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

It's Not Enough to Get Through the Open Door: Inequalities by Social Background in Transfer from Community Colleges to Four-Year Colleges


by Kevin J. Dougherty & Gregory S. Kienzl - 2006

The growing policy interest in community colleges as gateways to the baccalaureate degree naturally raises the question of how equitably transfer opportunities are distributed by student background and what factors may explain background differences that might be found. We analyze two nationally representative data sets to examine how the likelihood of transfer is affected by social background, precollege academic characteristics, external demands at college entrance, and experiences during college. We find that high-SES students have significantly higher transfer rates, in part because of advantages in precollege academic preparation and educational aspirations. Older college entrants are much less likely to transfer than students entering college right out of high school, and a significant portion of this age gap is more often due to having children, lower educational aspirations, and a vocational major, and being enrolled part time. Though women and nonwhites differ from men and whites in transfer rates, these differences are not statistically significant. But there is an important caveat: blacks tend to have higher educational aspirations than whites of the same socioeconomic background. When we control for educational aspirations, thus removing this black aspirational advantage, the black-white gap in transfer rates widens considerably, becoming statistically significant in one of our samples but not the other.

INTRODUCTION


From the beginning of the community college, one of its fundamental missions has been to facilitate the attainment of baccalaureate degrees by providing the early stages of a baccalaureate education and aiding transfer to four-year colleges (Brint and Karabel 1989; Cohen and Brawer 2003; Dougherty 1994). This role not only continues today but also promises to become increasingly important (Dougherty 2002; Wellman 2002).


But during the years between 1960 and 1990, the transfer mission was eclipsed. Community colleges shifted their attention to expanding occupational education and continuing education, and transfer rates declined (Brint and Karabel 1989; Dougherty 1994; Dougherty and Bakia 2000). For example, among students entering community college right out of high school, the rate of transfer to four-year colleges within four years of entering college dropped from 29% for those entering community college in 1972 to 20% for those entering community college in 1980 (Grubb 1991, 202).i


However, in the last 10 years, interest in the transfer function has strongly revived. The reasons are various. State governments have encouraged students eligible for the state universities to begin at community colleges and then transfer to four-year colleges because this saves the states considerable money at a time when university enrollments have been sharply rising but state finances have been badly battered by a stagnant economy (Dougherty 2002; Ignash and Townsend 2001; Mercer 1992; Wellman 2002, 4). For example, during the first decade of the new millennium, college enrollments are projected to increase 21% in California, 12% in Texas, 26% in Florida, and 20% in North Carolina (Wellman 2002, 4). Moreover, community college transfer attracted the attention of policy makers in states such as California and Texas, where affirmative action in admissions had been outlawed, but universities hoped to retain a diverse student body (Hebel 2000). Finally, advocates, scholars, and policy makers concerned about issues of social stratification have highlighted the importance of transfer, noting that minority and working-class students are increasingly relying on community colleges for access to the baccalaureateii because of sharply rising four-year college tuition, stagnating need-based student aid,iii slumping incomes for less advantaged families, and the reduction of remedial education in public four-year colleges (Callan 2003; Dougherty 2002, 315-33; Lumina Foundation 2005; McPherson and Schapiro 1998, 44-46; McPherson and Schapiro 1999, 6-7, 19-24; Wellman 2002, 4-7; Young 2002).


In focusing on transfer, we are not suggesting that other outcomes of the community college are not important. Certainly, many students enter the community college interested not in transfer but in acquiring short-term training or a terminal degree, whether a certificate or an associate’s degree. And in fact, students do secure valued outcomes other than transfer. For students who entered community colleges in 1995-1996, 10% had secured a certificate, and 16% had received a terminal associate’s degree by June 2001 (Berkner, Ho, and Cataldi 2002, 12). And these degrees do bring significant payoffs: compared with high school graduates with similar social and academic characteristics, terminal associate degree holders enjoy a 15%-30% income advantage, and certificate degree holders a 5%—15% advantage (Grubb 2002; Kane and Rouse 1999; Kienzl 2004; Marcotte et al. 2005).


Yet transfer still remains important. Many students enter the community college with the expectation of achieving at least a baccalaureate degree. For example, an analysis of the National Education Longitudinal Study of the 8th Grade in 1988 (NELS: 88) found that among those students who graduated from high school in 1992 and entered a community college within the subsequent two years, 63% stated that they were aiming at a baccalaureate degree or higher (Hoachlander, Sikora, and Horn 2003, 10). This figure is echoed by an analysis of the most recent Beginning Postsecondary Students Longitudinal Study (BPS:96), which found that, among first-time college students entering public two-year colleges in 1995-1996, 78% planned for at least a bachelor’s degree (Kojaku and Nunez 1998, 7). Clearly, there is reason to question how firm and realistic these high educational ambitions are, but they cannot simply be dismissed.iv Moreover, there is strong warrant for community college entrants to pursue a baccalaureate degree. Although some holders of terminal occupational degrees, such as nursing, can secure incomes that exceed those of certain baccalaureate degree holders, the fact remains that on average, baccalaureate degrees still confer significantly higher economic returns than do terminal associate degrees or certificates. On average, with all other things being equal, baccalaureate degree holders enjoy a 30%-40% advantage in yearly income over high school graduates, considerably higher than the income advantage for the average terminal associate degree or certificate holder (Grubb 2002; Kane and Rouse 1999; Kienzl 2004).


This renewed importance of transfer raises the question of how equitably transfer opportunities are distributed by student background. The answer to this question carries major implications for state policies that encourage more baccalaureate aspirants to begin college at community colleges. If major differences exist in transfer rates by social background, the pursuit of equal access cannot stop simply with getting minority and working-class baccalaureate aspirants into the community college. It is also important to make sure that students have an equal chance to transfer.v


REVIEW OF THE LITERATURE


Transfer has been the subject of a considerable amount of study over the years.vi The research literature is dominated by institutional studies that examine the transfer rate for a particular community college. However, a fair number of studies of national data sets exist: Velez and Javalgi’s 1987 analysis of the National Longitudinal Survey of the High School Class of 1972 (NLS-72); Lee and Frank’s 1990 study of High School and Beyond (HS&B); Grubb’s 1991 study of both NLS-72 and HS&B; Surette’s 2001 analysis of the National Longitudinal Survey of Youth (NLSY); the studies by McCormick (1997) and Bradburn, Hurst, and Peng (2001) of the Beginning Postsecondary Students Longitudinal Study (BPS:90) of 1989-1994; and the studies by Hoachlander, Sikora, and Horn (2003) of NELS:88 and the Beginning Postsecondary Students Longitudinal Study of 1995-1996 (BPS:96).


The studies of the NLS-72, HS&B, and NLSY found that for students entering community colleges in the early 1970s and early 1980s, socioeconomic status (SES), race-ethnicity, and gender all significantly affected transfer rates. Students who were female, black, or had lower-SES parents were significantly less likely to transfer than were students with the obverse characteristics (Lee and Frank 1990; Surette 2001; Velez and Javalgi 1987).


Despite the quality of these national studies, there is good reason for further analysis. For one, these studies are restricted to the 1970s and 1980s, yet the question remains as to whether the patterns found hold for the 1990s. During the late 1980s and 1990s, major efforts were made to raise the transfer rate, particularly for minority and lower-SES students. For example, the Ford Foundation sponsored the Urban Community College Transfer Opportunity Program, and several states, such as California and Florida, put extra resources into enhancing transfer rates (Dougherty 1994, 254-60; Ford Foundation 1988). Moreover, the 1990s witnessed a shift in college-going rates by gender, with women eclipsing men.


