Abstract

One way administrators and policy makers have responded to the complexity of the community college transfer process has been to create articulation agreements between two-year and four-year colleges in a state or region. Our study examines the effects of one such statewide articulation and transfer policy, the Ohio transfer module (TM). This agreement is intended to allow individuals who successfully complete the TM at one institution to transfer all of these credits to a receiving institution. We find that students who complete the TM are more likely to transfer to a four-year institution and earn associate's degrees than observationally similar peers who did not complete a TM. We also find suggestive evidence that students who complete the TM are able to bring more credits with them when they transfer. However, students who complete the TM also take slightly longer to complete their bachelor's degrees. Thus, although the TM may improve the probability that students will transfer, it may be inefficient for students, leading them to spend more time enrolled in college. Moreover, because only a small number of students complete the TM, this policy may not be far-reaching enough to dramatically improve Ohio's community college transfer rate.

1.  Introduction

In order to meet former President Obama's stated goal of eight million new college graduates by 2020, policy makers will have to focus more attention on America's community colleges, which now enroll a larger share of undergraduates than any other sector in higher education. With over 1,100 institutions across the United States, community colleges served approximately 42 percent of the total undergraduate population in 2014, or approximately 7.3 million students (American Association of Community Colleges 2016; Ma and Baum 2016). The demand to serve low-income and underrepresented groups falls on community colleges because they are open access, lower cost, and more widespread. These institutions serve large numbers of first-generation and racial/ethnic minority students: 56 percent of Latino, 44 percent of black, and 40 percent of Asian students enrolled in higher education (Ma and Baum 2016). With the mission of providing access to higher education for all, community colleges provide a critical point of entry for students who may not otherwise be able to afford or be admitted to a four-year institution (Jenkins and Fink 2015).

Data from the National Center for Educational Statistics suggest that over 80 percent of community college students express a desire to one day complete a bachelor's degree (Horn and Skomsvold 2011). However, only about 24 percent of community college students ever transfer to a four-year institution (Shapiro et al. 2015), and there is evidence that these transfer rates are even lower for black and Latino students (Sengupta and Jepsen 2006; Gandara et al. 2012). Moreover, among students who start at a community college, only 17 percent completed a Bachelor of Arts degree (BA) within six years (Shapiro et al. 2013). In comparison, among full-time students who first enrolled at a four-year college in 2007, 59 percent had completed a BA six years later, although this percentage varies widely by institutional selectivity (Kena et al. 2015). Although research suggests higher odds of BA completion when beginning at four-year institutions (Doyle 2009; Long and Kurlaender 2009; Reynolds 2012), the reality is that the majority of students will begin their education at a community college. Thus, the transfer pathway from two-year to four-year colleges is critically important to improving BA completion rates nationwide (Goldrick-Rab 2010).

The transfer pathway, however, can be fraught with challenges for community college students, particularly when the requirements for transfer are not clear. In an effort to offer students a range of course options and major choices as part of a flexible college experience, many community colleges offer hundreds of different degree programs (Scott-Clayton 2011), yet these institutions often have significantly fewer resources than four-year colleges to provide advising and counseling to students as they navigate these choices. Helping students successfully transition from two-year to four-year institutions is an important mission of community colleges, but it is still not clear how to best design policies to support students in this transition.

One way state policy makers have responded to the complexity of the college transfer process has been to develop articulation agreements between public institutions in a state or region. These agreements are designed to assure that credits completed at one institution will be accepted into a degree program upon transfer to another institution. Our research focuses on one such statewide policy, the Ohio transfer module (TM). First developed and implemented in 1990, the Ohio TM was created to “address the articulation problems associated with students transferring from state-assisted technical and community colleges to state-assisted universities” and to “assist a student in advising and transferring” (Ohio Transfer Module 1990). It is a set of courses that satisfy the general education requirements common to all Ohio institutions. Each public college is allowed to determine which specific courses they will include in their TM, but all must contain a minimum of 36 semester hours of courses distributed between (1) English; (2) mathematics; (3) arts/humanities; (4) social and behavioral sciences; and (5) natural sciences. Students who successfully complete the TM at a two-year institution in Ohio are guaranteed to be able to transfer those credits to any other public college in the state, contingent on their admission, and are thus, in theory, able to transition more seamlessly and efficiently across institutions without confusion or disruption. In this study, we seek to understand if and how articulation agreements, like the Ohio TM, can assist students in successfully making the transition from two-year to four-year colleges. We ask the following research questions:

  1. What are the effects of completing the Ohio TM on transferring to a four-year college?

  2. Among those students who transfer from a two-year to a four-year college, how does the completion of the TM impact the likelihood of completing a bachelor's degree, the number of terms to a bachelor's degree, and the number of credits earned at the time of degree?

Ohio is one of many states that have adopted an articulation agreement between the state's public colleges. As of 2014, 42 states had adopted some type of statewide policy designed to smooth the transition of credits for students transferring between institutions (Education Commission of the States 2016). These articulation policies can be largely divided into two types: (1) policies that guarantee admission to a bachelor's degree program with full credit transfer (typically amounting to junior-level standing upon transfer) for students who complete their associate's degree at a community college (31 states, including Ohio1); and (2) policies that guarantee the transferability of a core, predefined set of general education courses, but with no guarantee of admission (36 states, including Ohio). In 2014, twenty-three states had policies pertaining to both types. Although both types of policies are designed to help students transfer as many of their prior college credits to their new institution as possible, students may transfer institutions at any time as long as they are able to secure admission to their new institution. States that guarantee admission to a four-year college upon completion of an associate's degree, however, incentivize completion of the associate's degree prior to transfer.

Despite the ubiquity of articulation agreements, little research has provided empirical evidence regarding their effectiveness in improving students’ postsecondary outcomes and college degree attainment rates in the transition from two-year to four-year colleges. A handful of studies have used national datasets to compare the probability of transfer for students in states with articulation policies to students in states without these policies and found little effect of the presence of an articulation agreement on transfer (Anderson, Sun, and Alfonso 2006; Roksa 2006; Roksa and Keith 2008; Gross and Goldhaber 2009). Because of the nature of a national dataset, however, these studies are only able to compare states with and without articulation agreements, not to evaluate the effects of a specific type of articulation policy on student success. If states have very different articulation agreements from one another, grouping states is likely to mask the effect an individual policy may have within its own state. Two recent working papers have attempted to address this issue by using state data to explore the effect of specific articulation policies that guarantee transfer with junior status for students who complete their associate's degrees before transfer (Crook, Chellman, and Holod 2012; Crosta and Kopko 2014). These studies find positive effects of this type of articulation agreement on BA completion. Our study, although similarly utilizing a state-level dataset, explores an articulation agreement that encourages students to complete a set of general education requirements before transferring.

This study contributes to the small body of literature on the effect of state articulation policies in several important ways. First, despite the large number of states with articulation policies, we still know little about the effectiveness of specific state policies on promoting transfer and reducing the number of course credits that are lost in the transfer process.

We are able to examine the effects of a statewide policy not only on BA attainment, but also on the probability of transfer, credit accumulation before transfer and at BA attainment, and the number of terms it takes students to acquire their BAs. Second, our analysis makes use of data from Ohio, one of the largest public higher education systems in the United States. The community college system in Ohio, the point of entry for the majority of students intending to transfer to a four-year college, has a student body similar to the national averages for community colleges in the United States (Appendix table A.1). Finally, our paper methodologically builds on previous research. Using longitudinal, student-level data from the Ohio Board of Regents, we use inverse propensity score weighting to compare the transfer and post-transfer outcomes for observationally similar students who did and did not complete the TM. In addition, we implement a sensitivity analysis to test the robustness of our estimates to potential confounders.

First, and perhaps most importantly, we find that very few students in our sample completed the TM. The small number of students who complete the TM suggests that this policy may not be far-reaching enough to dramatically improve Ohio's community college transfer rate. Second, we find that students who complete the TM are more likely to transfer to a four-year institution and earn associate's degrees than observationally similar peers. We also find suggestive evidence that students who complete the TM are able to bring more credits with them when they transfer than those who do not. However, students who complete the TM also take slightly longer to complete their BAs. Thus, although the TM may improve the probability that students will transfer, it may be inefficient for students, leading them to spend more time enrolled in college.

2.  Literature Review

The Potential Benefits of Articulation Agreements

Articulation agreements may improve transfer rates, the number of credits that transfer, and/or eventual bachelor's degree attainment. The goal of a formal state or system articulation policy may be to simply increase the number of students transferring from two-year to four-year colleges, and/or to help students who do transfer to reduce the number of credits that are lost during the transfer process. Articulation agreements address this latter goal by ensuring that any credits a student successfully completes under the agreement will be automatically accepted at the receiving institution, thus leading to a greater efficiency in the credit transfer process.

These agreements may benefit students, particularly those who intend to transfer, by making the pathway to transfer clearer and easier to understand. By outlining a pathway or a core set of courses for students, it may be the case that students are more likely to see transfer as an actual possibility and begin to take steps toward transfer that they may not have previously considered. Simply through the existence of a pathway, students may be encouraged to transfer to a four-year institution, particularly if they are assured that their courses will count toward graduation at their new institution. In this way, these agreements may help to promote not only transfer, but BA degree attainment as well.

A clearer pathway for course-taking can also lead to greater efficiency in the transfer of credits from one institution to another, which may, in turn, lead students to spend less time in college repeating credits they had already completed elsewhere. Academic momentum, or the speed at which students are able to complete their college coursework, is an important predictor of bachelor's degree completion (Adelman 1999, 2006; Attewell, Heil, and Reisel 2012). As the amount of time students spend successfully earning credits each term increases, the likelihood of graduation also increases (DesJardins, Ahlburg, and McCall 2006). This greater likelihood of degree completion can be explained by students’ developing greater connections with the institution over time (Tinto 1993), not exhausting their financial aid prior to graduation (Perna 1998), or a host of other reasons. To the extent that students are able to continue making academic progress toward a degree, the more likely they are to stay enrolled at their current institution or to transfer to another.

Articulation agreements may also help students to lower the cost of receiving a bachelor's degree, not only through reducing the number of excess credits they complete, but through tuition savings associated with completing portions of their coursework at a community college. In 2015–16, the national average tuition and fees for a full-time student at public two-year institutions was $3,435, compared with $9,410 at public four-year colleges (College Board 2015). Among BA degree recipients, enrolling in a community college for the first several terms or years may significantly lower the tuition costs of attaining that degree. In this paper, we test to determine whether any of these potential benefits are observed for students completing the Ohio TM. Below we summarize the existing prior research for these outcomes.

Articulation Agreements and Student Transfer

Much of the prior research on transfer policies does not find significant positive associations between these policies and increasing rates of student transfer (Roksa 2009). Using logistic regression and a national sample from the 1990–1994 Beginning Postsecondary Students Longitudinal Study (BPS:90/94), one study found that statewide articulation policies do not increase students’ likelihood of transferring, although the sample was restricted to students who responded to all three waves of the BPS survey and did not transfer for at least one year (Anderson, Sun, and Alfonso 2006). Similarly, Roksa and Keith (2008) examined students’ likelihood of transfer using the National Education Longitudinal Study of 1988 (NELS:88), and the Postsecondary Education Transcript Study (PETS), in combination with data they collected on the transfer and articulation policies of the 50 states, and found, on average, no effect of state articulation policies on individual student transfer rates. These results were similar to those found by Roksa (2006) in her analysis of the NELS:88 data. However, it is possible that the use of national datasets may not account for variation across states in the specific design of articulation policies, or the wide range of academic goals among students starting at community colleges. In response, Gross and Goldhaber (2009) used NELS:88, PETS, and the 1992 Integrated Postsecondary Education Data System data, along with the 1999 Survey of State Transfer and Articulation Policies conducted by Ignash and Townsend (2001), to explore more specifically whether different types of articulation policies were associated with different patterns of student transfer from two-year to four-year colleges. They found only small effects, concentrated among Latino students, that state transfer policies influence student transfer. However, their analysis is based on an analytic sample of only 3,621 students enrolled in 836 different institutions. It is possible that this sample was not sufficient to pick up effects.

Efficiency of Credits Transferred

Articulation agreements may help to reduce the number of credits that are lost during the transfer process. Recent evidence suggests that the problem of students completing more credits than they need for their degree program may be larger than previously thought. A 2011 report by Complete College America (CCA 2011) found that students completing associate's degrees had earned an average of 79 credits when only 60 are typically required for graduation, and bachelor's degree recipients earned 136.5 credits when only 120 credits were typically required. Recent research suggests that 12 percent to 14 percent of credits earned by BA degree holders were in excess of degree requirements (Auguste et al. 2010; Zeidenberg 2012), with substantial variation among students by major.

This issue of excess credits at graduation may be further exacerbated among transfer students if some of the credits students complete at one college do not carry forward to their new institution. Using a nationally representative dataset (BPS:04/09), Monaghan and Attewell (2014) document the widespread loss of credits that occur after students transfer from a community college to a four-year college. The authors found that only 58 percent of students who transferred from a two-year to a four-year college were able to keep 90 percent or more of the credits they earned while in community college. According to a 2014 report from the National Center for Education Statistics, of students who transferred from a two-year to a four-year college, approximately 39 percent transferred no credits, 28 percent transferred some credits, and 32 percent transferred all previously earned credits. On average, students lost approximately 13 credits following the first transfer (Simone 2014).

