Home » About MHCC » Research and Planning »  Research Studies and Publications » Student Attendance Patterns And Performance In An Urban Post-Secondary Environment

Student Attendance Patterns And Performance In An Urban Post-Secondary Environment

S. K. Bach, M. T. Banks, M. K. Kinnick, M. F. Ricks, J. M. Stoering and R. D. Walleri, "Student Attendance Patterns and Performance in an Urban PostSecondary Environment." Research in Higher Education, 41 (3), June 2000 (pp. 315-330).

Interest in student transfer/attendance patterns remains high. There is considerable evidence that the traditional perception of transfer from a single community college directly to university is becoming less the norm than in the past. In a continuation of a research consortium project begun in 1992, an urban university and three community colleges investigated the complex nature of student attendance/transfer patterns and their affects on student achievement in terms of credits transferred, GPA performance after transfer, degree completion, and time to degree. While previous research has documented the varied patterns of student attendance/transfer, the goal of this phase of the study was to identify what effects, if any, these varied patterns have on student performance.

Studies of the student transfer process from the community college to the four-year institution have focused primarily on students who take a direct path from the former to the latter. Studies that assume this one-way, linear migration of students may be examining a smaller and smaller portion of the transfer patterns actually occurring. In fact, there is increasing evidence that students' actual patterns of attendance at these institutions are more convoluted, more a "swirl" than a straight line (Barkley, 1993, p.40; de los Santos & Wright, 1990; Matthews & Mellow, 1996, p.88). Part-time attendance and the increasing mobility of students contribute to a "more occasional and ad hoc" relationship with higher education (Adelman, 1992, p.31). Such a relationship confirms that in their paths towards the baccalaureate, students "use community colleges on their own terms, rarely following the institutionally prescribed associate's degree path" (Palmer & Pugh, 1993, p.54).

This study was designed to provide more information about the movement of students in an urban postsecondary environment and to identify the implications of specific patterns of attendance for the transfer process and performance at the baccalaureate institution. Information of this nature is critical as state legislatures and educational governing bodies consider policies and procedures to influence articulation and transfer (Kintzer, 1996, pp.11-12).

The design of this study differed significantly from previous studies of student transfer because it focused on student behavior and the collective role of a group of organizationally unrelated institutions in delivering post-secondary education to a metropolitan community. The contention at the heart of this study was that the linear progression of students from a community college to a university should not be assumed, but is a hypothesis in need of additional testing.

Research on the community college transfer process has been extensive and often critical of the community college contribution (Lee, Mackie-Lewis & Marks, 1993). A primary concern has been with how the community college prepares students to achieve the bachelor's degree. The assumption that the pattern of attendance is simple and linear has been the basis of most research despite the findings reported by Adelman (1992, pp.4-6) that there are at least ten distinct patterns of two/four year attendance. Adelman's patterns are defined as: (1) transfer with 2 degrees (3.4%); (2) transfer with bachelor's degree (3.3%); (3) transfer with associate's degree (1.7%); (4) terminal associate's degree with incidental at 4 year (5.7%); (5) no degree, 2-year and 4-year, non-incidental at both (2.7%); (6) no degree, non-incidental, 2-year only (15.6%); (7) no degree, incidental, 2-year (7.6%); (8) 4-year college only (49.3%); (9) no 2-year or 4-year college (6.7%); and (10) other patterns (4.1%). "Incidental" is defined as earning more than 10 credits and "non-incidental" as earning 10 credits or fewer.

Moreover, early in the history of community colleges the view of the transfer function was confined almost totally to the vertical transfer from high school to junior colleges to universities (Kintzer, 1996). This linear paradigm of the relationship continues to shape the research, so much so that Conklin (1995) and Ronco (1996) considered their findings of a convoluted path to baccalaureate degree attainment a "surprise." Similarly, when Piland (1995) explicitly tested the linear model using a sample of community college transfer students, he found that prior to transfer students attended multiple institutions, stopped out, and often enrolled part-time. He concluded, therefore, that the vertical linear progression paradigm "is a myth" (p. 40).

Only one study, of undergraduate students who had attended two or more post-secondary institutions prior to transferring to a public urban university, examined the relationship between student transfer patterns and select demographic and student performance variables (Kearney, Townsend & Kearney, 1995). The study identified four distinct transfer paths which differed significantly on demographic, academic and interaction/attitudinal variables ( p.328-329). The four paths, where "2" refers to a two-year college and "4" to a four-year college or university, were: 4-2-4 (38%), 2-4-4 (17%), 2-2-4 (20%), and 4-4-4 (12%). No distinct path was identified for 13% of the sample.

