Intensity and Attachment

Abstract

This study examines the relationship between community college enrollment patterns and student outcomes—credential completion and transfer to a 4-year institution—introducing a new way of visualizing the various attendance patterns of community college students. Patterns of enrollment intensity (full- or part-time status) and continuity (enrolling in consecutive terms or skipping one or more terms) are graphed and then clustered according to their salient features. Using data on cohorts of first-time community college students at five colleges in a single state, the study finds astounding variation in student enrollment patterns. Clustering these patterns reveals two relationships: the first is a positive association between enrollment continuity and earning a community college credential, and the second is a positive association between enrollment intensity and likelihood of transfer.

Keywords

enrollment patterns intensity community college persistence cluster visualization

The study of student pathways through community college is an important part of understanding the student experience. Student pathways are the time-ordered series of courses that students complete as they advance toward their education goals, typically program completion with a credential or transfer to a bachelor’s degree program. Centrally related to student pathways are students’ enrollment patterns—both the intensity of enrollment as measured by full- and part-time statuses and the continuity or attachment of enrollment as measured by the consecutiveness of attendance. Few students who enroll in public 2-year colleges go on to complete an award within 2 years of study (Radford, Berkner, Wheeless, & Shepherd, 2010). A key reason is that community college student pathways and enrollment patterns are anything but traditional; students routinely switch into and out of full- and part-time statuses, and they frequently skip terms.¹

Precisely how diverse student enrollment patterns are among students and the extent to which they are correlated with postsecondary outcomes have yet to be thoroughly documented. Although previous research considers the relationship between starting as a full- or part-time student and educational outcomes (O’Toole, Stratton, & Wetzel, 2003) or describes the circumstance of mixed enrollment intensity (McCormick, Geis, & Vergun, 1995), investigators have not fully considered the extent of diversity in enrollment patterns. It is important for institutions to track students and understand when they are at risk of abandoning their studies, but colleges have not yet developed the ability to distinguish between normal variations in students’ education pathways and signs of potential dropout. This study addresses two research questions:

What are the enrollment patterns generated by community college students?

How are characteristics of these patterns related to postsecondary outcomes, such as earning a credential and transferring to a 4-year institution?

Using data on two cohorts of students at five colleges in a single state, the investigation presented here reveals the diversity of enrollment patterns in terms of intensity and continuity that are generated by community college students along their educational pathways. The study uses a novel graphical technique to illustrate these patterns. In addition, the study aggregates thousands of enrollment patterns into six distinct types using a cluster analysis that combines patterns based on their main features. These clusters are correlated with the probabilities that students will earn credentials and transfer to a bachelor’s degree-granting institution.

Review of the Literature

Organizing, describing, and analyzing community college students by their patterns of enrollment and course-taking behavior is a useful exercise to better understand diverse student educational pathways (Adelman, 2005; Bahr, 2010; Hagedorn, Cabrera, & Prather, 2010).² In Moving Into Town, Adelman (2005) used rich data from the nationally representative National Education Longitudinal Study of 1988 (NELS:88) to provide an organizing framework for traditional-age community college students. The report considered students as “residents” who move into and out of the “town” of community college, focusing on their pathways in a series of portraits: the link between high school and college (the event portrait), what happens at the college (the residence history portrait), and patterns of exit (the graduates portrait). In the residence history portrait, for example, Adelman used course-taking behavior to classify students as “homeowners,” “visitors,” or “tenants” based on the number of credits they earn from community colleges. Combining all three portraits ultimately led to a list of six distinct traditional-age populations served by the community college.

Incorporating enrollment intensity into a broad classification of student types, Bahr (2010) developed a typology of community college students using a cluster analytic technique. He built on earlier classification work (Ammon, Bowman, & Mourad, 2008; Hagedorn & Prather, 2005; VanDerLinden, 2002) that sought to identify broad types of community college students by combining behavioral data on course-taking and enrollment. Bahr used a very large sample of credit and non-credit students and focused on 13 enrollment and behavioral characteristics, such as units attempted in several subjects of study, enrollment intensity (mean units attempted per semester), course success ratios, and persistence (number of terms and years enrolled). Taxonomies such as this serve to illustrate the main types of students who are enrolling in these multiple-mission-oriented institutions and for what purpose. As Bahr noted, such an understanding can “assist policy makers, administrators, practitioners, and other stakeholders in directing and optimizing the use of limited resources to maximize the benefits received by students. . . . Additionally, the increasing attention of institutional accountability . . . has drawn attention to the need to distinguish students who enroll for differing objectives or desired benefits” (2010, p. 726).

Much of the existing literature related to enrollment patterns and enrollment intensity, however, does not take a broad descriptive strategy but rather focuses on the relationship between students’ first-semester experiences (their intensity in the first term) and educational outcomes. Additional national work by Adelman (1999, 2006) found that students who attended full-time less frequently, who did not enroll continuously, or were unable to earn at least 20 credits by the end of the first calendar year were much less likely to earn a bachelor’s degree. Driscoll (2007) in analyzing course-taking patterns over 6 years found strong correlations between high levels of first-term enrollment intensity and the likelihood of returning for the spring semester as well as transfer and bachelor’s degree earning rates. This descriptive analysis, however, relies on simple correlations and is not able to account for other confounding factors that may explain the observed relationships.

Stratton, O’Toole, and Wetzel (2004, 2006) and O’Toole et al. (2003) have performed various analyses of the 1990/1994 Beginning Postsecondary Students (BPSs) Longitudinal Survey to explore the reasons behind enrollment intensity differences and the relationship between intensity and dropping out of college among students in 2- and 4-year colleges. In an effort to understand why students enroll full- or part-time, Stratton et al. (2004) concluded that older individuals and those in states with lower unemployment rates are less likely to enroll full-time. Studying the relationships between initial enrollment intensity and intensity over time, O’Toole et al. revealed that using initial (first term) full-/part-time status underestimates the incidence of part-time enrollment intensity, as about one quarter of their sample stopped out or attended part-time for at least one term but still managed to graduate or continue to enroll at the end of 5 years.