There are several ways in which coursework completed at one institution may not transfer to another. The first may be caused by a misalignment in course requirements between the same major at two-year and four-year colleges. In order to align major requirements, faculty across institutions have to agree on the equivalence of prerequisite, electives, and major requirements across colleges. If this is not done uniformly across institutions, students may end up taking versions of similar courses at both campuses. Even if these courses are aligned as a matter of policy, there may also be confusion about degree requirements in differing institutions through lack of information from advisors or counselors. Confusion over the different course requirements for a certain major at a two-year college versus a four-year college may be one of the key factors explaining why community college students accumulate excess credits (Zeidenberg 2012), which could in turn lead to lower levels of degree completion. Other reasons credits may not transfer may include unclear pathways through a degree program, change of majors, late declaration of major, or multiple transfers across two-year colleges before transferring to a four-year college. Even when general education courses are guaranteed to transfer across institutions, it is possible that these courses may not count toward the elective or major requirements at the new institution. For example, more technical majors such as engineering or one of the sciences may require specific math classes such that the math classes specified by the TM on a given campus may not fulfill the math requirement for the student's major.

Although it may seem that the primary function of an articulation agreement is to aid students in their transfer from one type of institution to the other, it is clear that an equally important purpose of the agreement is to help students avoid taking unnecessary courses and credits after they have transferred. According to Ohio state code, the transfer module in Ohio is designed to ensure transfer “in the most efficient and effective manner” possible (Ohio Board of Regents 2003). As such, policies like Ohio's are expected to reduce the number of credits lost in the transfer process, which would reduce the time to degree and potentially increase bachelor's degree attainment rates. Factors such as excess credits, that lengthen the time to degree, have been shown to reduce the probability of degree completion (Bailey 2009). We expand upon the prior research by exploring whether students who transfer with the TM complete their degrees with fewer excess credits than their peers who transferred without the TM.

Degree Completion

Transfer policies may aid students not only in transferring to a four-year college, but also in completing an associate's and/or bachelor's degree. Certainly, policies perfectly aligned with the requirements for an associate's degree are likely to result in completion of that degree for participating students. For programs like the Ohio TM, which only requires a portion of the credits needed for an associate's degree, it is not clear whether students would be encouraged to complete the TM and remain at the two-year college to earn an associate's degree before transferring, or to transfer prior to completing an associate's degree. Programs like this may encourage students to spend more time at the two-year college prior to transfer in an effort to complete the necessary requirements for the program. We test whether students who complete the TM and subsequently transfer earn more credits at the community college than their peers who did not complete the TM but also transferred.

Completing an associate's degree prior to transfer may help students be more successful in earning a bachelor's degree, post-transfer, given that students with an associate's degree often have more success in transferring credits from the community college to the four-year college (Jenkins and Fink 2015). The greater the loss of the credits in the transfer process, the lower the chances of completing a BA (Monaghan and Attewell 2014). Research suggests that students seeking baccalaureate degrees who begin college at four-year institutions have a 15 percent to 20 percent greater chance of completing those degrees than do their observationally similar peers who began at two-year institutions (Long and Kurlaender 2009; Shapiro et al. 2013). This penalty is explained, in part, by a loss in course credits that students typically experience during the transfer process (Doyle 2009; Long and Kurlaender 2009; Monaghan and Attewell 2014). Among transfer students in the BPS:96/01 and 1996 National Postsecondary Student Aid Survey, 82 percent earned a bachelor's degree when their four-year college accepted all of their community college credits, but only 42 percent of students earned a BA when the institution accepted only some of their credits (Doyle 2009).

In their 2008 study on the effects of statewide articulation policies, Roksa and Keith argue that the primary purpose of articulation agreements is to improve the efficiency of transfer. The authors advocate for examining the effects of these policies on students who have already transferred, and determining whether these policies help students to graduate with fewer credits and in less time than their peers. Using national data (NELS:88 and PETS) the authors find that the presence of an articulation agreement is not related to the number of credits needed to earn a bachelor's degree, the time to a degree, or the probability of earning a BA.

Two recent working papers make use of administrative datasets to examine the effects of articulation policies that guarantee junior status for students who transfer from a community college to a four-year institution after completing their associate's degree. Crook, Chellman, and Holod (2012) identify a sample of 4,549 students who transferred from a City University of New York associate's degree program to a bachelor's program in the 2003–04 academic year. They use a probit regression to examine whether the articulation policy affects students’ likelihood of earning a bachelor's degree and find that students subject to this policy who complete an academic, as opposed to vocational, associate's degree are more likely to go on to earn their BA than students who did not complete their degree before transferring. Similarly, Crosta and Kopko (2014) make use of administrative data from an unnamed state to examine the effect of another articulation agreement that guarantees junior status for students who transfer after completing their associate's degrees. Making use of matching methods, they find that students who earn their associate's degree before transferring are 1.8 percentage points more likely to earn their BA within four years of entering community college, 3.1 percentage points more likely to earn it within five years, and 4.0 percentage points more likely to earn it within six years. Both of these papers explore the long-term outcome of BA completion after students have already transferred. We expand upon this research by examining the effects of a specific state policy, not only on associate's and bachelor's degree completion rates, but also several intermediate outcomes, such as the likelihood of transfer, the credits earned prior to transfer, and the time and number of credits students accumulate on the way to earning a BA.

3.  The Ohio Transfer Module

The Ohio TM is a statewide transfer agreement that was first developed and implemented in 1990, with the original goal of helping to assist students in advising and transferring across institutions, particularly from two-year to four-year colleges (Ohio Transfer Module 1990). Although the Ohio TM has undergone changes since its adoption, its primary goal remains increasing the ease and efficiency with which students can transfer from one public college in the state to another.2 Efficiency in this case is measured by the number of credits that transfer from one institution to another. The original intention of the Ohio TM was to reduce the number of credits students lost in the transfer process, thereby allowing them to maintain their academic momentum despite changing institutions.

The TM allows students who complete a set of academic courses at one institution to transfer those courses to any receiving institution in the state. Each public college in Ohio is required to create a TM that specifies which courses at their institution satisfy the general education requirements common to all Ohio institutions. If a student completes any of these courses, she is assured that, upon transfer, each will count for credit at the receiving institution. Both the credits and the course are guaranteed to transfer to any other public college in the state, meaning the credits will transfer, as well as any requirements that the course fulfills. Each institution has identified its own institution-specific TM requirements, although there is a high degree of overlap across campuses. This overlap is due in large part to the requirement that the TM must contain a minimum of thirty-six semester hours of courses in English, mathematics, humanities, social and behavioral sciences, and natural sciences. Courses for the TM must be between 100- and 200-level general education courses. Essentially, completion of the TM fulfills the general education core of an Associate of Arts or Science degree or a Bachelor of Arts or Science degree in Ohio. The overwhelming majority of all associate's degrees in Ohio require 60 credits and bachelor's degrees require approximately 120 credits, 36 of which are general education requirements. For the majority of the institutions in the sample, completing the full TM requires somewhere between 36 and 40 semester credits. For example, as part of the Sinclair Community College transfer module requirements in the 2002–03 academic year, students were required to take at least 3 credits of mathematics. However, the mathematics courses that can be used to fulfill this requirement range from 3 to 5 credit hours. All TM courses must be approved by the Ohio Board of Regents and be made publicly available to students. Anecdotal evidence from Ohio suggests that the extent to which colleges publicize the TM to their students varies, although all colleges must have the requirements of the TM available to students in some form, most commonly the course catalog or on the college Web site.

Given that the TM is a nonbinding pathway, students may choose to take as many or as few courses in the TM as they wish before transferring. A student may complete the requirements in one subject, several subjects, or all five. Additionally, a student need not complete the entire set of courses within a given subject in order for that course or courses to transfer. For example, a student may take only one of their three required general education English courses, but as long as that course is designated as meeting the TM requirements in English, the individual course is guaranteed to transfer. In other words, all courses that make up the TM at the institution are protected upon transfer, regardless of how many were completed. However, students may benefit from completing the full TM prior to transfer as they are then assured that they have completed all of their general education requirements for an associate's degree, as well as most bachelor's degrees. A student is considered to have met the full TM requirement when he has completed all of the course requirements within all five of the aforementioned subject areas. For the sample institutions in our study we observe only 15 percent of students who express a desire to transfer actually completing the full module. This could be due to the visibility of the program, the role of advisors, the number of credits required, or potential confusion over the specific requirements. For a sample of the variation in the courses that fall under each of the subject areas within the TM, see Appendix table A.2.

4.  Empirical Strategy

Data

The data for this study were provided by the Ohio Board of Regents and consist of administrative files from all public institutions in Ohio. The public university system in Ohio is one of the largest comprehensive public systems of higher education in the nation, serving almost 600,000 students with fourteen public universities, twenty-four regional four-year branch campuses, and thirty-one community and technical colleges distributed across the state. The dataset includes enrollment files with student demographics, term-by-term credit completions and grade point averages (GPAs), and transcript information. Financial aid records report students’ adjusted gross income (AGI), as well as their estimated family contribution (EFC), as reported by the Free Application for Federal Student Aid (FAFSA). The dataset allows us to track the enrollment of students who first enrolled at a community college in 2002, 2003, and 2004 until the spring of 2010. In order to track students across cohorts for an equivalent amount of time, we use a tracking period of six years, which is sufficiently long enough to make inferences about degree attainment as well as transfer behavior, while still taking into consideration that many community college students attend part-time.3

Sample

We restrict the sample to first-time college students beginning at one of nine community colleges in Ohio in the fall of 2002, 2003, or 2004. We include only the nine institutions where we observe at least 1 percent of students completing the TM in our sample years.4 Although it is true that all public institutions should be offering some version of the TM, we only observe critical numbers of students actually completing the full module at nine institutions. There may be two primary reasons for these differences across colleges. First, anecdotal evidence from the colleges suggests that the TM appears to be more widely promoted at some campuses than others. Also, the requirements of the TM are more closely aligned with an Associate of Arts or general education curriculum and would likely be of little interest or advantage to students pursuing technical degrees. In fact, those institutions offering primarily technical degrees report almost no students completing the TM. By including only those colleges where we observe at least some students successfully completing the TM, we are more confident that our comparison group includes students who had an equal opportunity to access the TM, as opposed to including students at colleges in which the program was either nonexistent, not widely promoted, or not well understood. These nine institutions are geographically diverse and represent both large and mid-sized community colleges in the state. Columns 3 and 4 of Appendix table A.1 compare the 2004 characteristics of the nine colleges in our sample with all thirty-one public community colleges in the state. Although the colleges in our sample appear to have a higher percentage of black students and a lower percentage of dependent students compared to the total group of community colleges in the state, these are neither large nor statistically significant differences.

In our primary sample, we limit our analysis to students who were first-time enrollees, and who had complete basic demographic information, such as gender and ethnicity, in their enrollment files. Given that we are interested in the effects of an articulation agreement on transfer, we also limit our sample to students who expressed an intention to transfer when they first enrolled. Of the 38,349 first-time students entering one of the nine community colleges in our sample in 2002, 2003 or 2004, approximately 39 percent stated when they enrolled that they intended to eventually transfer to a four-year institution.5 Because some students attend community college never intending to transfer, we exclude those students from the sample and instead focus only on students for whom completing a TM would make the most sense. Students’ intention to transfer or their commitment to see a goal through to the end makes a substantial contribution to whether or not they transfer and ultimately persist in college (Bers and Smith 1991; Cabrera, Nora, and Castañeda 1993). Therefore, when examining barriers to transfer, it is important to consider only those students who express a desire to transfer, so as to distinguish among students with differing academic aspirations.

Finally, we limit our sample to students who took the ACT before enrolling in community college. We believe that taking the ACT signals that these students are serious about pursuing a four-year degree and are less likely to be stating that they intend to transfer with no understanding of what that might entail. Additionally, it is important in the creation of our comparison group that we have some measure of student ability prior to beginning college and making the TM decision. Because students are not required to submit ACT scores in order to attend community college in Ohio, almost half of all students attending community colleges do not have reported ACT scores at the time of first entry. Excluding these students from the sample requires excluding some students who completed the TM; however, this is necessary in order to obtain a uniform measure of student ability at the time of college entry. Because student selection into the TM is correlated with prior academic ability and potentially academic motivation, we need to be able to control for it. We caution that our results should be interpreted as applying only to those students who completed the TM after taking the ACT prior to enrolling in community college. Appendix table A.3 compares the demographic, high school, and college characteristics of students with reported ACT scores to those without in order to describe the students excluded from this study due to missing ACT scores.

Our final analytic sample includes 7,051 first-year students across all three cohorts (first-time, first-year students in 2002, 2003, and 2004), among which 939, or approximately 15 percent, complete the full TM. This ranges from 12.2 percent at Edison Community College to 20.9 percent at Lorain County Community College. Completing the full TM requires a minimum of 36 credits across several subject areas. Although many more students complete individual portions of the TM, managing the full requirements across all five subject areas requires substantial time and foresight. Appendix table A.2 contains further information on the specific TM requirements within each of the five subject areas at four of our sample colleges. The greatest numbers of students complete the English composition requirement, followed by the Social and Behavioral Sciences component. The Mathematics component of the TM is the least completed subject, despite being one of the subjects with the fewest requirements.

