed.). (2012). Handbook of structural equation modeling. New York, NY: The Guilford Press.

33.

Hurtado

Carter

D. F.

(1997). Effects of college transition and perceptions of the campus racial climate on Latino students’ sense of belonging. Sociology of Education, 70(4), 324–345.

34.

Inkelas

K. K.

Weisman

J. L.

(2003). Different by design: An examination of student outcomes among participants in three types of living-learning programs. Journal of College Student Development, 44(3), 335–368.

35.

Keith

T. Z.

(2015). Multiple regression and beyond: An introduction to multiple regression and structural equation modeling. (2nd ed.). New York, NY: Routledge.

36.

Klempin

(2014). Redefining full-time in college: Evidence on 15-credit strategies. Retrieved from http://ccrc.tc.columbia.edu/media/k2/attachments/redefining-full-time-in-college.pdf

37.

Kline

R. B.

(2011). Principles and practice of structural equation modeling. (3rd ed.). New York, NY: The Guilford Press.

38.

Kuh

G. D.

(2001). Assessing what really matters to student learning. Change, 33(3), 10–17.

39.

Kuh

G. D.

(2003). What we’re learning about student engagement from NSSE. Change, 35(2), 24–33.

40.

Kuh

G. D.

(2009). The national survey of student engagement: Conceptual and empirical foundations. New Directions for Institutional Research, (141), 5–20. doi: 10.1002/ir.283

41.

Kusurkar

R. A.

Cate

T. J.

Vos

C. M.

Westers

Croiset

(2012). How motivation affects academic performance: A structural equation modeling analysis. Advances in Health Science Education, 18, 57–69. doi:10.1007/s10459-012-9354-3

42.

Marti

C. N.

(2008). Latent postsecondary persistence pathways: Educational pathways in American two-year colleges. Research in Higher Education, 49(3), 317.

43.

Maurer

Allen

Gatch

D. B.

Shankor

Sturges

(2012). Students’ academic motivation in allied health classes. Internet Journal of Allied Health Service and Practices, 10(1), 1–12. Retrieved from http://jahsp.nova.edu

44.

McClenney

K. M.

Waiwaiole

E. N.

(2005). Focus on student retention: Promising practices in community colleges. Community College Journal, 75(6), 36–41.

45.

Muthen

L. K.

Muthen

B. O.

(2015). Mplus user’s guide. (7th ed.). Los Angeles, CA: Authors.

46.

Nakajima

M. A.

Dembo

M. H.

Mossler

(2012). Student persistence in community colleges. Community College Journal of Research and Practice, 36(8), 591–613. doi:10.1080/10668920903054931

47.

National Center for Education Statistics. (n.d). Quickstats. Retrieved from http://nces.ed.gov/datalab/quickstats/default.aspx

48.

National Conference of State Legislatures. (2015). Performance-based funding for higher education. Retrieved from http://www.ncsl.org/research/education/performance-funding.aspx

49.

O’Reilly

(2014). Correlations among perceived autonomy support, intrinsic motivation, and learning outcomes in an intensive foreign language program. Theory and Practice in Language Studies, 4(7), 1313–1318.

50.

Pascarella

E. T.

Terenzini

P. T.

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

51.

Phillips

E. D.

(2013). Improving advising using technology and data analytics. Change: The Magazine of Higher Learning, 45(1), 48–55. Retrieved from http://www.changemag.org/Archives/Back%20Issues/2013/January-February%202013/improving-advising-full.html

52.

Reason

R. D.

Terenzini

P. T.

Domingo

R. J.

(2006). First things first: Developing academic competence in the first year of college. Research in Higher Education, 47(2), 149–175. doi: 10.1007/s11162-005-8884-4

53.

Ryan

M. P.

Glenn

P. A.

(2003). Increasing one-year retention rates by focusing on academic competence: An empirical odyssey. Journal of College Student Retention, 4(3), 297–324.

54.

Ryan

R. M.

Deci

E. L.

(2000). Intrinsic and extrinsic motivations: Classic definitions and new directions. Contemporary Educational Psychology, 25, 54–67.

55.

Schuetz

(2005). UCLA community college review: Campus environment: A missing link in studies of community college attrition. Community College Review. 32(4), 60–80. doi:10.1177/009155210503200405

56.

Schuetz

(2007). Influences of campus environment on adult community college student engagement (Doctoral dissertation). Retrieved from ProQuest Dissertations. (UMI No. 3288232).

57.

Schuetz

(2008). Developing a theory-driven model of community college student engagement. New Directions for Community Colleges, (144), 17–28. doi:10.1002/cc.342

58.

Schumaker

R. E.

Lomax

R. G.

(2010). A beginner’s guide to structural equation modeling. (3rd ed.). New York, NY: Routledge Taylor & Francis Group.

59.

Shapiro

Dundar

Wakhungu

P. K.

Yuan

Nathan

Hwang

(2015). Completing college: A national view of student attainment rates—Fall 2009 cohort (Signature Report No. 10). Retrieved from National Student Clearinghouse Research Center: https://nscresearchcenter.org/wp-content/uploads/SignatureReport10.pdf

60.

Skomsvold

(2014). Profile of undergraduate students: 2011–12. Retrieved from https://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2015167

61.

St. John

E. P.

Paulsen

M. B.

Starkey

J. B.

(1996). The nexus between college choice and persistence. Research in Higher Education, 37(2), 175–220.

62.

Taylor

McAleese

(2012). Beyond retention: Using targeted analytics to improve student success. Retrieved from http://er.educause.edu/articles/2012/7/beyond-retention-using-targeted-analytics-to-improve-student-success

63.

The White House. (n.d). Building American skills through community colleges. Retrieved from http://whitehouse.gov/sites/default/files/100326-community-college-fact-sheet.pdf

64.

Tinto

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

65.

Ullman

J. B.

(2006). Structural equation modeling: Reviewing the basics and moving forward. Journal of Personality Assessment, 87(1), 35–50.

66.

Wilds

Ebbers

(2002). Rethinking student retention in community colleges. Community College Journal of Research and Practice, 26, 503–519. doi:10.1080/02776770290041864