The Answer Is Yes

Abstract

Objective: The study assesses the impact of dual enrollment participation on remediation and completion for traditional first time, full-time freshmen at a community college in Northeast Tennessee. Method: This study began with the full population of 1,232 students who enrolled between 2008 and 2012 at a community college in northeast Tennessee the fall semester after finishing high school. The population was required to have American College Testing (ACT) scores, completely fill out the Free Application for Federal Student Aid (FAFSA), enroll full-time as a degree-seeking student, and complete the first fall semester. Propensity score matching was utilized to eliminate self-selection bias and enable parametric comparisons using optimal matching of dual enrollment participants and non-participants while controlling for a range of covariates. Results: The analyses showed that community college students who participated in dual enrollment were (a) 9% or nearly 3.4 times less likely to take remediation, (b) 26% or nearly 2.5 times more likely to graduate in 2 years, and (c) 28% or nearly 1.5 times more likely to graduate in 3 years. Contributions: This study contributes to the literature showing that dual enrollment reduces remediation rates and assists in timely completions for community college students. Policy recommendations are to increase equitable participation, normalize dual enrollment for students academically able to do college coursework, align state terminology with the nation, and improve data for future research.

Keywords

dual enrollment remediation completion community college students propensity score matching

In April 2014, Tennessee garnered national attention and merited presidential endorsement by becoming the first state in two decades to promise free community college to all high school graduates beginning in the fall of 2015. The Tennessee Promise restructured the state lottery-based scholarships to provide five semesters of tuition-free community college for Tennessee high school graduates entering college the fall semester after graduation. This program was the latest in a series of educational reforms in Tennessee that implemented an agenda emphasizing economic development through college readiness and completion of postsecondary credentials. The success of Tennessee Promise as an engine of economic development depends upon the ability of students to use community colleges as legitimate pathways to universities and complete college in a timely manner because the funding is finite and it is the completed credentials that will drive economic development. However, the completion of community college is often not timely. Rosenbaum and Rosenbaum (2015) have shown that 46% of students who begin at community college do not complete in 8 years, only 33% complete an associate degree, and only 20% earn a bachelor’s degree. These findings highlight the necessity of studies that help stakeholders understand community college students and factors that inhibit or promote student success.

Dual enrollment participation,¹ the earning of college course credit during high school, has been shown to benefit students in being college ready (An, 2013b; Rodriguez, Hughes, & Belfield, 2012) and in completing college (An, 2013a; Giani, Alexander, & Reyes, 2014; Speroni, 2011a, 2011b; Struhl & Vargas, 2012; Swanson, 2008; Taylor, 2015)—the two factors essential in making the Tennessee Promise succeed. Furthermore, the participation in dual enrollment courses across the nation has grown tremendously (Kleiner & Lewis, 2005; Marken, Gray, & Lewis, 2013; Thomas, Marken, Gray & Lewis, 2013; Waits, Setzer, & Lewis, 2005). However, the literature does not address the benefit of dual enrollment participation on reducing remediation specifically at the community college except for one correlational study (Kim & Bragg, 2008). Similarly, studies examining benefit of dual enrollment participation on the completion of associate degrees are nearly as rare (Speroni, 2011b; Struhl & Vargas, 2012), with only one study examining timely completions at 3 years (Struhl & Vargas, 2012). There are no studies examining the impact of dual enrollment participation for on-time completion at the community college.

The purpose of this study is to determine the contribution of dual enrollment participation to community college completion as measured by remediation and completion rates at 2 and 3 years at a community college. Using data that predate the beginning of the Tennessee Promise program from one Tennessee community college, we examined the outcomes for committed first time, full-time public high school graduates matriculating between 2008 and 2012. The entire study population numbered 1,232 students prior to the matching procedure. Commitment in this study is defined as taking the ACT prior to college, completing the FAFSA, enrolling full-time by the first fall semester after public high school at community college, and finishing the first semester without withdrawing or dropping to part-time status. The following research questions guided this study:

Research Question 1: What is the average effect of dual enrollment participation on remediation rates?

Research Question 2: What is the average effect of dual enrollment participation on completion rates at 2 and 3 years?

Propensity score matching was utilized to create a quasi-experimental comparison of treatment and control groups based on dual enrollment participation or non-participation. A sensitivity analysis was used to assess the influence of unobserved variable bias.

Studies of Dual Enrollment Impacts on Remediation and Completion

Remediation at the beginning of college is antithetical to timely completion (Complete College America, 2012). Adelman (2006) found that rigorous educational courses during high school overcame any statistical tendencies to achieve less academically—whether gender, race, socioeconomic status, or family structure. Dual enrollment offers rigorous academic experience to high school students and connects the high school student to higher education. However, the literature on remediation reduction through dual enrollment is limited. Karp, Calcagno, Hughes, Jeong, and Bailey (2007) reported that policy makers and others seeking to influence educational policy previously had assumed that dual enrollment would reduce remediation. For example, in a survey of state policies regarding reform of secondary education in the United States, Martinez and Bray (2002) explicitly stated that dual enrollment participation would reduce the need for remediation. Yet, remediation or college readiness (defined as no remediation) has rarely been examined. Most of the subsequent literature on dual enrollment does not address remediation (Allen & Dadgar, 2012; An, 2013a; An & Taylor, 2015; D’Amico, Morgan, Robertson, & Rivers, 2013; Giani et al., 2014; Kanny, 2014; Morrison, 2008; Pretlow & Wathington, 2014; Speroni, 2011a, 2011b; Struhl & Vargas, 2012; Swanson, 2008; Taylor, 2015). In 2008, Kim and Bragg declared that they were perplexed “that no studies were found that controlled for prior academic performance while investigating the influence of dual credit on remediation” (p. 135).