In addition, the earlier studies ignored the impact of variations in age. The surveys used by earlier studies focused on students entering college right out of high school, thus disallowing any examination of the transfer rates of students who delayed their college entrance by two or more years. Yet, the issue of the impact of age differences on transfer rates has become quite important as increasing numbers of older people enter college. To be sure, many are not first-time college entrants but returning college students. Still, many first-time college entrants are of nontraditional age. For example, among first-time college students entering community college in 1995-1996, 26% were 24 years of age and older (Kojaku and Nunez 1998, 7). Evidence that age is likely to be associated with differences in transfer comes from the Transfer and Retention of Urban Community College Students, an ongoing study of transfer in the Los Angeles Community College District. The project has found significant differences by age in transfer preparation—that is, in the taking of state-designated courses designed to prepare students for transfer (Hagedorn et al. n.d.)


All these reasons suggest the desirability of studying transfer patterns during the 1990s, particularly using data sets that allow us to examine older first-time college entrants. Two national longitudinal surveys allow us to do this. The National Educational Longitudinal Study of the 8th Grade (NELS:88) is a long-term follow-up of students who were in the eighth grade in 1988. They were subsequently followed through the year 2000. A key advantage of NELS:88 is the availability of college transcripts, which allows us to precisely measure college attendance and track transitions between postsecondary institutions. The main limitation of the NELS:88— that it cannot capture the experience of older first-time college students—is rectified by the Beginning Postsecondary Students Longitudinal Study of 1989-90 (BPS:90), which examines first-time college students of any age who entered college in 1989 and were followed up in 1994.


The data on transfer rates in the BPS:90 sample have been ably analyzed by McCormick (1997) and Bradburn, Hurst, and Peng (2001). However, neither study provides a full-blown multivariate analysis of the impact of student background on transfer. McCormick analyzed the impact of SES, gender, and age (but not race), controlling for educational aspirations, college enrollment status (full time or part time), college GPA, receipt of financial aid, and overall satisfaction with the first institution (43). These are good variables to control, but many are also left out, including racial-ethnic background, high school academic performance, marital and parental status at time of college entry, working during college, and college major. Hence, a more extensive analysis of the impact of social background on transfer and the mechanisms by which that impact is transmitted is needed.


RESEARCH QUESTIONS AND METHODS THE PREVIOUS CONSIDERATIONS PROMPTED US TO ASK TWO RESEARCH QUESTIONS.


1. To what degree do transfer rates vary by student social background, and how have those patterns changed over time?


Previous studies have found that SES, race-ethnicity, and gender all significantly affect transfer rates (Lee and Frank 1990; Velez and Javalgi 1987). However, these studies analyzed data from the 1970s and the 1980s (NLS-72 and HS&B, respectively). We investigated whether in the 1990s, with the renewal of interest in and support for transfer, inequality of transfer by class, race, and gender changed in its extent or form. Moreover, we also investigated the impact of a variable that those earlier studies could not examine: age.


2. How are the effects of social background transmitted?


In this article, we focus on the impact of three sets of mediating variables: precollege personal characteristics (academic preparation in high school and educational and occupational aspirations); external demands as the student enters college (marital and parental status, extent and intensity of work, and enrollment status); and experiences during college (major or college program, degree of academic and social integration into the college).


DATA


For this study, we analyzed two national data sets: the National Education Longitudinal Study of the 8th Grade (NELS: 88) and the Beginning Post-secondary Students Longitudinal Study (BPS:90). Both focus on students entering college around the same time, but they bring different strengths to this analysis. NELS:88 gives us a larger sample of community college entrants, a better measure of SES (see below), and a better set of variables that measure precollege academic preparation. However, because NELS:88 focuses on younger students, it did not allow us to examine the impact of age on transfer. Moreover, it lacks measures of academic and social integration during college. BPS:90 rectified both omissions by examining students of any age who are entering college in 1989-1990 and by providing a large number of academic and social integration variables. By analyzing these data sets together, we can get a much better sense of the extent of the impact of social background on transfer and the means by which that impact is exerted.


In the case of NELS:88, we focused on students who first entered a community college in the period between 1992 (when most of the respondents would have graduated from high school) and 1994. This allowed for a few delayers to be eligible for our analysis. All students in our NELS:88 sample responded to the 1990, 1992, 1994, and 2000 follow-ups. In the case of BPS:90, we focused on students who entered a community college in 1989 and had responded to the 1992 and 1994 follow-ups. These data sets are described in the appendix.


DEPENDENT VARIABLE


Our dependent variable is transfer status. It is a binary variable that measures whether community college entrants transferred to a four-year college at any point after their initial year in postsecondary education (PSE). For BPS:90, we used the student’s primary postsecondary institution in the 1989-1990 academic year to anchor our analysis, and we restricted our sample to those stating that they attended a public two-year college as their primary institution in their first year. For the next five years, students were asked to identify their primary institution each year. If a student indicated attendance at a four-year college during any year subsequent to 1989, the student is considered to have transferred. In a small number of cases, community college students may have reported attending a four-year college and then switching back to a two-year school. We still consider them as having transferred to a four-year college. For NELS:88, we relied on transcript information to identify initial postsecondary entry and subsequent transitions between institutions. To be counted as having transferred in our NELS:88 analysis, a student’s referent institution must be a public two-year college, and transfer credits must be observed on the student’s four-year college transcript. Community college students who were simultaneously enrolled in a four-year college were not regarded as transfer students unless the intensity of their four-year attendance dominated in terms of credits earned in that period. Overall, students in BPS:90 had five years to transfer to a four-year college, while students in NELS:88 had, at most, eight years.


We examined the transfer status of all community college entrants, with no restriction pertaining to intentions. Much discussion has gone into what is the appropriate denominator for any measure of transfer. Many community college advocates have correctly pointed out that many students enter the community college with no intention to transfer, or if they state such an intention, it is only weakly and unrealistically held. Hence, they have called for calculating transfer rates only for those who have clearly established potential to transfer. One frequently used formula is to restrict the denominator of calculations of transfer rates to those who have accumulated 12 or more credits within four years of entering community college (Bradburn, Hurst, and Peng 2001; Cohen and Brawer 2003, 56).


Such restrictions of the denominator introduced some important distortions of analysis. First, they failed to consider as transfer eligible many students who do indeed end up transferring. For example, Bradburn and Hurst found that in the Beginning Postsecondary Students Longitudinal Study of 1989-94, 45% of those who had transferred by 1994 did not meet the 12-credit criterion proposed by Cohen and Brawer (Bradburn, Hurst, and Peng 2001, 123). Another problem with restricting the denominator to purportedly transfer-oriented students is that it hinders our ability to measure the importance of factors that affect transfer. Focusing on students with, say, baccalaureate aspirations or certain patterns of course taking hinders our ability to examine how powerfully educational aspirations or certain course-taking patterns shape whether students transfer. Moreover, measures that build transfer propensity into the denominator suggested that transfer propensity is a trait that students bring into community college and one that cannot be changed; this undercuts awareness of the possibility and desirability of changing that incoming propensity, such as by “warming up” student motivation to transfer. For all these reasons, we chose to examine the transfer status of all community college entrants regardless of incoming characteristics. We could then explicitly bring into the analysis the question of the relationship between transfer propensity and having a certain level of educational aspirations upon entering college.


INDEPENDENT VARIABLES


We examined the impact on transfer of four sets of independent variables:


• Social background: SES, race-ethnicity, gender, age


• Other precollege personal characteristics: Academic preparation in high school, educational and occupational aspirations


• External demands as the student enters college: Marital and parental status at the time of college entrance, extent and intensity of work


• Experiences during college: Enrollment status, major or college program, degree of academic and social integration into the college


Our focus is on the social background variables. In addition to establishing their total unique effects, we are interested in how much of this effect is indirect—transmitted through or mediated by the other variables in the model.


