• Introduction
• Data and Measurement
• Results
• Current Results in the Context of Previous Studies
• Conclusion
• References

• share studies and research that are developmental in nature;
• share results of studies that are on the cutting edge of methodological developments; and
• participate in discussions of emerging issues of interest to researchers.

A large proportion of undergraduates attend community colleges (public 2-year institutions) seeking a wide range of services, from a place to experiment with postsecondary education to a structured vocational certificate or associate's degree program (Grubb 1988, 1991). Although the course offerings and degree programs of many community colleges can accommodate diverse student interests and goals, preparing students to transfer to a 4-year college remains a central characteristic of community colleges (Brint and Karabel 1989). This preparation is key to the community college's role in higher education because it affirms the community college's claim to a collegiate, academic identity and to a role in broadening access for those historically excluded from a college education. Moreover, transfer is a component of most community college students' educational aspirations (Grubb 1991, pp. 19596).

Despite, or perhaps because of, the importance of transfer from 2- to 4-year institutions, calculating the percentage of community college students who transfer has proven to be somewhat problematic. At first glance, the transfer rate seems relatively unambiguous: it is the number of students who transfer to a 4-year college divided by the number of potential transfer students. However, the numerator and especially the denominator can both be defined in a number of different ways, each having a significant impact on the transfer estimate. The purpose of the present study is to use nationally representative community college data to examine several ways of defining the population of potential transfer students, the relationship of these definitions to student background characteristics, and the relationship of each definition to the resulting transfer rate. This report consists of three sections. The first section describes the data set used in the analysis and the measurement issues implicated in the study of transfer. The second section presents the selected indicators of the key concepts in the study and the results of the analysis. The report concludes with a discussion of the results in the context of other studies of community college students and transfer.

Although a considerable amount of research has investigated community college transfer rates, many of these analyses have used data that are either limited to a cohort of recent high school graduates, such as the National Education Longitudinal Study of 1988 (NELS:88), or are not nationally representative. This study uses the National Center for Education Statistics (NCES) 1990 Beginning Postsecondary Students Longitudinal Study (BPS:90/94), a nationally representative sample of all students who enrolled in postsecondary education for the first time between July 1, 1989, and June 30, 1990. Follow-up interviews were conducted in spring 1992 and 1994. BPS is particularly appropriate for the study of community college students because it is representative of all beginning postsecondary students, not just recent high school graduates.

The approach of this report is similar to that used in analyses of individual community colleges or districts, particularly that of Spicer and Armstrong (1996). Holding the numerator constant, variously restrictive definitions of the denominator are employed based on the different approaches to specifying the transfer population found in the literature.

This report defines transfer as follows: initial enrollment at a community college followed by subsequent enrollment at any 4-year institution within the 5-year study period. Potential transfer refers to being eligible for transfer or "at risk" of transfer. The broadest definition of potential transfer used in this analysis includes all first-time, beginning community college students, although students only taking courses for which they receive no credit are excluded from the BPS sample. The pool of potential transfer students is then restricted using eight additional definitions of the denominator. They are referred to as "increasingly restrictive" because the total proportion of the sample that is included generally decreases, although the more restrictive groups are not necessarily subsets of the less restrictive groups. These definitions were selected to approximate measures commonly used in previous research, from explicit student goals to behaviors often thought to indicate intent to transfer or commitment to postsecondary education. They are as follows:

1. Expected to complete bachelor's degree or higher;
2. Enrolled in an academic program;
3. Enrolled continuously in 1989–90;
4. Enrolled anytime in academic year 1990–91;
5. Pursuing academic major or taking courses toward a bachelor's degree or both;
6. Enrolled for 12 or more credit hours;
7. Taking courses toward a bachelor's degree in 1989–90; and
8. Pursuing academic major and taking courses toward a bachelor's degree.

The analysis begins with estimating the percentage of the 1989–90 cohort of beginning community college students who meet each of these definitions. The relationship of these criteria to various other student characteristics is then explored. The first issue to be examined is whether the composition of the pool of potential transfer students varies as the definitions become more restrictive. Then, consideration is given to whether different subgroups of students are more or less likely to meet each definition. Finally, a transfer rate is calculated for each definition of potential transfer students, and the relationships of these definitions to transfer are explored.

Overall, 71 percent of 1989­90 beginning community college students responded that they anticipated earning a bachelor's degree or higher when asked, "What is the highest level of education you ever expect to complete?" (figure A). Also, the majority of students were enrolled in an academic program, enrolled continuously in 1989–90, and enrolled during the 1990–91 academic year. Less than half of the students met the other definitions, with 11 percent of the students both having an academic major and taking courses leading toward a bachelor's degree.

Restricting the group of potential transfer students according to these definitions may alter the composition of the group, since these educational characteristics are themselves associated with other background variables (Berkner, Cuccaro-Alamin, and McCormick 1996). Across increasingly restrictive definitions of potential transfer, the percentage of the pool that was in the highest socioeconomic status (SES) quartile increased from 30 percent of all beginning community college students to 51 percent of beginning community college students with an academic major and taking courses leading toward a bachelor's degree. Furthermore, none of the students in this particular sample who met the most restrictive definition of potential transfer were Black, compared to 6 to 10 percent Black students in each of the other potential transfer groups.^*  In general, restricting the pool of potential transfer students systematically altered the composition of the group to include more traditional students (younger, dependent students who do not work full time).

Figure A.—Among 1989-90 beginning postsecondary students enrolled at public 2-year institutions, the percentage of the initial cohort meeting each definition of potential transfer

In addition to examining how the composition of the population of potential transfer students changed as the definitions became more restrictive, the report also compares the likelihood of meeting each definition of potential transfer across various subgroups of students. For example, students 22 years or older were generally less likely than younger students to meet the various definitions of potential transfer. In general, the higher the SES, the higher the percentage of students who met the criteria for each specification. Students who reported taking at least 1 credit hour of remedial mathematics instruction during 1989–90 were generally about as likely to fit each definition as students who did not take any remedial mathematics instruction. Students who were enrolled full time were generally more likely to meet the various specifications than those who were enrolled less than full time.