Kim and Bragg (2008) examined differing correlations between academic dual enrollment, career and technical education (CTE) dual enrollment, and articulated credit examining college readiness at four community colleges. Kim and Bragg noted a significant correlation between taking academic dual credit coursework and college readiness (no remediation) in mathematics in three consortia: Texas (r = .15), Florida (r = .34), and Oregon (r = .19). The Oregon consortium alone had a significant correlation between academic dual credit coursework and college readiness in writing but not in reading. Kim and Bragg filled a gap in the literature by completing the only study that addressed remediation effects on typical dual enrollment populations at community colleges. However, the study was correlational. The lack of a quasi-experimental framework and control for significant observable covariates (e.g., socioeconomic status or race/ethnicity) limits the impact of the study given the emergence of more rigorous methods.

Of the two more statistically rigorous studies that assessed dual enrollment effects on remediation (An, 2013b; Rodriguez et al., 2012), neither study reported the remediation rates specifically at the community college. An (2013b) studied the impact of dual enrollment participation on remediation using propensity score matching (PSM). The study combined two nationally representative data sets, the Beginning Postsecondary Students Longitudinal Study and the 2009 Postsecondary Education Transcript Study. An (2013b) found that dual enrollment students across postsecondary institutions, aggregating community colleges with 4-year institutions, were 6% less likely to need any remediation upon entering college than non-dual enrollment students. Rodriguez et al. (2012) analyzed a data set from the California Partnership for Achieving Student Success (Cal-PASS), a voluntary statewide collaborative of primary, secondary, and postsecondary institutions including all public 2-year institutions and over half of public 4-year institutions. The analysis included the outcomes of a 3-year, grant-funded research study on the benefits of career-based dual enrollment for a highly disadvantaged population with 60% being students of color and 40% from non-English speaking homes. In addition to coursework, the project provided counseling and other supports to encourage student success. Aggregating the remediation rates to include both community college and 4-year institutions, Rodriguez et al. (2012; Table 7) found that the effect of program participation reduced remedial coursework from 41% for non-participating peers to 30% for participants in the class of 2009 and from 46% for non-participating peers to 32% for participants in the class of 2010, though the authors noted that some dual enrollment participants might have delayed remedial courses or enrolled in technical programs that did not require remediation before beginning coursework.

Completion has often been studied in the dual enrollment literature, but rarely has the analysis emphasized the completion of associate degrees earned at the community college and very rarely has the interval for completion examined been suitable to assess timely completions at the community college. Several studies utilizing large data sets examined dual enrollment effects on completion rates and have affirmed positive benefits for students (An, 2013a; Giani et al., 2014; Speroni, 2011a, 2011b; Struhl & Vargas, 2012; Swanson, 2008; Taylor, 2015), but rarely has the analysis emphasized the completion of associate degrees. The major studies used different methods of analysis coupled with various time to completion periods from 3 to 8 years. In addition, the types and combinations of credentials evaluated varied from one study to another. Two studies focused on completion rates for associate degrees without aggregating the associate degrees with certificates, bachelor’s degrees, or both (Speroni, 2011b; Struhl & Vargas, 2012). Speroni (2011b) studied Florida data using regression discontinuity to compare students just above and below the state grade point average (GPA) requirement for dual enrollment. Speroni found that dual enrollment participants who took a rigorous college algebra course were 23% more likely to finish an associate degree within 5 years than non-participants were. However, a 5-year completion period is 250% of degree time and would not be indicative of timely completion at the community college. Only one study examined completion of associate degrees within a suitable 3-year time frame (Struhl & Vargas, 2012). That study examined college outcomes for students in Texas. The analysis utilized PSM and controlled for a rich set of covariates finding that dual enrollment students were 1.83 times more likely to finish an associate degree within 3 years. The significant difference in completion rates between these studies may be attributable to the large difference in the study populations and completion intervals, 5 years for Speroni (2011b) and 3 years for Struhl and Vargas (2012). However, even in the work of Struhl and Vargas (2012), there was no examination of on-time completion at the community college. In fact, on-time completion is examined for bachelor’s degrees only once in the literature (Giani et al., 2014). See Table 1 for a detailed listing of studies on dual enrollment focusing on completion including the completion interval and the certificate and degree combinations.

Table 1.

Rigorous Studies of Dual Enrollment Effects on Completion.

Author(s)	OR/MD	Years to completion	Scope	Method
An (2013a)
Bachelor’s	7%	8	National	PSM
Bachelor’s and associates	8%	8	National	PSM
Giani, Alexander, and Reyes (2014)
Bachelor’s (began any institution)	1.54	4	Texas	PSM
Bachelor’s (began at 4 year)	1.37	6	Texas	PSM
Bachelor’s (began any institution)	1.72	6	Texas	PSM
Bachelor’s, associates, and certificates	1.52	6	Texas	PSM
Speroni (2011a)
Bachelor’s (non-minorities)	8%	5	Florida	Regression discontinuity
Bachelor’s (minorities)	6%	5	Florida	Regression discontinuity
Speroni (2011b)
Associates (marginal students taking college algebra)	23%	5	Florida	Regression discontinuity
Bachelor’s (marginal students taking college algebra)	24%	5	Florida	Regression discontinuity
Bachelor’s and associates (marginal students taking college algebra)	23%	5	Florida	Regression discontinuity
Struhl and Vargas (2012)
Associates	1.83	3	Texas	PSM
Bachelor’s	1.46	6	Texas	PSM
Bachelor’s and associates	1.66	6	Texas	PSM
Taylor (2015)
Bachelor’s (full sample)	1.79	7	Illinois	PSM
Bachelor’s (minorities only)	1.87	7	Illinois	PSM
Bachelor’s (low SES only)	1.89	7	Illinois	PSM
Swanson (2008)
Bachelor’s	1.16	8	National	Linear regression
Associates and certificates	1.89	8	National	Linear regression

Note. When the analyses specified a population beyond degree attainment, it is noted in parentheses. OR/MD = odds ratio or mean difference of marginal effects; PSM = propensity score matching; SES = socioeconomic status.