Social Background


Our analysis of the impact of social background on transfer rates spotlights four background characteristics:


• Socioeconomic status (SES): In the case of NELS:88, this is an index combining the education, occupations, and incomes of parents. The scale is in centiles. Because of major deficiencies in the SES index available in BPS:90,vii we instead use family income, in log form, and the educational level of the parent with the most education. Parental education is measured in the form of five dummy variables: less than high school, high school graduate only, some college, baccalaureate degree recipient, post-baccalaureate training.


• Gender: Binary variable, female 5 1, male 5 0.


• Race and ethnicity: Binary variables for black, Hispanic, and Asian backgrounds that compare the transfer rates for students of these backgrounds with those for white students. No Asian American students were in our BPS:90 community college sample.


• Age: Age at first enrollment in college, coded as four binary variables at the following intervals: 16-18, 19-20, 21-30, and 31+.


Much of the impact of social background on transfer is indirect, operating through other intervening or mediating variables. In this article, we focus on three sets of mediating variables.


Other Precollege Personal Characteristics


• Academic preparation coming out of high school: In the case of NELS:88, we used scores on reading and math tests taken in the 12th grade. The scale is in centiles. Unfortunately, BPS:90 does not have test score data for most respondents, so instead we used four other binary variables: self-rating of academic ability; whether students received a regular diploma; and whether students took remedial math or reading in college. The self-rating of academic ability was based on whether students rated themselves above average in academic ability “compared with the average person of your age.”


• Educational aspirations: For both NELS:88 and BPS:90, we used a binary variable indicating whether the respondent aspired to a baccalaureate degree or higher.


• Occupational aspirations: For NELS:88, we used two variable indicating whether the respondent aspired to a professional or managerial occupation or to a lower white-collar or skilled blue-collar job. The comparison group is those aspiring to unskilled or semiskilled blue-collar jobs. Unfortunately, because a comparable question was not asked in the BPS:90 survey, we could not construct a comparable set of variables for BPS:90 and therefore relied on a single binary variable. However, as presently coded, the highest levels of occupational expectations in both data sets are roughly analogous.


External Demands as Students Enter College


Students vary considerably regarding the extent of outside demands—including family obligations and work—that they encounter as they enter college. We hypothesize that the lower these demands, the more likely students are to transfer, either because they are less likely to drop out of higher education without a credential or because they are less likely to immediately work and not pursue a baccalaureate degree if they receive a subbaccalaureate credential.


• Marital status: For both NELS:88 and BPS:90, we used whether the respondent was single, never married, at the time of college entrance. Married, separated, and divorced people were put in the excluded category. We are hypothesizing that single never-married people have fewer family obligations that may impede prolonged education or deter moving to a four-year college.


• Parental status: For both NELS:88 and BPS:90, this is defined as whether the respondent was childless at the time of college entrance. Respondents who had a child prior to college entrance were put in the excluded category.


• Work involvement during college: We hypothesize that students are more likely to transfer if they have fewer work obligations during college, either because those obligations reflect income demands incompatible with staying in college or because they provide a competing alternative to obtaining a baccalaureate degree. In both NELS:88 and BPS:90, we used a measure of work intensity that indexes the average number of hours a student worked during those weeks that he or she was working. It takes the form of three dummy variables: no work, working 20 hours or less, and working between 21 and 39 hours (NOWORK, HRSLTPT, HRSPT). The excluded category is working 40 hours or more (HRSFT).


Experiences During College


It is not just the characteristics and experiences that students bring to college that may affect transfer but also their experiences during college. Hence, we looked at a variety of collegiate experiences. One that many observers have pointed to is the major or college program that students choose. It has been argued that students majoring in vocational subjects are considerably less likely to transfer to four-year colleges (Brint and Karabel 1989; Dougherty 1994). Moreover, because dropout interferes with transfer, we examined several different variables that have been found to affect college persistence. In particular, we focused on full-time enrollment and degree of academic and social integration into the college. The data on these within-college experiences are considerably richer in BPS:90 than NELS:88, so we rely more on the former data set at this point.


Unfortunately, we cannot capture with NELS:88 or BPS:90 data a number of other variables that have been identified as having a considerable impact on the likelihood of transfer. These include the extent to which a student’s community college is highly vocationalized, its degree of commitment to transfer, whether students are accepted by four-year colleges into their preferred programs and campuses, how many credits are accepted for transfer, and how much financial aid would-be transfer students receive from four-year colleges (Dougherty 1994, 2002).


Enrollment status


We hypothesize that students are more likely to persist in college and therefore are more likely to transfer the closer to full-time their student status. Full-time enrollment both betokens fewer external demands and perhaps shows a stronger commitment to college, but it also strengthens students’ academic and social integration by making students more available to the academic and social influence of faculty and fellow students. In the analysis, our enrollment status variable is dichotomous, indicating whether a student was enrolled full time.


Major


We hypothesize that students’ majors or programs in the community colleges will affect the likelihood of transfer through differential exposure to faculty and staff who urge and facilitate transfer to four-year colleges and by shaping students’ degrees of academic and social integration. In the case of NELS:88, we coded students’ self-reported major as academic, vocational, or undecided,viii with undecided being the reference category. Unfortunately, because of peculiarities in the BPS:90 variables, we could not isolate those who were undecided. Hence, we simply coded students as academic or vocational, with the latter serving as the reference category.


Academic integration


Drawing on the work of Tinto (1993) and Braxton (2000), we used several measures of a student’s integration into the academic life of his or her community college during the first year of college. These include the student’s GPA (as a measure of successful integration) and several measures of conditions facilitative of a commitment to the academic life: whether students talked to their academic advisors (RTALK-ADV), talked about academic matters with faculty outside of class (RTALKFAC), attended career related lectures (RLECTUR1), and had academic contact with students in the form of participation in study groups (RSTDYGRP). In all cases, these variables were coded in binary form—that is, whether students have had the experiences in question. These measures are available only in BPS:90.


Social integration


We used several different measures of students’ social integration into the social life of their community colleges during the first year. These measures include whether students had informal contact with faculty outside of class (CONTACT), participated in school clubs (CLUBS), had campus friendship ties as marked by going places with friends from school (GOPLACES), and used student assistance programs on campus, such as counseling, remediation, and health programs (CENTERS). These variables, all coded in binary form, are available only in BPS:90.


See Table 1 for the means and standard deviations of these and other variables used in the analysis.


We estimated the impact of these characteristics on transfer via a set of logistic regressions. This set of regressions is described in Table 2. As can be seen, the baseline equation—Model 0—regresses transfer status on each of the four social background variables by themselves. In Model 1, we then put in all four social background variables together to give us the total unique effect of each background variable. In succeeding equations, we added various possible mediating variables, thus allowing us to explore the indirect effects of social background on transfer.


These potential mediating variables were added in an order dictated by our judgment regarding where they fall in a causal sequence connecting social background with transfer. If the addition of one of these potential variables substantially reduces the coefficient for a background variable, we know that new variable is carrying part of the influence on transfer of that background variable.


REGRESSION METHOD


Because our dependent variable is binary, we estimated the impact of these variables on transfer by means of logistic regression. The logistic model is


[39_12332.htm_g/00001.jpg]


where y is the outcome variable, transfer to a four-year college, and X is a vector of social background and educational characteristics. Unlike ordinary least squares (OLS) regression, the coefficient of a logistic regression cannot be interpreted as the probability of attaining a positive outcome—that is, the marginal effect on the dependent variable from a one-unit increase in the independent variables. Thus, in logistic regression, a separate calculation is made of the marginal effect. It is expressed as


[39_12332.htm_g/00002.jpg]


and is usually reported at the mean values of the explanatory variables. This statistic is shown along with the regression coefficient, allowing a more straightforward interpretation of the impacts of the independent variables.


