Catching Up in Community Colleges: Academic Preparation and Transfer to Four-Year Institutions
by Josipa Roksa & Juan Carlos Calcagno - 2010
Background/Context: Transfer from community colleges to four-year institutions remains a contentious issue in higher education, with proponents showing that students do indeed transfer to four-year institutions and opponents arguing that starting in community colleges hinders baccalaureate degree attainment. One particularly salient issue in this debate is academic preparation. Although virtually all studies of transfer control for academic preparation, there is a dearth of research focusing on whether and how academically unprepared students can catch up in higher education.
Research Questions: We address two research questions: To what extent do academically unprepared students transfer to four-year institutions? And, can successful completion of intermediate outcomes, such as passing college-level math and writing courses, meeting specific credit thresholds, and earning an associate’s degree, diminish the role of initial preparation and increase the probability of transfer?
Research Design: Using event history techniques, we estimate the likelihood of transfer from community colleges to four-year institutions. Analyses include 20,900 first-time degree-seeking students who enrolled in Florida community colleges in the fall of 1998. Student enrollment is tracked through the summer of 2003.
Results: Community colleges can indeed serve as an alternative road of access to four-year institutions, even for academically unprepared students: Almost 20% of students in our sample who entered community colleges unprepared for college-level work made the transition to four-year institutions. Moreover, we found that successful completion of intermediate outcomes, such as passing college-level math and writing courses, meeting specific credit thresholds, and earning an associate’s degree, enhances the probability of transfer. However, the ability of community colleges to mitigate the negative effects of inadequate academic preparation on transfer is limited; regardless of the intermediate outcome completed, academically unprepared students continued to lag substantially behind their more prepared counterparts.
Conclusion: Community colleges can serve as a democratizing force in higher education; however, their ability to overcome inadequate academic preparation with which some students enter higher education is limited. Improving academic preparation in K–12 is thus a crucial component of enhancing transfer.
American education displays a remarkable degree of flexibility, providing a range of second chances and alternative pathways to success. This flexibility fits the norms about mobility and the ideal of educational opportunity (Turner, 1960) and has granted American education the perception of unparalleled openness. But it is exactly this flexibility that Clark (1960) criticized in his classic article on cooling out, arguing that the American educational system only provides an illusion of openness, preferring to sidetrack unpromising students rather than have them fail (p. 571).
Community colleges played a prominent role in Clarks (1960) critique, which was later echoed by Brint and Karabel (1989). With their open admission policies, community colleges admit all students interested in attending. However, Clark argued that faced with the reality of limited opportunities, community colleges develop elaborate ways to dampen students expectations and discourage them from aspiring to transfer to four-year institutions. In addition to Clark, Brint and Karabel emphasized the ways in which community colleges divert students from transfer pathways and channel them into short-term and vocational programs. Indeed, transfer from community colleges to four-year institutions has remained a contentious issue since these original critiques, with proponents showing that students do indeed transfer to four-year institutions and opponents arguing that starting in community colleges hinders baccalaureate degree attainment (for reviews, see Bailey & Morest, 2004; Dougherty, 1991, 2001).1
In this study, we return to the underlying concern in Clarks (1960) critique, which is whether students who are academically unprepared at one level of education can succeed at the next level. If students leave high school academically unprepared, can they succeed in higher education? We examine this question in the context of transfer from community colleges to four-year institutions. Although virtually all studies of transfer control for academic preparation, there is a dearth of research focusing on whether and how students who leave high school academically unprepared can catch up in higher education. Relying on Floridas unit record data of first-time community college students, we address two questions: To what extent do academically unprepared students transfer to four-year institutions? And, can successful completion of intermediate outcomes, such as passing college-level math and writing courses, meeting specific credit thresholds, and earning an associates degree, diminish the role of initial preparation and increase the probability of transfer?
The results suggest that indeed a substantial proportion of students who enter community colleges without adequate academic preparation do transfer to four-year institutions. Moreover, successful completion of intermediate outcomes is related to an increase in the probability of transfer. However, successful completion of intermediate outcomes does not eliminate the negative consequences of entering higher education unprepared. Even when academically unprepared students complete the most demanding intermediate outcome (the associate of arts [AA] degree), they continue to lag behind their academically prepared peers in transfer to four-year institutions. Thus, although community colleges can serve as a democratizing force, their ability to overcome the poor academic preparation with which some students enter higher education is limited.
These findings not only provide a better understanding of students educational trajectories but also have important policy implications. The extent to which students in higher education can compensate for inadequate preparation in high school helps to illuminate the likelihood of success of different policy solutions aimed at improving educational attainment. If students can catch up, then investing extra resources in information, counseling, developmental education, and other forms of support in community colleges may be effective. If they cannot catch up, then it is crucial that students be adequately prepared before they enter community colleges. The results of this study shed light on the extent to which policy makers can count on the effectiveness of either of these approaches. This may assist future decision-making aimed at facilitating the transition of students from community colleges to four-year institutions.
ACADEMIC PREPARATION AND TRANSFER
Early critics of community colleges (e.g., Brint & Karabel, 1989; Clark, 1960) argued that these institutions provide only an illusion of opportunity and contribute more to social reproduction than to social mobility.2 Whether community colleges serve as a democratizing force in higher education has often rested on the question of whether they provide an alternative route to a bachelors degree. Approximately two thirds of community college students express a desire to transfer to four-year institutions (Lee & Frank, 1990; National Center for Education Statistics [NCES], 2001, 2003a; Roksa, 2006); however, most of them do not actually make this transition (e.g., Brint & Karabel; Cohen, 1994; Grubb, 1991; Lee & Frank, 1990; Surette, 2001). Previous research suggests that baccalaureate degree attainment is hindered by starting in a community college as opposed to a four-year institution, even when comparing students who are similar in many background and academic characteristics (e.g., Alfonso, 2006; Dougherty, 1992; Monk-Turner, 1995; Whitaker & Pascarella, 1994). Moreover, recent studies based largely on descriptive statistics imply that once community college students transfer to four-year institutions, they complete bachelors degrees at similar rates as their counterparts at four-year colleges and universities (Adelman, 1999, 2005; Melguizo & Dowd, 2006). Although additional studies of baccalaureate completion are needed, current research indicates that transfer presents a critical juncture in the educational attainment of community college students who aspire to a bachelors degree.