Figure B shows estimated transfer rates for all community college entrants and for the eight increasingly restrictive definitions of potential transfer arranged in order. The results show that, in general, the transfer rate increased for more restrictive definitions. The lowest rate of ever enrolling in a 4-year institution by spring 1994, 25 percent, was found for all 1989–90 beginning community college students, compared to 52 percent for students meeting the most restrictive definition (both pursuing an academic major during 1989–90 and taking courses leading toward a bachelor's degree). That is, the transfer rate for the most restrictive definition was at least twice the rate for all students. However, figure C demonstrates that the percentage of actual transfer students meeting the criteria for inclusion in the denominator declined significantly as the definitions of potential transfer became more restrictive. In other words, attempts to include only those students most likely to transfer actually exclude a sizable proportion of students who transfer anyway, without meeting those criteria. For example, four out of five actual transfer students did not meet the most restrictive definition considered.

Additional exploratory analyses examined the percentage of beginning community college students who ever transferred to a 4-year institution for each potential transfer definition, by selected student background characteristics. In several cases, relationships of student characteristics to transfer rates generally persisted even when the analyses were restricted to students meeting the various potential transfer definitions. For example, in general, the older the age group, the lower the percentage of students who transferred, regardless of the definition of potential transfer that was used. Also, regardless of the potential transfer definition used, higher SES was generally associated with a higher transfer rate.

Figure B.—Among 1989-90 beginning postsecondary students enrolled at public 2-year institutions, the percentage who transferred to 4-year institutions by spring 1994, for various definitions of potential transfer

The results of this study can be placed in the context of the literature about two questions: what percentage of students in community colleges have educational expectations that include a bachelor's degree, and what is the transfer rate for community college students? This contextual information is not intended to constitute a statistical comparison across studies. In general, however, BPS estimates of the percentage of beginning community college students whose expectations included a bachelor's degree or higher, as well as the percentage of students who transferred to a 4-year institution, are higher than estimates based on other data sets. Dougherty (1987, 1992), for example, reviewed several studies and concluded that 30 to 40 percent of all community college entrants aspire to a bachelor's degree, while the present study found that 71 percent of community college students in BPS expect to complete a bachelor's degree or higher. Similarly, while the overall transfer rate found in this study is comparable to the average estimate of 22 percent found by the Transfer Assembly project (Cohen and Sanchez 1997), both the numerator and denominator of the Transfer Assembly project are more restrictive. Although BPS data do not facilitate use of such a restrictive definition, an approximation of it using BPS data yielded a transfer rate of 33 percent, somewhat higher than the rate resulting from the Transfer Assembly project.

However, any comparisons between the estimates presented in this report and those from other studies must be considered in light of differences in how the overall population of community college students is defined across studies. For example, BPS excluded students taking courses only for remedial or avocational purposes without receiving credit, while other estimates may include these students. As a result, the typical amount of remediation for students in BPS may underestimate, or otherwise differ from, the amount of remediation found among community college students in generala factor that might be associated with transfer to a 4-year institution. Furthermore, this report focuses only on students enrolled in public 2-year colleges; including other less-than-4-year colleges, particularly less-than-2-year institutions, may lower the estimates. In addition, BPS data are restricted to first-time beginning postsecondary students; colleges conducting their own studies of transfer may include entering students who are not first-time beginners as defined in the BPS study. Finally, it could also be that student aspirations change appreciably from one cohort to the next and that estimates therefore could depend in part on when the survey was administered.

This examination of alternative ways of defining potential transfer was undertaken in part to inform research at the design stage. Which definition (or definitions) is (are) most appropriate for addressing a specific research question? The decision is not straightforward. This report illustrates the trade-off between restricting the pool of potential transfer students and excluding substantial portions of the initial cohort. For example, including only students who have an academic major and are taking courses leading toward a bachelor's degree results in a high transfer rate (52 percent), but no more than about 1 in 10 community college students meets this definition, and it excludes 4 out of 5 transfer students. Restricting the pool to the 71 percent of students who expect to earn a bachelor's degree or higher yields a transfer rate of 36 percent, but fully 95 percent of all transfers have this expectation.

Just as any statistic depends on the specific variables used to indicate the underlying concepts, the transfer rate for community college students is sensitive to the specification of potential transfer. The most complete picture is provided by using multiple indicators, but this approach is not always practical. If data collection costs or other constraints only permit one definition, one strategy is to define the group of potential transfer students broadly enough that it still reflects community college students somewhat generally, while not so broadly as to include students who never harbor plans to transfer to a 4-year college. The results of this study present several alternatives with different advantages and disadvantages.

Overall, the results provide national estimates of community college students' academic expectations and transfer activity. These estimates refine and update our understanding of students' intentions and paths to transfer. Selecting an approach to defining potential transfer is a necessary first step in any effort to analyze the impact of institution type on persistence and attainment. This report has taken a step back and analyzed the definition itself by examining several alternative approaches using a complete nationally representative sample. While the results do not demonstrate the superiority of any single definition, they sharpen one's appreciation for the consequences of measurement decisions and build a firmer foundation for future work on this population.

Figure C.—Among 1989–90 beginning postsecondary students enrolled at public 2-year institutions who transferred to 4-year institutions by spring 1994, the percentage who met the various definitions of potential transfer

Spicer, S.L., and Armstrong, W.B. (1996). Transfer: The Elusive Denominator. New Directions for Community Colleges, 24 (4):4554.