Prevailing research on the transfer process is limited and dated (Lee, Mackie-Lewis & Marks, 1993), with much of it based on the National Longitudinal Study of 1972 (Adelman, 1992). Given the greater diversity of today's post-secondary student population, transfer patterns in the 1990s may be quite different from the 1970s or 1980s. Moreover, there should be no assumption of linearity or even of a particular direction of movement between two- and four-year institutions. Although evidence of "transfer swirl" (de los Santos & Wright, 1990) is growing (Adelman, 1992; Kearney, Townsend & Kearney, 1995; Kinnick, et. al., 1998; Matthews & Mellow, 1996), relatively little is known about patterns of student attendance among institutions in an urban setting or about the implications of such patterns on articulation and performance at the four-year college.

To address the need for more information about transfer patterns, an inter-institutional team called the University/Community College Research Consortium (CRC) was formed in 1992. The CRC's mission was to "conduct research designed to strengthen the transfer role of the metropolitan community colleges and the transfer process such that student educational success is enhanced" (Kinnick 1994, p.3).To that end, the CRC undertook to answer the question, "How well is the student transfer process working in this urban post-secondary environment?"

From the literature review and in an effort to be responsive to issues identified by local administrators and policy makers, the CRC identified four specific research questions:

1. What are the patterns of student attendance between the community colleges and the university?
2. How effective is the articulation between the community colleges and the university? That is, do students lose credits when they transfer from the community colleges to the university? If so, what is the nature and extent of the loss?
3. How well do community college students perform academically after transferring to the university?
4. Is there a relationship between patterns of student attendance and credit loss, performance (GPA and degree attainment) and time to degree?

In keeping with these questions and recent studies, the CRC sought to identify the nature of actual student attendance patterns before attempting to assess the effectiveness of the transfer process. The study design arose from questions asked by presidents and administrators at the three metropolitan community colleges and the public university in what we defined for the purposes of this study as an urban post-secondary system (UPS) (Robertson, 1992). The participating institutions provided the funding and a specific research agenda: investigate the movement of students from community colleges to the university and their success after transfer. To that end, the study population was confined to students who left the community colleges at the end of the 1990-91 academic year and who did not re-enroll at the same school the following year.

The UPS used as the basis for this study was treated as a closed system which included only the four public institutions. The authors wish to acknowledge that in reality the UPS is neither a formal system nor is it closed, as there are many private two- and four-year institutions in the area which students may have attended. In addition, no attempt was made to include in the attendance patterns enrollments at public or private institutions outside the designated metropolitan area. Defining the data in such detail was beyond the scope of the study and was not necessary to address the major research questions. The data structure of the study also ignored movement of students among the three community colleges. Moreover, the study made no attempt to compare native university students with transfer students because historically, more than 90% of admitted undergraduate students at the university have presented some transfer credit upon admission. Finally, since the sample originated with listings of students who departed the community college, it was not reflective of the entire population of the university.

The CRC database (Kinnick et al., 1998) consisted of 504 student records containing demographics, transcript data from each UPS institution, and transfer evaluation data from the university. The cases in the database were randomly selected strata by 1) university status (admitted undergraduate, non-admitted undergraduate, post-baccalaureate, and graduate) and 2) community college of students who completed at least three credits at one of the community colleges in 1990-91, did not return to the same community college in the following year and had a record of enrollment at the university. University enrollment included any term pre-, post- or concurrent with 1990-91 enrollment at the community college. The population was 5,057. University records used were current through spring term 1995; degree information was updated one year later and was current though spring term 1996. The sample subset for this study consisted of 336 students who were undergraduates at their first university enrollment and who were not concurrently enrolled at a community college and the university for their first UPS enrollment. This sample is similar to the population in general demographics such as age, gender and ethnic classification. Students who began their UPS experience with concurrent community college and university enrollment were excluded from this study because there were too few (n = 7) to make comparisons with other pattern groups worthwhile.

Developing a coding structure to track the complex attendance patterns evident in student transcripts resulted in an attendance coding structure which provided a fair but limited estimate of student movement to and from the university. Because of the CRC's charge and the definition of the UPS, the coding structure did not account for enrollment at institutions outside the UPS nor did it reflect attendance at multiple community colleges. That is, a student who enrolled consecutively at two or more of the three participating community colleges and then transferred to the university looked the same for attendance coding purposes as the student who enrolled first at one community college, then transferred directly to the university.