There are gaps in the literature regarding dual enrollment effects on remediation and completion at the community college. This study contributes to filling those gaps by examining dual enrollment effects on a population at one community college, adding to the work of Kim and Bragg (2008) by examining remediation rates using PSM and adding to the work of Struhl and Vargas (2012) for completions rates at 150% of degree time and reporting for the first time in the literature on completion rates at 100% of degree time. In addition, this study includes remediation and completion effects in the same study, which has not appeared in the dual enrollment literature previously.

Conceptual Model

The conceptual model for this study is a modified version of the definition of college readiness offered by Conley (2007), who defined college readiness as “the level of preparation a student needs in order to enroll and succeed—without remediation—in a credit-bearing course at a postsecondary institution that offers a baccalaureate degree or transfer to a baccalaureate program” (p. 5). An emphasis on the baccalaureate degree pervades the dual enrollment literature to date, but associate degrees, the focus of this study, often lead to affluence and fulfilling careers and thus deserve consideration apart from transfer to a 4-year institution. Conley (2007) defined success as passing introductory coursework or making progress in a sequence of courses. Educational reforms predicating economic development driven by completions emphasize success as earning a credential or degree—not simply completing courses. With these qualifications in mind, we define community college readiness as the level of academic preparation a student needs to enter community college—without remediation—and to complete college in a timely manner. The treatment under examination is dual enrollment participation. The research questions connect to the study’s conceptual model by examining whether participation in dual enrollment decreases the frequency of remediation and increases the frequency of timely completions in the treatment groups in comparison with the matched control group. If dual enrollment participation increases community college readiness, then the treatment group should have fewer students taking remediation and have more timely completions than the control group. It is important to note that this is a “black box” study like that of Taylor (2015, p. 16), meaning we are not addressing why dual enrollment participation may increase community college readiness, but whether it increases such readiness.

Research Design

Policy Context

The policy framework for dual enrollment in Tennessee is complex. Unlike some states, Tennessee does not require that dual enrollment opportunities be offered in high schools. However, the state has provided a lottery-funded dual enrollment grant since 2005. In addition, the funding formula for community colleges provides incentive to offer dual enrollment. State policy lays out the definitions and basic eligibility requirements for dual enrollment students while leaving room for institutional requirements for specific programs or courses (Tennessee Board of Regents, 2016). Most of the public high schools in the service area of the community college in this study have dual enrollment classes at the site. All courses offered by the college are taught by college faculty using the same curriculum as the courses at the college campuses and online. Some individuals included in the study may have taken dual enrollment on the college campus or at another college or university campus and transferred those credits upon admission.

Data Sources and Sample

The data set used in this study was obtained from Northeast State Community College in Blountville, Tennessee. Our interest in selecting data was to prevent an inflated estimation of dual enrollment impacts that may result from including students who were not committed to going to college. For this reason, we required that all participants in the study be first time, full-time traditional freshmen entering college by the fall after finishing public high school. To establish commitment in the population, we required that each student have ACT scores prior to college, complete the FAFSA, be degree seeking, and complete the first semester without withdrawing or dropping to part-time status. This last provision was to avoid including what Adelman (1999) termed false starts—students who are not committed to college (p. 42). These requirements yielded an initial population of 1,232 students, 246 of whom had participated in dual enrollment.

Variables

The 11 independent variables used to generate the propensity score were selected to capture many facets of identity and ability that would affect dual enrollment participation. The variables included gender (male, female), minority status (White or non-White), academic performance in high school reflected by GPA, college readiness as indicated by ACT composite score and all subscores, financial dependence on parents for financial aid purposes (true, false), socioeconomic status as indicated by the level of PELL award (full, partial, or none), and whether or not at least one parent had a college degree. Dependent variables were participation in remediation and completion at 2 years, 100% degree time, and 3 years, 150% degree time.

Analytic Strategy

This study utilizes PSM as the basis of analytic strategy. Rosenbaum and Rubin (1983) developed PSM to facilitate quasi-experimental comparisons of observational data. PSM uses a logistic regression model to create a propensity score for each participant in the data set. The propensity score encapsulates the likelihood of an individual to volunteer for the treatment under study. After creating the propensity scores, a matching algorithm selects a control group from the untreated group to minimize the covariate differences between the groups. For this reason, a logistic regression alone would not be sufficient to create a quasi-experimental study. The essence of the technique is to equalize observed characteristics between treatment and control groups removing observed bias from the comparison.

The steps of analysis were to (a) choose matching algorithm, (b) use software to estimate propensity scores and create a matched set of cases with the minimum amount of bias, (c) assess covariate balance and bias reduction, (d) estimate the average treatment effect on the treated with effect size, and (e) conduct sensitivity analysis. The analysis was completed using the R statistical software (R Core Team, 2015) through the RStudio graphical environment (RStudio Core Team, 2015) including the following critical packages: MatchIt (Ho, Imai, King, & Stuart, 2013), optmatch (Hansen et al., 2015), and stats (R Core Team, 2016). Because of the binary nature of the outcomes studied and a lack of an applicable package in R, the last stage of sensitivity analysis utilized an Excel spreadsheet that is preconfigured to assist in such analyses (Love, 2008).

The first step was to select a matching algorithm. The matching algorithm used in the procedure was 1 to 1 optimal matching without replacement. Optimal matching is a form of nearest neighbor matching that minimizes the absolute distance between matches for the entire sample (Ho, Imai, King, & Stuart, 2011). The second step of the analysis was completed by using software discussed above. MatchIt (Ho et al., 2013) by default uses a logistic regression to calculate the propensity scores. The MatchIt package then calls the optmatch package automatically for optimal pair matching. The following equation generated the propensity score:

{PS}_{i} {= β}_{0} {+ β}_{1} X_{i}

Where PS is the probability of an individual to participate in dual enrollment, the intercept is β₀, the vector of covariates is X_i, and the parameter estimate is β_1. The following command in R generated the propensity score and matched one control case to each treatment case:

\begin{array}{l} matchit (formula = DE_PARTICIPATION ~ IS_FEMALE + IS_WHITE + \\ HSGRAD_GPA + ACT_ENGLISH + ACT_MATH + ACT_READING + \\ ACT_SCIENCE + ACT_COMPOSITE + DEPENDENT_STATUS + \\ PELL_AWARD_LEVEL + AT_LEAST_ONE_PARENT_WITH_ \\ COLLEGE_OR_MORE, data = DISSERTATION_DATA, method = " optimal ", \\ ratio = 1, replace = FALSE) \end{array}

The third step, assessing covariate balance and standardized bias reduction, is reported in Table 2. This table reports comparisons of each covariate used in the matching procedure via proportion or numerical score before and after matching for the treatment and control groups. Significance of the difference between treatment and control groups for each variable was assessed by employing chi-square procedures for proportional comparison of qualitative variables and t-test procedures for quantitative variables. The right side of Table 2 reports the percentage of standardized bias before and after matching with the percentage of reduction of standardized bias given in the last column. MatchIt summary reported a 99.9% reduction of bias as seen through the propensity score.