Previous studies of transfer have illuminated how a range of different factors, such as structural features of community colleges (Brint & Karabel, 1989; Clark, 1960; Dougherty, 2001), articulation policies (Anderson, Sun, & Alfonso, 2006; Roksa, 2006), and student characteristics (Bailey, 2004; Dougherty & Kienzl, 2006; Lee & Frank, 1990; Palmer, 1991; Surette, 2001), influence this educational transition. However, extensive extant research has rarely considered whether disadvantages with which students enter higher education will inevitably lead to failure. Existing explanations would predict that students who enter community colleges academically unprepared will have a very low likelihood of transfer to four-year institutions. However, there is a dearth of research addressing this hypothesis or examining whether and how students who leave high school academically unprepared can catch up on their journeys through higher education. Although one could consider other factors, we focus on academic preparation because it is susceptible to policy influence and has become a prominent part of educational reforms in both K12 and higher education sectors. Moreover, academic preparation is a formidable concern in community colleges.
Although there is much variation among students, many community college students are not well prepared academically, and they are less well prepared than their counterparts in four-year institutions (Bailey, 2004; Lee & Frank, 1990; Monk-Turner, 1995; NCES, 1998, 2003a, 2003b). For example, a report from the NCES (2003a) noted that of the 1992 high school graduates who entered community colleges, 44% were at or below the lowest level of reading proficiency, and 30% were at or below the lowest level in math proficiency.3 A third of the students did not complete even the basic high school curriculum, and two thirds were deemed either not qualified or minimally qualified to attend a four-year college.
At the same time, academic preparation is one of the most important predictors of educational success. Community college students who are less well prepared academically are less likely to experience successful outcomes in higher education, including the completion of credentials and/or transfer to a four-year institution (Bailey, 2004; Calcagno, Crosta, Bailey, & Jenkins, 2007; Dougherty & Kienzl, 2006; Lee & Frank, 1990; NCES, 2003a; Palmer, 1991; Roksa, 2006; Surette, 2001). Earning an associates degree, transferring to a four-year institution, and earning a bachelors degree are all related to academic preparation in predictable ways: Students who complete a basic high school curriculum, who have greater math and reading proficiency, and who are overall deemed academically qualified to attend four-year institutions are more likely to succeed in attaining any of those outcomes (NCES, 2003a). The importance of academic preparation is also apparent in multivariate models. Even net of a range of controls for demographic characteristics and attendance patterns in higher education, students with higher academic achievement, particularly in math, are more likely to transfer (e.g., Dougherty & Kienzl; Lee & Frank; Roksa, 2006).
These two patterns, showing that a substantial proportion of community college students are not well prepared academically and that academic preparation is crucial for educational success, raise the question of whether community colleges are providing opportunities to academically unprepared students or whether they are simply delaying failure, as suggested by Clark (1960). This question is particularly relevant in the context of transfer from community colleges to four-year institutions. In principle, students who are not well prepared academically in the K12 system could use community colleges as an alternative path to a bachelors degree. Following this line of reasoning, community colleges would be expected to help students compensate for their poor academic preparation; instead of reproducing the preexisting pattern of academic disadvantage, community colleges would assist students in moving up the educational ladder. Although virtually all studies of transfer control for some aspect of academic preparation, they rarely consider whether academically unprepared students indeed do transfer to four-year institutions and whether specific events occurring while students are in community colleges can help to alleviate the negative consequences of inadequate academic preparation.
THE PRESENT STUDY
We begin this study by estimating the average gap in transfer rates between academically prepared and unprepared students. We then explore whether successful completion of intermediate outcomes can help students compensate for the initial disadvantage associated with inadequate academic preparation. We define intermediate outcomes as passing college-level math and writing courses, meeting specific credit thresholds, and earning an associates degree. We consider these outcomes because previous research demonstrates that completion of these intermediate steps increases the probability of overall educational success. McCormick (1999) proposed that credit accumulation provides a basic measure of progress through postsecondary education and showed that four-year students who completed a certain number of credits in the first year were much more likely to accumulate more credits overall, as well as complete their degrees (see also Adelman, 1999). Earning credits has been perceived as so important to degree attainment that some analysts exclude students who do not earn a certain number of credits from their models. Adelman (2003, 2005, 2006), for example, defined students who have earned less than 10 credits as incidental and removed them from his analyses of attainment. These students are considered not adequately committed to earning postsecondary degrees to warrant inclusion in analyses. Indeed, exclusion of students below a certain credit threshold is a common practice in estimating transfer rates, which we discuss in more detail in the section on data and methods.
In addition to credit accumulation, some scholars have noted that passing gatekeeper classes, such as remedial and initial college-level courses, can substantially increase the probability of earning educational credentials (Adelman, 2006; Bailey & Alfonso, 2005). Examining the educational attainment of community college students, Calcagno and his colleagues (2007) showed that passing the first college-level writing and math courses is related to students likelihood of completion (including completion of both degrees and certificates). Successfully passing these initial courses thus seems to have consequences for eventual educational attainment.
In the context of transfer from community colleges to four-year institutions, another important intermediate outcome is earning an associates degree. AA degrees are usually the only credentials included in statewide articulation policies and thus serve as an established pathway to four-year institutions (Ignash & Townsend, 2000, 2001). The importance of AA degrees for transfer is particularly pronounced in Florida, where we base our analyses. Whereas nationally, only one third of transfer students have associates degrees,4 in Florida, more than two thirds of students earn AA degrees before transferring to the four-year system (Florida Department of Education, 2003; Florida Postsecondary Education Planning Commission, 1999). Because of state-mandated articulation policies by which community college students who complete AA degrees are guaranteed admission to the four-year system as juniors, students have strong incentives to earn AA degrees before transfer. Thus, in the context of Florida, earning an AA degree is a crucial step in gaining access to four-year institutions.
Although previous research has demonstrated that completion of intermediate outcomes influences student success, scholars have yet to examine the relationship between these intermediate outcomes and academic preparation. More specifically, previous research has not examined the extent to which successful completion of intermediate outcomes can help compensate for inadequate academic preparation or whether completing intermediate outcomes can be particularly beneficial for students who enter community colleges academically unprepared. We examine these possibilities by using data on first-time community college students in Florida. We use event history modeling, which is specifically designed to study the occurrence and timing of events (Allison, 1984; Singer & Willett, 2003) and is thus appropriate to examining the dynamic nature of students passage through higher education (DesJardins, 2003). Although not common in the literature on community colleges in general or on transfer in particular, this methodology allows us to dynamically measure the impacts of intermediate outcomes on transfer, better approximating the reality of complex and varied trajectories that occur in student progression through higher education.