Stratton et al. (2006) focused on differences in attrition rates among students who begin postsecondary education on a part- or full-time basis. Their analysis recognized that factors correlated with initial enrollment intensity may be correlated with the decision to drop out, leading to bias in simply estimating the effect of initial enrollment intensity on the probability of attrition. Modeling both the choice to enroll as a full- or part-time student and the decision to drop out as separate but related processes revealed that observable factors associated with drop-out behavior differ by initial enrollment intensity. That is, there are different observable factors associated with attrition depending on whether a student begins as a full- or part-time student. The authors found that parental education, timing of enrollment, college grade point averages, and local economic conditions are associated with attrition for full-time students but not for part-time students. Therefore, it is not initial part-time status per se that is correlated with attrition, but the underlying differences in observable factors (e.g., parental education, household characteristics, racial and ethnic characteristics) that determine the correlation between full- or part-time status and attrition.

More recently, Attewell, Heil, and Reisel (2012) used NELS:88 data to conduct a study from the academic momentum perspective, which posits that students who accumulate credits more quickly improve their chances of completing a college degree independent of academic readiness or socioeconomic status. Credit accumulation is intimately related to enrollment intensity, as full-time students more often accumulate credits more quickly. Attewell et al. examined four categorical indicators of momentum (no delay between high school and college, attending part-time in the first semester, taking 18 or more credits in the first semester, and enrolling in the first summer after freshman year) and used propensity score matching to identify average treatment effects of momentum on attaining a college credential. They found that graduation rates are lower for students who delay college entry and who take part-time course loads but found somewhat weaker positive effects on graduation for taking a large course load or enrolling in the first summer after freshman year.

Contributions of This Study

This study contributes to the literature on enrollment patterns in multiple ways. First, it focuses on a relatively recent cohort of students solely from community colleges rather than using an aging nationally representative data set that combines college sectors or focuses on 4-year colleges. A better understanding of students in 2-year colleges can be obtained by studying them in isolation. Because they are more likely to have family responsibilities, to work full-time, to have greater financial constraints, and to be more academically underprepared than their counterparts in 4-year institutions (Horn & Nevill, 2006), community college students attend college erratically and vary greatly in the rate at which they earn college credits.

Second, this study uses a longitudinal approach in identifying enrollment patterns. Most studies consider enrollment intensity in the student’s first term as the most important aspect of intensity, and none have described the subsequent variation in intensity revealed as students’ progress along their pathways. Third, we introduce a new method for describing enrollment patterns that provides a visual representation of the entire diversity of enrollment patterns. To aid interpretation, the resulting patterns are clustered by intensity and continuity features, which in turn link enrollment decisions (such as switches from full- to part-time attendance) to postsecondary outcomes. This kind of representation could be useful in facilitating communication among faculty and other stakeholders about how community college students attend college, and it could help illustrate the link between aspects of enrollment patterns and postsecondary outcomes.

Data Overview

This study uses student-level data from five community colleges located in a single state in the United States. As part of a centralized state system, the colleges participate in a common course numbering system and offer a similar set of degrees and certificates. Each college uses a semester system in which an academic year is defined as the fall and spring terms followed by a shorter summer term. The analysis sample comprises first-time-in-college (FTIC) students who began at one of the five institutions in the 2005-2006 or 2006-2007 school year. The students are followed through the 2010-2011 school year (18 terms or six academic years for the 2005 entering cohort, and 15 terms or 5 years for the 2006 entering cohort). The data collected are extensive, including both student demographic information and full community college transcripts. The colleges provided credential attainment. Data on transfer to other colleges came from the National Student Clearinghouse.

The sample consists of 14,429 degree- or transfer-seeking students who generally intended to earn a certificate, diploma, or associate degree, distinguishing them from the block of community college students who enroll in shorter, non-credit vocational or adult basic skills programs.³ Students in the sample are considered degree- or transfer-seeking if they took placement exams or, when placement exam results were unavailable, they enrolled in credit-bearing courses and did not meet any of the following criteria: enrolled in non-credit vocational courses, English as a Second Language (ESL), Adult Basic Education and Graduate Equivalency Degree (ABE/GED) programs, or enrolled dually in high school and college. Students who just seek to take some courses and not pursue a degree are generally not required to take placement tests. Student data across the five colleges were aggregated.⁴

Empirical Analysis

Patterns of Enrollment

This section begins by introducing a framework for analyzing patterns of enrollment, focusing on the characteristics of intensity and continuity. Intensity distinguishes between full- and part-time enrollments, where full-time is defined as attempting 12 or more credit hours in the fall and spring terms and 6 or more credit hours in the summer term.⁵ In general, a full-time course load means four courses in a 16-week fall or spring semester. Our decision to include summer terms led to a few non-trivial decisions. Taking at least one course in the summer is not uncommon, and it represents an important continuity in enrollment.⁶ To include summer enrollment and not downplay its contribution to credit accumulation, this study treats it as a term like fall and spring. About 8.2% of courses and 8% of credits are attempted during summer terms. In addition, 37% of students take at least one summer course. These numbers are sufficient to justify considering the summer term as an important data point in a study of enrollment attachment and intensity.

To aid exposition, all terms have been numbered so that each individual’s first term of enrollment is term one, regardless of whether it is fall, spring, or summer. Of the three potential first terms of enrollment during the academic year, 68% of the sample started in fall, 24% in spring, and 8% in summer. Each subsequent term is numbered incrementally from the student’s first term. This approach does result in some blurring of enrollment, as one student’s Term 3 will be fall and another student’s will be summer. However, as community college students attend so haphazardly, this left-shifted numbering (making a student’s first semester as Term 1) should not distort conclusions about the diversity of patterns or the clusters generated from the patterns. Nevertheless, there are some differences among students who started in different terms. Spring entrants were more likely than fall or summer entrants to skip their immediate second term, and summer entrants were more likely to enroll consecutively in the first three terms. That said, subsequent patterns for each group are still greatly diverse.