Table 1 presents the differences in demographic, high school, and college information variables for the students in our analytic sample, separating those who completed the full TM (column 1) from those who did not (column 2). All students included in table 1 expressed an intention to transfer to a four-year college. We see that students who completed the TM are more likely to be white or Asian, are slightly younger, and are more likely to be financially dependent on their parents than students who did not complete the TM. In addition, TM completers report higher AGI on their FAFSAs than those who did not complete the TM. Completers also have slightly higher ACT composite scores on average, but significantly higher ACT math and English scores, as well as higher high school GPAs (as reported to ACT) than those who do not. These patterns of higher academic performance in high school are perhaps not surprising, given the complexity of completing the TM. Students with more academic preparation, as measured by the ACT, may be more drawn to complete the TM than those with lower scores. Indeed, in the college information panel of this table, we see that students who complete the TM are more academically successful than those who do not; at the end of the first semester they have acquired more credits (12.7 versus 11.2) and have higher GPAs (3.02 versus 2.49).6 Below, we further explore the issue of academic ability and selection into completing the TM.

Table 1.
Descriptive Statistics for First-Time Freshmen Beginning at an Ohio Community College in the Fall of 2002, 2003, or 2004 with an Intention to Transfer
Completed Transfer Module (1)Did Not Complete Transfer Module (2)
Demographics   
Female 0.55 0.51 
White 0.83 0.81 
Black 0.06 0.10 
Latino 0.03 0.02 
Asian 0.02 0.01 
Dependent 0.96 0.93 
Age at first enrollment 18.6 18.9 
 (1.1) (1.8) 
Age <24 years 0.99 0.98 
Pell grant eligible 0.36 0.37 
Married 0.01 0.01 
Instate 0.99 0.99 
Average AGI 48,124 46,316 
 (21,944) (21,775) 
EFC 8,907 8,189 
 (7,341) (7,668) 
Mother graduated from HS 0.54 0.57 
Father graduated from HS 0.54 0.56 
Mother has at least some college 0.40 0.38 
Father has at least some college 0.39 0.37 
High school variables   
ACT composite score 20.3 19.1 
 (3.6) (3.6) 
ACT math score 20.0 18.8 
 (3.9) (3.7) 
ACT English score 19.4 18.0 
 (4.6) (4.6) 
ACT science score 20.6 19.6 
 (3.6) (3.8) 
ACT reading score 20.7 19.4 
 (5.2) (5.1) 
HS GPA 5.61 5.24 
 (1.03) (1.09) 
College information   
Enrolled part-time first semester 0.22 0.40 
Ever took a remedial course 0.62 0.67 
Took a remedial course first semester 0.52 0.57 
Ever worked during school year 0.99 0.98 
Worked first semester 0.78 0.86 
Number of semesters worked 4.63 2.78 
 (1.94) (1.80) 
Math major 0.04 0.04 
Science major 0.04 0.02 
Engineering major 0.05 0.08 
Social Science major 0.12 0.05 
Arts & Humanities major 0.39 0.32 
Health major 0.08 0.15 
Education major 0.07 0.05 
Business major 0.16 0.17 
Other professional major 0.01 0.04 
No major 0.07 0.10 
Attended more than one CC 0.05 0.11 
Number of terms at another CC 3.62 3.52 
 (3.49) (2.98) 
 [45] [685] 
Credits after first semester 12.7 11.2 
 (3.6) (4.5) 
GPA after first semester 3.02 2.49 
 (0.72) (1.06) 
  [6,041] 
N 939 6,112 
Completed Transfer Module (1)Did Not Complete Transfer Module (2)
Demographics   
Female 0.55 0.51 
White 0.83 0.81 
Black 0.06 0.10 
Latino 0.03 0.02 
Asian 0.02 0.01 
Dependent 0.96 0.93 
Age at first enrollment 18.6 18.9 
 (1.1) (1.8) 
Age <24 years 0.99 0.98 
Pell grant eligible 0.36 0.37 
Married 0.01 0.01 
Instate 0.99 0.99 
Average AGI 48,124 46,316 
 (21,944) (21,775) 
EFC 8,907 8,189 
 (7,341) (7,668) 
Mother graduated from HS 0.54 0.57 
Father graduated from HS 0.54 0.56 
Mother has at least some college 0.40 0.38 
Father has at least some college 0.39 0.37 
High school variables   
ACT composite score 20.3 19.1 
 (3.6) (3.6) 
ACT math score 20.0 18.8 
 (3.9) (3.7) 
ACT English score 19.4 18.0 
 (4.6) (4.6) 
ACT science score 20.6 19.6 
 (3.6) (3.8) 
ACT reading score 20.7 19.4 
 (5.2) (5.1) 
HS GPA 5.61 5.24 
 (1.03) (1.09) 
College information   
Enrolled part-time first semester 0.22 0.40 
Ever took a remedial course 0.62 0.67 
Took a remedial course first semester 0.52 0.57 
Ever worked during school year 0.99 0.98 
Worked first semester 0.78 0.86 
Number of semesters worked 4.63 2.78 
 (1.94) (1.80) 
Math major 0.04 0.04 
Science major 0.04 0.02 
Engineering major 0.05 0.08 
Social Science major 0.12 0.05 
Arts & Humanities major 0.39 0.32 
Health major 0.08 0.15 
Education major 0.07 0.05 
Business major 0.16 0.17 
Other professional major 0.01 0.04 
No major 0.07 0.10 
Attended more than one CC 0.05 0.11 
Number of terms at another CC 3.62 3.52 
 (3.49) (2.98) 
 [45] [685] 
Credits after first semester 12.7 11.2 
 (3.6) (4.5) 
GPA after first semester 3.02 2.49 
 (0.72) (1.06) 
  [6,041] 
N 939 6,112 

Notes: Data provided by the Ohio Board of Regents. Sample is conditioned on students having stated an intention to transfer and taken the ACT and includes first-time students beginning in the fall of 2002, 2003, or 2004 at Cincinnati State Technical and Community College (CC), Columbus State CC, Edison CC, Lakeland CC, Lorain County CC, Sinclair CC, Southern State CC, Terra CC, Washington State CC. Category “Did Not Transfer to a Four-Year College” includes students who completed the transfer module but did not transfer. “Dependent” refers to students who are financially dependent on their parents. “Pell grant eligible” indicates that students have low-enough income levels (parental income for dependent students) to qualify for the federal Pell grant. “EFC” refers to the estimated family contribution calculated by the Free Application for Federal Student Aid and is a measure of family wealth based on both income and assets. For students missing financial aid files and thus “Pell grant eligible,” “Average AGI,” and “EFC” variables, we used regression imputation to recover the missing data. The number in brackets indicates the sample size, if different from the overall. AGI: adjusted gross income; GPA: grade point average; HS: high school.

The College Information panel of table 1 provides descriptive information on students’ postsecondary behaviors. Students who complete the TM are less likely than those who do not to enroll part-time in their first semester, and slightly less likely to have ever taken a remedial course (62 percent versus 67 percent). Additionally, only 5 percent of students completing the TM attended more than one community college compared with 11 percent for noncompleters. This makes sense given that the TM consists of a list of courses to be completed at a student's primary institution, which might not have clear parallels at other institutions. Additionally, more TM completers declare in the first semester a desire to major in arts and humanities (39 percent versus 32 percent) and the social sciences (12 percent versus 5 percent) than non-TM completers.

Outcomes

We examine seven outcomes in this analysis. The first is whether the TM has any impact on students transferring from a two-year to a four-year college. Given that a primary goal of the TM is to help students successfully transition between these two types of institutions, we code students as “transferred” when they attend any four-year campus for two consecutive semesters. The term at which they transferred is coded as the first of these two consecutive terms. This excludes those students who attend a four-year institution for a semester and ultimately stop-out or return to a two-year college. The other outcome we examine for all students is receipt of an associate's degree within six years. Because the TM closely aligns with a little over half of the general education credits typically required for an associate's degree, we hypothesize that students completing the TM may complete associate's degrees at higher rates than their observationally similar peers who did not complete the TM.

We then examine, for those students who transferred, two additional outcomes. We first look at the number of credits they had accumulated in the term prior to transfer to determine if students completing the TM were staying longer at the community college than their peers who transferred to a four-year college without completing the TM. Next, we examine the bachelor's degree completion rate, after six years, for students who transferred. Finally, for students who complete BAs, we also examine the number of terms they spent at a four-year institution, along with the number of terms and the cumulative number of credits earned at the time of degree completion. If TM completers can more efficiently transfer their general education credits across institutions, these students may report graduating with bachelor's degrees in less time and with fewer excess credits. For students who do not complete bachelor's degrees within six years, both the number of terms and number of credits are recorded as the last observable record in the data.

Table 2 provides descriptive statistics for these seven outcomes over a six-year tracking period. Descriptively, students completing the TM are much more likely to transfer to a four-year college (57 percent versus 26 percent), and to complete an associate's degree (81 percent versus 16 percent). Students completing the TM who transfer earn more credits at the community college before transferring (92 versus 54 credits for non-TM completers) than their peers not completing the TM. Students who earn a TM also spend more time enrolled in college than those who do not (15.3 terms versus 14.7 terms). Unfortunately, the dataset does not allow us to observe the number of credits that students are able to officially transfer when they transition from a two-year to a four-year college. However, the number of credits students have when they finish their bachelor's degree gives us some indirect evidence related to this question. Bachelor's degrees in the Ohio public university system typically require just over 120 credits for completion. However, many transfer students complete their BAs with fewer than the required number of credits, suggesting these students were able to transfer in credits from the two-year college where they began. Students who complete the TM and go on to earn a BA acquire fewer credits at the four-year institution by the time they graduate than those who did not complete the TM (118 versus 135 credits). We continue to explore the question of efficiency in the Methods section.

Table 2.
College Outcomes for First-Time Freshmen Beginning at an Ohio Community College in the Fall of 2002, 2003, or 2004 with an Intention to Transfer
Completed Transfer Module (1)Did Not Complete Transfer Module (2)
College Outcomes   
Transferred to a 4-year institution 0.57 0.26 
Completed associate's degree 0.81 0.16 
Credits in the term before transfer 91.86 53.97 
 (35.991) (33.955) 
 [539] [1,563] 
Completed BA degree 0.55 0.43 
 [539] [1,563] 
Number of terms at a 4-year college 7.21 9.30 
 (2.243) (2.945) 
 [298] [678] 
Number of terms to BA degree 14.67 14.36 
 (2.872) (3.033) 
 [298] [678] 
Number of credits at BA degree 117.91 134.57 
 (53.357) (44.011) 
 [298] [678] 
N 939 6,112 
Completed Transfer Module (1)Did Not Complete Transfer Module (2)
College Outcomes   
Transferred to a 4-year institution 0.57 0.26 
Completed associate's degree 0.81 0.16 
Credits in the term before transfer 91.86 53.97 
 (35.991) (33.955) 
 [539] [1,563] 
Completed BA degree 0.55 0.43 
 [539] [1,563] 
Number of terms at a 4-year college 7.21 9.30 
 (2.243) (2.945) 
 [298] [678] 
Number of terms to BA degree 14.67 14.36 
 (2.872) (3.033) 
 [298] [678] 
Number of credits at BA degree 117.91 134.57 
 (53.357) (44.011) 
 [298] [678] 
N 939 6,112 

Note: See notes for table 1.

Methods

We cannot estimate the effect of completing the TM on student outcomes by simply comparing the outcomes of students who complete the TM to those who do not. Students who complete the TM may be different from those who do not in ways that affect both their likelihood of completing the module and their educational outcomes. Several prior studies have attempted to address the selection problem inherent in comparing two-year students with four-year students through the use of quasi-experimental methods (Rouse 1998; Alfonso 2006; Sandy, Gonzalez, and Hilmer 2006; Doyle 2009; Long and Kurlaender 2009; Melguizo, Kienzl, and Alfonso 2011; Reynolds 2012; Shapiro et al. 2013), yet few studies have explored specifically what factors contribute to the community college students’ likelihood of transferring to a four-year college (Dougherty and Kienzl 2006; Laanan, Starobin, and Eggleston 2011; National Center for Public Policy and Higher Education 2011).

In order to address issues of selection, we utilize inverse propensity score weighting (IPSW) to analyze the effects of completing the full TM on the seven outcomes of interest. The goal of IPSW is to weight observations by the inverse of their probability of treatment (in this case, completion of the TM) and then compare students who completed the TM with those who did not. We first estimate the probability of treatment by calculating a propensity score for each student in the data, using the available covariates. We estimate propensity scores using the following logistic regression model:
P(Yijk=1)=11+e-(β0+β1Xijk+γk+θj+εijk),
(1)
where P(Y) is the probability of a student i, in college j, in cohort k, completing the TM given a set of observed characteristics (Y) that we theorize determines whether or not a student decides to complete the courses in the module. These variables include demographic characteristics (ethnicity, gender, age at entry) socioeconomic variables (AGI, high school academic information, EFC from the FAFSA), and variables that indicate students’ goals at the time of first enrollment, including the major declared on their enrollment file. We assume that some of these covariates directly influence selection into treatment, such as a student's academic intentions, academic background, major choice, whereas others, such as ethnicity and age at entry, indirectly influence selection. The logistic regression model also includes cohort fixed effects, denoted by γk, and college fixed effects, denoted by θj, so that we are estimating propensity scores for students within a given cohort, within a given college. Figure 1 graphically represents the overlap in estimated propensity scores for TM completers and noncompleters.7
Figure 1.