It is important to note that we are not concerned specifically with remedial education in this study. Low academic preparation is often linked to remedial education, and much research has been conducted to examine the effects of remedial education on educational outcomes (for reviews, see Goldrick-Rab, 2007; Levin & Calcagno, 2007; Schwartz & Jenkins, 2007). Although researchers sometimes use remedial education as a proxy for low academic preparation, this tends to confuse two different issues: (1) students who are not prepared, and (2) students who are receiving a particular treatment to remedy their lack of preparation. What we explore in this study is the extent to which academically unprepared students can make the transition to four-year institutions and, in particular, the extent to which completing certain intermediate outcomes plays a role in this process. Remedial education is one common institutional response to the overall lack of academic preparation, and, to the extent that it facilitates students transitions through the community college curriculum, it may be an effective strategy for improving educational attainment. Examining institutional responses and ways in which they can encourage students to advance along the road to transfer is beyond the scope of this study. However, this remains an important area for future research, which we highlight in the conclusion.
In this study, we use unit record transcript data for 37,623 first-time, degree-seeking Florida community college students who enrolled in a college-credit course at one of Floridas 28 community colleges in the fall of 1998.5 Student enrollment is tracked through the summer of 2003 (for a total of 15 terms, which includes fall, spring, and summer terms). The data set includes demographic characteristics and College Placement Test scores, as well as detailed information on enrollment, coursework, and outcomes (e.g., degree completion, transfer, and so on) for each term.
Because of our focus on transfer, we restricted the sample to students who completed at least 12 nonremedial credits. It is recommended that a certain threshold of credits be employed in the assessment of transfer (e.g., Clagett & Huntington, 1992; Spicer & Armstrong, 1996). Establishing a credit threshold eliminates incidental students who enroll for a limited number of courses only and who do not intend to transfer or earn a degree. These students are not likely candidates for transfer, so including them would substantially bias transfer estimates. Whereas Adelman (2003, 2005) used 10 credits as the threshold for inclusion in his sample, studies specifically focusing on transfer, such as the Transfer Assembly Project, have restricted the estimation of transfer to students who have earned 12 or more credits (Cohen, 1994, 1996). Using this latter restriction reduces our sample to 24,572 students. We then excluded those cases with missing information for independent variables (3,388 cases were deleted because of missing test score data and another 284 because of missing high school diploma information), which produced the final sample of 20,900 students with complete records.
It is important to note that the sample is not restricted to students who expected to earn a bachelors degree. The administrative data employed in this study do not provide information on long-term educational expectations. Moreover, educational expectations tend to shift over time and are not entirely reliable (see Bailey, Leinbach, & Jenkins, 2006). For example, almost 20% of traditional-age students in the National Education Longitudinal Study (NELS 1988-2000) sample who transferred to four-year institutions did not have an expectation of earning a bachelors degree.6 Restricting the sample to students who expected to earn a bachelors degree would have missed such students. Moreover, the current restrictions employed in this study (degree seeking, earned 12 or more credits) create a sample that represents a reasonable risk set for students who may wish to transfer. Indeed, an NCES report (2001) comparing transfer rates across different groups of students noted that transfer rates for students expecting to earn a bachelors degree were remarkably similar to those for students enrolled in degree-granting programs who completed 12 or more credits.
The dependent variable examined in this study is transfer to a four-year institution in the Florida higher education system. As noted, the sample includes first-time, degree-seeking7 students who have completed at least 12 credits in community colleges. When estimating transfer rates, previous research also frequently restricts the amount of time students have to transfer. For example, the Transfer Assembly Project considers whether students transferred within 4 years of enrollment in community college. Although it is likely that some students will transfer after this time period, the likelihood of transfer decreases precipitously over time. NCES (1997) reported that among transfers, students spent, on average, 20 months in a community college. And in a sample of traditional-age community college students, the average time to transfer was slightly more than 3 years, with 80% of students transferring within 4 years of entering community college.8 The data in our study includes information on 15 terms (or 5 years), which should thus capture the majority of students who transfer from community colleges to four-year institutions.
Our analyses include two sets of key independent variables: academic preparation and intermediate outcomes. The first key variable is a dummy variable indicating whether a student is academically prepared for college-level work. We define academic preparedness using Florida State Board of Education guidelines. According to these guidelines, students who score below a certain cutoff point are deemed to be not academically prepared for college-level work. The thresholds of preparedness depend on the tests that students submit for consideration: The SAT threshold is 420 for verbal and 440 for math;9 the College Placement Test (CPT) threshold is 83 for verbal and 72 for math. Our category of academically unprepared students includes those who have fallen below the threshold on both reading and math. This category includes almost one third (28 percent) of our sample. Academically unprepared students scored, on average, 346 in math and 371 in reading/writing, which is significantly below the average for the whole sample, p < 0.01 (422 for math and 465 for verbal).10
The second set of key independent variables refers to intermediate outcomes that were identified in previous research as important stepping stones in progression toward degree completion: (1) whether students passed their first college-level math and writing courses; (2) whether students met specific credit thresholds 24, 36, or 48 credits; and (3) whether students completed an AA degree. Students who complete these intermediate outcomes would be expected to have a higher likelihood of transfer. Intermediate outcomes are entered as time-varying indicators, which take on different values over the course of the 15 terms of enrollment. They turn on when a student reaches a given step, and stay on in all periods thereafter.
Finally, all models in our study include a set of time-invariant control variables. These variables control for student demographic characteristics (age, gender, and race/ethnicity), background (whether students are U.S. citizens and whether they received federal financial aid), and educational attributes (whether students completed a high school diploma, whether they enrolled full time, and whether they took SAT/ACT entrance exams). Although our data set does not include a measure of educational expectations, we control for whether students took SAT or ACT entrance exams. In the Florida community college system, students are permitted to either report their SAT/ACT scores or take the Florida-administered CPT. However, only SAT/ACT scores are accepted in the State University System (SUS). We thus include dummy variables denoting students who used their SAT/ACT math and verbal scores. These measures may serve as proxies of students expectations and motivations to transfer to four-year institutions.
TRANSFER TO FOUR-YEAR INSTITUTIONS
Figure 1 reports the cumulative proportions of community college students who transferred to public four-year institutions in Florida within different time frames. Focusing on the last set of bars first, this figure presents what might be considered a notable success of academically unprepared students: Almost 20% of students who entered community college with poor academic preparation (i.e., they were deemed not prepared for college-level work at the point of entry into community college) transferred to a four-year institution within 15 terms (5 years).11 However, academically unprepared students lagged behind their more academically prepared counterparts, 34% of whom transferred within 5 years.
Figure 1. Proportion of Students Transferring to Four-Year Institutions
Note. Differences between academically prepared and unprepared students are significant at p < 0.01.