I describe enrollment patterns by creating a vector (or character string) for each student that consists of a series of zeroes, ones, and periods. The ith location of the vector is a 1 if the student enrolled in term i full-time, 0 if enrolled part-time, and a period if not enrolled. This analysis combined data from the 2005 and 2006 cohorts and thus the maximum vector length is 18 for the 2005 cohort and 15 for the 2006 cohort, reflecting the amount of data provided for each cohort.⁷ Combining the cohorts may inflate the total number of patterns, but it does not change the substantive interpretation. As an example of this vector representation, consider a traditional student who begins in the fall and follows an idealized 2-year degree track—enroll full-time in the first two terms and skip the summer term for two consecutive years. That student’s vector would appear as,

“11.11.…………”

A student who enrolls intermittently with different degrees of intensity may have a vector that appears as,

“01010..0..0..1.…”

Over the 18 observed terms and 14,429 students in the sample, there are 4,594 distinct patterns of full-time, part-time, and non-enrollment. Although it is impractical to tabulate all of them, the 10 most common types are shown on the top panel of Table 1 and 10 of the least common patterns are presented on the bottom panel.⁸ Of more than 4,500 distinct patterns, the 10 most frequent types account for 44% of students. The two most common patterns are for students who enroll either part-time only or full-time only in the first term (28% of students). These 4,000 students are the earliest leavers—potentially for a number of reasons—and their frequency is striking. Although most are no longer enrolled in higher education after the first term, some of them do earn short-term certificates (19 students) or transfer to a 4-year institution sometime after that first term (595 students).

Table 1.

Enrollment Vector Patterns.

Pattern	Frequency	Percent (%)	Cumulative Percent (%)
0.…………….	2,810	19.5	19
1.…………….	1,190	8.2	28
00…………….	744	5.2	33
11…………….	561	3.9	37
10…………….	291	2.0	39
11.11………….	175	1.2	40
01…………….	145	1.0	41
0.0……………	140	1.0	42
11.1.………….	125	0.9	43
00.0.………….	115	0.8	44
100……0..0.….	1	0.007	99
00.00.11.11.1.….	1	0.007	99
11111.1100.…….	1	0.007	99
00.01.00.00…….	1	0.007	99
10.1..11.11.1.….	1	0.007	99
00……….0..0..	1	0.007	99
10.00.10001.100…	1	0.007	99
00..0.0.00.….0..	1	0.007	99
10.11111111…….	1	0.007	99
11.1..11011.1.….	1	0.007	100

Note. In these vector patterns, 1 indicates enrolled full-time, 0 indicates enrolled part-time, and a period (.) indicates non-enrollment. The position of the number in the vector indicates the term number, from 1 to 18.

Although many of the students in the sample enroll sparsely, as suggested by the top panel, there are thousands of students who generate unique enrollment intensity patterns over a long period of time, as illustrated by those in the bottom panel. These students are characterized by several matriculation periods, gaps in enrollment, longer persistent states of attendance or non-attendance, and frequent switching among full-time, part-time, and non-enrollment statuses. In general, common patterns are short and unique patterns are long. The longer a student stays enrolled, the more likely the student’s pattern will be unique. Few students who stay enrolled relatively longer do so in identical ways.

Importantly, students usually stop enrolling after they graduate or transfer; yet, the pattern representation used here does not provide any explicit way to capture formal exit, such as graduation or transfer. Among students captured in the top panel, 65 (about 1%) earned a certificate or associate degree. Certificate awards are concentrated in the second pattern and associate degree awards in the sixth pattern.

Visualization of Enrollment Intensity and Continuity

To better understand the entire range of enrollment patterns without tabulating every distinct type requires a graphical approach. A well-organized image can be constructed from the pattern vectors, providing a broad overview of the patterns and how they relate to graduation and transfer. Each student’s vector shown in Table 1 is stacked on top of each other and sorted to create a matrix of 14,429 students by 18 terms. In the images that follow (and located on the URLs), student vector patterns are represented by thin bands of color instead of zeros, ones, and periods; each row corresponds to one student. The height of each resulting block of color in each term is proportional to the number of students it represents. The intuition is that in Term 1 some students attend full-time and others attend part-time. In Term 2, the full-time students in Term 1 attend full-time, attend part-time, or do not enroll; first-term part-time students also split among these three options in Term 2. Students continue to be divided in this manner, term by term, and the resultant graphic representation uses three different colors for full-time, part-time, and non-enrollment statuses to illustrate the patterns.

Figure 1a presents an image of the enrollment patterns for all 14,429 students. White space indicates non-enrollment (no attempted credits), blue is part-time enrollment, and orange is full-time enrollment.⁹ In Term 1, all students are either enrolled full-time (represented by the orange block of color in the top portion of the first column) or part-time (represented by the blue block of color in the bottom portion of the first column). As the terms progress and students switch their enrollment statuses, these blocks of color are subdivided to represent students’ divergent enrollment patterns. Scanning across the image allows for the visualization of the wide variation of enrollment intensity and continuity after the first few terms. For understanding the image, it may be useful to consider the bottom of the image, with a dark blue bar followed by white space. This depicts students who enrolled part-time in Term 1 and never returned. The students represented by blue bars in Terms 1 and 2 and white space after (just below the midpoint of the image) are those who enrolled part-time in Terms 1 and 2 and never returned.

Figure 1.