Overlap in Estimated Propensity Scores for Treatment and Comparison Groups

Notes: Data provided by the Ohio Board of Regents. Propensity scores were estimates using a logistic regression model that regressed the treatment indicator (coded 1 for students completing the TM and zero otherwise) on a set of covariates including demographic characteristics (gender, race, age at entry), socioeconomic variables (Pell grant eligibility, AGI, EFC), background characteristics (ACT scores, high school GPA, marital status, mother's and father's education), and observed behaviors at the time of first enrollment (enrollment status, hours working, declared major), as well as institutional and cohort-year fixed effects.

Figure 1.

Overlap in Estimated Propensity Scores for Treatment and Comparison Groups

Notes: Data provided by the Ohio Board of Regents. Propensity scores were estimates using a logistic regression model that regressed the treatment indicator (coded 1 for students completing the TM and zero otherwise) on a set of covariates including demographic characteristics (gender, race, age at entry), socioeconomic variables (Pell grant eligibility, AGI, EFC), background characteristics (ACT scores, high school GPA, marital status, mother's and father's education), and observed behaviors at the time of first enrollment (enrollment status, hours working, declared major), as well as institutional and cohort-year fixed effects.

Then, using the inverse of the propensity scores as weights, we run the following regression:
yijk=β0+β1(HasTM)ijk+β2(X)ijk+γk+θj+εijk,
(2)
in which yijk are our seven outcomes of interest (transfer, completion of an associate's/ bachelor's degree, the number of credits pre-transfer, the number of terms spent at a four-year institution, and the number of total terms and credits upon BA completion) and the covariates and fixed effects are the same as in equation 1. Using IPSW allows some control units to count more than others when conducting the analysis. Control units with high propensities are weighted more than those with lower propensities.

Table 3 compares students who did and did not complete the TM in the unweighted and weighted samples. This table also includes p-values for a t-test of the hypothesis that the difference in covariate values between treatment and comparison groups is not equal to zero. The weighting improves balance on all of the covariates we theorize to be most relevant to predicting our outcome variables. In the weighted sample, students have very similar AGIs ($48,123 in the treatment group and $48,012 in the comparison group), are of equivalent age (18.6 years old), and have similar levels of parental education. In the weighted sample, the average ACT composite score in the treatment group is 20.3, compared with 20.4 in the comparison group. After weighting, similar percentages of students in the treatment and comparison groups declare an intention to major in similar fields. In addition, weighted treatment and comparison groups both have approximately 22 percent of students enrolling part-time their first semester and approximately half of all students enrolling in a remedial course in their first semester. The p-values suggest that, when weighting the sample using the inverse of the propensity scores, the difference in covariate values between the treatment and comparison groups are not statistically significantly different from zero. This also indicates that the bias existing in the unweighted treatment and comparison groups has been reduced.8

Table 3.
Pre- and Post-Weighting Sample Characteristics
No WeightingWeighted Sample
Completed Transfer Module (1)Did Not Complete Transfer Module (2)p-value (3)Completed Transfer Module (4)Did Not Complete Transfer Module (5)p-value (6)
Demographics       
Female 0.55 0.51 0.03 0.55 0.55 0.99 
White 0.83 0.81 0.30 0.83 0.83 0.81 
Black 0.06 0.10 <0.01 0.06 0.06 0.97 
Latino 0.03 0.02 0.20 0.03 0.02 0.55 
Asian 0.02 0.01 0.02 0.02 0.02 0.85 
Dependent 0.96 0.93 <0.01 0.96 0.96 0.91 
Age at first enrollment 18.6 18.9 <0.01 18.6 18.6 0.99 
Age <24 years 0.99 0.98 <0.01 0.99 0.99 0.99 
Pell grant eligible 0.37 0.37 0.62 0.37 0.36 0.87 
Married 0.01 0.01 0.09 0.01 0.01 0.93 
Instate 0.99 0.99 0.45 0.99 0.99 0.94 
Average AGI 48,123 46,316 0.02 48,123 48,012 0.89 
Worked first semester 0.78 0.86 <0.01 0.78 0.78 0.87 
EFC 8,908 8,189 <0.01 8,908 8,945 0.90 
Mother graduated from HS 0.54 0.57 0.02 0.54 0.54 0.95 
Father graduated from HS 0.54 0.56 0.08 0.54 0.53 0.90 
Mother has at least some college 0.40 0.38 0.21 0.40 0.40 0.79 
Father has at least some college 0.39 0.37 0.08 0.39 0.39 0.75 
High school variables       
ACT composite score 20.3 19.1 <0.01 20.3 20.4 0.64 
ACT math score 20.0 18.8 <0.01 20.0 20.1 0.56 
ACT English score 19.4 18.0 <0.01 19.4 19.5 0.75 
ACT science score 20.6 19.6 <0.01 20.6 20.7 0.78 
ACT reading score 20.8 19.4 <0.01 20.7 20.8 0.72 
HS GPA 5.61 5.24 <0.01 5.61 5.62 0.73 
College information       
Enrolled part-time first semester 0.22 0.40 <0.01 0.22 0.22 0.95 
Took a remedial course first semester 0.52 0.57 <0.01 0.52 0.51 0.61 
Math major 0.04 0.04 0.84 0.04 0.04 0.94 
Science major 0.04 0.02 0.01 0.04 0.04 0.81 
Engineering major 0.05 0.08 <0.01 0.05 0.05 0.97 
Social Science major 0.12 0.05 <0.01 0.12 0.12 0.94 
Arts & Humanities major 0.39 0.32 <0.01 0.39 0.39 0.78 
Health major 0.08 0.15 <0.01 0.08 0.08 0.99 
Education major 0.07 0.05 0.04 0.07 0.07 0.86 
Business major 0.16 0.17 0.89 0.16 0.16 0.95 
Other professional major 0.01 0.04 <0.01 0.01 0.01 0.99 
No major 0.07 0.10 <0.01 0.07 0.07 0.99 
N 939 6,112  939 6,112  
No WeightingWeighted Sample
Completed Transfer Module (1)Did Not Complete Transfer Module (2)p-value (3)Completed Transfer Module (4)Did Not Complete Transfer Module (5)p-value (6)
Demographics       
Female 0.55 0.51 0.03 0.55 0.55 0.99 
White 0.83 0.81 0.30 0.83 0.83 0.81 
Black 0.06 0.10 <0.01 0.06 0.06 0.97 
Latino 0.03 0.02 0.20 0.03 0.02 0.55 
Asian 0.02 0.01 0.02 0.02 0.02 0.85 
Dependent 0.96 0.93 <0.01 0.96 0.96 0.91 
Age at first enrollment 18.6 18.9 <0.01 18.6 18.6 0.99 
Age <24 years 0.99 0.98 <0.01 0.99 0.99 0.99 
Pell grant eligible 0.37 0.37 0.62 0.37 0.36 0.87 
Married 0.01 0.01 0.09 0.01 0.01 0.93 
Instate 0.99 0.99 0.45 0.99 0.99 0.94 
Average AGI 48,123 46,316 0.02 48,123 48,012 0.89 
Worked first semester 0.78 0.86 <0.01 0.78 0.78 0.87 
EFC 8,908 8,189 <0.01 8,908 8,945 0.90 
Mother graduated from HS 0.54 0.57 0.02 0.54 0.54 0.95 
Father graduated from HS 0.54 0.56 0.08 0.54 0.53 0.90 
Mother has at least some college 0.40 0.38 0.21 0.40 0.40 0.79 
Father has at least some college 0.39 0.37 0.08 0.39 0.39 0.75 
High school variables       
ACT composite score 20.3 19.1 <0.01 20.3 20.4 0.64 
ACT math score 20.0 18.8 <0.01 20.0 20.1 0.56 
ACT English score 19.4 18.0 <0.01 19.4 19.5 0.75 
ACT science score 20.6 19.6 <0.01 20.6 20.7 0.78 
ACT reading score 20.8 19.4 <0.01 20.7 20.8 0.72 
HS GPA 5.61 5.24 <0.01 5.61 5.62 0.73 
College information       
Enrolled part-time first semester 0.22 0.40 <0.01 0.22 0.22 0.95 
Took a remedial course first semester 0.52 0.57 <0.01 0.52 0.51 0.61 
Math major 0.04 0.04 0.84 0.04 0.04 0.94 
Science major 0.04 0.02 0.01 0.04 0.04 0.81 
Engineering major 0.05 0.08 <0.01 0.05 0.05 0.97 
Social Science major 0.12 0.05 <0.01 0.12 0.12 0.94 
Arts & Humanities major 0.39 0.32 <0.01 0.39 0.39 0.78 
Health major 0.08 0.15 <0.01 0.08 0.08 0.99 
Education major 0.07 0.05 0.04 0.07 0.07 0.86 
Business major 0.16 0.17 0.89 0.16 0.16 0.95 
Other professional major 0.01 0.04 <0.01 0.01 0.01 0.99 
No major 0.07 0.10 <0.01 0.07 0.07 0.99 
N 939 6,112  939 6,112  

Note: See notes for table 1.

Though weighting by the inverse of the propensity score does some work to reduce the bias in our estimates due to endogenous selection into the “treatment” of completing the TM, our estimates are still biased by unobserved and unobservable variables. For example, our dataset might not allow us to sufficiently control for students’ academic abilities before entering college because our only measure of high school grades are the self-reported GPAs recorded by the ACT. In addition, our estimates may be biased by student characteristics such as levels of determination and motivation, which are not observable. These two sources of bias are discussed further in section 5. Because we are not able to fully account for selection into treatment, we interpret our results, not as causal but rather as suggestive evidence of the impact of completing the TM on student success.9

5.  Results

Table 4 presents the estimated effect of completing the TM on our seven outcomes of interest. In the odd numbered columns we report estimates from our ordinary least squares model and, in the even columns, the IPSW estimates. Models estimating the effect of completing the TM on the first two outcomes (transferred to a four-year college and completed an associate's degree) include all students. Estimates of the number of credits completed before transfer and BA completion are conditional on students transferring to a four-year college, and estimates for the last three outcomes (number of terms spent at a four-year institution, number of terms to BA and the number of credits completed at BA) are conditional on students having completed a bachelor's degree. We believe that the IPSW model does a better job of controlling for selection into the TM and, thus, is our preferred specification.

Table 4.
Treatment Effects for Students Completing the Transfer Module (TM) Compared with Those Not Completing the TM
Transferred to a 4-year CollegeCompleted Associate DegreeNumber of Credits in Term Before TransferCompleted BANumber of Terms at 4-yearNumber of Terms to BA DegreeNumber of Credits at BA Degree
OLS (1)IPSW (2)OLS (3)IPSW (4)OLS (5)IPSW (6)OLS (7)IPSW (8)OLS (9)IPSW (10)OLS (11)IPSW (12)OLS (13)IPSW (14)
Completed TM 0.224 0.207 0.647 0.648 37.633 35.934 0.077 0.047 −1.859 −1.876 0.500 0.769 −14.679 −15.448 
 (0.026)*** (0.020)*** (0.035)*** (0.013)*** (6.642)*** (2.177)*** (0.017)*** (0.026)* (0.371)*** (0.198)*** (0.188)** (0.206)*** (7.459)* (3.780)*** 
N 7,051 7,051 7,051 7,051 2,102 2,102 2,102 2,102 979 979 979 979 979 979 
R2 0.247  0.326  0.32  0.14  0.286  0.426  0.272  
Transferred to a 4-year CollegeCompleted Associate DegreeNumber of Credits in Term Before TransferCompleted BANumber of Terms at 4-yearNumber of Terms to BA DegreeNumber of Credits at BA Degree
OLS (1)IPSW (2)OLS (3)IPSW (4)OLS (5)IPSW (6)OLS (7)IPSW (8)OLS (9)IPSW (10)OLS (11)IPSW (12)OLS (13)IPSW (14)
Completed TM 0.224 0.207 0.647 0.648 37.633 35.934 0.077 0.047 −1.859 −1.876 0.500 0.769 −14.679 −15.448 
 (0.026)*** (0.020)*** (0.035)*** (0.013)*** (6.642)*** (2.177)*** (0.017)*** (0.026)* (0.371)*** (0.198)*** (0.188)** (0.206)*** (7.459)* (3.780)*** 
N 7,051 7,051 7,051 7,051 2,102 2,102 2,102 2,102 979 979 979 979 979 979 
R2 0.247  0.326  0.32  0.14  0.286  0.426  0.272  

Notes: Data provided by the Ohio Board of Regents. Columns 5—8 conditional on students having transferred. Columns 9—14 conditional on students having transferred and completed a BA. Covariates include demographic characteristics (gender, race, age at entry), socioeconomic variables (Pell grant eligibility, Adjusted Gross Income, Estimated Family Contribution), background characteristics (ACT scores, high school grade point average, marital status, mother's and father's education), and observed behaviors at the time of first enrollment (enrolment status, hours working, declared major), as well as institutional and cohort-year fixed effects. OLS: ordinary least squares; IPSW: inverse propensity score weighting.