Examination of the various patterns resulting from the coding showed two distinct groups: those that began at the community college (pattern codes 1-x-x...), and those that began at the university (pattern codes 2-x-x...). The traditional linear transfer path from community college to university (pattern 1-2) was indicated in 52% of 1-x-x... codes and was designated as the Linear URban Transfer, or LURT, pattern. Further examination of the LURT pattern transcripts showed that 37% of the LURT pattern had post-secondary experience outside the UPS; these LURT patterns were labeled False LURT (F-LURT) while the remainder were considered True LURT (T-LURT). All other, more complex patterns were designated Complex URban Transfer, or CURT, with subsets of CURT-C, community college first enrollment (1-x-x...), and CURT-U, university first enrollment (2-x-x...). (see Figure 1).

Overall, the CURT-C pattern averaged 3.31 switches within the UPS while the CURT-U pattern averaged 2.79 switches. By definition, the LURT patterns contained only one switch.

A look at the demographic characteristics of the four groups showed a fairly similar gender distribution of 56% female overall. The F-LURT pattern showed the largest proportion of females. Racial or ethnic minorities comprised 15% of the sample with the CURT-C pattern showing a minority representation of 25%, the highest of the groups, and the F-LURT pattern the lowest at 9.6%. The average age at entry into post-secondary education for the entire sample was 19.5 years. The T-LURT pattern age distribution showed 32% of students in the sixteen to twenty-one age bracket, the highest percentage of all pattern groups for that bracket. The F-LURT pattern had the narrowest range of ages at entry, none of which were below sixteen years of age or above thirty.

Categories of non-transferable credits identified in an earlier CRC study (Kinnick, et. al., 1998) are further defined as:

• Developmental: Non-transferable courses below the post-secondary level.
• Vocational: Non-transferable post-secondary courses at the community college.
• Low grade: Transferable post-secondary courses with unacceptably low grades.
• Duplicate: Transferable post-secondary courses transcripted more than once.
• Over maximum: Credits in transferable post-secondary courses that exceed the 108 generally accepted for transfer by the university.
• Other: Courses not included in any previous category.

Credit transferability is a function of the type of course transcripted, as in the case of developmental and vocational. The other categories involve transferable courses but are not accepted on the basis of other institutional policies.

Credit transferability at the university is based only on submitted transcripts and include transcripts from non-UPS colleges as well as transcripts from the UPS community colleges. University policy is to require transcripts from all previous postsecondary schools attended for all admitted undergraduates. However, some students in the sample did not submit their community college transcripts upon enrollment at the university and these were not evaluated for transferrability. Transcript evaluation findings exclude four CURT-U pattern students with only one switch (reverse transfers).

Measures of academic performance after transfer are based on grade point average (GPA) changes and degree completion. For changes in GPA, the first term post-transfer to the university GPAs are compared with cumulative GPAs earned at the community colleges for students in the LURT patterns. Due to the complex nature of CURT attendance patterns, the authors felt that an analysis of first term post-transfer to the university was not appropriate for those groups, so only the cumulative GPAs are compared. For students who attended more than one community college in the UPS, cumulative GPA was calculated using weighted averages of credits and grades earned at each community college. GPAs earned at institutions outside the UPS are excluded from comparison. Degree completion measures student attainment of the baccalaureate degree. Time to degree measures elapsed time in years from first post-secondary enrollment to the baccalaureate degree.

The findings address research questions on credit transferability, academic performance after transfer, the relationship of attendance patterns to credit transferability, and academic performance.

Students fell into forty-eight discrete attendance patterns, with six of those patterns representing 80% of students in the sample (see Table 1). Not surprisingly, 77% of the patterns began with community college attendance and 23% began with the university. The pattern codes contain as many as eight "switches" (i.e., a change in attendance from one classification to another) with a mean of 1.99 switches.

The majority of students in the sample (97%) submitted transcripts for evaluation by the university. The university did not accept some credits from approximately 85% of these students. Students submitted an average of 101.5 credits for evaluation, with an average of 81.3 credits accepted by the university. Of all credits submitted on transcripts, 80% were accepted (see Table 2).

In pattern group comparisons, average credits submitted and average credits accepted were highest for F-LURT and CURT-C pattern students, suggesting that these two groups were more heavily vested in post-secondary enrollment outside the university. However, the percent of credits accepted for transfer was higher for T-LURT and CURT-U pattern students, suggesting that these two groups may have been more purposeful in their choice of courses. In terms of the discrepancy between total credits earned outside the university and transferable credits, a one-way analysis of variance (ANOVA) indicated significant differences between pattern groups; F(3, 307) = 5.35, p = .0013. Utilizing a Least Significant Difference (Bonferroni) Modification, CURT-C pattern students had a greater average discrepancy between total and transferable credits than either T-LURT or CURT-U pattern students.