Please see the companion web site for full color images: https://github.com/pmcrosta/eps/tree/master/ccr_images. In all figures, blue indicates part-time enrollment; orange indicates full-time enrollment; white space indicates non-enrollment. The height of each block of color in each term is proportional to the number of students it represents. In Figure 1c, an indicator mark shows that a student earned a credential. Legend entries for indicator marks, some of which overlap, are Short-term Certificate (244 students), Long-term Certificate (157), Associate of Arts (538), Associate of Science (56), and Associate of Applied Science (658). Some students have award dates in terms in which they have no enrollment record, resulting from late filing of award paperwork or a delay in recognizing transfer credit, among other reasons. In Figure 1d, each of the 2,656 purple dots, some of which overlap, indicate a student’s first term enrolled in a four-year institution.

Figure 1b shows what an enrollment patterns graphic might look like if students followed standard pathways that begin with fall enrollment. The graphic is organized from the top down to express the following three enrollment intensity patterns, separated by black horizontal lines:

“110110…………”

“11011.…………”

“11.11.…………”

Figure 1b thus shows what Figure 1a would look like if all students follow some version of the conventional 2-year pathway through community college. These patterns are in reality particularly rare (even after including summer- and spring-entrant students). However, it is useful to contrast the homogeneity of Figure 1b with the heterogeneity of Figure 1a to highlight the difference between the canonical student pathways (Figure 1b) and actual student pathways (Figure 1a). Few students exhibit these canonical pathways, which form the basis of curriculum guides and financial aid policies.

An immediate concern when looking at patterns like those of Figure 1a is that some students graduate or transfer out of the community colleges, and their doing so results in periods of non-enrollment using the descriptive technique applied here. Figure 1c updates Figure 1a by adding indicator marks to show when students have earned a degree or certificate, and Figure 1d shows when students transfer by adding indicator marks (purple dots) that represent students’ first enrollment term in a 4-year school.¹⁰ Perhaps not surprisingly, credentials are most heavily clustered along the top of Figure 1c, where students have more consecutive terms of full-time enrollment.¹¹ However, there are several examples of persistent students who manage to earn a credential after 12 or even 15 terms of part-time or intermittent enrollment.

The indicator marks in Figure 1d that represent transfer show a few clusters. Some students transfer after one term of community college study, perhaps due to deferral from a 4-year institution, spring admission to a 4-year institution, or perhaps even co-enrollment at a 4-year institution. Of the 148 students who attended in Term 1 and transferred by Term 2, 4% began community college in the spring and 32% began in the summer. Others arrive at the 4-year college during Term 4, which would likely be the first fall term after a full year of community college study. Terms 7 and 10 have clusters of transfers, a pattern expected of fall entrants who transfer to a 4-year institution in a following fall term. Remarkably, the transfer patterns suggest a high degree of non-continuous postsecondary enrollment. Although most transferees leave after one or two community college terms, many students depart community college and then wait years before enrolling in a bachelor’s degree-granting institution. Still others engage in concurrent enrollment (as seen by transfer indicator marks inside of the blue or orange bars).

Summary of Vector and Graphical Analysis

The method presented here provides a readily available tool for describing student progress both quantitatively and qualitatively. Some key insights emerge from the vector and graphical analyses taken together. First, they reveal that students generate many patterns due to intermittent enrollment and frequent switching between full- and part-time statuses (4,585 distinct patterns for 14,429 students). Some students still enroll alternately full- and part-time well into their 6th year of study (17%). About 1% of students follow the traditional fall-spring, fall-spring pattern (with a break for summer) during the first six terms, followed by no additional enrollment in subsequent terms. Some 28% of students have only one term of community college enrollment, and over one quarter of them never return after that first term. Almost 40% of students enroll in one term or in two consecutive terms and never return to either a 2-year or 4-year institution within the study’s tracking period. Except for those who leave the institution early into their postsecondary careers, few remaining students have the same enrollment patterns in college.

Second, and related to the first, there is much switching between full-and part-time statuses. In general, those who begin as full-time students are more likely to attend full-time subsequently, suggesting a much quicker rate of credit accumulation than for those who start part-time. However, students frequently switch between full- and part-time attendances (43% of students do so at least once.) About 69% percent of full-time starters who returned at least once had at least one part-time term. Half of part-time starters who returned at least once had at least one full-time term. This finding challenges the notion that starting intensity is indicative of future enrollment intensity (and it reinforces the findings of O’Toole et al. [2003] discussed earlier). Similarly, among students who enrolled in more than one term, 17% attended only full-time, 22% attended only part-time, and 61% percent attended a mix of part-and full-time. About one quarter of students had two or more switches between full-and part-time statuses, and 32% of students had consecutive part-time enrollment. The high degree of switching challenges the common assumption that students can be identified as full- or part-time based on their status upon entering college.

Third, the patterns help explain why colleges have difficulty getting students into and through programs of study (Jenkins & Cho, 2012). The enrollment intensity figures reveal that students who persist are quite likely to experience a range of enrollment intensities over their college careers. Very few community college students follow a traditional fall–spring–fall–spring pattern with full-time enrollment in all terms (1.2%), the pathway that is often advertised by colleges as standard and that can be seen in suggested curriculum guides on college websites. Few students earn an associate degree in expected 2 years (3.5%). Over a 6-year/18-term horizon, many students leave after their first contact with the college (28%). Only a handful of them complete short-term certificates or transfer to a 4-year institution (15% of the 28%). In the cohorts under study, the typical student attended full-time in about 44% of the terms attended. Finishing a 2-year degree within 2 years is bound to be uncommon when full-time enrollment is this low.

The enrollment patterns identified by the current study are remarkably varied and can be even described as chaotic; they raise several questions about the nature of the patterns. Why do full-time students switch to part-time and vice versa? Are there any differences in academic achievement between students who attend full-time consecutively compared with those who switch to part-time? What about differences in demographics or financial aid awards in the second term? Perhaps students simply cannot get into desired courses. Of course, a range of other factors (as well as the eventual attainment of postsecondary outcomes) will have an impact on whether, when, and how intensely students enroll.