*p < 0.1; **p < 0.05; ***p < 0.01.

The estimate displayed in column 2 of table 4 suggests that students who completed the TM are approximately 21 percentage points more likely to transfer to a four-year institution, on average, than those who do not complete the TM. They are also approximately 65 percentage points more likely to attain their associate's degrees than students who don't complete the TM, on average. This suggests that, at least for some students, completing the TM may just be a step along the way to completing an associate's degree before transfer.

The estimate in column 6 suggests students who transfer and also complete the TM earn approximately thirty-six more credits before transitioning to a four-year institution, on average, than students who also transfer but do not complete the TM. In other words, transfer students who don't complete the TM are more likely to move on to a four-year institution sooner. The same students who transferred to a four-year college and completed the TM are 4.7 percentage points more likely to earn their BAs than transfer students who did not complete the TM, but this difference is not statistically significant.

In addition to the effect on transfer and degree attainment, it could also be the case that students who complete the TM transfer with more credits, and/or earn their degree in less time than students who don't complete the TM (Roksa and Keith 2008). We find mixed results related to whether completing the TM improves the efficiency of transfer. On the one hand, when we condition on having transferred and completed a BA and compare students who did and did not complete the TM, we find that TM completers spend almost two fewer terms at the four-year institution, on average, but close to a full additional term enrolled in any institution (0.769 additional terms). On the other hand, students who earn the TM earn approximately fifteen fewer credits at the four-year institution than those without the TM. This is not surprising, given that TM completers appear to stay longer at the two-year colleges prior to transferring (column 6).

Unfortunately, our dataset does not allow us to directly observe the number of credits that students are able to transfer from their starting institution to their receiving four-year institution. Therefore, we make use of the number of credits students have accumulated by the time of BA completion to infer how completing the TM affects students’ ability to transfer credit from the two-year to the four-year institution. Columns 13 and 14 of table 4 show the estimates of the effect of completing the TM on the number of credits accumulated at the time of BA completion, defined as the number of credits accumulated at the four-year institution by the semester of BA receipt. For many transfer students, this number is less than the total number of credits required for BA completion at their degree-granting institution,10 suggesting that these students are able to transfer some of the credit they earned at the community college before transferring to the four-year institution. Moreover, the estimate shows that students who complete the TM have accumulated fewer credits at the four-year institution, on average, by the time they earn their BAs, and this estimate is statistically significant. This is what we would expect to see if students who complete the TM before transferring are able to transfer more of the credit they earned at the community college before transfer compared to non-TM completers.

However, this estimate may be confounded by two parallel issues: first, students who complete the TM spend fewer semesters, on average, than students who do not complete the TM at the four-year, BA-granting institution. Second, students who do not complete the TM accumulate less credit, on average, at the two-year colleges where they start, before transferring. In order to overcome the former, we reestimate the effect of completing the TM on the number of credits accumulated at the time of BA, limiting the sample to TM completers and non-completers who spent the same number of terms at the four-year institution. These estimates are displayed in columns 1–9 of table 5. In general, the signs on these coefficients support the story told by the estimated effect of completing the TM on the number of credits at BA degree (column 14) in table 4: Students who complete the TM need to complete fewer credits at the four-year institution in order to earn their BAs than non-TM completers, suggesting that they are able to transfer more credit hours from their initial two-year institutions. However, perhaps because of the small sample sizes created by comparing treatment and comparison-group students who have completed the same number of terms at the four-year institution, these estimates are not statistically significant.

Table 5.
Robustness Checks: The Effect of Completing the Transfer Module (TM) on the Number of Credits at BA Completion for Students Who Spent the Same Number of Terms at the Four-Year Institution and the Effect of Completing the TM on a Lower Bound Estimate of Transfer Credit
4 Terms (1)5 Terms (2)6 Terms (3)7 Terms (4)8 Terms (5)9 Terms (6)10 Terms (7)11 Terms (8)12 Terms (9)Outcome: Minimum Transfer Credit (10)
Completed TM −4.273 47.604** 6.231 −6.043 −9.274 −10.593 −18.295 −12.395 −38.276* 0.583*** 
 (11.188) (17.805) (8.710) (6.986) (8.374) (7.081) (11.156) (20.454) (20.986) (0.142) 
Observations 47 63 117 138 118 126 93 92 65 979 
4 Terms (1)5 Terms (2)6 Terms (3)7 Terms (4)8 Terms (5)9 Terms (6)10 Terms (7)11 Terms (8)12 Terms (9)Outcome: Minimum Transfer Credit (10)
Completed TM −4.273 47.604** 6.231 −6.043 −9.274 −10.593 −18.295 −12.395 −38.276* 0.583*** 
 (11.188) (17.805) (8.710) (6.986) (8.374) (7.081) (11.156) (20.454) (20.986) (0.142) 
Observations 47 63 117 138 118 126 93 92 65 979 

Notes: Estimates are conditional on students having transferred and completed a BA. “Minimum transfer credit” is defined as the difference between the number of credits accumulated at the time of BA receipt and the number of credits required for the BA over the number of credits accumulated at the two-year institution in the term prior to transfer. Covariates include demographic characteristics (gender, race, age at entry), socioeconomic variables (Pell grant eligibility, Adjusted Gross Income, Estimated Family Contribution), background characteristics (ACT scores, high school grade point average, marital status, mother's and father's education), and observed behaviors at the time of first enrollment (enrollment status, hours working, declared major), as well as institutional and cohort-year fixed effects.

*p < 0.1; **p < 0.05; ***p < 0.01.

In a further attempt to overcome the fact that students who do not complete the TM complete fewer credits at their starting two-year institution than TM completers, we estimate the effect of completing the TM on the proportion of credits students were able to transfer. As stated above, in the data we observe many transfer students who earn a BA graduating with fewer credits than the degree-granting institution actually requires for a BA. Thus, for students with these “missing” credits, we subtract actual credits accumulated from total credits required for the BA and divide by the number of credits accumulated before transfer. For students who accumulate more than the number of credits required for their BA, this number is replaced with zero. Given that many college students tend to acquire more credits than they need to graduate (Monaghan and Attewell 2014), the numerator of this proportion likely underestimates the number of credits students were able to transfer.

For example, some BA completers who have more than the required number of credits at graduation may have been able to transfer some of the credit from their initial two-year institution, but they accumulated excess credits by the time of graduation and, thus, receive a zero in our calculation. Column 10 of table 5 displays the estimated effect of completing the TM on the proportion of credits students are able to transfer from their starting institution. This estimate suggests that students who complete the TM are able to transfer almost 60 percent more of their credits, on average, than students who do not complete the TM, and this estimate is statistically significant. This estimate supports the story told by the effect of completing the TM on the number of credits accumulated by the time students earn their BAs; it appears that TM completers are able to transfer more credit hours from the community college than students who do not complete the TM. However, because students who graduate with the correct number of credits or with excess credit have zeros for this outcome, they may still have been able to transfer in some credit that we cannot observe. Therefore, this coefficient provides the lower bound for the effect of completing the TM on the proportion of credits transferred from the two-year institution.

Finally, we may be concerned that students who complete the TM before earning their BAs have fewer credits at graduation than those who do not complete the TM because they are less likely to choose majors such as science and engineering, which may require more courses. When we compare major at graduation for BA-completers who did and did not complete the TM, we find that non-TM completers are slightly more likely to be science majors (6 percent of the sample compared with 5 percent of TM-completers) or engineering majors (8 percent of the sample compared with 6 percent of TM-completers). However, when we use t-tests to compare these differences, we find they are not statistically significantly different from zero. We therefore conclude that the differences we find in credit accumulation at graduation are not simply masking differences in choice of major. It could also be the case that students who do not complete the TM are more likely to choose to double-major than TM-completers—although we are not able to observe this in our data.

Sensitivity Analysis

The use of propensity scores to control for selection into treatment relies on the conditional independence assumption, which states that treatment assignment should be determined only by factors that the researcher can observe. Ichino, Mealli, and Nannicini (2008) suggest a sensitivity analysis that tests this assumption by reestimating the propensity scores including some simulated, unobserved covariate that may affect both assignment to treatment and the outcomes of interest. The only potentially “killer” confounder would be one that had both an effect on the treatment assignment and an effect on the differences we observe in the outcome (Ichino, Mealli, and Nannicini 2008). If a confounder affects assignment to treatment but not the outcome, then it cannot be driving any treatment effects we observe. Likewise, if a potential confounder has an effect on the outcome, but is present in equal quantities in both treatment and comparison groups, then it would not have an impact on treatment effects. The authors propose using the sample distribution of key covariates to simulate these unobserved covariates.

One important limitation of our study is that we do not have many measures of students’ precollege academic ability. We hypothesize that students’ academic abilities could affect both their likelihood of completing the TM and their academic outcomes. Therefore, for the simulation, we chose variables that we hypothesize may be correlated with students’ academic ability, including measures of parents’ education and the binary variable indicating whether the student took a remedial course in the first semester. We also hypothesize that students who are working during college may be less likely to complete the TM. These students may also fare worse on outcome measures such as likelihood of transfer and BA completion if working while enrolled causes students to spend less time or effort on their studies. Therefore, in our sensitivity analysis, we also include a simulated indicator variable for students who worked during the first semester.11 The distributions of the variables used to simulate the unobserved confounders are displayed in Appendix table A.4.

Table 6 displays the results of the sensitivity analysis. As stated above, only confounders that have both an outcome effect and a selection effect (indicated by a value of 1 or higher in the first two columns of table 6) pose a potential problem.12 In all cases when a potential confounder has both an outcome and selection effect, we find that our estimates are robust to the inclusion of these confounders, with the exception of models exploring the robustness of our estimates related to the number of terms students are enrolled before they complete their bachelor's degrees. In other words, in most cases, when we reestimate propensity scores including the simulated covariate, the estimated treatment effects are very close to our main estimates in table 4. However, a potential confounder distributed like the covariate “placed into remediation” has both an outcome and selection effect in a model estimating the effect of having completed the TM on the number of terms it takes students to complete their bachelor's degrees. When we include this potential confounder, the estimated effect is much smaller than in our main models (0.174 additional terms) and no longer statistically significant. Thus, it is possible that there may not be a statistically significant difference in the time it takes to complete a BA for TM completers and noncompleters.

Table 6.
Sensitivity Analysis
Outcome Effect (1)Selection Effect (2)ATT (3)SE (4)
Outcome: Transferred 
Mother has at least some college 1.215 1.044 0.306 0.028 
Father has at least some college 1.465 1.105 0.308 0.027 
Placed into remediation 0.647 0.862 0.306 0.027 
Worked first semester 0.658 0.594 0.297 0.029 
Outcome: Completed Associate Degree 
Mother has at least some college 0.881 1.079 0.651 0.023 
Father has at least some college 0.924 1.117 0.652 0.022 
Placed into remediation 0.646 0.836 0.651 0.022 
Worked first semester 0.705 0.561 0.649 0.022 
Outcome: Number of Credits Before Transfer 
Mother has at least some college 0.948 1.093 37.194 2.978 
Father has at least some college 1.121 1.009 37.278 2.954 
Placed into remediation 1.256 0.97 36.961 2.966 
Worked first semester 1.015 0.843 37.043 2.976 
Outcome: Completed BA Degree 
Mother has at least some college 1.137 1.106 0.109 0.04 
Father has at least some college 1.524 0.98 0.111 0.041 
Placed into remediation 0.587 1.017 0.112 0.042 
Worked first semester 0.896 0.85 0.11 0.041 
Outcome: Number of Terms Spent at a Four-Year Institution 
Mother has at least some college 0.913 0.913 −2.1 0.317 
Father has at least some college 1.557 0.791 −2.117 0.316 
Placed into remediation 1.653 1.205 −2.112 0.315 
Worked first semester 0.894 0.863 −2.104 0.32 
Outcome: Number of Terms to BA Degree 
Mother has at least some college 0.918 0.909 0.199 0.339 
Father has at least some college 1.191 0.795 0.214 0.331 
Placed into remediation 2.261 1.258 0.174 0.337 
Worked first semester 0.798 0.86 0.201 0.341 
Outcome: Number of Credits Earned at BA Degree 
Mother has at least some college 1.103 0.948 −16.329 5.267 
Father has at least some college 1.377 0.826 −16.292 5.183 
Placed into remediation 1.900 1.152 −17.051 5.203 
Worked first semester 0.799 0.857 −16.402 5.530 
Outcome Effect (1)Selection Effect (2)ATT (3)SE (4)
Outcome: Transferred 
Mother has at least some college 1.215 1.044 0.306 0.028 
Father has at least some college 1.465 1.105 0.308 0.027 
Placed into remediation 0.647 0.862 0.306 0.027 
Worked first semester 0.658 0.594 0.297 0.029 
Outcome: Completed Associate Degree 
Mother has at least some college 0.881 1.079 0.651 0.023 
Father has at least some college 0.924 1.117 0.652 0.022 
Placed into remediation 0.646 0.836 0.651 0.022 
Worked first semester 0.705 0.561 0.649 0.022 
Outcome: Number of Credits Before Transfer 
Mother has at least some college 0.948 1.093 37.194 2.978 
Father has at least some college 1.121 1.009 37.278 2.954 
Placed into remediation 1.256 0.97 36.961 2.966 
Worked first semester 1.015 0.843 37.043 2.976 
Outcome: Completed BA Degree 
Mother has at least some college 1.137 1.106 0.109 0.04 
Father has at least some college 1.524 0.98 0.111 0.041 
Placed into remediation 0.587 1.017 0.112 0.042 
Worked first semester 0.896 0.85 0.11 0.041 
Outcome: Number of Terms Spent at a Four-Year Institution 
Mother has at least some college 0.913 0.913 −2.1 0.317 
Father has at least some college 1.557 0.791 −2.117 0.316 
Placed into remediation 1.653 1.205 −2.112 0.315 
Worked first semester 0.894 0.863 −2.104 0.32 
Outcome: Number of Terms to BA Degree 
Mother has at least some college 0.918 0.909 0.199 0.339 
Father has at least some college 1.191 0.795 0.214 0.331 
Placed into remediation 2.261 1.258 0.174 0.337 
Worked first semester 0.798 0.86 0.201 0.341 
Outcome: Number of Credits Earned at BA Degree 
Mother has at least some college 1.103 0.948 −16.329 5.267 
Father has at least some college 1.377 0.826 −16.292 5.183 
Placed into remediation 1.900 1.152 −17.051 5.203 
Worked first semester 0.799 0.857 −16.402 5.530 

Notes: The estimates displayed are from running models identical to those displayed in table 4, but also including an additional, unobserved covariate distributed similarly to the covariates listed on the left. ATT: average treatment effect on the treated; SE: standard error.