Although there is a significant gap in transfer rates between students who entered community college with and without adequate academic preparation, it is possible that this pattern reflects the relatively short time period examined in this study. It could be that students who enter community college with poor academic preparation take a longer time to transfer. By considering transfer within a span of 15 terms, we may be underestimating the ability of academically unprepared students to catch up with their more prepared counterparts in the transition to four-year institutions.
To consider this possibility, we examine the trend in transfer over the 15 terms of available data. Figure 1 shows that the difference in the proportion of students transferring in the first periodwithin three termsis negligible because a very small proportion of students transfer in such a short amount of time. What is notable about the other time periods is that the gap in the proportion of students who transferred within 6, 9, 12, or 15 terms remains relatively stable: 12%16% fewer academically unprepared students transfer as compared with their academically prepared counterparts.12 Thus, although considering longer time periods may increase the total proportion of students who transfer to four-year institutions, the gap in transfer between academically prepared and unprepared students is not notably altered with the passing of time. Moreover, the proportion of students transferring increases rapidly at the beginning and then begins to level off. The majority of students transferred in their second or third year of community college enrollment. The growth in the proportion of students transferring slows down after the ninth term and is quite small after the 12th term. These findings imply that lengthening the observation period is not likely to either substantially increase the proportion of academically unprepared students who transferred or lead to a convergence in transfer rates of the two groups of students.
STUDENT CHARACTERISTICS AND INTERMEDIATE OUTCOMES
Although academically unprepared students lagged behind their more academically prepared peers in transfer to four-year institutions, one may hypothesize that at least two sets of factors may help to explain this difference: (1) student characteristics and (2) completion of intermediate outcomes. Table 1 presents descriptive statistics for a range of student characteristics in the first term of community college enrollment.13 The pattern of results indicates that the academically unprepared students tended to be more disadvantaged: They were more likely to be members of racial minority groups and to be from less advantaged family backgrounds (i.e., they were more likely to have received federal financial aid). At the same time, they were less likely to be enrolled full time and less likely to have taken SAT/ACT exams. The difference in whether students took SAT/ACT exams is particularly notable, but it is not surprising given that the first group of students includes those who were not well prepared for college.
Table 1. Characteristics of Students in the First Term
In light of these differences in individual characteristics, most studies examining the likelihood of transfer control for a range of similar attributes. However, previous research has shown that even net of these controls, academically unprepared students tend to have less desirable educational outcomes. For example, even after controlling for a host of individual characteristics, academic preparation (as measured by 12th-grade test scores) is significantly associated with the probability of transfer (Dougherty & Kienzl, 2006; Roksa, 2006). One of the explanations may be that academically prepared and unprepared students differ not only in what they bring to college but also in how they advance through college. Thus, in addition to considering a common set of student characteristics, we also examine a range of intermediate outcomes capturing varying dimensions of student progress through community college.
Table 2 presents descriptive statistics for different intermediate outcomes, all of which are time varying, reflecting the proportion of students who completed specific transitions sometime during the observation period. Although the gaps vary across these measures, the pattern is consistent: Academically unprepared students were less likely to successfully complete any of the intermediate outcomes. With respect to the first college-level courses, the gap is more pronounced for math than for writing. This is notable considering that previous research suggests that math plays a crucial role in persistence and success in postsecondary education (Adelman, 1999, 2003, 2006). The gap between academically prepared and unprepared students is greatest for the most demanding transitioncompletion of an AA degree. Whereas 38% of academically prepared students completed an AA degree, only 21% of academically unprepared students reached the same intermediate outcome. Because earning an AA degree is the main pathway to transfer in Florida, this gap in AA degree completion may be particularly consequential for understanding differences in transfer between academically prepared and unprepared students.
Table 2. Descriptive Statistics for Intermediate Outcomes
Considering these descriptive results, which reveal statistically significant differences between academically prepared and unprepared students in both individual characteristics and completion of intermediate outcomes, it may not be surprising that academically unprepared students have lower transfer rates. Because most previous research on transfer includes controls for individual characteristics, we focus in particular on intermediate outcomes, asking whether the gap in transfer rates between students who are academically prepared and those who are not decreases or disappears after accounting for the completion of intermediate outcomes. This is a crucial question because it illuminates the extent to which students can catch up in community colleges and suggests avenues for intervention. If the gap between academically prepared and unprepared students (measured as the regression coefficient for academically unprepared students) disappears after controlling for intermediate outcomes, then there is a possibility that students can make up for poor academic preparation. In this case, educational leaders and policy makers could develop ways to provide students with information, support, and other resources that might allow students to successfully complete these intermediate steps and thus enhance their educational attainment. In contrast, if the gap in transfer remains after controlling for intermediate outcomes, that would indicate that even students who successfully navigate community college curricula are not able to make up for the low academic preparation with which they entered higher education. In this case, assistance at the community college level may have only marginal effects, implying that interventions need to focus on academic preparation before college entry. We turn to event history modeling to consider these different possibilities.
EVENT HISTORY ANALYSES
To estimate the likelihood of transfer for our sample of community college students, we rely on event history models, and, in particular, we estimate single-risk discrete-time hazard models (Allison, 1984; DesJardins, 2003; Singer & Willett, 2003). To estimate these models, the data are organized in a person-period format, with a maximum of 15 observations for each studentone for each term in which the student was observed after initial enrollment. Time-invariant variables remain constant for each person in each period, whereas time-varying variables can take on different values in different time periods.
The event we are modeling is transfer from a community college to a four-year institution in the Florida public higher education system.14 Each student in the sample is considered at risk of transfer when first enrolled. Once a student experiences an event, his or her observations in later time periods are discarded, effectively preventing him or her from reentering the risk set. We use data for 15 terms (5 years), thus observing outcomes only for these 15 terms, and the 15 terms are referred to as the event time or event period. The beginning of the event is the first term of enrollment in a Florida community college (fall 1998), and the end is summer 2003. Students who did not transfer by the 15th term had unknown outcomes at the end of the event period and thereafter.