Clusters and Their Relationship to Postsecondary Outcomes

Although there are thousands of distinct patterns of enrollment intensity, they are all generated from the same basic components: students attempt different course loads at different points in time. This section describes the pattern clusters based on the features of the patterns that indicate degrees of intensity and continuity, enabling the production of a typology of enrollments, a more parsimonious way of thinking about the student behavior observed. Variation in postsecondary outcomes among clusters provides a way to correlate the features of patterns with outcomes. The study uses a k-means clustering algorithm that generates six clusters of enrollment patterns. The algorithm seeks to maximize the differences between clusters and minimize the differences within clusters through an iterative process that moves students among groups. Clusters are created solely from the information gleaned from enrollment intensity patterns and do not include other academic or demographic characteristics.¹²

Importantly, the analysis presented here is related to but different from that of Bahr (2010) and his predecessors, in that the emphasis is on aggregating the longitudinal patterns created only by variation in intensity and continuity of enrollment. The goal, however, remains to identify student types and provide a more parsimonious way of describing the enrollment patterns presented previously.¹³Table 2 presents examples of patterns found in each cluster using the method described above. Students are unevenly spread across clusters, as Cluster 5 has 5% of students and Cluster 2 has 35%. This spread of students into clusters is not unexpected as 44% of students generate the top 10 patterns of enrollment. Below, I describe the clusters and provide descriptive names.¹⁴

Table 2.

Sample of Vector Patterns Found in Clusters.

Cluster	Example	Most frequent
Full-Time Persisters	111111…………	11…………….
Early Leavers	0…0………….	0.…………….
Early Persistent Switchers	11.01………….	10…………….
Mostly Part-Timers	0..00………….	00…………….
Early Attachers	11011..0……….	11011………….
Later Attachers	11.11.11.10.00….	11.10.00……….

Note. Each example pattern is one of several chosen for illustrative purposes.

Cluster 1:Full-Time Persisters (n = 2,858; 20%). These students enroll primarily full-time and for an average of 4.5 terms. They begin full-time and remain full-time, or begin part-time and change to full-time, where they remain. They have relatively few changes in attendance statuses compared with students in other clusters with a similar number of enrolled terms. For many of these students, their first part-time enrollment was followed by a long spell of non-enrollment.

Cluster 2: Early Leavers (n = 4,998; 35%). This largest cluster captures the students who enroll for the fewest number of terms (usually only one). Later enrollments usually occur well after the first enrollment term if at all, and there is virtually no consecutive enrollment. These students are thus characterized by very sparse enrollment.

Cluster 3: Early Persistent Switchers (n = 1,958; 14%). These students attend for four terms on average, about 50% of which are full-time. Almost all change intensities between the first two terms. They are likely to switch from full-to part-time attendance and then remain part-time, though they occasionally revert back to full-time. They have a relatively high number of switches between full- and part-time attendances. These students consecutively enroll in the first two terms but then have sporadic enrollment over the remainder of the time frame.

Cluster 4: Mostly Part-Timers (n = 2,376; 16%). These primarily part-time students have very few intensity changes. Much of this group might be described as first-year experimenters, enrolling only for two part-time consecutive terms, although some do persist into later terms. A few Mostly Part-Timers start full-time, but quickly lower their intensity and maintain a lengthy trail of part-time enrollment.

Cluster 5: Early Attachers (n = 728; 5%). This smallest cluster is characterized by almost nine terms of enrollment on average along with frequent switching between full- and part-time intensities. These students do not interrupt enrollment until the eighth term on average, into the third year of study, and most of the enrollment is full-time. Their enrollment is front-loaded in the earliest terms and highly consecutive. These students consistently attempt to earn credits term after term at any intensity possible.

Cluster 6: Later Attachers (n = 1,511; 10%). Students in this group also enroll for a long period of time—over nine terms, on average—but attend full-time less often than the Early Attachers. The students have a similar number of full-time to part-time switches, but experience their first enrollment interruptions earlier on, generally in their third terms. That is, Later Attachers follow a more traditional approach of “two terms on, one term off,” and they also have a high degree of persistence. This group is more likely to switch from part- to full-time attendance than the opposite, but have a lower level of consecutive full-time enrollment due to more interruptions.

The six clusters identify some student types that appear elsewhere in the community college literature. For example, the Early Leavers cluster is similar to the drop-in cluster of Bahr (2010), but perhaps with less favorable success rates. Like Bahr’s clusters, those presented here also stratify along some demographic lines (though no demographic or environmental characteristics were considered in their creation). Table 3 presents demographic characteristics by cluster that show how some enrollment intensity and continuity clusters are correlated with individual characteristics.¹⁵Full-Time Persisters, Early Persistent Switchers, and Later Attachers tend to be the youngest at about age 21, whereas Early Leavers and Mostly Part-Timers are age 25 to 26 on average. These findings are consistent with evidence that older students have different enrollment trajectories than younger students (Calcagno, Crosta, Bailey, & Jenkins, 2007). All of the clusters hover around a composition of 60% White students, ranging from 55% for Mostly Full-Time Switchers to 64% for Later Attachers. Early Full-Time Persisters and Later Attachers have the smallest Black student representation, at about 18%. Secondary education attainment is somewhat stratified across clusters, as students in Full-Time Persisters, Early Attachers, and Later Attachers are more likely than students in other clusters to have traditional high school diplomas (rather than GEDs or no diploma). Large differences in first-term financial aid are found as well, as only 21% of students in the Mostly Part-Timers and Early Leavers clusters received aid compared with 45% of students in the Full-Time Persisters cluster.

Table 3.

Student Characteristics Within Each Enrollment Pattern Cluster.