Limitations

Although the sensitivity analysis suggests that the majority of our estimates are robust to the inclusion of potential confounders, this study still has limitations. We are primarily concerned that we are not able to sufficiently control for students’ incoming academic ability. It may be the case that students who complete the TM are more academically prepared than their peers who do not; if this is the case, we will have overestimated the effects of completing the TM on transfer, degree completion, and transfer efficiency. In fact, descriptively we observe that students who completed the TM had higher high school GPAs and ACT scores than those who did not, which is not surprising given the rigorous requirements for completing the full TM. Although we attempt to control for this potential source of bias by matching on ACT composite, math, English, science, and reading scores, we recognize that ACT scores is only one measure of academic ability.13 Future research should test the robustness of our results in a dataset that allows for a more complete composite of academic ability; for example, by including multiple measures of students’ abilities from high school administrative data.

Along with this observed measure of academic ability, we are also limited in this study by two main unobserved covariates: motivation and college knowledge. If students who complete the TM are simply more motivated to be successful in college than those who do not, we will be overestimating the treatment effects. Some of the success we are attributing to the TM policy would, instead, be due to unobserved levels of motivation among TM completers. Additionally, students who complete the TM may be more institutionally savvy when it comes to understanding the rules and requirements for transfer and degree completion. These students may be better able to navigate administrative requirements, which enables them to not only complete a rather complicated TM, but to have greater success in navigating their way through multiple institutions more generally. For example, we do not observe whether more academically prepared students seek out advising that both increases their likelihood to pursue the TM and has a positive effect on academic outcomes, such as successfully transferring from a two-year to a four-year institution. It may also be the case that some campuses provide more advising support to some students than others. Although we are not able to control for unobserved confounders, such as motivation or college knowledge that may result in some students completing the TM over others, we are able to control for time-invariant differences across campuses through institutional fixed effects.

6.  Discussion and Conclusions

In this paper we explore the effectiveness of a statewide articulation agreement to bring new evidence to bear on a type of policy that has been adopted by several state higher education systems across the United States, but about which to date there is little research. Although we do not have exogenous variation to exploit for causal evidence, the results from this paper contribute to our understanding of how these policies may promote transfer or the efficiency of transfer. The findings from our study comparing the transfer and post-transfer outcomes of students who completed the Ohio TM to similar students who did not complete it suggest that fulfilling these requirements is associated with higher rates of transfer and associate's degree completion. We also find suggestive evidence that completing the TM is associated with greater portability of credits. However, for those who go on to earn a BA, the TM may increase the time it takes students to complete their degrees.

Although our estimates suggest that there is an effect of completing the TM on subsequent transfer and associate's degree completion, we caution against interpreting these results as causal, as we are not able to fully control for all issues of selection, namely, academic motivation, among those who complete the TM. In addition, our sample is limited to students who have taken the ACT and who indicate, when they first enter one of the sample community colleges, that they intend to eventually transfer to a four-year institution. Because of these sample limitations, our findings cannot be generalized to all students who enter community colleges and attempt to complete a transfer degree or a core of classes that an articulation agreement has designated as transferable.

Insofar as the goal of the Ohio TM is to mandate that four-year institutions accept the credits students earn at community colleges, we find suggestive evidence that students who complete the TM and go on to earn a BA are able to bring more credits with them when they transfer than those who do not. While the dataset does not include a variable that expressly defines the number of transfer credits, we are able to compare the number of credits upon BA completion and the length of time students spend at both two-year and four-year colleges. These comparisons suggest that TM completers transfer more credits in the transition between institutions than non-TM completers. However, on average, students who complete the TM also appear to take approximately a term longer to complete their BAs. Thus, while the TM may improve the probability that students will transfer and complete associate's degrees, it is not as clear if it leads students to be more efficient in the transfer process.

Given the popularity of articulation policies in today's policy environment, this study would support the existence of these programs as a tool by which to encourage students to transfer to a four-year college and/or complete an associate's degree. For states focused on increasing the educational attainment of their adult population, programs requiring even some portion of the general education credits needed for an associate's degree appear to help students to complete these degrees at higher rates. It should be noted that our analysis focused exclusively on students who expressed an intention to transfer—these students may be differentially motivated than students whose goal is to earn an associate's degree. Regardless of their educational goals, however, research suggests that people are able to handle decisions better if presented with limited categories of choices (Keller et al. 2011). Although the Ohio TM is not one specified set of courses across the entire state (as illustrated by the sample course choices in Appendix table A.2), it does attempt to help students outline a pathway, or even just the general educational requirements component of that pathway, toward earning an associate's degree and/or transferring to a four-year college. For students who complete this pathway, the rates of associate degree completion and transfer are higher than for observationally similar peers who did not.

Importantly, we also observe very few students completing the TM in the three cohorts of our analysis. Across our sample of nine community colleges, an average of 15 percent of students who expressed an intention to transfer completed the TM. So, while the TM does appear to benefit students in the short term, these low numbers suggest that better advising or information should be provided to students on the front end. Though articulation policies that define a transferable core may be the first step in defining pathways from two-year colleges to four-year colleges, students are still left facing a considerable amount of complexity when navigating their way through this set of courses. Articulation agreements alone may be necessary but not sufficient conditions for improving transfer, and particularly for improving post-transfer success. Given the complexity of the transfer process, further research is needed to test this hypothesis.

It is also important to consider whether articulation agreements that require students to complete a general education core may have any unintended negative consequences, such as reducing the number of students who pursue science, technology, engineering, and math (STEM) majors. If the math and science components of the general education core do not align with requirements for these majors, some students who entered community college with the intention of majoring in a STEM field may elect to choose a different major after transfer. We do not find strong evidence for this hypothesis in our data; nevertheless, future research should explore how starting at a community college affects students’ major choices, as well as whether articulation agreements mediate these effects.

Among students who transfer to a four-year college, we do not see effects of completing the TM on BA degree attainment, although we see some suggestive evidence that students who complete the TM are able to transfer more credits from the two-year to the four-year college. The portability of credits across institutions is a critically important issue in today's political environment. As increasing numbers of states consider policies such as “free” community college, akin to the Tennessee Promise, the question of how to help students make the transition from two-year to four-year colleges becomes all the more relevant.

Important benefits to students enrolled in institutions or states with articulation policies may include both the transferability component of the program and the cost savings associated with beginning one's college coursework at a two-year college. The guarantee that courses successfully completed at one institution will transfer to another appeals to students and legislators, but requires considerable coordination on behalf of the public colleges in the state/region. The tradeoff between simplicity across the state and individuality across campuses is an important implementation issue to be considered. While greater simplicity in terms of courses and options may encourage more students to pursue the TM, it may also require more complicated negotiations across institutions with disparate requirements and degree programs.

Additionally, there are tuition savings for students who complete portions of their BA degree requirements at a community college. Although we find that, among BA recipients, students who completed the TM spent nearly one more term in college than non-TM completers (table 4, column 10), we also find that these students spent almost two more terms enrolled at a community college over a four-year institution (table 4, column 12). With an average annual tuition across Ohio's community colleges of $3,171 in 2007–08, and an average in-state tuition across regional public four-year colleges of $5,664 (Ohio State University 2007; Ohio Board of Regents 2012), the tuitions savings from enrolling for two additional terms at a community college over a four-year institution outweighs the cost of potentially spending an additional term to complete a BA degree. Articulation agreements, particularly those that encourage or require a certain number of credits prior to transfer, have the potential to lower tuition costs significantly for degree completers.

This study contributes to a growing body of evidence suggesting that articulation policies that define a transferable core of courses or give junior status to students who complete an associate's degree before moving on to the four-year institution may fulfill their mission to improve the seamlessness and efficiency of transfer (Crook, Chellman, and Holod 2012; Crosta and Kopko 2014). However, policies that more directly address the challenges facing students both before and after transfer may be necessary in order to significantly increase the rates of degree completion post-transfer.

Acknowledgments

We thank Bridget Long and the Ohio Board of Regents for the data used in this study. We also thank Josh Goodman, Jessica Howell, Mark Long, and James Palmer for helpful comments on early drafts of this paper, as well as participants at the 2014 annual meetings of the Association of Education Finance and Policy, the American Educational Research Association, and the 2015 annual meeting of the Association of Public Policy and Management. All errors, omissions, and conclusions are our own.