Analysis based on credit categories indicated that roughly half (49%) of students in the sample submitted transcripts containing non-transferable developmental credits (see Table 3). Credits for which students earned unacceptably low grades were submitted by about 43% of students and vocational credits were submitted by 38% of students. Students in the T-LURT pattern were most likely to have submitted credits in the developmental category (64%), while students in the CURT-U pattern were least likely. T-LURT pattern students were also least likely to have submitted credits in the low grade (32%) and duplicate course (6%) categories. CURT-C pattern students had the highest incidence of credits submitted in the vocational category, while CURT-U students had the lowest incidence of credits in excess of the maximum allowable.

Students in the LURT subgroups experienced an average drop in GPA during the first term of post-transfer enrollment at the university of approximately one third of a grade point. T-LURT pattern students (-0.27) were slightly better than F-LURT pattern students (-0.34). Cumulative university GPAs indicate some recovery from this initial transfer shock (2.77 to 2.83 for T-LURT pattern and 2.85 to 2.90 for F-LURT pattern). University GPAs for all pattern subgroups are lower than their community college GPAs. The average difference in cumulative GPAs overall was -0.27 with T-LURT pattern students experiencing the least difference (-0.21) and CURT-U pattern students experiencing the greatest (-0.32).

By Spring 1996, 48% of all students in the sample had completed baccalaureate (BA/BS) degrees. Approximately one-third of baccalaureate recipients (38%) also completed a degree or certificate at another institution, typically at a community college (see Table 4).

Based on a statewide articulation agreement between public 2-year and 4-year institutions, the AAOT degree ensures that community college graduates will meet all general education lower division requirements for junior standing at the university. For all groups, significantly higher proportions of students who complete the AAOT degree attained a baccalaureate degree than did other community college transfers. For the combined population, 68.4% of AAOT completers (26 out of 38) also completed a BA/BS at the university compared to 47.3% of the students who earned another degree or certificate (35 out of 74) and 44.6% of the students who earned no community college degree (100 out of 224). CURT-C pattern students who earned no degree at the community college were least likely to complete a baccalaureate degree (35.8%) than the other groups. Among students who earned degrees other than the AAOT, F-LURT (53.3%) and CURT-C pattern students (52.0%) were more likely to complete baccalaureate degrees than their T-LURT (40.9%) or CURT-U (41.7%) counterparts.

The average elapsed time to degree for all pattern groups was 9.65 years. A one-way analysis of variance revealed differences between pattern groups; F(3, 157) = 5.63, p= .0011. A Least Significant Difference (Bonferroni) Modification indicated that T-LURT pattern students took significantly less time than CURT-C pattern students to complete baccalaureate degrees (see Figure 2).

The one-way, linear paradigm of the student transfer process fails to portray the post-secondary attendance patterns of two-thirds of the students in the study sample. The study identified 48 discrete patterns estimated to be a conservative reflection of student movement in the UPS. For that reason, the data support Piland's conclusion that a linear and simplistic view of the transfer process "is a myth." (1995, p.40). We found that the process of transfer in an urban post-secondary environment might better be conceptualized as swirl which maximizes students' interests, capabilities and financial resources. Breaking the two main patterns, LURT and CURT, into subgroups T-LURT, F-LURT, CURT-C and CURT-U, proved useful in revealing some statistically significant differences in the pattern subgroups related to the transferability of credits and courses, academic performance after transfer and time to degree.

T-LURT pattern students, 32% of the sample, experienced the second highest transferable credit yield at 83%; made greater use of developmental education courses in the community colleges; experienced less credit loss due to low grades; experienced less transfer shock; and took less time to earn a BA/BS (seven elapsed years) than students in the other pattern subgroups. F-LURT pattern students, 19% of the sample, had the highest average community college GPA; had the highest percentage (20%) of transferable but duplicate courses on their transcripts; and took an average of 10.2 years to earn the BA/BS. More than half earned the baccalaureate.

CURT-C pattern students, 25% of the sample, made an average of 3.3 switches within the UPS, more than any other subgroup. They made greater use of non-transferable vocational courses at the community college, which resulted in the lowest (75%) transferable credit yield; had the lowest GPAs at both the community college and the university; and took the most time to earn the BA/BS, 12.2 years. Fewer CURT-C pattern students (43.5%) earned the baccalaureate.

CURT-U pattern students, 23% of the sample, switched schools an average of 2.8 times; had the highest transferable credit yield (84%); were the least likely to have taken developmental education courses or to have taken more credits than the university accepts for transfer. They took about the same amount of time (10.3 years) to earn the BA/BS as F-LURT pattern students.