Table 2.

Covariate Imbalance Check for 1:1 Optimal Match Without Replacement.

Variables	Before: Proportion or M		After: Proportion or M		Chi-square or t test—association between X and T		Percent bias		Bias reduced (%)
Variables	DE	Non-DE	DE	Non-DE	Before	After	Before	After	Bias reduced (%)
ACT composite	22.2	20.9	22.2	22.3	6.77 (p < .001)	−0.55 (p = .58)	51.7	−5.5	89.3
ACT English	22.5	20.9	22.5	22.7	5.93 (p < .001)	0.86 (p = .39)	44.1	−8.2	81.4
ACT math	21.6	20.6	21.6	21.6	4.09 (p < .001)	0.22 (p = .82)	29.6	−2.1	93.1
ACT reading	23.3	21.7	23.3	23.4	5.71 (p < .001)	0.22 (p = .82)	42.2	−2.2	94.8
ACT science	22.2	21.0	22.2	22.3	5.64 (p < .001)	0.26 (p = .80)	42.6	−2.5	94.1
High school GPA	3.43	3.16	3.43	3.45	8.74 (p < .001)	−0.29 (p = .77)	65.1	−2.6	96.1
Sex					10.85 (p < .001)	0.03 (p = .85)
Male	0.42	0.54	0.42	0.41			−24.2	2.5	89.8
Female	0.58	0.46	0.58	0.59			24.2	−2.5	89.8
Race					0.49 (p = .48)	0.09 (p = .77)
Non-White	0.03	0.04	0.03	0.02			−7.3	4.9	32.9
White	0.97	0.96	0.97	0.98			7.3	−4.9	32.9
Dependent status					0.36 (p = .55)	0.00 (p = 1.00)
True	0.95	0.96	0.95	0.95			−5.0	0.0	100
False	0.05	0.04	0.05	0.04			5.0	0.0	100
PELL award					15.26 (p < .001)	0.65 (p = .72)
No award	0.65	0.51	0.65	0.63			28.9	4.2	85.3
Partial award	0.18	0.26	0.18	0.17			−21.3	2.1	90.1
Full award	0.18	0.23	0.18	0.20			−5.7	−2.9	49.6
At least one parent with college or more					2.61 (p = .11)	2.36 (p = .12)
True	0.57	0.51	0.57	0.50			12.1	14.8	−22
False	0.43	0.49	0.43	0.50			−12.1	−14.8	−22
Propensity score	26.2	18.4	26.2	26.2	−9.84 (p < .001)	0.55 (p = 59)	69.6	0.0	99.9

Note. DE = dual enrollment; ACT = American College Testing; GPA = grade point average.

The fourth step was to assess statistical significance of the outcomes. McNemar’s chi-square test with continuity correction was used to compare the treatment and control groups after matching. In addition, we calculated the odds ratios, standard error of the odds ratio, marginal effects of the mean difference, and the standard error of the marginal effects. The fifth and final step of analysis was to conduct a sensitivity analysis to determine whether the results were robust to unobserved variable bias. P. R. Rosenbaum (2002) developed the sensitivity test that assesses how large the effect of an unobserved confounding variable would have to be to change the outcome of the analysis significantly. The larger the sensitivity score, represented by Γ, the less likely it is that a hidden bias is present. R does not have a sensitivity package for binary outcomes. Therefore, we used a sensitivity analysis spreadsheet by Love (2008), which calculates a range of sensitivity increments between 1 and 6 with associated p values. It is possible from this range to arrive at the Γ level through these calculations.

Treatment

All of the students in this study participated in dual enrollment and earned college credit as a result. Participants were required to be current high school students and apply to the college for admission. In addition, applicants provided a high school transcript and ACT or Scholastic Assessment Test (SAT) scores. Prior to taking dual enrollment, students enrolling in college English were required to have an ACT subscore of 18 in English and 19 in reading. Students enrolling in college math were required to have an ACT subscore of 19 in math. SAT requirements were a 920 minimum composite score with scores of at least 460 for math or verbal. Students had to have parental consent, have worked with a school counselor, and signed a Family Educational Rights and Privacy Act (FERPA) release prior to acceptance.

The dual enrollment grant required that the student be admissible to the college, but did not require any additional academic qualifications. However, to access grant funding for a second course during the same semester, students were required to be Hope Scholarship eligible by having an ACT composite score of 21 or a 3.0 high school GPA.

Limitations

The most significant limitation of this study was the lack of covariates that may also influence educational decisions. There were no data on school counseling, family size or structure, language spoken at home, or participation in extracurricular activities (Taylor, 2015). The specific availability of rigor in the students’ respective high schools was also not available (Adelman, 1999). There are temporal issues with high school GPAs that include grades during and potentially after dual enrollment participation. The location of the class and the status of the instructor could also be significant in assessing dual enrollment effects (Speroni, 2011a). Some researchers (An, 2013b; Speroni, 2011b) have examined outcomes based on taking a particular course; the data for this study did not include courses taken. Because of all of these missing variables and others not listed, it must be asserted that findings be considered tentative. Another limitation is the scope of the study. Data regarding transfer to other institutions were not available for this study. Transfer data may have indicated more or less success for students who did not persist at the community college.