Characteristic	Full-Time Persisters	Early Leavers	Early Persistent Switchers	Mostly Part-Timers	Early Attachers	Later Attachers
Female (%)	49	48	50	54	52	53
Age	20.5	26.0	21.2	25.1	22.4	20.4
Age ≤ 19 (%)	81	51	75	54	70	82
Age 20-26 (%)	10	18	13	17	12	9
Age ≥ 27 (%)	9	31	12	29	18	9
White (%)	62	56	55	56	58	64
Black (%)	26	31	29	28	17	18
American Indian (%)	0	1	0	0	0	1
Asian (%)	2	2	3	2	6	3
Hispanic (%)	3	4	4	5	4	4
Mixed Race (%)	1	1	0	1	1	1
Non-resident Alien (%)	6	6	7	7	14	9
HS diploma (%)	93	87	90	85	93	94
GED (%)	3	7	6	6	4	3
No HS diploma (%)	4	6	4	9	3	3
Received financial aid in first term (%)	45	21	36	21	33	29
Received Pell grant in first term (%)	34	15	27	16	22	23
SES index	3.37	3.44	3.40	3.50	3.53	3.48
Fall entrants (%)	83	52	72	73	72	76
Spring entrants (%)	12	36	19	19	15	19
Summer entrants (%)	4	11	9	8	13	5

Note. HS = high school; GED = Graduate Equivalency Degree; SES = socioeconomic status.

Table 4 focuses on how clusters differ by college readiness as measured by developmental placement, considering both the number of subjects to which students are referred and the depth of developmental placement in math, English, and reading. It has been well documented that developmental education impacts persistence, as students often do not complete the required sequences that enable enrollment in college-level coursework (Bailey, Jeong, & Cho, 2010). Students in the Early Attachers and Full-Time Persisters clusters had slightly higher overall college-readiness rates (11% and 10%, respectively). Students in the Early Leavers and Mostly Part-Timers clusters had the lowest college-readiness rates overall and in each individual subject. In fact, Early Leavers have a higher percentage of students who are placed two or more levels below college math, English, and reading than the other five clusters. These findings are congruent with the notion that better-prepared students are more likely to persist and rapidly accumulate credits than their less prepared counterparts who enroll with hesitation (part-time) and become discouraged quickly (leave community college early).

Table 4.

Developmental Education Student Placement by Cluster.

	Full-Time Persisters (%)	Early Leavers (%)	Early Persistent Switchers (%)	Mostly Part-Timers (%)	Early Attachers (%)	Later Attachers (%)
Developmental placement
College-ready	10	8	9	8	11	9
Developmental education in one subject	38	35	36	38	39	39
Developmental education in two subjects	22	23	24	25	23	25
Developmental education in three subjects	30	33	31	29	27	27
Math developmental placement
College-ready	13	9	11	8	15	13
One level below	17	12	15	13	16	18
Two+ levels below	70	79	74	78	68	69
English developmental placement
College-ready	56	51	55	54	57	57
One level below	23	25	23	24	27	24
Two+ levels below	20	24	21	22	16	20
Reading developmental placement
College-ready	52	50	49	51	55	53
One level below	29	30	32	29	28	29
Two+ levels below	19	20	19	20	17	18

Note. Within each panel, columns sum to 100%.

Of more interest for this study is how postsecondary outcomes such as earning a credential or upward transfer correspond to clusters of enrollment patterns. Figure 2 shows the community college credential-earning rates (within 5 years for the 2006 cohort and 6 years for the 2005 cohort) for each of the 6 clusters. Credentials earned include short- and long-term certificates, associate of arts, associate of science, and associate of applied science degrees. The lowest rates, perhaps not surprisingly, are found among the cluster of Early Leavers (1%), the Mostly Part-Timers (5%), and the Early Persistent Switchers (6%). Early Attachers have the highest graduation rate (43%) and the Later Attachers are not far behind (37%). The group of Full-Time Persisters has a credential-earning rate that is somewhat lower than what one might expect for students who have so much full-time attendance (18%), mostly because they are transferring before earning a 2-year credential.

Figure 2.

Percentage of students who earned a credential within 5 or 6 years by enrollment pattern cluster.

Differences in transfer behavior in relation to enrollment pattern clusters help explain some of the variation in credential-earning rates as well. As Figure 3 shows, transfer rates are higher than credential-earning rates for all clusters except Early Attachers and Later Attachers. Notably, the first cluster of Full-Time Persisters has the second highest transfer rate (29%), suggesting that students in this group, who have more intense enrollment, seek to transfer without first obtaining a credential. Students in the Early Attachers cluster have the highest transfer rates and graduation rates (33% and 43%, respectively). Students in this highly attached group are earning a lot of credits early on, earning a 2-year degree, and then transferring upward. Their outcomes are markedly different than those of the Later Attachers, who are earning a 2-year credential at a slower pace and are less inclined to transfer to a 4-year college within the observed time frame (14% transfer). The Early Leavers, Early Persistent Switchers, and Mostly Part-Timers have about the same transfer rates as the Later Attachers (14%-15%), but Later Attachers have a much higher graduation rate of 37%.

Figure 3.

Percentage of students who transferredto a four-year institution within 5 or 6 years by enrollment pattern cluster.

Figure 4 shows how the clusters vary for the two outcomes: earning a credential or transferring to a 4-year institution. Early Attachers, Later Attachers, and Full-Time Persisters having the highest likelihood of one of the outcomes. A combination of attached, intense enrollment with few breaks is associated with the greatest probability of transfer or earning a credential.

Figure 4.

Percentage of students who transferred or earned a credential within 5 or 6 years by enrollment pattern cluster.

Taken together, the six clusters support the fourth main finding of this article: Students in groups characterized by high levels of enrollment continuity (Early Attachers and Later Attachers) are more likely to earn a credential than students in groups with low levels of continuity. Moreover, students in groups characterized by high levels of intensity and consecutive full-time enrollment (Full-Time Persisters and Early Attachers) are more likely to transfer to a 4-year college than students in groups with low levels of enrollment intensity. Although not causal, these relationships suggest that taking breaks in enrollment (discontinuous enrollment) may be particularly harmful for students who desire to earn a credential and that part-time enrollment may be particularly harmful for students who desire to transfer. For credential seekers, it is important to maintain consecutive enrollment; for transfer seekers, it is important to earn credits early. Although it does not appear that the frequency of switching between full- and part-time states is detrimental, groups identified by mostly part-time or discontinuous enrollment have lower credential-earning and transfer rates. Continuity of enrollment and full-time study are critical for student success.