Based on the course work normally defined as transferable, these data suggest no major, overall problem in course transferability from an institutional perspective. Of course, the student perspective of the experience cannot be determined from these data. Also, until comparative data are available from other studies, we cannot determine whether course transferability patterns found can be considered either typical or distinctive, or high or low. The study clearly shows, however, that inherently non-transferable developmental education, vocational, and low grade courses account for the largest proportion of "lost credits" in the transfer process. Further, this study confirmed the transfer shock and recovery phenomenon reported by Diaz (1992).

The AAOT was designed to facilitate transfer to the university. Despite the evidence that very few students earned the AAOT (16%), those who did demonstrated a higher graduation rate than those who did not. Also, since the AAOT program was in its infancy in 1990-91 (the base year of the sample), later student populations may show an even higher rate of degree completion. Between 1994-95 and 1996-97, the number of AAOT degrees awarded by all the community colleges in the state has increased from 1,867 to 2,031. The increasing number of students earning the AAOT combined with the performance of students in this study who eventually obtained a Bachelor's degree, suggests that efforts to improve articulation can have meaningful results in terms of degree completion.

Finally, this study shows that the traditional linear paradigm for degree completion -four to five years- is a false one for most students. Overall, students in the study took an average of 9 ½ years to complete a Bachelor's degree from their initial post-secondary enrollments (at any post-secondary education institution). Even the T-LURT pattern students took an average of seven years to graduate.

The findings of this study have implications for how student transfer is framed and studied in higher education; the bases on which student learning is inferred or assessed; and how inter-institutional cooperation might enhance the transfer process.

We have found clear, strong evidence of the transfer "swirl" identified by de los Santos & Wright (1990), as well as Adelman's (1992) observation of student post-secondary attendance as "more occasional and ad hoc." While studies that focus on the LURT pattern of attendance are important and currently dominate, researchers should be on notice that such studies are neglecting a sizeable portion of college students. Additionally, policy deliberations and educational practices which are shaped by research based on the narrow LURT paradigm may ultimately be misinformed. This is especially significant in the computation of transfer rates and the use of such rates to gauge the effectiveness of the community college.

More research is needed that identifies actual patterns of student movement among institutions and the possible consequences associated with such movement. As in this study, the use of student transcripts as an associated data source in future studies will be essential in broadening the understanding of attendance patterns. Future studies may address the advantages and disadvantages of the complex attendance patterns over the linear pattern to resolve issues related to inter-institutional mobility and the development of policies and services which respond to and are designed to support the educational success of students who follow these patterns. Despite the shorter time to degree completion for the T-LURT subgroup, more information is needed to know what advantages the shorter elapsed time represents in terms of economics and the relationship between education and the workforce. These data are open to alternate interpretations: that excessive time pursuing a degree is evidence of inefficiency; or, that education is a process of discovery, change and adaptability in which students use educational resources as needed.

As student attendance patterns become more complex the task of evaluating transcripts submitted on transfer becomes increasingly difficult. It suggests problems with the articulation process itself, and raises questions about the effectiveness - in terms of student costs and institutional accountability - of a system that counts strings of courses and credits rather than more directly assessing student learning. In evaluating student transcripts we found a sizable proportion of courses which were non-transferable vocational, particularly among students in the CURT-C pattern group. This seems to indicate that renewed attention to the transferability of applied knowledge and skills is needed to strengthen articulation between two- and four-year institutions. Moreover, as bachelor's degree completion rates for students who earned the AAOT indicate that the articulation the AAOT provides is working, perhaps a similar form of articulation could be designed for vocational programs.

In his 1995 presidential address at the annual meeting of the Association for the Study of Higher Education, Patrick Terenzini called for "practice-and policy-oriented research" (1996, p. 8) and warns that "It is the general tendency away from action, practice- and policy-relevant research toward the more scholastic that should concern us if the research we produce is not to become what Keller (1985) characterized as 'a literature without an audience' (p. 8)." In terms of practical implications, the CRC studies have had a direct influence on collaborative efforts between the university and the community colleges. Based on how students are actually using the UPS, leaders at the institutions now see themselves as part of a whole with shared responsibility for student success.

In an effort to improve articulation and facilitate the transfer process, the university and the community colleges have recently developed a co-admissions program. Benefits to students include a single admissions application, university advising on the community college campus, library privileges at both the university and community college, and coordinated financial aid among others. In addition, faculties at the university have worked with faculty at the community colleges to design and implement a "transition" course for sophomores at the community colleges to better prepare these students for the academic experience at the university. Additional research will be needed to determine if these efforts bear fruit in terms of performance outcomes.