Results

Table 3 reports the descriptive results for the research questions. The outcomes of dual enrollment participants were superior to those of non-participants. Less than 4% of dual enrollment participants were placed in remediation, but over 30% of participants completed college in 2 years and over 45% finished college in 3 years. In contrast, over 11% of non-participants needed remediation while only 15% completed in 2 years. Even at 3 years, the non-participants lag participants by 9%. Although causation cannot be proved, there is a stark difference in outcomes after PSM. These findings are in harmony with the dual enrollment literature in finding positive effects on remediation (An, 2013b; Rodriguez et al., 2012) and completion (An, 2013a; Giani et al., 2014; Speroni, 2011a, 2011b; Struhl & Vargas, 2012; Swanson, 2008; Taylor, 2015).

Table 3.

Descriptive Comparison of Remediation and Completion Rates.

Dependent variable	Total (492)	Dual enrollment (246)	Non-dual enrollment (246)
Remediation	7.5% (37)	3.6% (9)	11.4% (28)
Completion in 2 years	23.2% (114)	30.9% (76)	15.4% (38)
Completion in 3 years	42.1% (207)	46.7% (115)	37.4% (92)

Matched Samples and Effect Heterogeneity

The odds ratios and marginal effects for the propensity score models are reported in Table 4. As with the descriptive comparison, the odds ratios show that dual enrollment participants were far less likely to require remediation and were far more likely to complete college in 2 or 3 years. The odds ratio of the logistic regression model for dual enrollment participants being placed in remediation was 0.30 (p < .001) or less than one third that of non-participants. The odds ratio of the logistic regression model for dual enrollment participants completing college in 2 years was 2.45 (p < .001) and in 3 years 1.47 (p < .01). The mean difference in remediation rates between the matched sample of dual enrollment participants and non-participants revealed that dual enrollment participants were 9% less likely to require remediation. Dual enrollment participants were 26% more likely to finish in 2 years and 28% more likely to graduate in 3 years. Results from the sensitivity analysis revealed significant Γ values of 1.70 for remediation, 1.78 for completion in 2 years, and 1.06 for completion in 3 years. This means that remediation and completion in 2 years were sensitive to a hidden bias. Completion in 3 years was very sensitive to a hidden bias.

Table 4.

Comparison of Odds Ratios and Marginal Effects.

Dependent variable	Naive odds ratio before matching^a	Odds ratio after matching	Marginal effects—mean difference
Remediation	0.16** (.35)	0.30** (.39)	−0.09** (.03)
Completion in 2 years	4.56** (.18)	2.45** (.22)	0.26** (.04)
Completion in 3 years	2.60** (.15)	1.47 * (.18)	0.28* (.04)

Note. Standard errors provided in parentheses. Significance was assessed with McNemar’s chi-square test with continuity correction.

Naive odds ratio before matching utilized the full data set of 1,232 cases (246 were dual enrollment participants).

p < .01. **p < .001.

Discussion and Implications

This study contributes to the literature on dual enrollment effects on remediation and college readiness by analyzing the benefit of participation for community college students. Kim and Bragg (2008) found a significant correlation between dual enrollment and being ready for community college. An (2013b), in a national study, reported that dual enrollment students were 6% less likely to require remedial coursework. However, An reported aggregated rates that did not separate the reduction rate for community college students from that of students at 4-year institutions. This aggregated rate does not reveal whether dual enrollment is more or less helpful to community college students. Logic suggests that because universities are more selective, dual enrollment participants who begin at the university may not benefit as much from participation. If that were true, dual enrollment participants who choose community college would experience a greater benefit over non-participating peers than dual enrollment participants who initially choose to attend a university. The findings of this study are favorable to this line of reasoning. Dual enrollment students in this study were 9% less likely (p < .001) to require remedial coursework. This reduction is 50% higher than that found by An (2013b). Yet, this finding is not quite as large as that identified by Rodriguez et al. (2012).

Rodriguez et al. (2012) reported an 11% reduction in remediation. The dual enrollment courses in their study focused on career skills instead of university parallel dual enrollment as in this study. Their study population was disadvantaged economically and unrepresented racially and ethnically in higher education. For these reasons, the dual enrollment coursework was supplemented with social support and counseling. The control group in the current study of first time, full-time committed students only had a 12% remediation rate in comparison with the 41% rate of the control group of Rodriguez et al. (2012). The treatment group in this study had a remediation rate of 3.6%—over 8 times less than the 30% remediation rate of the treatment group of Rodriguez et al. (2012). Rodriguez et al. (2012) also expressed some concerns that their student population may have avoided remediation by seeking certificates rather than degrees upon entering college. Dual enrollment participation remained highly beneficial to the participants in this study.

The benefits of dual enrollment participation on completion have been frequently demonstrated in the literature (see Table 1). However, only two prior studies examined the completion rates for associate degrees without grouping other degrees or certificates with the 2-year degrees (Speroni, 2011b; Struhl & Vargas, 2012). Speroni (2011b) studied a statewide data set from Florida comparing marginally eligible dual enrollment students who participated in a rigorous college algebra course with students who were marginally ineligible for dual enrollment participation. Under these strict delimitations, the dual enrollment students were 23% more likely to finish an associate degree than non-participating students were within 5 years. In comparison, this study found that dual enrollment participants were 28% more likely to complete in 3 years. The difference in rates could result from the 5-year interval for completion used by Speroni (2011b). This longer interval may have allowed more non-dual enrollment participants to complete. Alternatively, it would be reasonable to postulate that a population composed entirely of students on the margin of participation may be slightly less likely to complete college than in general. Even were this so, a 5-year completion period is 250% of degree time and would not be indicative of timely completion at community college.