Formally, we are modeling the risk of transfer in each term, called the hazard. This is the conditional probability that an individual will obtain an outcome in time period j, given that he or she did not do so in an earlier time period and given that he or she was in the risk set. To be in the risk set in a given term, the student must be enrolled. The general population discrete-time hazard can be conceptualized as:
where Tk = j indicates student k’s outcome in term j, and the condition Tk ≥ j ensures that an outcome for student k has not occurred before time period j and that the student is enrolled in (observed in) time period j. The hazard in this case is the conditional probability that student k completes in term j, given that he or she did not complete before j and was in the risk set. The model also includes an indicator variable, G, for being an academically unprepared student, a vector X of student characteristics, and a vector Z of intermediate outcomes. The occurrence of these intermediate events raises or lowers the hazard of the final outcome; hence, we could enter them in the model as time-varying explanatory variables (Allison, 1984). Algebraically, we can write this relationship as:
In Equation 2, we took the logit of the hazard15 and defined a linear relationship between the conditioning data and logit hazard, where Dj is a vector of dummy variables indexing each term of enrollment, and α, δ , β and γ are parameters to be estimated. Taking an inverse transformation of both sides, we derive
which now describes a nonlinear relationship between the predictors and the hazard, analogous to the standard logistic regression routine (Singer & Willett, 2003). Once the data were put in a person-period data set, we could estimate parameters that maximized the likelihood of observing the sample data, assuming a logistic distribution.
Because event history analysis has a large temporal component, it is important to discuss our treatment of time in the presented equations. Our discrete-time models assumed a fully flexible spline for the specification of time that is entered as a series of t dummy variables indicating each term of enrollment as D'jα j. That is, no explicit functional restrictions are placed on how time affects the probability of transfer. This allows the baseline hazard to take on any shape and thus captures the effect of enrollment patterns through time, or the profile of hazard over time.
The nonexperimental nature of this studys design has methodological strengths and weaknesses. Although the event history model allows us to examine the dynamic nature of students passage through higher education, it is important to note that the estimates obtained from nonexperimental data are limited in terms of making causal inferences. Our models include controls for student characteristics and demographic background, and we use a fully flexible spline for the specification of time,16 but we are limited by the variables contained in the data set. Given that we do not have a randomized design, we are unable to control for all important preexisting characteristics in the models, and our findings may therefore be affected by unmeasured factors that are not accounted for in our models.
Most notably, previous nonexperimental studies on transfer have shown that students from more privileged backgrounds are more likely to make the transition to four-year institutions (e.g., Dougherty & Kienzl, 2006; Lee & Frank, 1990; Roksa, 2006). However, we do not have comprehensive measures of students family background (such as parental education, occupation, income, or composite socioeconomic status). In the absence of direct measures of family background, we include a variable indicating whether students received federal aid. This measure, which comprises mostly Pell Grant recipients (and thus encompasses low- and middle-income students), is used as a proxy of students relative income level.17 Future research should seek additional control variables and use experimental and quasi-experimental designs to establish a causal relationship among academic preparation, intermediate outcomes, and transfer rates.
DOES SUCCESSFUL COMPLETION OF INTERMEDIATE OUTCOMES DECREASE THE GAP IN TRANSFER?
We estimate several hazard models to examine the relationship between academic preparation and transfer. Table 3 represents the first set of baseline models, including odds ratios and standard errors derived from a maximum likelihood estimation of logistic regression parameters. The first model presents a simple baseline hazard model with the time-invariant dummy indicating whether a student was academically prepared at the point of entry into community college. The results indicate that academically unprepared students were approximately 80% less likely to transfer to four-year institutions in any given term than their academically prepared counterpartsor, stated differently, the results indicate that a lack of academic preparation is associated with an 80% decrease in the odds of transfer to four-year institutions in any given term.
Table 3. Estimated Odds Ratios for Hazard Models, With Transfer as the Outcome
Note. The model includes 15 time dummies.
*p < 0.05. **p < 0.01.
The second model includes a list of control variables from Table 1. As might be expected, the disadvantage of entering higher education with poor academic preparation is now smaller, although still notable: Students who were not academically prepared were 52% less likely to transfer in any given term, even net of the controls. The results for control variables on the probability of transfer mirror previous research: Older students, students from racial minority groups and less privileged family backgrounds, and students who were not U.S. citizens were less likely to make the transition to four-year institutions. Conversely, students who were enrolled full time and who took SAT/ACT exams were more likely to make this transition. The only characteristic that is not statistically significant is gender. This supports other findings based on nationally representative data indicating gender equality in transfer for recent cohorts (e.g., Dougherty & Kienzl, 2006; Roksa, 2006).
Models in Table 4 examine whether completing different intermediate outcomes mediates the relationship between academic preparation and the probability of transfer. The results indicate that students who passed a college-level math course were more than twice as likely to transfer as those who did not complete this intermediate outcome. However, passing a college-level reading course does not seem to be consequential for this educational outcome. After controlling for these intermediate outcomes, the coefficient for being academically unprepared changes only slightly: Students who were not academically prepared were still almost 50% less likely to make the transition to four-year institutions than their academically prepared peers.
Table 4. Estimated Odds Ratios for Hazard Models, Intermediate Outcomes
Remaining models show a strong and statistically significant relationship between each of the credit thresholds and the likelihood of transfer, with students completing 48 credits being almost 15 times as likely to transfer as students who did not pass this credit threshold. After controlling for credit thresholds, the disadvantage associated with poor academic preparation declines. Among students who passed the 48-credit threshold, academically unprepared students were 38% less likely than their academically prepared counterparts to make the transition to a four-year institution. Comparing the model with the 48-credit threshold with the baseline model (Model 2 in Table 3), the gap in transfer between academically prepared and unprepared students decreases by 14 percentage points (from 52% to 38%).
A similar finding is reported in the final model in Table 4. Students who completed AA degrees were almost 40 times as likely to transfer as those who did not. The magnitude of this finding is likely unique to the Florida context, reflecting the role of AA degrees in the states articulation policy. Because of the state-mandated articulation policy, students with AA degrees from Floridas community colleges are guaranteed a place in the SUS. The AA degree is thus one of the key gatekeepers for access to four-year institutions. Among students who earned an AA degree, the gap in the probability of transfer between students who were academically prepared and those who were not is smaller than in the baseline model.
Although the gap between academically prepared and unprepared students decreases as students successfully complete intermediate outcomes, it remains substantial. The persistence of the gap is perhaps most remarkable among students who completed an AA degree. Even among students who earned this crucial credential, academically unprepared students were 41% less likely than their more prepared counterparts to make the transition to four-year institutions. Because earning an AA degree essentially guarantees transfer, it is crucial to explore in future research why academically unprepared students continued to lag behind. These results may reflect individual differences not captured by our control variables. For example, it is possible that these students did not intend to transfer in the first place, that they changed their transfer aspirations and decided to pursue a different path, or that they did not get accepted to the school/program of their choice. Although we do not have the data to explore these possibilities, this is an area ripe for future research.