FINDINGS


THE IMPACT OF SOCIAL BACKGROUND


Our analysis of transfer in the 1990s using NELS:88 and BPS:90 arrived at findings about the impact of social background on transfer that both converge with and diverge from studies of transfer in the 1970s and 1980s, using the National Longitudinal Survey of the High School Class of 1972 (Velez and Javalgi 1987) and High School and Beyond (Cabrera, LaNasa, and Burkum 2001; Lee and Frank 1990). In particular, we found, as did earlier studies, a very strongly significant impact of parental SES. We also found a powerful impact of age, a variable that the earlier studies could not address because of the nature of their data sets. However, unlike earlier studies, we did not find a statistically significant impact of gender and race-ethnicity on transfer rates.


[39_12332.htm_g/00003.jpg]
click to enlarge


[39_12332.htm_g/00004.jpg]
click to enlarge


[39_12332.htm_g/00005.jpg]
click to enlarge


[39_12332.htm_g/00006.jpg]
click to enlarge


The Impact of Socioeconomic Status


Our analysis of NELS:88 and BPS:90 found that in the 1990s, as in earlier decades, the SES of the parents of students was strongly and significantly associated with whether those students transferred to four-year colleges. As can be seen in Models 0 and 1 in Tables 3 and 4, SES has a sizable and statistically significant impact on transfer rates, both before and after controlling for gender, race-ethnicity, and age. For example, in the NELS:88 analysis—comparing students who are in the top and bottom 10% in SES (87th and 16th percentiles, respectively) but of the same race, gender, and age—the gap in transfer rate is 45 percentage points (55% vs. 10%).


The Impact of Age


A major benefit of analyzing the Beginning Postsecondary Students Longitudinal Study (BPS:90) is that it allowed us to examine the impact of age. Unlike NELS:88 and its predecessors (HS&B and NLS-72), BPS:90 is not built around the evolving experiences of a particular age cohort, whether 8th graders, 10th graders, or 12th graders. Rather, BPS:90 focuses on all students entering college for the first time in a given year, regardless of age.ix Hence, it allows us to analyze students who are first entering college at much more advanced years than the traditional college-age students analyzed by previous studies of transfer. Analyzing BPS:90, we found that age does have a major impact on transfer. Controlling for other background characteristics, students who are older than 18 when they enter community college are significantly less likely to transfer, with the size of this negative age effect increasing as students get older. For example, students entering community college between ages 21 and 30 are 15% less likely to transfer, and students aged 31 and older are 20% less likely to transfer than are students entering college below age 19 (see Model 1 in Table 4).


This finding is of great importance given the growing number of students who are first entering college well after they leave high school. Although community colleges have made major efforts to facilitate the success of older students, the fact remains that they are much less likely to transfer than are younger students. This begs the question of why; this question will be addressed when we turn to analyzing the factors that transmit the influence of the social background variables on transfer.


The Impacts of Race-Ethnicity and Gender


Unlike studies of transfer in the 1970s and 1980s (Lee and Frank 1990, 190; Velez and Javalgi 1987, 86), we did not find that race-ethnicity has a statistically significant impact on transfer rates in the 1990s. To be sure, compared with whites, blacks and Hispanics are less likely to transfer, and Asians are more likely to transfer. For example, in NELS:88, the transfer rates for blacks and Hispanics are, respectively, 13 and 5 percentage points lower than for whites, while that for Asians is 13 points higher (see Model 0 in Table 3).


[39_12332.htm_g/00007.jpg]
click to enlarge


[39_12332.htm_g/00008.jpg]
click to enlarge


[39_12332.htm_g/00009.jpg]
click to enlarge


[39_12332.htm_g/00010.jpg]
click to enlarge


[39_12332.htm_g/00011.jpg]
click to enlarge


However, unlike the case for data from the 1970s and 1980s, these differences are not statistically significant. Regardless of whether other background variables are controlled, blacks, Hispanics, and Asians do not demonstrate statistically significant differences in transfer rates compared with whites (see Models 0 and 1 in Tables 3 and 4). However, there is an important caveat to this finding in the case of blacks. As we note next, if we compare blacks and whites with similar educational aspirations (and high school academic preparation), the black-white gap in transfer rates grows sharply, becoming statistically significant in the BPS:90 analysis.


In the case of gender, earlier studies of transfer using national data sets found that males had a significantly higher transfer rate than females (Lee and Frank 1990, 184, 190; Velez and Javalgi 1987, 85, 88). For example, analyzing NLS-72, Velez and Javalgi found that in the 1970s, males were 18% more likely to transfer than females, even after controlling for differences in social background, academic preparation, and aspirations. However, using data from the 1990s, although we did find that women are still less likely to transfer (a gap of six percentage points), this gender gap is much smaller than before and no longer statistically significant. Gender is no longer strongly associated with transfer, whether before or after controlling for SES, race-ethnicity, and age (see Models 0 and 1 in Tables 3 and 4.)


That we diverge from previous studies in not finding statistically significant impacts of race and gender on transfer rates could be attributed to peculiarities of the data that we used. However, it is noteworthy that the lack of significant impact showed up in both data sets that we analyzed. Moreover, it is important to note that the impact of race and gender on transfer was weaker in data from the 1980s (Lee and Frank 1990) compared with the 1970s data (Velez and Javalgi 1987). This suggests that a major temporal shift has occurred in the impact of race and gender to the point that they have ceased to play a major role in affecting the magnitude of transfer. This temporal shift could be due to the major efforts in the 1980s and 1990s to reduce race and gender gaps in college access and success. However, because those efforts were also directed toward reducing the social class gap in transfer, it is puzzling that the class gap remains large and statistically significant.x


HOW ARE SOCIAL BACKGROUND EFFECTS TRANSMITTED?


It is not enough to document that social class and age have a significant impact on transfer. It is important to determine how that impact is exerted. To that end, we look at three sets of possible mediating variables.


• Other precollege personal characteristics: Academic preparation in high school, educational and occupational aspirations


• External demands as students enter college: Marital and parental status at the time of college entrance, and extent and intensity of work


• Experiences during college: Enrollment status, major or college program, degree of academic and social integration into the college


As we show next, we found that these variables, with the notable exception of most of the academic and social integration variables, do substantially mediate the impact of social background on transfer.


Precollege Academic Traits


We examined three sets of precollege characteristics that might mediate the impact of social background on transferring: academic preparation in high school and educational and occupational aspirations before college entrance. Models 2 and 3 in Tables 3 and 4 summarize the impact of adding these variables to the baseline social background equation. A sharp decrease in the coefficient of a background variable as another variable (such as educational aspirations) was added indicates that the new variable plays a major role in transmitting the impact of the background variable on transfer.xi


Academic preparation coming out of high school


Our analysis found that transfer in the 1990s continued the pattern of earlier decades of transfer being strongly affected by degree of academic preparation coming out of high school (Cabrera et al. 2001, 22, 24; Lee and Frank 1990, 190; Velez and Javalgi 1987, 88). Moreover, we also found that academic preparation plays a major role in mediating the impact of social background on transfer.