Struhl and Vargas (2012) provided the only analysis of dual enrollment effects on the completion of associate degrees in 3 years in the literature. Struhl and Vargas (2012) in a statewide study of Texas students found dual enrollment participants were 1.83 times more likely to finish in 3 years. This present study found 46% of dual enrollment participants completed in 3 years, while 33% of non-participants completed in 3 years. Dual enrollment students were 28% or 1.5 times more likely to finish in 3 years. The study of Texas students had a much larger and richer data set; it is possible that the richness and size of the data set gave a more accurate measure of the effect of dual enrollment participation. It is also possible that the delimitations of the present study population eliminated students who completed in 3 years but did not enroll full-time immediately following high school or finished the first semester without dropping to part-time or withdrawing. Other factors such as differences in student transfers may also have accounted for the difference in the rates of completion.

The studies of dual enrollment effects on completion feature extended times to degree from 150% up to 250% or more. The least common completion interval examined is on-time completion. The literature reviewed featured only one examination of on-time completion for any type of degree and that analysis focused on bachelor’s degrees (Giani et al., 2014). While Struhl and Vargas (2012) provided the only timely completion interval for associate degrees (3 years), there were no studies that examined on-time completion for associate degrees. This study fills that gap by providing the first analysis of on-time completion for associate degrees in the dual enrollment literature. Dual enrollment participants in this study completed at a rate of 31% in 2 years while 15% of non-participants completed in 2 years. Dual enrollment participants were 26% or 2.5 times more likely to complete in 2 years.

Dual enrollment participants are less likely to require remediation and more likely to complete in a timely manner—particularly on-time completion. However, the findings are rather sensitive to an unknown confounding factor. A richer, larger data set may render less sensitive findings as did Taylor (2015). Nevertheless, the preponderance of the literature combined with this focused study on a community college population indicates that participation in dual enrollment increases student success. The findings affirm that dual enrollment participation increases community college readiness defined as the level of academic preparation a student needs to enter community college—without remediation—and to complete college in a timely manner. The increase of timely completions contributes to the overarching goal of the Drive to 55 in Tennessee to increase educational attainment among the citizens of Tennessee (Drive to 55 Alliance, 2014). However, this good result is met with challenges around the policies and priorities of dual enrollment in the state.

Policy Goals and Priorities

Policy Goal 1: Increase Equity

In undertaking this study, we accepted the Tennessee agenda to focus on completion uncritically. The question of increasing the efficiency of preparing students for college and getting them to completion is important in its own right. Our first recommendation for policy would be to increase equitable participation in dual enrollment. A highly detailed report on the lottery-based scholarships by the Tennessee Higher Education Commission (THEC; 2012) reveals those families that are most underrepresented in higher education and would most benefit from participation in dual enrollment are either not being given opportunities to participate through their local high school or are not taking advantage of those opportunities. Ninety-nine percent of dual enrollment grantees have at least one parent with an associate degree. Nearly a third of grantees came from households with incomes in excess of US$96,000 and nearly two-thirds came from households with incomes above US$48,000. While Black Tennesseans make up about 20% of the population, only about 9% of dual enrollment grantees are Black. Students in rural areas have very little opportunity for dual enrollment because participation is directly affected by distance to the nearest college. Increasing equity may mean expanding the dual enrollment grant to embrace a needs-based component to cover fees or textbooks. The grant itself is accessed through a web portal. It may be that access to computers and the Internet are barriers to participation for the disadvantaged.

Policy Goal 2: Normalize Dual Enrollment

As Karp (2013) indicated, Tennessee must have more students prepared academically, enabled financially, and willing to enroll in postsecondary education. Currently, the remediation rate for entering freshman straight from high school in Tennessee is 63% (THEC, 2014). Of that cohort, about 68% will complete remediation work and finish college in 150% of degree time (THEC, 2014). This is in stark contrast to Tennessee students who take dual enrollment. When comparing dual enrollment grant recipients with Tennessee Educational Lottery Scholarship (TELS) recipients, dual enrollees tend to have a higher GPA, more often have both a 3.0 high school GPA and at least a 21 on the ACT, are more often eligible for a merit scholarship (8% vs. 5%), and are less likely to require remediation (12% vs. 19%; THEC, 2012). Dual enrollment grant recipients are also more likely to retain their TELS scholarships in the second year of college.

A close examination of the 986 students in the base data of this study who did not participate in dual enrollment revealed that only 75 were not qualified to take either college math or English by ACT subscores. Those eligible for college math numbered 734, while those eligible for college English numbered 706. Dual enrollment participation could help far more students avoid remediation and complete college. For that reason, we recommend that the state consider a long-term strategy to normalize dual enrollment for all college preparatory classes where students have demonstrated the academic ability to do college-level coursework. In so doing, many pipeline and equity issues could be resolved. Students who are capable of college-level work should be doing so as early as possible.

Policy Goal 3: Align Dual Credit Terminology to the Nation

Tennessee definitions of dual credit differ from common federal and state definitions. In most of the literature, dual credit is defined as a specific arrangement within dual enrollment parameters by which high schools award credit for college coursework (Waits et al., 2005). However, the Tennessee Department of Education (2015) defined dual credit as a local high school course with a comprehensive exam available for a fee created by a committee of high school teachers and college instructors at the state level. We recommend that the definition of dual credit in Tennessee be aligned with the rest of the nation to mean an arrangement whereby high school credit is awarded upon the completion of dual enrollment coursework. Such alignment would reduce confusion and facilitate research comparing Tennessee with other states or the nation as a whole.

Policy Goal 4: Improved Data for Research

One of the challenges in researching dual enrollment is the temporal compromise of data. Most of the academic markers available to researchers were recorded during or after dual enrollment participation. Neither ACT nor even a cumulative GPA from high school can indicate the student performance prior to dual enrollment participation. Capturing GPAs prior to and early in high school can provide researchers with a benchmark for prior performance to dual enrollment participation. Other standardized tests, including state performance testing, could enrich the data for analysis. We propose that a statewide educational attainment database be established to track students throughout their education from preschool through postdoctoral work. This recommendation arose from the literature review. Several states have such databases. Research has been produced to inform public policy in those states and nationally (Karp et al., 2007; Struhl & Vargas, 2012). Other data points of interest for dual enrollment evaluation include specific coursework, the status of the course instructor, and the mode and location of the course.