ARE INTERMEDIATE OUTCOMES MORE CONSEQUENTIAL FOR ACADEMICALLY UNPREPARED STUDENTS?
The results presented thus far raise another question: Are intermediate outcomes more consequential for academically unprepared students? Although academically unprepared students lagged behind their more prepared counterparts, even conditional on completing different intermediate outcomes, there may be an interaction between academic preparation and intermediate outcomes. Table 5 considers this possibility. The first two rows report odds ratios and standard errors for academically unprepared students and intermediate outcomes, respectively. These two rows replicate the previously reported results: Academically unprepared students were less likely to make the transition to four-year institutions, whereas completing intermediate outcomes increased the odds of making this transition.
Table 5. Estimated Odds Ratios for Hazard Models, Interactions Between Academic Preparation and Intermediate Outcomes
The third row presents the odds ratios and standard errors for interaction terms. An interaction term close to 1 would indicate that there is not much observed difference between students who were academically prepared and those who were not. However, interaction terms for all intermediate outcomes are positive and significant, indicating that successful completion of intermediate outcomes is significantly more important for academically unprepared students.
The odds ratios for the interaction terms indicate the advantage associated with passing a certain intermediate outcome for academically unprepared students. For example, the interaction term of 2.265 for passing the first college-level math course indicates that the benefit associated with passing this course was approximately twice as large for academically unprepared students. Stated differently, an academically prepared student who passed the first college-level math course was 1.9 times as likely to transfer as an academically prepared student who did not pass this course. However, an academically unprepared student who passed the first college-level math course was 4.3 times as likely (1.905 × 2.265) to transfer as an academically unprepared student who did not pass this course. The same pattern of results holds for other intermediate outcomes. For example, the odds ratio for passing a 36-credit threshold is more than 5 times as large for students who were not academically prepared as it is for those who were. In addition, earning an AA degree is associated with an increase in the probability of transfer for all students, but the odds ratio is almost twice as large for students who entered community college with low levels of academic preparation.
We would like to emphasize that reported interaction results need to be interpreted with caution. The results are based on observational data and thus cannot establish causality. Positive interaction terms in Table 5 do not imply a causal link between completing intermediate outcomes and transfer. Stated differently, the results presented do not imply that completing an intermediate outcome causes an increase in the probability of transfer. Interaction results are useful in that they reveal a complex relationship between academic preparation and intermediate outcomes in the data, illuminating an important area for future research. For example, developing an experimental study in which a group of academically unprepared students are offered additional assistance in completing intermediate outcomes while another group of students simply follow the regular curriculum could provide crucial insights into whether and how community colleges could use intermediate outcomes to facilitate the success of academically unprepared students.
American higher education is characterized by a remarkable degree of flexibility, consequences of which remain contested. Although flexibility can provide alternative pathways to success, it can also serve as an illusion of opportunity that simply delays failure. Community colleges play a critical role in this debate; some authors have argued that their role is to manage student ambitions in light of limited opportunities (Brint & Karabel, 1989; Clark, 1960). This management of ambition should be particularly pronounced for academically unprepared students. We thus examined the probability that students who enter community college academically unprepared transfer to four-year institutions. Moreover, we explored the extent to which successful completion of intermediate outcomes, such as passing college-level math and writing courses, meeting specific credit thresholds, and earning an associates degree, diminishes the role of initial preparation and increases the probability of transfer.
The results presented in this study indicate that community colleges can indeed serve as an alternative road of access to four-year institutions, even for academically unprepared students: Almost 20% of students in our sample who entered community colleges unprepared for college-level work made the transition to four-year institutions. Moreover, we found that successful completion of intermediate outcomes enhances the probability of transfer. However, the ability of community colleges to mitigate the negative effects of inadequate academic preparation on transfer is limited; regardless of the intermediate outcome completed, academically unprepared students continued to lag substantially behind their more prepared counterparts.
These findings have important policy implications. They highlight the importance of taking a holistic view of education. Traditionally, transfer between community colleges and four-year institutions has been viewed as a community college issue. However, some factors influencing transfer, such as academic preparation, precede students entry into community college. Although community colleges can provide opportunities for transfer, there is a high cost to entering higher education without adequate academic preparation, which reverberates throughout students educational careers. Even when students complete the most demanding intermediate outcome (the AA degree), they remain hindered by the poor academic preparation with which they enter higher education.
One important approach for enhancing transfer would be to improve academic preparation in the K12 sector. Crucially, such an approach ought not to be considered simply a K12 issue, but should rather become a joint endeavor of education leaders across the system, including community colleges and four-year institutions. Indeed, many states are developing a range of initiatives aimed at aligning expectations and preparation of different segments of public education. Florida, for example, has initiated an effort to build one seamless K20 educational system. This would provide an opportunity for a systemwide dialogue about academic preparation and the development of policies and programs to ensure that students entering higher education are indeed prepared for college-level work. Recognizing the importance of a holistic approach to education, which aims to improve academic preparation across the entire system, will likely be most beneficial in helping students successfully navigate the community college curricula and transfer to four-year institutions.
A number of questions regarding the relationship between academic preparation, intermediate outcomes, and transfer remain to be addressed in future research. Although we demonstrate the importance of successful completion of intermediate outcomes for transfer, it is beyond the scope of this study to explore why completing intermediate outcomes does not fully compensate for the lack of academic preparation. This may result from students motivation and personal characteristics, as well as inadequate instruction and support in community colleges. Moreover, we do not directly assess the trade-offs between education provided in K12 and community colleges. Students may not complete a certain intermediate outcome because they are not academically prepared or because they are not receiving adequate support in community colleges, among other reasons. Based on our results, preparation in K12 has notable consequences for student success, but some academically unprepared students make substantial progress in community colleges. This suggests the importance of conducting careful studies of community college practices to ascertain how they can best help academically unprepared students succeed in higher education.
One particularly promising avenue of research in this regard is to study variation across community colleges. Transfer rates vary remarkably across institutions (e.g., Cohen 1994, 1996; Florida Community College System, 2001). Understanding why certain community colleges are more successful at facilitating students completion of intermediate outcomes and eventual transfer could provide valuable insights into ways to enhance student success. Moreover, examining relationships between community colleges and four-year institutions, including formal and informal ties as well as tuition differentials, could suggest other avenues for improving transfer. Relationships between community colleges and four-year institutions could be particularly consequential for students who have completed intermediate outcomes. Future research would also benefit from replicating our study in other contexts. Although Floridas transfer rates do not differ substantially from the national average, Florida has a uniquely structured higher education system with a strong articulation policy. The benefits of academic preparation and successful completion of intermediate outcomes could have different consequences in other contexts.
Another important area of future research is to investigate what factors help studentsparticularly those who are not academically preparedcomplete intermediate outcomes at community colleges. Are those factors individual in nature (such as family and work obligations), related to networks (such as peers), or associated with institutional practices (e.g., counseling, advising, faculty characteristics, and other resources)? Thus far, one common response to inadequate student preparation has been the provision of remedial education. The extent to which providing students with different types of remedial services can enhance their probability of successfully completing intermediate outcomes is an important area of investigation. In this study, we have shown the critical role played by intermediate outcomes in the transition to four-year institutions. Future research would benefit from uncovering ways in which state and institutional policies could facilitate successful completion of these intermediate transitions.
Funding for the development of this article was provided in part by Lumina Foundation for Education through the Achieving the Dream: Community Colleges Count initiative. The authors thank Greg Kienzl, Davis Jenkins, and Tom Bailey for their insightful comments and Patricia Windham and Judith Thompson of the Florida Department of Educations Division of Community Colleges and Workforce Development for sharing the data used in the research.
1. Calculating transfer rates is highly contentious and depends in part on the definition used. Transfer rates can range from a low of 25% for all first-time community college students to a high of 52% for students who enroll in an academic major and take courses toward a bachelors degree (NCES, 2001; see also Spicer & Armstrong, 1996).
2. Recent studies examining democratization and diversion effects of community colleges provide a more balanced account (e.g., see Leigh & Gill, 2003; Roksa, 2008; Rouse, 1995, 1998).
3. Reported percentages refer to students who were at or below Level 1 in reading and math proficiency. Reading proficiency is divided into three levels: Level 1simple reading comprehension, including reproduction of detail and/or the authors main thought; Level 2ability to make relatively simple inferences beyond the authors main thought and/or to understand and evaluate relatively abstract concepts; Level 3ability to make complex inferences or evaluate judgments that require piecing together multiple sources of information from the passage. Math proficiency is divided into five levels, ranging from simple arithmetical operations on whole numbers (Level 1) to proficiency in solving complex, multistep word problems and/or the ability to demonstrate knowledge of the material found in advanced mathematics courses. For more information on these classifications, see NCES (1995).
4. Authors calculations are based on the National Education Longitudinal Study (NELS 19882000).
5. The sample does not include students who were formerly in dual high schoolcollege enrollment programs.
6. Authors calculations..
7. A student is considered degree-seeking if the college deems him or her to be studying in any of the following programs: associate of arts degree, associate of science degree, vocational certificate, general freshman, linkage (programs in more than one school), associate of science certificate, associate of applied science degree, applied technology diploma, or advanced technical certificate.
8. Authors calculations based on the NELS 19882000.
9. Starting with the fall 2000 semester, the verbal SAT threshold changed to 440.
10. Scores for placement tests were all converted to an SAT scale (200800) using the test makers formulas.
11. Although the average transfer rates reported in this study may appear low, they are within the range of national averages (Cohen, 1994, 1996; NCES, 2001).
12. Differences between academically prepared and academically unprepared students are statistically significant at p < 0.01.
13. In Table 1, all variables are time invariant; they are restricted to the first term of community college attendance. However, full-time enrollment is allowed to vary over time in the models.
14. Of the students who completed at least 12 credits, 8.1% transferred to institutions outside the Florida SUS (i.e., to private institutions in Florida or to institutions out of state). These students are not captured in our analyses.
15. The logit approach in an event history framework is the discrete-time version of the continuous-time Cox model. It assumes proportional odds rather than proportional hazards. A complementary loglog link also could have been used, but it did not seem necessary to assume proportional hazards given that the data process is occurring in continuous time that is interval censored (Singer & Willett, 2003).
16. Empirical evidence suggests that the misspecification problems due to unobserved heterogeneity that is uncorrelated with observed covariates are reduced when researchers use a fully flexible spline for the specification of time (Dolton & van der Klaauw, 1995).
17. Having a more comprehensive measure of family background would likely attenuate the effect of academic preparation on transfer. However, previous research provides ample evidence that academic preparation has a significant impact on educational transitions, even net of controls for family background. Thus, although the magnitude of the coefficients may change, we expect that the overall patterns of the results would persist even if we could control more adequately for students family background.
Adelman, C. (1999). Answers in the tool box: Academic intensity, attendance patterns, and bachelors degree attainment. Washington, DC: U.S. Department of Education.
Adelman, C. (2003). Postsecondary attainment, attendance, curriculum, and performance. Washington, DC: U.S. Department of Education.
Adelman, C. (2005). Moving into town and moving on: The community college in the lives of traditional-age students. Washington, DC: U.S. Department of Education.
Adelman, C. (2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education.
Alfonso, M. (2006). The impact of community college attendance on baccalaureate attainment. Research in Higher Education, 47, 873903.
Allison, P. (1984). Event history analysis: Regression for longitudinal event data. Thousand Oaks, CA: Sage.
Anderson, G. M., Sun, J. C., & Alfonso, M. (2006). Effectiveness of statewide articulation agreements on the probability of transfer: A preliminary policy analysis. Review of Higher Education, 29, 261291.
Bailey, T. (2004, April). Community college students: Characteristics, outcomes, and recommendations for success. CCRC Currents, 13
Bailey, T., & Alfonso, M. (2005). Paths to persistence: An analysis of research on program effectiveness at community colleges. Indianapolis, IN: Lumina Foundation for Education.
Bailey, T., Leinbach T., & Jenkins D. (2006). Is student success labeled institutional failure? Student goals and graduation rates in the accountability debate at community colleges (CCRC Working Paper No. 1). New York: Columbia University, Teachers College, Community College Research Center.
Bailey, T., & Morest, V. S. (2004). The organizational efficiency of multiple missions for community colleges. New York: Columbia University, Teachers College, Community College Research Center.
Brint, S., & Karabel, J. (1989). The diverted dream: Community colleges and the promise of educational opportunity in America, 19001985. New York: Oxford University Press.
Calcagno, J. C., Crosta, P., Bailey, T., & Jenkins, D. (2007). Stepping stones to a degree: The impact of enrollment pathways and milestones on older community college students. Research in Higher Education, 48, 775801.
Clagett, C. A., & Huntington, R. B. (1992). Assessing the transfer function: Data exchanges and transfer rates. Community College Review, 19(4), 2126.
Clark, B. R. (1960). The cooling-out function in higher education. American Journal of Sociology, 65, 569576.
Cohen, A. M. (1994). Analyzing community college transfer rates. In A. M. Cohen (Ed.), Relating curriculum and transfer: New directions for community colleges, No. 86 (pp. 7179). San Francisco: Jossey-Bass.
Cohen, A. M. (1996). Orderly thinking about a chaotic system. In T. Rifkin (Ed.), Transfer and articulation: Improving policies to meet new needs. New directions for community colleges, No. 96 (pp. 2534). San Francisco: Jossey-Bass.
DesJardins, S. (2003). Event history methods: Conceptual issues and an application to student departure from college. In J. Smart (Ed.), Higher education: Handbook of theory and research (pp. 421471). Boston: Kluwer Academic Press.
Dolton, P., & van der Klaauw, W. (1995). Leaving teaching in the UK: A duration analysis. Economic Journal, 105, 431444.
Dougherty, K. J. (1991). The community college at the crossroads: The need for structural reform. Harvard Educational Review, 61, 311336.
Dougherty, K. J. (1992). Community colleges and baccalaureate attainment. Journal of Higher Education, 63, 188214.
Dougherty, K. J. (2001). The contradictory college: The conflicting origins, impacts, and futures of the community college. Albany: State University of New York Press.
Dougherty, K. J., & Kienzl, G. (2006). Its 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, 452487.
Florida Community College System. (2001). AA transfers to the SUS. Tallahassee: Author.
Florida Department of Education. (2003). 2+2 in Florida. Tallahassee: Author.
Florida Postsecondary Education Planning Commission. (1999). Evaluation of Florida's two-plus-two articulation system. Tallahassee: Author.
Goldrick-Rab, S. (2007). Promoting academic momentum at community colleges: Challenges and opportunities (CCRC Working Paper No. 5). New York: Columbia University, Teachers College, Community College Research Center.
Grubb, W. N. (1991). The decline of community college transfer rates: Evidence from national longitudinal surveys. Journal of Higher Education, 62, 194222.
Ignash, J. M., & Townsend, B. K. (2000). Evaluating state-level articulation agreements according to good practice. Community College Review, 28(3), 121.
Ignash, J. M., & Townsend, B. K. (2001). Statewide transfer and articulation policies: Current practices and emerging issues. In B. K. Townsend & S. B. Twombly (Eds.), Community colleges: Policy in the future context (pp. 173192). Westport, CT: Ablex.
Lee, V. E., & Frank, K. A. (1990). Students' characteristics that facilitate the transfer from two-year to four-year colleges. Sociology of Education, 63, 178193.
Leigh, D. E., & Gill, A. M. (2003). Do community colleges really divert students from earning bachelor's degrees? Economics of Education Review, 22, 2330.
Levin , H., & Calcagno, J. C. (2007). Remediation in the community college: An evaluators perspective (CCRC Working Paper No. 9). New York: Columbia University, Teachers College, Community College Research Center.
McCormick, A. (1999). Credit production and progress toward the bachelors degree: An analysis of postsecondary transcripts for beginning students at 4-year institutions. Washington, DC: U.S. Department of Education.
Melguizo, T., & Dowd, A. C. (2006). The study of economic, informational, and cultural barriers to community college student transfer access at selective institutions. Section I in Transfer access to elite colleges and universities in the United States: Threading the needle of the American dream (Report for the Jack Kent Cooke Foundation). Lansdowne, VA: Jack Kent Cooke Foundation.
Monk-Turner, E. (1995). Factors shaping the probability of community vs. four-year college entrance and acquisition of the B.A. degree. Social Science Journal, 32, 255264.
National Center for Education Statistics. (1995). Psychometric report for the NELS:88 base year through second follow-up (Report prepared by D. A. Rock, J. M., Pollack, & P. Quinn, NCES 95-382). Washington, DC: U.S. Department of Education.
National Center for Education Statistics. (1997). Transfer behavior among beginning postsecondary students: 19891994 (Report prepared by A. C. McCormick & C. D. Carroll, NCES 1997-266). Washington, DC: U.S. Department of Education.
National Center for Education Statistics. (1998). Descriptive summary of 199596 beginning postsecondary students, with profiles of students entering 2- and 4-year institutions (Report prepared by L. K. Kojaku, A.-M. Nunez, & A. G. Malizio, NCES 1999-030). Washington, DC: U.S. Department of Education.
National Center for Education Statistics. (2001). Community college transfer rates to 4-year institutions using alternative definitions of transfer (Report prepared by E. M. Bradburn, D. G. Hurst, & S. Peng, NCES 2001-197). Washington, DC: U.S. Department of Education.
National Center for Education Statistics. (2003a). Community college students: Goals, academic preparation, and outcomes (Report prepared by G. Hoachlander, A. C. Sikora, L. Horn, & C. D. Carroll, NCES 2003-164). Washington, DC: U.S. Department of Education.
National Center for Education Statistics. (2003b). Postsecondary attainment, attendance, curriculum, and performance (Report prepared by C. Adelman, B. Daniel, I. Berkovits, & J. Owings, NCES 2003-294). Washington, DC: U.S. Department of Education.
Palmer, J. (1991). What do we know about transfer? Transfer Working Papers, 2(4), 38.
Roksa, J. (2006). Does vocational focus of community colleges hinder students educational attainment? Review of Higher Education, 29, 499526.
Roksa, J. (2008). Structuring access to higher education: The role of differentiation and privatization. Research in Social Stratification and Mobility, 26, 5775.
Rouse, C. E. (1995). Democratization or diversion? The effect of community colleges on educational attainment. Journal of Business and Economic Statistics, 13, 217224.
Rouse, C. E. (1998). Do two-year colleges increase overall educational attainment? Evidence from the states. Journal of Policy Analysis and Management, 17, 595620.
Schwartz, W., & Jenkins, D. (2007). Promising practices for community college developmental education (Report prepared for the Connecticut Community College System). New York: Columbia University, Teachers College, Community College Research Center.
Singer, J., & Willett, J. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press.
Spicer, S. L., & Armstrong, W. B. (1996). Transfer: The elusive denominator. In T. Rifkin (Ed.), Transfer and articulation: Improving policies to meet new needs. New directions for community colleges, No 96 (pp. 4554). San Francisco: Jossey-Bass.
Surette, B. J. (2001). Transfer from two-year to four-year college: An analysis of gender differences. Economics of Education Review, 20, 151163.
Turner, R. H. (1960). Sponsored and contest mobility and the school system. American Sociological Review, 25, 855867.
Whitaker, D. G., & Pascarella, E. T. (1994). Two-year college attendance and socioeconomic attainment: Some additional evidence. Journal of Higher Education, 65, 194210.