As mentioned, NELS:88 provided us with the best measures of academic preparation coming out of high school. We found that a student’s score on the NELS:88 12th-grade math test has a strong impact on transfer likelihood, though reading test scores do not. In the case of BPS:90, test scores were not available for most respondents, so we used four variables to measure academic preparation: self-rating of academic ability; whether a student received a regular diploma; and whether a student took remedial math or reading in college. We found that a self-rating as above average in academic ability has a strongly significant positive impact on transfer. Surprisingly, however, we also found that taking remedial math increases transfer likelihood. Two possible causes could be at work. Those taking remedial math may benefit from skill improvement and therefore transfer potential. Alternatively, those taking remedial math—because a considerable element of self-selection is involved (Perin and Charron, forthcoming)—may also be those who are more motivated to begin with to achieve greater academic success and thus transfer.


The addition of academic preparation coming out of high school sharply reduces the impact of the social background variables, indicating that a considerable part of the impact of social background on transfer is indirect, operating through differences in precollege academic preparation. For example, in the NELS:88 analysis, when high school test scores are added to the social background variables, the coefficient of SES drops 15%, that of Black background drops 80%, and that of Hispanic background drops 76% (see Models 1 and 2 in Table 3). Clearly, a considerable part of the impact of class and racial-ethnic background on transfer is due to differences in academic preparation coming out of high school.


However, differences in high school academic preparation seem to play little role in accounting for the impact of age on transfer. In the BPS:90 analysis, we found minimal drops in the effect of the age variables when we controlled for the four high school preparation variables (see Model 2 in Table 4). We should not put undue weight on this finding given that the measures of academic preparation that we used in the BPS:90 analysis are less than ideal. For example, the four academic preparation variables in BPS:90 mediate much less of the influence of the SES variables (family income and parental education) on transfer than do the test score variables in NELS:88. Still, it is striking that the addition of the academic preparation variables in BPS:90 causes a much smaller drop in the coefficients for age than for the BPS:90 SES and race variables.


Educational aspirations


Educational aspirations continued in the 1990s to have the same strong impact on transfer that they did in the 1970s and 1980s (Cabrera et al. 2001, 22, 24; Lee and Frank 1990, 190; Velez and Javalgi 1987, 190). In both NELS:88 and BPS:90, educational aspirations have a large and statistically significant impact even when social background and high school academic preparation are controlled (see Model 3 in Tables 3 and 4).


Differences in educational aspirations play a key role in mediating the impact of social background on transfer. For example, in NELS:88, the addition of educational aspirations drops the coefficient for SES by another 24%. And in BPS:90, the coefficient for the oldest age group (31 and older) drops 18%, though the other coefficients for the other age variables are much less affected (see Model 3 in Tables 3 and 4).


But if class and age inequality in transfer is in part due to associated differences in educational aspirations, the opposite is the case with race. Controlling for educational aspirations does not reduce the coefficient for the black background variable but sharply increases it. In fact, in the BPS:90 analysis, the black background variable becomes statistically significant. This suggests that educational aspirations work as a suppressor variable for blacks. That is, the negative impact of black background on transfer is kept from being as big as it might otherwise be because blacks have higher educational aspirations than whites of similar social class, gender, and age.xii But once that aspirational advantage is nullified by controlling for educational aspirations, the negative impact on transfer of being black becomes considerably larger.


Occupational aspirations


Occupational aspirations do not have a statistically significant impact on transfer in either NELS:88 or BPS:90. As a result, occupational aspirations mediate very little of the impact of SES, race-ethnicity, or age on transfer.


External Family and Work Demands as Students Enter College


We have hypothesized that students are more likely to transfer if they encounter fewer external demands, whether from work or family, as they enter college. In Model 4, we add three sets of variables: marital status, parental status, and degree of involvement with work. To ease interpretation, we measured the variables in terms of the hypothesized absence of external demands—that is, we are comparing single to ever-married people; those without children to those with children; and those who do not work or work part time to those who work full time.


Being single does not have a significant impact on transfer in either NELS:88 and BPS:90. However, being childless at time of college entry has a substantially large, positive effect in both BPS:90 and NELS:88 (see Model 4 in Tables 3 and 4).


Not working or working fewer than 40 hours a week proves to have a positive impact on transfer in NELS:88 and BPS:90, but this impact is statistically significant only in NELS:88 (see Model 4 in Tables 3 and 4). Curiously, in NELS:88, though not BPS:90, working fewer than 20 hours a week has a more positive impact on transfer than not working at all.


The family status and work intensity variables substantially mediate the impact of age on transfer. In BPS:90, controlling for these variables reduces the coefficient of the age 21 to 30 variable by 78% and of the 31 years and older variable by 90%. However, the picture is less clear with regard to SES. Controlling for the external demands variables reduces the impact of family income in BPS:90 by 67%, and it substantially reduces two of the three dummy variables pertaining to students with parents with more than a high school education. But in NELS:88, controlling for the external demands variables actually increases the coefficient for SES.


Experiences During College


Our three sets of measures of collegiate experiences proved to have quite mixed effects. Enrolling in an academic major proved to have a positive and significant impact on transfer. In the case of enrollment status, being enrolled full-time increased the probability that a student would transfer, with the impact being statistically significant in NELS:88 but not BPS:90. Finally, almost all the measures of academic and social integration in the BPS:90 analysis proved statistically insignificant.


Enrollment status


As expected, the closer to full time a student’s enrollment, the more likely he or she is to transfer. This is clearly seen in the NELS:88 analysis, in which the size of the effect is large and statistically significant. In the BPS:90 analysis, however, the magnitude of the effect is roughly half that of NELS:88, and the variable just missed being significant at the 5% level (see Model 5 in Tables 3 and 4).


Enrollment status mediated a significant portion of the impact of black background (in NELS:88 but not BPS:90) and being age 21 and older in BPS:90, as evidenced by the sharp drops in the coefficients for these variables when enrollment status was controlled (see Models 5 and 4 in Tables 3 and 4).


Major


Enrolling in an academic program had a positive and significant impact (at the 0.05 level) in both NELS:88 and BPS:90. In NELS:88, the comparison group was those students who had not declared a major in the first year of college. In BPS:90, the comparison was occupational students because we were not able to isolate those students who did not have a major (see Model 6 in Tables 3 and 4).


Differences in college major appear to play a significant role in transmitting the impact of race and age on transfer. Controlling for students’ college major sharply reduces the negative coefficients for those age 19 and older in BPS:90. However, choice of college major does not seem to be a significant channel through which socioeconomic background affects transfer. Controlling for major actually slightly increases the SES coefficient in NELS:88 and has inconsistent impacts on the various SES variables in BPS:90.


That enrolling in an academic program has a positive impact on transfer is important because the argument has been made that enrolling in an occupational major is no barrier to transfer. This is true if one means that it is not an absolute barrier, but it still remains an important relative barrier, reducing the probability that students will transfer.


Academic and Social Integration


The academic and social integration variables proved to have very little impact on transfer between community colleges and four-year colleges in BPS:90. (These variables were not available in NELS.) Only one, involvement in a study group with other students, had a statistically significant impact. Moreover, several of the variables, such as academic contact with an advisor and with faculty members, had signs opposite to what we expected. Their impact on transfer was negative when we had expected positive impacts (see Model 7 in Table 4).


The weak effect and sometimes unexpected signs of the academic and social integration variables may be attributable to their positive impact on two different sets of students: students who go on to transfer and students who persist long enough to acquire a subbaccalaureate degree but then do not transfer. Greater contact with a faculty member or advisor may increase the persistence of both groups but promote the transfer of only the first group. Such a spread of effects across the transfer-nontransfer divide would considerably weaken the impact of the integration variables on transfer likelihood.xiii


Whatever the case, controlling for the academic and social integration variables did not affect the coefficients for the SES, race, and age variables in any clear pattern.