Implications for Future Research

As this study has shown, dual enrollment participation has an impact on reducing remediation and increasing timely completions for committed students at one community college. Future researchers may consider widening the scope to statewide studies of dual enrollment participation in community college systems including comparisons of timely completion rates of students who began at community college with those who began at a 4-year institution. Richer data sets including course taking, instructor status, course location, and other data may help researchers understand the effects of dual enrollment participation better. Research into the prevalence of students eligible for dual enrollment participation by academic ability but who do not participate in dual enrollment or other forms of accelerated study would also be beneficial.

As shown in the literature review, the completion interval of most dual enrollment studies is determined by the data available. This is understandable when the data cover a short time interval. However, when a large data set covers a period of time far longer than the certificates and degrees included require, including on-time completion intervals and 150% intervals would be beneficial in building the literature further. In addition, associate degrees are rarely the unit of analysis for credentials by themselves (see Table 1). This puts community colleges at a disadvantage to 4-year institutions in the analysis by obscuring the effect of dual enrollment at 2-year institutions.

The Niswonger Foundation (2015) administered an Investing in Innovation Grant from 2010 to 2015 in east Tennessee. While dual enrollment was only part of the grant, there was an explosion of participation in partner schools. If the work of this grant improved college readiness and fueled college completion, especially among underrepresented populations, the findings could justify further investment in dual enrollment.

Adelman (2006) found that students who had earned 20 credit hours by the end of their freshman year of college were significantly more likely to complete college. Swanson (2008) combined the “nest egg” threshold developed by Adelman with two other measurable benchmarks: enrollment within 7 months of finishing high school and no more than one semester of taking time off during college (p. 82). These three measurable markers constitute academic momentum. Swanson demonstrated that students who participated in dual enrollment were more likely to have academic momentum and complete college at greater rates. Future studies might seek to confirm whether dual enrollment does help earn a nest egg of credit and contribute to academic momentum. Such findings could shape public policy and strengthen the pipeline between secondary and postsecondary institutions.

Conclusion

Dual enrollment is a well-established practice to prepare students for college that is supported by research and stakeholders in education. The study underscores the benefits that dual enrollment participants who choose to attend community college have over non-participating peers. Participants more often avoid remediation and enjoy a higher rate of timely completions. As completion continues to become a defining measure of the value of higher education in Tennessee and elsewhere, dual enrollment participation should be considered an important strategy to help high school students connect to higher education and succeed.

Footnotes

Acknowledgements

We are grateful to the editorial staff of the Community College Review and the anonymous reviewers who have helped shape this article. We would also like to acknowledge Northeast State Community College for providing the data for this research.

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.

Notes

Author Biographies

John M. Grubb is the Director of Planning and Assessment at Northeast State Community College. Prior to working in Institutional Effectiveness, Dr. Grubb worked in academic libraries for 25 years.

Pamela H. Scott is Associate Professor and Program Coordinator for School Leadership at East Tennessee State University. Dr. Scott is former chair of the Department of Educational Leadership and Policy Analysis.

Donald W. Good is Professor in the Educational Leadership and Policy Analysis Department at East Tennessee State University. He has held faculty and administrative positions, including Dean and Vice President for Academic Affairs, at various colleges and universities.

References

Adelman

(1999). Answers in the toolbox: Academic intensity, academic patterns, and bachelor’s degree attainment. Washington, DC: U.S. Department of Education. Retrieved from https://www2.ed.gov/pubs/Toolbox/toolbox.html

Adelman

(2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education. Retrieved from http://www2.ed.gov/rschstat/research/pubs/toolboxrevisit/toolbox.pdf

Allen

Dadgar

(2012). Does dual enrollment increase students’ success in college? Evidence from a quasi-experimental analysis of dual enrollment in New York, NY. New Directions for Higher Education, 2012, 11-19. doi:10.1002/he.20010

B. P.

(2013a). The impact of dual enrollment on college degree attainment: Do low-socioeconomic status students benefit? Educational Evaluation and Policy Analysis, 35, 57-75. doi:10.3102/0162373712461933

B. P.

(2013b). The influence of dual enrollment on academic performance and college readiness: Differences by socioeconomic status. Research in Higher Education, 54, 407-432. doi:10.1007/s11162-012-9278-z

B. P.

Taylor

J. L.

(2015). Are dual enrollment students college ready? Evidence from the Wabash National Study of Liberal Arts Education. Education Policy Analysis Archives, 23, 1-26. doi:10.14507/epaa.v23.1781

Complete College America. (2012). Remediation: Higher education’s bridge to nowhere. Retrieved from http://www.completecollege.org/docs/CCA-Remediation-final.pdf

Conley

D. T.

(2007). Redefining college readiness. Eugene, OR: Educational Policy Improvement Center. Retrieved from http://www.epiconline.org/redefining-college-readiness/

D’Amico

M. M.

Morgan

G. B.

Robertson

Rivers

H. E.

(2013). Dual enrollment variables and college student persistence. Community College Journal of Research and Practice, 37, 769-779. doi:10.1080/10668921003723334

10.

Drive to 55 Alliance. (2014). Drive to 55 Tennessee. Available from http://driveto55.org/

11.

Giani

Alexander

Reyes

(2014). Exploring variation in the impact of dual-credit coursework on postsecondary outcomes: A quasi-experimental analysis of Texas students. The High School Journal, 97, 200-218. doi:10.1353/hsj.2014.0007

12.

Hansen

Fredrickson

Buckner

Errickson

Solenberger

Bertsekas

D. P.

Tseng

(2015). optmatch: Functions for optimal matching, including full matching (R package version 0.9-5). Retrieved from https://cran.r-project.org/web/packages/optmatch/optmatch.pdf

13.

D. E.

Imai

King

Stuart

E. A.

(2011). MatchIt: Nonparametric preprocessing for parametric causal inference. Boston, MA: Harvard University. Retrieved from http://r.iq.harvard.edu/docs/matchit/2.4-20/matchit.pdf

14.