REFERENCES

Adelman
,
Clifford
.
1999
.
Answers in the toolbox: Academic intensity, attendance patterns, and bachelor's degree attainment
.
Available
www2.ed.gov/pubs/Toolbox/toolbox.html.
Accessed 23 December 2016
.
Adelman
,
Clifford
.
2006
.
The toolbox revisited: Paths to degree completion from high school through college
.
Washington, DC
:
U.S. Department of Education
.
Alfonso
,
Mariana
.
2006
.
The impact of community college attendance on baccalaureate attainment
.
Research in Higher Education
47
(
8
):
873
903
. doi:10.1007/s11162-006-9019-2.
American Association of Community Colleges (AACC)
.
2016
.
2016 fact sheet
.
Washington, DC
:
AACC
.
Anderson
,
Gregory
,
Jeffrey C.
Sun
, and
Mariana
Alfonso
.
2006
.
Effectiveness of statewide articulation agreements on the probability of transfer: A preliminary policy analysis
.
Review of Higher Education
29
(
3
):
261
291
. doi:10.1353/rhe.2006.0001.
Attewell
,
Paul
,
Scott
Heil
, and
Liza
Reisel
.
2012
.
What is academic momentum? And does it matter
?
Educational Evaluation and Policy Analysis
34
(
1
):
27
44
. doi:10.3102/0162373711421958.
Auguste
,
Byron
,
Adam
Cota
,
Kartik
Jayaram
, and
Martha C. A.
Laboissière
.
2010
.
Winning by degrees: The strategies of highly productive higher education institutions
.
New York
:
McKinsey and Company
.
Bailey
,
Thomas
.
2009
.
Challenge and opportunity: Rethinking the role and function of developmental education in community college
.
New Directions for Community Colleges
145
:
11
30
. doi:10.1002/cc.352.
Bers
,
Trudy H.
, and
Kerry E.
Smith
.
1991
.
Persistence of community college students: The influence of student intent and academic and social integration
.
Research in Higher Education
32
(
5
):
539
556
. doi:10.1007/BF00992627.
Cabrera
,
Alberto F.
,
Amaury
Nora
, and
Maria B.
Castañeda
.
1993
.
College persistence: Structural equations modeling test of an integrated model of student retention
.
Journal of Higher Education
64
(
2
):
123
139
.
College Board
.
2015
.
Trends in college pricing
.
New York
:
College Board Trends in Higher Education Series
.
Complete College America (CCA)
.
2011
.
Time is the enemy: The surprising truth about why today's college students aren't graduating … and what needs to change
.
Washington, DC
:
CCA
.
Crook
,
David
,
Colin C.
Chellman
, and
Aleksandra
Holod
.
2012
.
Does earning an associate degree lead to better baccalaureate outcomes for transfer students
?
Working Paper, CUNY Office of Policy Research
.
Crosta
,
Peter M.
, and
Elizabeth
Kopko
.
2014
.
Should community college students earn an associate degree before transferring to a four-year institution
?
Community College Research Center Working Paper No. 70, Teachers College, Columbia University
.
DesJardins
,
Stephen L.
,
Dennis A.
Ahlburg
, and
Brian P.
McCall
.
2006
.
The effects of interrupted enrollment on graduation from college: Racial, income, and ability differences
.
Economics of Education Review
25
(
6
):
575
590
. doi:10.1016/j.econedurev.2005.06.002.
Dougherty
,
Kevin J.
, and
Gregory S.
Kienzl
.
2006
.
It's not enough to get through the open door: Inequalities by social background in transfer from community colleges to four-year colleges
.
Teachers College Record
108
(
3
):
452
487
. doi:10.1111/j.1467-9620.2006.00658.x.
Doyle
,
William R.
2009
.
The effect of community college enrollment on bachelor's degree completion
.
Economics of Education Review
28
(
2
):
199–206
. doi:10.1016/j.econedurev.2008.01.006.
Education Commission of the States
.
2016
.
50-state comparison: Transfer and articulation policies
.
Available
https: //www.ecs.org/transfer-and-articulation-policies-db/.
Accessed 23 December 2016
.
Gandara
,
Patricia
,
Elizabeth
Alvarado
,
Anne
Driscoll
, and
Gary
Orfield
.
2012
.
Building pathways to transfer: Community colleges that break the chain of failure for students of color
.
Los Angeles
:
UCLA Civil Rights Project
.
Goldrick-Rab
,
Sara
.
2010
.
Challenges and opportunities for improving community college student success
.
Review of Educational Research
80
(
3
):
437
469
. doi:10.3102/0034654310370163.
Gross
,
Betheny
, and
Dan
Goldhaber
.
2009
.
Community college transfer and articulation policies: Looking beneath the surface
. Working Paper No. 2009–1,
Center on Reinventing Public Education
,
University of Washington
.
Horn
,
Laura
, and
Paul
Skomsvold
.
2011
.
Community college student outcomes: 1994–2009 (NCES 2012–253)
.
Washington, DC
:
U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics
.
Ichino
,
Andrea
,
Fabrizia
Mealli
, and
Tommaso
Nannicini
.
2008
.
From temporary help jobs to permanent employment: What can we learn from matching estimators and their sensitivity
?
Journal of Applied Econometrics
23
(
3
):
305
327
. doi:10.1002/jae.998.
Ignash
,
Jan M.
, and
Barbara K.
Townsend
.
2001
.
Statewide transfer and articulation policies: Current practices and emerging issues
. In
Community colleges: Policy in the future context
,
edited by
Barbara K.
Townsend
and
Susan B.
Twombly
, pp.
173
192
.
Westport, CT
:
Ablex Publishing
.
Jenkins
,
Davis
, and
John
Fink
.
2015
.
What we know about transfer
.
New York
:
Community College Research Center, Teachers College, Columbia University
.
Keller
,
Punam A.
,
Bari
Harlam
,
George
Loewenstein
, and
Kevin G.
Volpp
.
2011
.
Enhanced active choice: A new method to motivate behavior change
.
Journal of Consumer Psychology
21
(
4
):
376
383
. doi:10.1016/j.jcps.2011.06.003.
Kena
,
Grace
,
Lauren
Musu-Gillette
,
Jennifer
Robinson
,
Xiaolei
Wang
,
Amy
Rathbun
,
Jijun
Zhang
,
Sidney
Wilkinson-Flicker
,
Amy
Barmer
, and
Erin Dunlop
Velez
.
2015
.
The condition of education 2015 (NCES 2015–144)
.
Washington, DC
:
U.S. Department of Education, National Center for Education Statistics
.
Laanan
,
Frankie S.
,
Soko S.
Starobin
, and
Latrice E.
Eggleston
.
2011
.
Adjustment of community college students at a four-year university: Role and relevance of transfer student capital for student retention
.
Journal of College Student Retention
12
(
2
):
175
209
. doi:10.2190/CS.12.2.d.
Long
,
Bridget T.
, and
Michal
Kurlaender
.
2009
.
Do community colleges provide a viable pathway to a baccalaureate degree
?
Educational Evaluation and Policy Analysis
31
(
1
):
30
53
. doi:10.3102/0162373708327756.
Ma
,
Jennifer
, and
Sandy
Baum
.
2016
.
Trends in community colleges: Enrollment, prices, student debt and completion
.
New York
:
College Board Research Brief
.
Melguizo
,
Tatiana
,
Gregory
Kienzl
, and
Mariana
Alfonso
.
2011
.
Comparing the educational attainment of community college transfer students and four-year college rising juniors using propensity score matching methods
.
Journal of Higher Education
82
(
3
):
265
291
. doi:10.1080/00221546.2011.11777202.
Monaghan
,
David B.
, and
Paul
Attewell
.
2014
.
The community college route to the bachelor's degree
.
Educational Evaluation and Policy Analysis
37
(
1
):
70
91
. doi:10.3102/0162373714521865.
National Center for Public Policy and Higher Education
.
2011
.
Affordability and transfer: Critical to increasing baccalaureate degree completion. Policy Alert.
San Jose, CA
:
The National Center for Public Policy and Higher Education
.
Ohio Board of Regents
.
2003
.
3333.16 Universal course equivalency classification system for state institutions of higher education
.
Available
http: //codes.ohio.gov/orc/3333.16.
Accessed 23 December 2016
.
Ohio Board of Regents
.
2012
.
Undergraduate tuition and fees: Annualized full-time in-state undergraduate tuition and fees charged to entering students, FY 2003 to FY 2012
.
Available
https://www.ohiohighered.org/files/uploads/reports/Undergrad_Tuition_Fees_FY2003-12.pdf.
Accessed 23 December 2016
.
Ohio State University
.
2007
.
Statistical summary
.
Available
https://www.osu.edu/osutoday/stuinfo07.php.
Accessed 23 December 2016
.
Module
,
Ohio Transfer
.
1990
.
The Ohio articulation and transfer policy: Appendix A: legislation
.
Available
https://www.ohiohighered.org/transfer/policy/appendices.
Accessed 23 December 2016
.
Perna
,
Laura W.
1998
.
The contribution of financial aid to undergraduate persistence
.
Journal of Student Financial Aid
28
(
3
):
25
40
.
Reynolds
,
C. Lockwood
.
2012
.
Where to attend? Estimating the effects of beginning college at a two-year institution
.
Economics of Education Review
31
(
4
):
345
362
. doi:10.1016/j.econedurev.2011.12.001.
Roksa
,
Josipa
.
2006
.
States, schools and students: Contextualizing community college outcomes
. PhD dissertation,
New York University, New York
.
Roksa
,
Josipa
.
2009
.
Building bridges for student success: Are articulation policies effective
?
Teachers College Record
111
(
10
):
2444
2478
.
Roksa
,
Josipa
, and
Bruce
Keith
.
2008
.
Credits, time, and attainment: Articulation policies and success after transfer
.
Educational Evaluation and Policy Analysis
30
(
3
):
236
254
. doi:10.3102/0162373708321383.
Rosenbaum
,
Paul
, and
Donald B.
Rubin
.
1985
.
Constructing a control group using multivariate matched sampling methods that incorporate the propensity score
.
American Statistician
39
(
1
):
33
38
.
Rouse
,
Cecilia E.
1998
.
Do two-year colleges increase overall educational attainment? Evidence from the states
.
Journal of Policy Analysis and Management
17
(
4
):
595
621
. doi:10.1002/(SICI)1520-6688(199823)17:4<595::AID-PAM1>3.0.CO;2-5.
Sandy
,
Jonathan
,
Arturo
Gonzalez
, and
Michael J.
Hilmer
.
2006
.
Alternative paths to college completion: Effect of attending a 2-year school on the probability of completing a 4-year degree
.
Economics of Education Review
25
(
5
):
463
471
. doi:10.1016/j.econedurev.2005.05.003.
Scott-Clayton
,
Judith
.
2011
.
The shapeless river: Does a lack of structure inhibit students’ progress at community colleges
?
Community College Research Center Working Paper No. 25, Teachers College, Columbia University
.
Sengupta
,
Ria
, and
Christopher C.
Jepsen
.
2006
.
California's community college students
.
California Counts
8
(
2
):
1
22
.
Shapiro
,
Doug
,
Afet
Dundar
,
Phoebe K.
Wakhungu
,
Xin
Yuan
, and
Autumn
Harrell
.
2015
.
Transfer & mobility
:
A national view of student movement in postsecondary institutions
, Fall 2008 cohort (Signature Report No. 9).
Herndon, VA
:
National Student Clearinghouse Research Center
.
Shapiro
,
Doug
,
Afet
Dundar
,
Mary
Ziskin
,
Yi-Chen
Chiang
,
Jin
Chen
,
Autumn
Harrell
, and
Vasti
Torres
.
2013
.
Baccalaureate attainment: A national view of the postsecondary outcomes of students who transfer from two-year to four-year institutions (Signature Report No. 5)
.
Herndon, VA
:
National Student Clearinghouse Research Center
.
Simone
,
Sean Anthony
.
2014
.
Transferability of postsecondary credit following student transfer or coenrollment
.
Available
https: //nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2014163.
Accessed 26 April 2018
.
Tinto
,
Vincent
.
1993
.
Leaving college: Rethinking the causes and cures of student attrition
. 2nd ed.
Chicago
:
University of Chicago Press
.
Zeidenberg
,
Matthew
.
2012
.
Valuable learning or “spinning their wheels”? Understanding excess credits earned by community college associate's degree completers
.
Community College Research Center Working Paper No. 44, Teachers College, Columbia University
.

Notes

1. 

Ohio adopted this policy in 2005.

2. 

The most significant changes occurred in 2005 and created discipline-specific pathways, and the guaranteed transfer admission to any public university for students who completed an associate's degree. This change occurred in the years after the cohorts we examine in this analysis.

3. 

For the 2002 and 2003 cohorts, we are able to track students for eight and seven years, respectively. For ease of interpretation, we conduct our analysis within six years for all cohorts but do explore the effects within these earlier cohorts for a longer period of time. The results (available upon request) are not statistically different from for the six-year window.

4. 

Cincinnati State Technical and Community College, Columbus State Community College, Edison Community College, Lakeland Community College, Lorain County Community College, Sinclair Community College, Southern State Community College, Terra Community College, Washington State Community College.

5. 

Gender and ethnicity are distributed very similarly in the subsample of students who intend to transfer compared to the full sample. However, students who intend to transfer are younger, on average (20 vs. 23 years) when they enter the community college, more likely to be financially dependent on their parents (67 percent vs. 84 percent), and slightly less likely to be low-income, as measured by Pell grant eligibility (38 percent vs. 41 percent).

6. 

The higher college GPA for TM completers may be explained by the Ohio Articulation and Transfer policy in the years of our study. Students needed to earn a C or higher in each class used to complete the TM in order for this course to be counted toward the TM. All non-TM courses were accepted for transfer credit with grades above a D/D–. In subsequent cohorts (beyond the years of this study), the minimum passing grade for TM courses was changed to D/D– to be equivalent across all courses.

7. 

Although many of the control group students (those not completing the TM) have a propensity score of less than 0.05, we see most of the overlap between the treatment and control groups with propensity scores higher than 0.2. We hypothesize that we do not see overlap of treatment and control students at higher propensity score levels because of not being able to include all of the variables that predict one's decision to complete the full TM.

8. 

We also calculated the bias reduction for our IPSW using Rosenbaum and Rubin (1985). Calculated percent reductions in bias are available from the authors upon request.

9. 

All of the model specifications are sufficiently powered, except perhaps for those conditional on BA completion. Given the smaller sample sizes for these models, and assuming a power of 0.8, the minimum detectable effect size for the IPSW model estimating the effect of completing the TM on obtaining a BA within six years is 0.073.

10. 

Credits required for BA degree: Bowling Green State University = 122; Central State University = 124; Cleveland State University = 120–130; Kent State University = 120; University of Miami = 128; University of Ohio = 120; Ohio State University = 121–128; Shawnee State University = 120; University of Akron = 120; University of Cincinnati = 120; University of Toledo = 124; Wright State University = 120–124; Youngstown State University = 124.

11. 

For each estimation of the average treatment effect on the treated (ATT) with the unobserved confounder, we performed the simulation 100 times. The reported ATT is the average over the repeated simulations.

12. 

Following Ichino, Mealli, and Nannicini (2008): outcome effect = P(Y=1|T=0,U=1,W)P(Y=0|T=0,U=1,W)P(Y=1|T=0,U=0,W)P(Y=0|T=0,U=0,W), selection effect = P(T=1|U=1,W)P(T=0|U=1,W)P(T=1|U=0,W)P(T=1|U=0,W), where Y is the outcome, T is treatment assignment, U is the unobserved confounder, and W are the observed covariates.

13. 

In an effort to further explore this possibility with the data available, we examine our primary results table for several different subgroups. These subgroups include students who have similar levels of academic ability and persistence in the early semesters of college as defined by: (1) students who were full-time enrollees in their first semester; (2) students who earned at least a 2.8 GPA in their first semester; and (3) students who completed at least 35 credits within their first four terms. These narrower samples allow us to generate comparison groups with academic profiles that more closely resemble those completing the TM. Although these restrictions do slightly change the magnitude of some of our outcome estimates, they do not differ dramatically from our main findings, suggesting that controlling for academic ability through ACT may be sufficient. These estimates are available upon request.