Discussion and Implications

The patterns of enrollment intensity and continuity identified in this study are the result of choices made at different points in time under different constraints. Students do not randomly switch between full- and part-time enrollments, but rather act rationally (though not always optimally) in accordance with particular circumstances. The clusters provide a way of looking at groups of students who made similar decisions and asking important questions about their behavior. For example, the Full-Time Persisters cluster contains many students who did not return to college after their first part-time enrollment. Does this behavior lead to a diagnosis of part-time status as a harbinger of dropout? The largest cluster of Early Leavers consists of students who make a similar decision to stop enrolling after very little time at the community college. What cost–benefit analyses are these students making that lead them to leave college so quickly? What factors contribute to this decision, and are the factors different for different groups of students?

Policy and Program Implications

Gaining a better understanding of the student experience is invaluable for various stakeholders. Informing college personnel about the types of enrollment patterns and their relative sizes illuminates the fact that too many students do not persist past the first term (e.g., Early Leavers) and few students choose the most efficient or recommended path toward earning a credential. Although it may be challenging for an institution to identify which students are in which clusters at the outset of college, performing this type of analysis on historical data would provide faculty, advisors, and student services professionals with a clearer understanding of student pathways. Faculty and administrators may not realize the wide variation in enrollment and the high likelihood that many students have significant interruptions in their enrollment; they depart from any type of “traditional” college pathway. For faculty, understanding interruptions may encourage more review during the first few lectures of a course. In addition, a focus on key subgroups, such as students who require various levels of developmental education and students in particular curricular programs, can lead to more effective interventions that are made when they are most needed. For example, if students in certain programs are likely to have interruptions in enrollment at certain points in time, college personnel may look at the program design to make changes or try different advising strategies. Similarly, if students in other subgroups are found to be systematically skipping the important fall or spring terms, student services and advising professionals may investigate causes and develop strategies to encourage more continuous enrollment. Chaotic and varied enrollment patterns provide challenges for college administrators tasked with scheduling classes and determining staffing and resource requirements and advisors charged with making recommendations to students. Making a college-wide effort to acknowledge this is a first step in improving student pathways.

Acknowledging the range of patterns is important for policy makers as well. Credential completion or transfer for community college students takes longer than 2 or 3 years, and that policies and incentives related to financial aid, tuition, and placement testing potentially reinforces the suboptimal pathways taken by students. For example, policies surrounding the number of terms that Pell Grant Awards are available may be appropriate for most 4-year college students but not for persisting community college students. The total number of terms for Pell eligibility was recently reduced to 12 from 18, a policy change that may hinder completion for many community college students.¹⁶ When designing metrics for evaluating college performance, policy makers should consider that college enrollment patterns range from those characterized by Early Leavers to Early Attachers—these different types of students make very different choices after their first contact with the college and may have different goals.

Implications for Future Research

The analysis presented here should give pause to researchers studying community college student behavior who are analyzing student pathways using short timelines as well as those using longitudinal designs. Many studies often use initial enrollment intensity when examining postsecondary student behavior and outcomes. First-term intensity may be a useful proxy for unobservable characteristics such as self-esteem and perceived academic ability as much as it is a function of financial and time constraints. However, it does not always indicate future enrollment intensity. Researchers studying transfer should be aware that the transition from 2- to 4-year college is often not immediate, requiring a close look at the timing of transfer.

One potential area for further research is the development of a model that can generate the observed enrollment patterns. Similar to that of Stange (2012) and Keane and Wolpin (1997), the strategy—though computationally complex—could model the dynamic decision-making process (enroll part-time, enroll full-time, work full-time, stop out, etc.) of these students over the life cycle. Researchers could then carry out policy simulations to study how enrollment decisions and postsecondary outcomes would change in response to changes in opportunity costs of going to college, tuition, self-assessment of academic ability (based on experiences at college), institutional structures, remediation placement policies, and so on.

A second area of research could explore changes in enrollment intensity more closely. Does switching between full- and part-time enrollment help or harm students? That is, is it better for students to have consistent enrollment of one type or to just accumulate credits in any way possible? Is a switch from full- to part-time attendance undesirable? Does such a switch imply greater part-time attendance in subsequent terms? Related to these questions is the issue of modeling changes in enrollment intensity. Can one predict when students are likely to have a gap in enrollment, change intensities, or simply enroll in the next semester full- or part-time? What characteristics are associated with these transitions?

A third important area of study concerns transfer. Many students enter the community college with the desire to transfer to a 4-year institution. However, as Figure 1c shows, transfer pathways vary considerably across individuals. Students transfer at many different points in time with varying numbers of transferable credits. There has been some research exploring the nature of upward transfer (see Long & Kurlaender, 2009), but there still remain several outstanding questions concerning enrollment patterns, transfer, and baccalaureate completion. For example, are the patterns associated with successful upward transfer related to completion of a bachelor’s degree as well? And how do disruptions between community college enrollment and 4-year college enrollment affect degree completion?

Summary and Conclusion

This article presents a way to conceptualize and visualize community college enrollment patterns and to cluster them by enrollment characteristics using student-level data from a sample of 14,429 degree- or transfer-seeking FTIC students from five community colleges located in a single state who began in the 2005-2006 or 2006-2007 school year. After 5 to 6 years, most of these students forged paths that are not highly productive or efficient. The diversity in individual patterns cannot be overstated—although nearly half of the students followed about 10 patterns (most of them associated with early attrition from college), the remaining students took thousands of distinct pathways involving full-time, part-time, and interrupted enrollment. Characterizations of students as either part- or full-time are thus largely inaccurate as they ignore the high degree of switching between these two enrollment statuses. The chaotic enrollment patterns of students illustrated in this study pose challenges for colleges and other stakeholders in helping students enter and complete programs of study. Clustering these enrollment patterns based on intensity, persistence, interruption, and frequency reveals six major pattern types. The most favorable graduation outcomes are associated with students who tend to enroll term after term with few breaks. The most favorable upward transfer outcomes are associated with students who tend to enroll full-time rather than part-time. Continuity of enrollment and full-time enrollment whenever possible are keys to community college success.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this research was provided by the Bill & Melinda Gates Foundation.