D. E.

Imai

King

Stuart

E. A.

(2013, June 27). MatchIt: Matching software for causal inference (Version 2.4–21). Boston, MA: Harvard University. Retrieved from http://gking.harvard.edu/matchit

15.

Kanny

M. A.

(2014). Forks in the pathway? Mapping the conditional effects of dual enrollment by gender, first-generation status, and pre-college academic achievement on first-year student engagement and grades in college (Doctoral dissertation, University of California, Los Angeles). Available from ProQuest Dissertations and Theses database. (UMI No. 3622646)

16.

Karp

M. M.

(2013). Dual enrollment for college completion: Policy recommendations for Tennessee. New York, NY: Community College Research Center, Columbia University. Retrieved from http://ccrc.tc.columbia.edu/publications/dual-enrollment-recommendations-tennessee.html

17.

Karp

M. M.

Calcagno

J. C.

Hughes

K. L.

Jeong

D. W.

Bailey

T. R.

(2007). The postsecondary achievement of participants in dual enrollment: An analysis of student outcomes in two states. St. Paul, MN: National Research Center for Career and Technical Education, University of Minnesota. Retrieved from http://ccrc.tc.columbia.edu/media/k2/attachments/dual-enrollment-student-outcomes.pdf

18.

Kim

Bragg

(2008). The impact of dual and articulated credit on college readiness and retention in four community colleges. Career and Technical Education Research, 33, 133-158. doi:10.5328/CTER33.2.133

19.

Kleiner

Lewis

(2005). Dual enrollment of high school students at postsecondary institutions: 2002-03 (NCES 2005-008). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from http://nces.ed.gov/pubs2005/2005008.pdf

20.

Love

T. E.

(2008). Spreadsheet-based sensitivity analysis calculations for matched samples. Cleveland, OH: Center for Health Care Research & Policy, Case Western Reserve University. Retrieved from http://chrp.org/propensity

21.

Martinez

Bray

(2002). All over the map: State policies to improve the high school. Washington, DC: National Alliance for the American High School, The Institute for Educational Leadership.

22.

Marken

Gray

Lewis

(2013). Dual enrollment programs and courses for high school students at postsecondary institutions: 2010–11 (NCES 2013-002). Washington, DC: National Center for Education Statistics. Retrieved from http://nces.ed.gov/pubs2013/2013002.pdf

23.

Morrison

M. C.

(2008). The benefits of acceleration: Graduation advantages. Mason City, IA: North Iowa Area Community College.

24.

Niswonger Foundation. (2015). I3 grant & partnerships. Greeneville, TN: Author. Retrieved from http://www.niswongerfoundation.org/partnerships/

25.

Pretlow

Wathington

H. D.

(2014). Expanding dual enrollment increasing postsecondary access for all? Community College Review, 42, 41-54. doi:10.1177/0091552113509664

26.

R Core Team. (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Available from http://www.R-project.org/

27.

R Core Team. (2016). The R stats package. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org/

28.

RStudio Core Team. (2015). RStudio: Integrated development for R. Boston, MA: RStudio. Retrieved from http://www.rstudio.com/

29.

Rodriguez

Hughes

K. L.

Belfield

(2012, July). Bridging college and careers: Using dual enrollment to enhance career and technical education pathways. An NCPR working paper. New York, NY: National Center for Postsecondary Research.

30.

Rosenbaum

(2015). The new forgotten half and research directions to support them. New York, NY: William T. Grant Foundation. Retrieved from http://wtgrantfoundation.org/library/uploads/2015/01/Summary-The_New_Forgotten_Half_and_Research_Directions_to_Support_Them.pdf

31.

Rosenbaum

P. R.

(2002). Design of observational studies. New York, NY: Springer.

32.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41-55.

33.

Speroni

(2011a). Determinants of students’ success: The role of advanced placement and dual enrollment programs. An NCPR working paper. New York, NY: National Center for Postsecondary Research.

34.

Speroni

(2011b). High school dual enrollment programs: Are we fast-tracking students too fast? An NCPR working paper. New York, NY: National Center for Postsecondary Research.

35.

Struhl

Vargas

(2012). Taking college courses in high school: A strategy guide for college readiness–the college outcomes of dual enrollment in Texas. Boston, MA: Jobs for the Future.

36.

Swanson

J. L.

(2008). An analysis of the impact of high school dual enrollment course participation on post-secondary academic success, persistence and degree completion (Doctoral dissertation, The University of Iowa). Available from ProQuest Dissertations and Theses database. (UMI No. 3323472)

37.

Taylor

J. L.

(2015). Accelerating pathways to college: The (in) equitable effects of community college dual credit. Community College Review, 43, 355-379. doi:10.1177/0091552115594880

38.

Tennessee Board of Regents. (2016, March 30). Admissions: 2:03:00:00. Nashville: Author. Retrieved from https://policies.tbr.edu/policies/admissions

39.

Tennessee Department of Education. (2015). Terms. Nashville: Author. Retrieved from http://www.tn.gov/assets/entities/education/attachments/cte_res_terms.pdf

40.

Tennessee Higher Education Commission. (2012). 2012 Tennessee education lottery scholarship special report: An examination of grant and loan forgiveness programs for special populations. Nashville: Author. Retrieved from https://www.tn.gov/assets/entities/thec/attachments/LM2012_2012%20Lottery%20Special%20Report.pdf

41.

Tennessee Higher Education Commission. (2014). Tennessee higher education fact book 2013-2014. Nashville: Author. Retrieved from https://tn.gov/assets/entities/thec/attachments/LM2014_2013-2014_FACTBOOK.pdf

42.

Thomas

Marken

Gray

Lewis

(2013). Dual credit and exam-based courses in U.S. public high schools: 2010-11 (NCES 2013-001). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from http://nces.ed.gov/pubs2013/2013001.pdf

43.

Waits

Setzer

J. C.

Lewis

(2005). Dual credit and exam-based courses in U.S. public high schools: 2002-2003 (NCES 2005-009). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from http://nces.ed.gov/pubs2005/2005009.pdf