SUMMARY AND CONCLUSIONS


Our analysis of the National Education Longitudinal Study of the 8th Grade (NELS: 88) and the Beginning Postsecondary Students Longitudinal Study (BPS:90) arrives at findings that both validate and break with previous studies. As with earlier studies, we found that the likelihood of transfer is strongly affected by parental SES. Students whose parents have higher incomes, more advanced education, and more prestigious and remunerative jobs have a very large and statistically significant advantage in transfer over less socioeconomically favored students. A significant portion of class advantage is transmitted through differences between more or less well-off students in precollege academic preparation and educational aspirations. However, even after controlling for these differences—and differences in occupational aspirations, marital and parental status at time of college entry, enrollment status, and college major—most of the impact of socioeconomic background on transfer is still left unexplained (about 85% in NELS:88 but less in BPS:90).


With regard to race and ethnicity, we found that in the 1990s, as in the 1970s and 1980s, blacks and Hispanics had lower transfer rates than did whites and Asians. However, our study breaks with earlier studies (Lee and Frank 1990, 190; Velez and Javalgi 1987, 86) in finding that this racial-ethnic disparity is not statistically significant, particularly if we compare black and Latino students with whites of similar gender, age, and above all, SES. But there is an important caveat to be made: when we control for educational aspirations, the black-white gap in transfer rates widens considerably, becoming statistically significant in the case of BPS:90, though not of NELS:88. Blacks have higher educational aspirations than whites of the same socioeconomic background, which serves to mitigate the negative impact of being black on transfer, keeping the black disadvantage smaller than it would otherwise be.


Another key difference from studies of transfer in the 1970s and 1980s is that we did not find powerful effects of gender on transfer rates in the 1990s. Although women are slightly less likely to transfer, that difference is no longer statistically significant both before and after controlling for other background differences.


These differences in findings about the impacts of race and gender between our study of transfer in the 1990s and earlier studies of transfer in the 1980s and 1970s could be an artifact of differences in data quality or analytic technique, but we doubt this. The patterns that we found hold across two different data sets: NELS:88 and BPS:90. Moreover, transfer in the 1980s appears to have been less affected by gender and race-ethnicity than was transfer in the 1970s (Cabrera et al. 2001; Lee and Frank 1990; Velez and Javalgi 1987). Moreover, a host of programs were initiated in the 1980s to reduce class and race differences in transfer (Dougherty, 1994, 254-60; Ford Foundation 1988). Hence, we believe that the absence of statistically significant gender and race/ethnic effects that we found in Models 0 and 1 of our analyses most probably reflects temporal changes in the impact on transfer of these social background characteristics rather than artifacts of data quality or analysis technique.


Another key difference between our findings and those from studies of transfer patterns in the 1970s and 1980s is that we have been able to analyze the impact of age, a variable that earlier studies were unable to study because of the age restrictions built into the cohort structure of the NLS-72 and HS&B longitudinal surveys (Cabrera et al. 2001; Lee and Frank 1990; Velez and Javalgi 1987). We found that age at college entry has a very powerful impact on transfer. Older college entrants, especially if they are over 30 years of age, are much less likely to transfer than students who enter college right out of high school. Differences in academic preparation among different age groups seem to explain little of the age gap in transfer.xiv This gap is explained mostly by differences by age in educational aspirations (particularly for those 21 and older at college entrance), external demands (particularly having

children),xv enrollment status, and college major.


That differences in social background, particularly SES, powerfully affect whether students transfer is of great concern, especially in light of the increasing role being given to community colleges as gateways to the baccalaureate. This inequality in transfer argues that it is not enough to get baccalaureate aspirants to the community college’s open door. It is also necessary to ensure equality of opportunity for community college success (Lumina Foundation 2005). To reduce or even eliminate differences in transfer rate according to social background, we need to investigate the precise channels by which social background affects transfer processes.


We have seen that differences in high school academic preparation,xvi educational aspirations,xvii having children, and college major play an important role in mediating class, age, and racial-ethnic differences in transfer. Still, we are far from exhausting our analysis of these mediating processes. We need to learn more about how precisely having children reduces the likelihood of transfer. Moreover, even with all the mediating variables that we have analyzed, most of the impact of social class on transfer is left unexplained in NELS:88. To fill in this uncharted territory, future analyses need to examine the impact of institutional variables that we have not been able to capture in our analysis of the NELS:88 and BPS:90 data—for example, the extent to which a student’s community college is committed both in word and deed to transfer; the readiness with which students are accepted by four-year colleges into their preferred programs and campuses; how many credits are accepted for transfer; and how much financial aid would-be transfer students receive from four-year colleges (Dougherty, 1994, 2002).


Although much analysis remains to be done before we fully understand why inequality of opportunity in transfer exists between community colleges and four-year colleges, we already have leads that policy makers should pursue. Clearly, we need to continue to improve high school preparation not only to ensure equality of access to college but also equality of opportunity within college (Dougherty 1996; Gladieux and Swail 2000; Rosenbaum 2001; Rothstein 2004). We also need to reduce the gap in aspirations at matriculation between community college entrants who are high and low in SES and younger and older in age (McDonough 1997; Terenzini, Cabrera, and Bernal 2001). In addition, we need to find ways to reduce the transfer gap between occupational majors and academic majors. Key in this regard is finding ways of providing better transfer counseling to occupational majors and making it easier for them to secure admission, financial aid, and credit acceptance at four-year colleges (Dougherty 2002, 326-27).


APPENDIX


DESCRIPTIONS OF DATA SETS USED


NATIONAL EDUCATION LONGITUDINAL STUDY OF THE 8TH GRADE (NELS:88)


NELS:88/2000 began with a nationally representative sample of eighth graders in 1988 and followed them up in 1990, 1992, 1994, and 2000. For the baseline data, questions were asked of students and their parents, teachers, and high school principals, and data such as high school transcripts were collected from school records. All dropouts were retained in the study. The 1992 follow-up occurred when most sample members were in the second term of their senior year. At that point, the NELS:88 sample was freshened by adding more respondents in order to represent the high school class of 1992, allowing trend comparisons to the high school classes of 1972 and 1980 that were studied in the National Longitudinal Survey of the High School Class of 1972 (NLS-72) and High School and Beyond (HS&B). The postsecondary transcript-based version of NELS:88/2000 allowed us to look more accurately at which colleges students entered, the college courses and program choices that students made, and transfer rates among colleges. Overall, there were more than 12,000 observations in the NELS:88 sample, but the size of the public two-year college sample is 2,660.


BEGINNING POSTSECONDARY STUDENTS SURVEY


Based on the 1990 National Postsecondary Student Aid Study (NPSAS:90), the Beginning Postsecondary Students Longitudinal Study (BPS:90) consisted of first-time beginners (FTBs) who enrolled in a postsecondary institution at any time between July 1, 1989, and June 30, 1990. The NPSAS:90 design involved a multistage probability sample of students enrolled in postsecondary institutions. To be eligible for participation in NPSAS:90, a postsecondary institution was required to satisfy several conditions. During the 1989-1990 academic year, the institution must have offered an educational program for persons who have completed secondary education; an academically, occupationally, or vocationally oriented program of study; access to persons other than those employed by the institution; more than just correspondence courses; and at least one program lasting at least three months or 300 contact hours. A student was NPSAS:90-eligible if he or she was enrolled in an eligible institution during the 1989-1990 academic year for one or more of the following purposes: taking course(s) for credit; participating in a degree or formal award program of at least three months’ duration; and taking part in a vocationally specific program of at least three months’ duration. Students were excluded regardless of whether they satisfied the above conditions if they enrolled solely in a high school program at an eligible postsecondary institution, enrolled only in correspondence courses or programs of less than three months’ duration, or were taking courses only for remedial or a vocational purposes without receiving credit.