Notes

Author Biography

Peter M. Crosta, PhD. is a Data Scientist at the Community College Research Center, Teachers College, Columbia University. His research interests lie within the economics of education and include quantitative analysis of educational policy change and visualization of educational data.

References

Adelman

(1999). Answers in the tool box: Academic intensity, attendance patterns, and bachelor’s degree attainment. Washington, DC: U.S. Department of Education. Retrieved from http://www.ed.gov/pubs/Toolbox/toolbox.html

Adelman

(2005). Moving into town and moving on: The community college in the lives of traditional-aged students. Washington, DC: Office of Vocational and Adult Education, U.S. Department of Education. Retrieved from http://www2.ed.gov/rschstat/research/pubs/comcollege/index.html

Adelman

(2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education. Retrieved from http://www2.ed.gov/rschstat/research/pubs/toolboxrevisit/toolbox.pdf

Ammon

B. V.

Bowman

Mourad

(2008). Who are our students? Cluster analysis as a tool for understanding community college student populations. Journal of Applied Research in the Community College, 16, 32-44.

Attewell

Heil

Reisel

(2012). What is academic momentum? And does it matter? Educational Evaluation and Policy Analysis, 34, 27-44.

Bahr

(2010). The bird’s eye view of community colleges: A behavioral typology of first-time students based on cluster analytic classification. Research in Higher Education, 51, 724-749.

Bailey

Jeong

D. W.

Cho

S. W.

(2010). Referral, enrollment, and completion in developmental education sequences in community colleges. Economics of Education Review, 29, 255-270.

Bean

J. P.

(1980). Dropouts and turnover: The synthesis and test of a causal model of student attrition. Research in Higher Education, 12, 155-187.

Calcagno

J. C.

Crosta

Bailey

Jenkins

(2007). Does age of entrance affect community college completion probabilities? Evidence from a discrete-time hazard model. Educational Evaluation and Policy Analysis, 29, 218-235.

10.

Driscoll

A. K.

(2007). Beyond access: How the first semester matters for community college students’ aspirations and persistence (Policy Brief 07-2). Berkeley: Policy Analysis for California.

11.

Hagedorn

L. S.

Cabrera

Prather

(2010). The community college transfer calculator: Identifying the course-taking patterns that predict transfer. Journal of College Student Retention: Research, Theory & Practice, 12, 105-130.

12.

Hagedorn

L. S.

Prather

(2005). The community college solar system: If university students are from Venus community college students must be from Mars. Paper presented at the 2005 Annual Forum of the Association for Institutional Research, San Diego, CA.

13.

Horn

Nevill

(2006). Profile of undergraduates in U.S. postsecondary education institutions: 2003-04: With a special analysis of community college students (NCES 2006-184). Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics.

14.

Jenkins

Cho

S. W.

(2012).Get with the program: Accelerating community college students’ entry into and completion of programs of study (CCRC Working Paper No. 32). New York, NY: Teachers College, Community College Research Center, Columbia University.

15.

Keane

M. P.

Wolpin

(1997). The career decisions of young men. Journal of Political Economy, 105, 473-522.

16.

Long

B. T.

Kurlaender

(2009). Do community colleges provide a viable pathway to a baccalaureate degree? Educational Evaluation and Policy Analysis, 31, 30-53.

17.

McCormick

A. C.

Geis

Vergun

(1995). Profile of part-time undergraduates in postsecondary education: 1989-90. Washington, DC: U.S. Department of Education, National Center for Education Statistics.

18.

Metz

G. W.

(2004). Challenge and changes to Tinto’s persistence theory: A historical review. Journal of College Student Retention Research Theory & Practice, 6, 191-207.

19.

Metzner

B. S.

Bean

J. P.

(1987). The estimation of a conceptual model of nontraditional undergraduate student attrition. Research in Higher Education, 27, 15-37.

20.

O’Toole

D. M.

Stratton

L. S.

Wetzel

J. N.

(2003). A longitudinal analysis of the frequency of part-time enrollment and the persistence of students who enroll part-time. Research in Higher Education, 44, 519-537.

21.

Pascarella

E. T.

Terenzini

P. T.

(1991). How college affects students. San Francisco, CA: Jossey-Bass.

22.

Radford

A. W.

Berkner

Wheeless

S. C.

Shepherd

(2010). Persistence and attainment of 2003-04 beginning postsecondary students: After 6 years (NCES 2011-151). Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics.

23.

Stange

(2012). An empirical examination of the option value of college enrollment. American Economic Journal: Applied Economics, 4, 49-84.

24.

Stratton

L. S.

O’Toole

D. M.

Wetzel

J. N.

(2004). Factors affecting initial enrollment intensity: Part-time versus full-time enrollment. Economics of Education Review, 23, 167-175.

25.

Stratton

L. S.

O’Toole

D. M.

Wetzel

J. N.

(2006). Are the factors affecting dropout behavior related to initial enrollment intensity for college undergraduates? (IZA Working Paper 1951). Bonn, Germany: Institute for the Study of Labor. Retrieved from http://ssrn.com/abstract=880426

26.

Tinto

(1975). Dropout from higher education: A theoretical synthesis of recent research. Review of Educational Research, 65, 89-125.

27.

VanDerLinden

(2002). Credit student analysis: 1999 and 2000. Annapolis Junction, MD: Community College Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

25.04 MB