Appendix A:  Additional Data

Table A.1.
Characteristics of First-Time, Full-Time Freshmen Attending a 2-Year College in the Ohio Public University System in 2004 Compared with National Samples
Fall 2003 Cohort: National Sample from BPS:03 (1)Fall 2004 Cohort: National Sample from NPSAS:04 (2)All Ohio Community Colleges (3)Sample Community Colleges (4)
Female 0.57 0.54 0.54 0.51 
White 0.60 0.67 0.83 0.75 
Black 0.15 0.12 0.09 0.16 
Latino 0.16 0.09 0.02 0.03 
Asian 0.06 0.07 0.01 0.02 
Dependent 0.61 0.81 0.77 0.81 
EFC = 0 0.25 0.23 0.18 0.16 
Pell grant eligible 0.33 0.35 0.50 0.42 
Married 0.16 0.19 0.07 0.05 
Average AGI 44,289.40 45,311.00 38,330.61 38,722.27 
N 2,102,394 2,147,546 49,969 23,450 
Fall 2003 Cohort: National Sample from BPS:03 (1)Fall 2004 Cohort: National Sample from NPSAS:04 (2)All Ohio Community Colleges (3)Sample Community Colleges (4)
Female 0.57 0.54 0.54 0.51 
White 0.60 0.67 0.83 0.75 
Black 0.15 0.12 0.09 0.16 
Latino 0.16 0.09 0.02 0.03 
Asian 0.06 0.07 0.01 0.02 
Dependent 0.61 0.81 0.77 0.81 
EFC = 0 0.25 0.23 0.18 0.16 
Pell grant eligible 0.33 0.35 0.50 0.42 
Married 0.16 0.19 0.07 0.05 
Average AGI 44,289.40 45,311.00 38,330.61 38,722.27 
N 2,102,394 2,147,546 49,969 23,450 

Notes: Computations for national samples by NCES PowerStats. Column 1 is limited to BPS-eligible students from NPSAS:04. Calculations for Ohio samples are by authors. The sample, “All Ohio Community Colleges” includes all 31 public community colleges in Ohio. The Sample Community Colleges includes: Cincinnati State Technical and Community College, Columbus State Community College, Edison Community College, Lakeland Community College, Lorain County Community College, Sinclair Community College, Southern State Community College, Terra Community College, and Washington State Community College. BPS: Beginning Postsecondary Students; NPSAS: National Postsecondary Student Aid Survey; EFC: Estimated Family Contribution.

Table A.2.
Sample Transfer Module Requirements by Select Institutions in the Sample
English CompositionMathNatural & Physical SciencesArts & HumanitiesSocial/ Behavioral SciencesTotalOther
Cincinnati State Technical and CC 6 hours (One 3-course sequence) 2.67 hours 8 hours 10 hours (5 courses from at least 2 areas) 10 hours (5 courses from at least 2 areas) 36.67 hours n/a 
Columbus State CC 5.33—7.33 hours (3.33—4 hours of English comp. and 2—3.33 hours of literature comp.) 3.33 hours 3 courses from option I (a 3-course sequence) or option II (select 3 courses from at least 2 areas) 3 courses from option I (a 3-course sequence) or option II (select 3 courses from at least 2 areas) 3 courses from option I (select 3 courses) or option II (select 3 courses from at least 2 areas) 38.67 hours n/a 
Edison CC 6 hours 3 hours 9 hours 6 hours 9 hours 40 hours 7 hours from all of the courses listed 
Southern State CC 5.33 hours 2 hours 6 hours (at least 1 course must be a lab course) 6 hours from at least 2 areas 6 hours from at least 2 areas 37.33 hours choose remaining from all of the courses listed 
English CompositionMathNatural & Physical SciencesArts & HumanitiesSocial/ Behavioral SciencesTotalOther
Cincinnati State Technical and CC 6 hours (One 3-course sequence) 2.67 hours 8 hours 10 hours (5 courses from at least 2 areas) 10 hours (5 courses from at least 2 areas) 36.67 hours n/a 
Columbus State CC 5.33—7.33 hours (3.33—4 hours of English comp. and 2—3.33 hours of literature comp.) 3.33 hours 3 courses from option I (a 3-course sequence) or option II (select 3 courses from at least 2 areas) 3 courses from option I (a 3-course sequence) or option II (select 3 courses from at least 2 areas) 3 courses from option I (select 3 courses) or option II (select 3 courses from at least 2 areas) 38.67 hours n/a 
Edison CC 6 hours 3 hours 9 hours 6 hours 9 hours 40 hours 7 hours from all of the courses listed 
Southern State CC 5.33 hours 2 hours 6 hours (at least 1 course must be a lab course) 6 hours from at least 2 areas 6 hours from at least 2 areas 37.33 hours choose remaining from all of the courses listed 

Note: All requirements in semester hours or number of courses. CC: community college; comp.: composition.

Table A.3.
Descriptive Statistics Comparing Students with Reported ACT Scores with Students with Missing ACT Scores for First-Time Freshmen, 2002—2004
Completed Transfer ModuleDid Not Complete Transfer Module
Has ACT DataMissing ACT Datap-valueHas ACT DataMissing ACT Datap-value
Demographics       
Female 0.545 0.527 0.497 0.506 0.505 0.928 
White 0.828 0.831 0.899 0.809 0.752 0.000 
Black 0.060 0.063 0.862 0.108 0.142 0.000 
Latino 0.025 0.025 0.931 0.020 0.026 0.031 
Asian 0.024 0.013 0.150 0.014 0.019 0.046 
Dependent 0.946 0.825 0.000 0.921 0.730 0.000 
Age at first enrollment 18.567 20.264 0.000 18.861 21.148 0.000 
Pell grant eligible 0.367 0.302 0.012 0.385 0.420 0.000 
Married 0.010 0.064 0.000 0.015 0.074 0.000 
Average AGI 46491.45 41626.93 0.000 44533.78 35663.84 0.000 
EFC 8540.55 7487.82 0.011 7766.616 5768.283 0.000 
Mother has at least some college 0.284 0.298 0.563 0.185 0.159 0.000 
Father has at least some college 0.280 0.295 0.550 0.176 0.141 0.000 
College information       
Ever worked during school year 0.983 0.958 0.004 0.950 0.911 0.000 
Worked first semester 0.777 0.768 0.685 0.835 0.802 0.000 
Number of semesters worked 4.680 4.470 0.047 2.836 2.258 0.000 
Ever took a remedial course 0.615 0.700 0.001 0.673 0.746 0.001 
Took a remedial course first semester 0.523 0.586 0.021 0.571 0.626 0.007 
Attended more than one CC 0.059 0.059 0.972 0.117 0.095 0.000 
Number of terms at another CC 3.464 3.613 0.835 3.621 3.736 0.513 
Credits after first semester 12.59 11.654 0.000 11.201 9.480 0.000 
GPA after first semester 3.021 3.061 0.324 2.476 2.29 0.000 
Number of credits before transfer 96.381 102.536 0.002 50.801 42.196 0.000 
Earned associate's degree 0.812 0.802 0.640 0.154 0.089 0.000 
Earned BA degree 0.386 0.329 0.029 0.124 0.049 0.000 
N 943 526  6308 6322  
Completed Transfer ModuleDid Not Complete Transfer Module
Has ACT DataMissing ACT Datap-valueHas ACT DataMissing ACT Datap-value
Demographics       
Female 0.545 0.527 0.497 0.506 0.505 0.928 
White 0.828 0.831 0.899 0.809 0.752 0.000 
Black 0.060 0.063 0.862 0.108 0.142 0.000 
Latino 0.025 0.025 0.931 0.020 0.026 0.031 
Asian 0.024 0.013 0.150 0.014 0.019 0.046 
Dependent 0.946 0.825 0.000 0.921 0.730 0.000 
Age at first enrollment 18.567 20.264 0.000 18.861 21.148 0.000 
Pell grant eligible 0.367 0.302 0.012 0.385 0.420 0.000 
Married 0.010 0.064 0.000 0.015 0.074 0.000 
Average AGI 46491.45 41626.93 0.000 44533.78 35663.84 0.000 
EFC 8540.55 7487.82 0.011 7766.616 5768.283 0.000 
Mother has at least some college 0.284 0.298 0.563 0.185 0.159 0.000 
Father has at least some college 0.280 0.295 0.550 0.176 0.141 0.000 
College information       
Ever worked during school year 0.983 0.958 0.004 0.950 0.911 0.000 
Worked first semester 0.777 0.768 0.685 0.835 0.802 0.000 
Number of semesters worked 4.680 4.470 0.047 2.836 2.258 0.000 
Ever took a remedial course 0.615 0.700 0.001 0.673 0.746 0.001 
Took a remedial course first semester 0.523 0.586 0.021 0.571 0.626 0.007 
Attended more than one CC 0.059 0.059 0.972 0.117 0.095 0.000 
Number of terms at another CC 3.464 3.613 0.835 3.621 3.736 0.513 
Credits after first semester 12.59 11.654 0.000 11.201 9.480 0.000 
GPA after first semester 3.021 3.061 0.324 2.476 2.29 0.000 
Number of credits before transfer 96.381 102.536 0.002 50.801 42.196 0.000 
Earned associate's degree 0.812 0.802 0.640 0.154 0.089 0.000 
Earned BA degree 0.386 0.329 0.029 0.124 0.049 0.000 
N 943 526  6308 6322  

Note: See Notes for table 1.

Table A.4.
Distribution of Variables Used to Determine Distribution of Unobserved Confounders in Table 6 
p_11p_10p_01p_00
T = 1, Y = 1T = 1, Y = 0T = 0, Y = 1T = 0, Y = 0
Outcome: Transferred 
Mother has at least some college 0.44 0.34 0.41 0.37 
Father has at least some college 0.42 0.34 0.42 0.33 
Placed into remediation 0.49 0.57 0.49 0.6 
Worked first semester 0.80 0.76 0.82 0.87 
Outcome: Completed Associate Degree 
Mother has at least some college 0.39 0.43 0.35 0.37 
Father has at least some college 0.40 0.41 0.36 0.39 
Placed into remediation 0.50 0.60 0.48 0.59 
Worked first semester 0.78 0.79 0.82 0.87 
Outcome: Number of Credits Before Transfer 
Mother has at least some college 0.42 0.42 0.44 0.41 
Father has at least some college 0.43 0.44 0.40 0.42 
Placed into remediation 0.53 0.31 0.52 0.47 
Worked first semester 0.79 0.84 0.83 0.82 
Outcome: Completed BA Degree 
Mother has at least some college 0.42 0.42 0.48 0.38 
Father has at least some college 0.41 0.47 0.43 0.40 
Placed into remediation 0.45 0.53 0.42 0.55 
Worked first semester 0.80 0.80 0.82 0.83 
Outcome: Number of Terms Spent at a Four-Year Institution 
Mother has at least some college 0.42 0.42 0.52 0.42 
Father has at least some college 0.45 0.39 0.41 0.45 
Placed into remediation 0.45 0.45 0.46 0.36 
Worked first semester 0.81 0.79 0.81 0.83 
Outcome: Number of Terms to BA Degree 
Mother has at least some college 0.49 0.35 0.49 0.46 
Father has at least some college 0.40 0.42 0.41 0.44 
Placed into remediation 0.59 0.31 0.50 0.33 
Worked first semester 0.81 0.79 0.80 0.84 
Outcome: Number of Credits Earned at BA Degree 
Mother has at least some college 0.46 0.33 0.49 0.47 
Father has at least some college 0.42 0.39 0.45 0.42 
Placed into remediation 0.50 0.34 0.42 0.42 
Worked first semester 0.80 0.78 0.77 0.84 
p_11p_10p_01p_00
T = 1, Y = 1T = 1, Y = 0T = 0, Y = 1T = 0, Y = 0
Outcome: Transferred 
Mother has at least some college 0.44 0.34 0.41 0.37 
Father has at least some college 0.42 0.34 0.42 0.33 
Placed into remediation 0.49 0.57 0.49 0.6 
Worked first semester 0.80 0.76 0.82 0.87 
Outcome: Completed Associate Degree 
Mother has at least some college 0.39 0.43 0.35 0.37 
Father has at least some college 0.40 0.41 0.36 0.39 
Placed into remediation 0.50 0.60 0.48 0.59 
Worked first semester 0.78 0.79 0.82 0.87 
Outcome: Number of Credits Before Transfer 
Mother has at least some college 0.42 0.42 0.44 0.41 
Father has at least some college 0.43 0.44 0.40 0.42 
Placed into remediation 0.53 0.31 0.52 0.47 
Worked first semester 0.79 0.84 0.83 0.82 
Outcome: Completed BA Degree 
Mother has at least some college 0.42 0.42 0.48 0.38 
Father has at least some college 0.41 0.47 0.43 0.40 
Placed into remediation 0.45 0.53 0.42 0.55 
Worked first semester 0.80 0.80 0.82 0.83 
Outcome: Number of Terms Spent at a Four-Year Institution 
Mother has at least some college 0.42 0.42 0.52 0.42 
Father has at least some college 0.45 0.39 0.41 0.45 
Placed into remediation 0.45 0.45 0.46 0.36 
Worked first semester 0.81 0.79 0.81 0.83 
Outcome: Number of Terms to BA Degree 
Mother has at least some college 0.49 0.35 0.49 0.46 
Father has at least some college 0.40 0.42 0.41 0.44 
Placed into remediation 0.59 0.31 0.50 0.33 
Worked first semester 0.81 0.79 0.80 0.84 
Outcome: Number of Credits Earned at BA Degree 
Mother has at least some college 0.46 0.33 0.49 0.47 
Father has at least some college 0.42 0.39 0.45 0.42 
Placed into remediation 0.50 0.34 0.42 0.42 
Worked first semester 0.80 0.78 0.77 0.84