Initially, the BPS:90 began with around 7,200 students of all ages entering college for the first time in fall 1989. These students were then followed up in spring 1994. Complete demographic and enrollment information up to and including the 1994 follow-up is available for 83% of the sample. Data restrictions, such as institution miscodings, reduce the sample to roughly 5,600 observations. Our analysis is focused on those students who entered public two-year colleges in fall 1989. As our measure of college entered, we used the colleges that students identified as their primary institution. This yielded a sample of 653 first-time community college students.


We would like to thank Thomas Bailey, Lisa Rothman, and Timothy Leinbach of the Community College Research Center for their research and administrative support. Earlier versions of this article were delivered at the 2002 annual meeting of the Association for the Study of Higher Education and the 2003 annual meeting of the American Educational Research Association. We wish to thank Madhabi Chatterji, David Karen, Frankie Laanan, Alexander McCormick, and Philo Washburn for comments and methodological advice.


References


Berkner, Lutz, Shirley Ho, and Emily Forrest Cataldi. 2002. Descriptive summary of 1995-96 Beginning Postsecondary Students: Six years later. Statistical analysis report. NCES 2003-151. Washington, DC: National Center for Education Statistics.


Bradburn, Ellen M., David G. Hurst, and Samuel Peng. 2001. Community college transfer rates to 4-year institutions using alternative definitions of transfer. Education Statistics Quarterly 3 (3): 119-25.


Braxton John M., ed. 2000. Reworking the student departure puzzle. Nashville, TN: Vanderbilt University Press.


Brint, Steven G., and Jerome B. Karabel. 1989. The diverted dream: Community colleges and the promise of educational opportunity in America, 1900-1985. New York: Oxford University Press.


Cabrera, Alberto F., Steven M. LaNasa, and Kurt R. Burkum. 2001. Pathways to a four-year degree: The higher education story of one generation. University Park, PA: Center for the Study of Higher Education, Pennsylvania State University.


Callan, Patrick. 2003. A different kind of recession. National Crosstalk. http://www.highereducation.org/crosstalk/ct0103/front.shtml (accessed November 21, 2005).


Cohen, Arthur M., and Florence B. Brawer. 2003. The American community college. 4th ed. San Francisco: Jossey-Bass.


Dougherty, Kevin J. 1994. The contradictory college: The conflicting origins, outcomes, and futures of the community college. Albany: State University of New York Press.


Dougherty, Kevin J. 1996. Opportunity to learn standards: A sociological critique. Special issue. Sociology of Education 68: 40-65.


Dougherty, Kevin J. 2002. The evolving role of the community college: Policy issues and research questions. In Higher education: Handbook of theory and research, ed. John Smart and William G. Tierney, 295-348. Dordrecht, Netherlands: Kluwer.


Dougherty, Kevin J., and Marianne Bakia. 2000. Community colleges and contract training. Teachers College Record 102: 197-243.


Ford Foundation. 1988. An assessment of Urban Community Colleges Transfer Opportunities Program. The Ford Foundation’s Second Stage Transfer Opportunity Awards. Final report. New York: Ford Foundation. ERIC ED 293573.


Gladieux, Lawrence, and W. Scott Swail. Beyond access: Improving the odds of college success. Phi Delta Kappan 81: 688-92.


Green, Patricia, Bernard L. Duggoni, and Steven J. Ingels. 1995. Trends among high school seniors, 1972-1992. National Educational Longitudinal Study of 1988. NCES 95380. Washington, DC: National Center for Education Statistics.


Grubb, W. Norton. 1991. The decline of community college transfer rates: Evidence from National Longitudinal Surveys. Journal of Higher Education 62: 194-217.


Grubb, W. Norton. 2002. Learning and earning in the middle, part I: National studies. Economics of Education Review 21: 299-321.


Hagedorn, Linda Serra, Hye Sun Moon, Scott Cypers, William E. Maxwell, and Jaime Lester. n.d. Transfer between community colleges and four-year colleges: The all-American game. Los Angeles: Transfer and Retention in Urban Community College Students, University of Southern California. http://www.usc.edu/dept/education/truccs/Papers/baseball.pdf (accessed November 21, 2005).


Hebel, Sara. 2000. States without affirmative action focus on community-college transfers. Chronicle of Higher Education 46 (38): A35-37.


Hoachlander, Gary, Anna C. Sikora, and Laura Horn. 2003. Community college students: Goals, academic preparation, and outcomes. NCES 2003-164. Washington, DC: National Center for Education Statistics.


Ignash, Jan, and Barbara Townsend. 2001. Statewide transfer and articulation policies. In Community colleges: Policy in the future context, ed. Barbara Townsend and Susan Twombly, 173-92. Westport, CT: Ablex.


Kane, Thomas J., and Cecilia R. Rouse. 1999. The community college: Educating students at the margin between college and work. Journal of Economic Perspectives 13 (1): 63-84.


Kienzl, Gregory S. 2004. The triple helix of education and earnings: The effect of schooling, work, and pathways on the economic outcomes of community college students. PhD diss., Teachers College, Columbia Univ.


Kojaku, Lawrence K., and Anne-Marie Nunez. 1998. Descriptive summary of 1995-96 Beginning Postsecondary Students. NCES 1999-030. Washington, DC: National Center for Education Statistics.


Lee, Valerie, and Kenneth Frank. 1990. Student characteristics that facilitate the transfer from two-year to four-year colleges. Sociology of Education 63: 178-93.


Lumina Foundation for Higher Education. 2005. Achieving the dream: Colleges count. http://www.luminafoundation.org/grants/atdfaqs.html (accessed November 21, 2005).


Marcotte, David E., Thomas Bailey, Carey Borkoski, and Greg S. Kienzl. 2005. The returns of a community college education: Evidence from the National Education Longitudinal Survey. Educational Evaluation and Policy Analysis 27: 157-75.


McCormick, Alex. 1997. Transfer behavior among Beginning Postsecondary Students 1989-94, NCES 97-266. Washington, DC: National Center for Education Statistics.


McDonough, Patricia. 1997. Choosing colleges: How social class and schools structure opportunity. Albany: State University of New York Press.


McPherson, Michael S., and Morton O. Schapiro. 1998. The student aid game. Princeton, NJ: Princeton University Press.


McPherson, Michael S., and Morton O. Schapiro. 1999. Reinforcing stratification in American higher education: Some disturbing trends. Stanford, CA: National Center for Postsecondary Improvement.


Mercer, Joye. 1992. States turn to community colleges as route to bachelor’s degree as 4-year campuses face tight budgets and overcrowding. Chronicle of Higher Education, May 6.


Mickelson, Roslyn Arlin. 1990. The attitude-achievement paradox among black adolescents. Sociology of Education 63: 44-61.


Pascarella, Ernest T., and Patrick T. Terenzini. 2005. How college affects students, 2nd ed. San Francisco: Jossey-Bass.


Perin, Dolores, and Carolyn Charron. Forthcoming. “Lights just click on every day”: Academic preparedness and remediation in community colleges. In Defending the community college equity agenda, ed. Thomas Bailey and Vanessa Smith Morest. Baltimore: Johns Hopkins University. Press.


Rosenbaum, James E. 2001. Beyond college for all: Career paths for the forgotten half. New York: Russell Sage Foundation.


Rothstein, Richard. 2004. Class and the classroom. American School Board Journal 191 (10): 17-21.


Surette, Brian J. 2001. Transfer from two-year to four-year college: An analysis of gender differences. Economics of Education Review 20: 151-63.


Terenzini, Patrick T., Alberto F. Cabrera, and Elena M. Bernal. 2001. Swimming against the tide: The poor in American higher education. New York: College Board.


Tinto, Vincent. 1993. Leaving college: Rethinking the causes and cures of student attrition. 2nd ed. Chicago: University of Chicago Press.


United States National Center for Education Statistics. 2002. Digest of education statistics, 2002. Washington, DC: Government Printing Office.


Velez, William, and Rajshekhar G. Javalgi. 1987. Two-year college to four-year college: The likelihood of transfer. American Journal of Education 96: 81-94.


Wellman, Jane V 2002. State policy and community college-baccalaureate transfer. San Jose, CA: National Center for Public Policy and Higher Education.


Young, Jeffrey. 2002. Public-college tuition jumps at highest rate in 10 years. Chronicle of Higher Education, October 22.

Endnotes

i Grubb analyzed the National Longitudinal Survey of the High School Class of 1972 (NLS-72) and the High School and Beyond Survey (HSB). The NLS-72 questioned 19,001 high school seniors nationwide in spring 1972 and followed them up in the fall of 1973, 1974, 1976, 1979, and 1986. The HSB study followed 10,583 high school graduates in 1982, 1984, and 1986 (Green, Duggoni, and Ingels 1995).

ii In 1981, the proportion of college entrants entering community colleges was 44.6% among students with family incomes below $20,000 and 41.4% among those with family incomes between $20,000 and $30,000. For the same groups (using inflation-adjusted constant dollars) in 1998, the percentages were 46.7% and 43.1%, respectively. At the same time, the proportion entering community colleges dropped sharply for all those with family incomes above $30,000. For example, among those with family incomes between $60,000 and $100,000 (in 1981 dollars), the proportion dropped from 31.2% to 25.7% (McPherson and Schapiro, 1999, 22-23).

iii In fiscal years 2002 and 2003, tuition at public four-year colleges rose by 10%, while state need-based aid rose only 7% (Callan 2003, 3A; Young 2002).

iv For example, in the BPS:96, only 37% of the community college entrants said that their primary purpose was transfer to a four-year college (Hoachlander, Sikora, and Horn 2003, 13). At the same time, it is important to keep in mind that even students who are focused on acquiring job skills or subbaccalaureate credentials may still harbor the desire to eventually acquire a bachelor’s degree. Among BPS:96 students who had a primary intention of receiving a subbaccalaureate degree, 19% also planned to transfer (15).

v But even if there is no difference by social background in transfer rates, there would still be the problem that baccalaureate aspirants who enter community colleges are less likely to eventually receive bachelor’s degrees than comparable students (in background, high school record, and educational and occupational aspirations) entering four-year colleges. For more on this, see Dougherty (1994, 2002).

vi For reviews of the literature on the extent and determinants of transfer, see Dougherty (1994, 2002) and Pascarella and Terenzini (2005).

vii The socioeconomic variable in BPS:90 includes a measure for family income that takes a radically different form according to the student’s dependency status. For students who are financially dependent on their parents, the family income value plugged in is that of the parents. But for students who are financially independent, the family income value plugged in is their own. As a result of this specification, the SES variable that emerges behaves quite oddly, showing no impact on transfer rates.

viii One of the benefits of NELS:88 is the availability of students’ postsecondary transcripts, which are supposedly less prone to recall and other reporting errors inherent in self-reported information. However, there is one potential problem with using transcripts, and this drawback is magnified for community college students. Students with abbreviated or incoherent enrollment spells or who attended college but did not attain a degree, which regrettably are commonly at public two-year colleges, are coded as "nonmajors" since their program of study could not be determined clearly from transcripts.

ix It should be noted that the BPS:90 data set does exclude students in noncredit remedial and avocational programs—that is, it excludes students who are not taking courses for credit, not enrolled in a degree or formal award program of more than three months’ duration, or not taking an occupational program of more than three months’ duration (Bradburn, Hurst, and Peng 2001, 2).

x One of our reviewers raised the interesting question of whether the disappearance of the gender gap in transfer rates is due to the fall in male transfer as much or more than it is due to a rise in the female transfer rate. Unfortunately, we cannot answer this question without having access to the transfer rates for each gender. We have them from our analysis of the 1990s, but the studies of transfer in the 1970s and 1980s that we have consulted do not provide gender breakdowns. However, if the gender dynamics of transfer over the last 20 years are the same as the gender dynamics of college access (U.S. National Center for Education Statistics 2002, 225), then one could argue that the erasure of the gap has been primarily due to a rise in women’s transfer rates rather than in a decline in men’s. A reviewer also raised the question of whether our null finding was due to the fact that, unlike previous studies, ours also controlled for age, which is correlated with gender. However, this is unlikely to be the cause of our divergence from previous studies; we did not control for age in our analysis of NELS:88 and also found a nonsignificant impact of gender in that analysis.

xi However, a sharp decrease in the coefficient of a variable may not necessarily be significant if, as in the case of gender, the baseline effect was not large to begin with.

xii When we regress educational aspirations in NELS:88 on race (black and Hispanic dummy variables), both variables are positive, though only the black variable is statistically significant. And when we control for SES, gender, and age, the coefficients stay positive and become even larger and statistically significant. This shows the degree to which blacks especially have an aspirational advantage over whites of similar social class and gender. In saying this, we do not want to gainsay the powerful argument that this aspirational advantage lies more in idealistic hopes for education than in realistic plans for securing it (Mickelson 1990). Still, it is noteworthy that controlling for these attitudes strongly changes the impact of race on transfer rates.

xiii We checked to see if the small coefficients and sometimes surprising signs of these variables might be due to a high degree of collinearity. However, the highest correlation between any two academic or social integration variables was 0.39, and most correlations were well below this, with the average being around 0.17.

xiv The tentative tone of our conclusion is due to the measures of academic preparation that we had available in BPS:90 not being as good as we would like and put a limit on how strongly we can say that academic preparation plays little role in accounting for age differences in transfer.

xv This is certainly the case for students in the “21 to 30” and “over 30” age groups. Not surprisingly, having children explains much less of the negative impact on transfer of being age 19 to 20 simply because most students of that age do not have children. But for older students, being a parent is much more common and imposes a myriad of obstacles—presumably of desire, time, energy, money and locational flexibility—that make it difficult to pursue a baccalaureate degree.

xvi Differences in academic preparation coming out of high school play an important role in explaining class and race differences in transfer but seemingly not age differences.

xvii Differences in educational aspirations play an important role in explaining class and age differences in transfer, but race does not appear to have an impact.




Cite This Article as: Teachers College Record Volume 108 Number 3, 2006, p. 452-487
https://www.tcrecord.org ID Number: 12332, Date Accessed: 10/23/2021 7:05:06 PM

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

Related Discussion
 
Post a Comment | Read All

About the Author
  • Kevin Dougherty
    Teachers College, Columbia University
    KEVIN DOUGHERTY is associate professor of higher education and senior research associate, Community College Research Center, Teachers College, Columbia University. He has published widely on the community college’s missions and history, its workforce preparation and economic development activities, educational and economic returns for community college students, and the impact of performance accountability on community colleges. He is now leading a research project on state policy affecting community college access and success for minority and low-income students, with funding from the Lumina Foundation’s Achieving the Dream initiative.
  • Gregory Kienzl
    American Institutes of Research
    GREGORY KIENZL is a research analyst with the American Institutes for Research. His research interests include estimating the economic benefits of postsecondary education for students who follow nontraditional educational pathways and examining the impact of labor markets on the educational achievement and economic outcomes of college students. His paper, “The Returns of a Community College Education: Evidence from the National Education Longitudinal Survey,” is forthcoming in Educational Evaluation and Policy Analysis (summer 2005).
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS