The Impact of Community Colleges on the School-to-Work Transition: A Multilevel Analysis.

In the past several decades scholars have expressed increasing concern about the school-to-work transition at all levels of education (William T. Grant Foundation 1988; Commission on the Skills of the American Workforce 1990; Hershey et. al 1999). Major policy initiatives, such as the School-to-Work Opportunities Act, recognized the community college as an essential component in helping subbaccalaureate students make the transition from school to work. A number of scholars have described the role of the community college as a "transitional institution" and as a "nexus" or a means of connection to the workplace (McCabe, 1998; Grubb, 1996; Grubb, et. al., 1997; Orr, 1998), especially in the rapidly changing technological and economic context of the late twentieth and early twenty-first century (Alfred et. al., 1999). The localized orientation of community colleges would seem to make them especially well-suited to respond to labor market demands and student goals.

Defenders of community college vocational education maintain that community college education applies directly to the demands of the labor market. According to Parnell (1985), the hallmark of high-quality community colleges is that "they [help to] trigger economic revitalization by matching skills to the needs of employers" (p. 95). Indeed, Shults (1997) recognized the need for community colleges to "secure the lead" in providing workforce training to their students. As suggested by Schuyler's (1997) review of training programs, community colleges are viewed as leaders in workforce development through their investment in the human capital of their students.

However, others have expressed concern about vocational training in community colleges and the school-to-work transition. Critics have claimed that community colleges are not fully accountable for student outcomes because they are not required to substantiate their positive claims about job availability, placement, and potential salaries (Diener, 1986). Pincus (1986) contended that community college vocational programs fail to meet students' expectations: "By and large, there is no good evidence that vocational education in community colleges delivers on the promises of secure employment, decent pay and ample career opportunities" (p. 49). More recently, Dougherty (1994) asserted that although community colleges do attempt to respond to industry needs, they do so inefficiently and "often under and overshoot the demands of the labor market, in many cases, training far more people than the labor market can absorb and in other cases, producing fewer workers than business would like" (p. 67).

This doubt about the effectiveness of community colleges in preparing students for work points to the need for additional research on the topic. However, a careful search into the educational literature reveals that few studies specifically address the role of community colleges in bridging the transition from school to work. The U.S. Department of Education (1994) has noted that a small number of community colleges have developed courses on education-labor linkages and school-to-work issues.

Perhaps the most relevant literature on the role of community colleges in linking school to work is the research on Technical Preparation (Tech Prep) programs. These partnerships between high schools and community colleges are geared toward preparing young adults for the work place and are seen to be an important component of school-to-work efforts (Orr, 1998). By 1993, nearly one half of all school districts had formulated Tech Prep articulation agreements with postsecondary institutions (U.S. Department of Education, 1994). Few of these programs, however, have been subject to comprehensive evaluations. The few that have been conducted show mixed findings regarding effectiveness (Parnell, 1985; Pollard, 1991; Silverberg, 1996; Farmer & Honeycutt, 1999).

This lack of attention to the bridging role of community colleges is reinforced by the inordinate attention given to community college students who transfer to four-year colleges and universities. Focusing on the "school-to-school" transition of community college students continues a general trend of favoring college-bound high school students (those who plan to attend a four-year college or university directly after high school graduation) over non-college-bound students. Some researchers contend that community colleges provide a high level of assistance for students who plan to transfer to four-year colleges or universities; however, little support is offered to "those who are making equally difficult transitions from school to work" (Cross, 1990, p. 9).

Although transfer issues are a major concern for many community college leaders, research indicates that most community college students do not intend to transfer to other educational institutions (Dougherty, 1987; Grubb, 1991). Indeed, some researchers speculated that the transfer rate would actually decrease in the 1990's and into the twenty-first century. A comprehensive longitudinal study on transfer rates suggests that only 22% of community college students transferred over an eight-year period (Cohen & Sanchez, 1997). In the end then, the literature on transfer students and the school-to-school transition does not address some key concerns about the many community college students making the transition directly from school to the workplace.

Another weakness in the literature on school-to-work transitions of community college students is methodological in nature. Studies that address the role of community colleges in facilitating students' entry into the labor market tend to conduct analyses at the individual (the student) level. These single-level models rarely consider the contributions to student outcomes from other levels (for example, classroom, school, and even state-level variables). The need for a more thorough analysis of the institutional context of community college education was recognized by the National Research Council (1994) in its report on vocational education.

In the past, research on educational effectiveness tended to focus on student background characteristics. This body of research, although important, does little to address the impact of the school context on student outcomes. An influential work in regard to the school effects literature was James Coleman's (1961) study, which attempted to determine the relationship between formal organizational properties of schools and student outcomes. This study, known as the Coleman Report, found that variations in student performance on an achievement tests were not explained by school characteristics. Jencks et. al. (1972) reached similar conclusions in their study of education and inequality. Their research indicated that school-related variables accounted for less than 4% of the variance in student outcomes. On the other hand, student background variables account for nearly 40% of the variance.

Other studies have alluded to the shortcomings of the Coleman studies. Immediate feedback from Bowles and Levin (1968) suggested that the Coleman's research was "biased in a direction that would dampen the importance of school characteristics" (p. 8). Likewise, van den Eeden and Huttner (1982) criticized Coleman's methodology along similar lines, especially for its failure to consider higher level contexts such as the school district or even state-level variables.

The literature on school-effects research offers insight into the importance of considering institutional-level variables when evaluating how well community college students transition into the workforce. This body of research focuses on additional school characteristics, reflecting both quantitative and qualitative measures, including curriculum, quality of teachers, administrative structure, socioeconomic composition, school policies, organizational characteristics, and student enrollments (Raudenbush & Bryk, 1988). Both quantitative (Berman & Weiler, 1990; Banks, 1992) and qualitative studies of community colleges (Valadez, 1994; Valadez, 1996; Shaw, Valadez, & Rhoads, 1999) have attempted to place student outcomes in an institutional context.

The current study was designed to illuminate the multilevel nature of the school-to-work transition. More specifically, the main objective was to investigate whether and how community college-level variables influence student outcomes. Using hierarchical linear modeling (HLM), the analyses goes beyond traditional regression equations by considering the multilevel nature of the school-to-work transition. The value of HLM is that it extends beyond the single-level multiple regression models that have often been used in the past to understand student outcomes.

One major premise underlying the research was the following: outcomes are influenced not only by various student background characteristics such as race, gender, and SES, but also by the context within which outcomes are achieved. Thus, the investigation centered around the following questions: Does the structure of a community college make a difference in how well a young adult makes the transition from school to work? Do certain structural characteristics of the community college affect certain types of students differently than others? That is, are each of these community college characteristics differentially effective for students on the basis of gender, race, and socioeconomic status (SES)?

Research Design and Methodology

This investigation relied on several sources of data for the multilevel analysis. The High School and Beyond Fourth Follow-Up and the Postsecondary Educational Transcript Study (PETS) were the source of national student-level data. Community college variables were obtained from the Integrated Postsecondary Data System (IPEDS) and the College Board Institutional Files. Together, these comprehensive data sets allowed for a study of the relationships among students, curricula, and community college correlates and how these relate to labor market outcomes.

Using actual student transcripts, a series of variables were constructed to summarize degree attainment and course-taking patterns. First, the highest degree earned was categorized as 0 for no degree, 1 for a certificate, and 2 for an associate's degree. Second, each student-level transcript was then categorized along 10 occupational clusters. Students who took at least 21 credit hours in an occupational cluster were categorized as vocational students. Those who took less than 21 hours were considered to be nonvocational students and were coded as 0. The PETS also provided information about students' GPAs. Table 1 provides a summary of the research variables used in the current analysis.

Table 1
Summary, of Variables Used in HLM Analyses

Dependent Variables

Wage Rates                        1991 annual wages

Use of Education on the Job       Opportunity to use past
                                    training and education on
                                    the job: 1=none; 2=some;
                                    3=a lot

Independent
 Variables: Student-Level

Gender                            A dummy variable representing
                                    respondent's gender:
                                    male=0; female=1

Race/Ethnicity                    A dummy variable representing
                                    student's race/ethnic
                                    background: nonminority=0;
                                    minority=1

Socioeconomic Status (SES)        Composite variable representing
                                    student's SES background:
                                    1=lowest quartile; 2=second
                                    quartile; 3=third quartile;
                                    4=highest quartile

Occupational Expectations         Response to question asking
                                    for respondents' expected
                                    job at age 30: 1=clerical;
                                    2=craftsman; 3=farmer; farm mgr;
                                    4=homemaker: 5=laborer; 6=mgr;
                                    7=mgr/admin; 8=military;
                                    9=operative; 10=professional;
                                    11=professional, doctor;
                                    12=proprietor, owner;
                                    13=protective service;
                                    14=sales; 15=school teacher;
                                    16=service; 17=technical

Vocational Concentration          A dummy variable representing
                                    student's course-taking pattern:
                                    0=nonvocational student;
                                    1=vocational student

Grade Point Average (GPA)         Student's GPA, ranging from
                                    0.0 to 4.0

Highest Degree Earned             Student's highest degree earned:
                                    0=no degree; 1=certificate,
                                    2=associate's degree

Independent Variables:
 Community College-Level

Enrollments                       Continuous variable representing
                                    total fall enrollment, 1987

Percentage of Faculty             Percentage of full-time
 Full-time                          instructional faculty

Transfer Rate                     Percentage of students who transfer
                                    to a 4-year institution

Career Counseling Placement Services
Dichotomous variable indicating availability of each of these
support services: 0=no; 1=yes

The analysis was limited to those individuals in the national data set who earned at least 30 credits at a community college, whether or not they earned an associate's degree. This credit limit is based on the research of Kane and Rouse (1993) which demonstrated that community college students with at least 30 hours of credits experienced positive outcomes. Second, students whose highest degree was an associate's degree and who had earned less than 20 credits at a four-year college or university were included in the sample. The sample does not include any students who attended a four-year college or university exclusively. The main objective in selecting the sample was to isolate those students whose major postsecondary educational experience was in a community college. The study focuses on those institutions that grant an associate's degree as their highest degree, including both private and public community colleges.

Dependent Variables

Wages. The measure of wages used for the current study was annual earnings in 1991, as recorded in the High School and Beyond survey. This variable, therefore, does not include other sources of income, such as public sources of income and family support. The latter could be particularly relevant to the sample under study. That is, young adults are probably more likely to rely on family support during the early stages of career development. It should be noted, therefore, that because of these limitations, the wage variable may actually underestimate student income. Additionally, when interpreting the results, it is important to note that wages were not standardized by number of hours worked or full-time work only or by differing wage levels by geographic location.

Use of Training. The literature on the returns to community college training suggests mixed results regarding whether graduates find training related to their field of study and whether there are benefits from doing so. Indeed, as Cohen and Brawer (1989) stated, "a person who receives a degree or certificate and who does not work in the field in which that certificate was earned does not represent an institutional indictment" against the community college (p. 358).

Investigators have also recognized the inherent difficulty in measuring the use of training on the job (Carter, 1975; Cohen & Brawer, 1989). The greatest challenge in measuring this variable is the fact that it is self-reported. Students may underestimate their use of training on the job, and they also may not understand just what "use of training" entails. The High School and Beyond question used for this particular dependent variable asked students how different their job was from the one they were trained or educated to do. The variable was originally coded as 0 = None; 1 =Some; 2=A lot. This coding scheme was reversed for the current analysis, as reflected in Table 1 (0=A lot; 1 =Some; 2=None). It is recognized that some students may not have been referring to their community college education when answering this question. Nevertheless, because the particular sample of students pursued their postsecondary education in community colleges, this is more than likely the education that they were referring to when answering the question.

Independent Variables: Community College Characteristics

As indicated above, much of the quantitative research on community college students inadequately addresses the influence of context on student outcomes. This project considers various contextual factors that prior research suggests are important in accounting for variations in student outcomes. In this way, it is easier to determine whether variation in the community college-level variables has an effect on student outcomes. The community college variables that were modeled for this investigation are described below and are referred to in the third panel of Table 1.

Community College Enrollments (Size). The sociological literature indicates that size of an organizational system may influence a system's performance. As systems increase in size, more levels of hierarchy are created and tasks become more complex to manage (Mintzberg, 1979; Hollingsworth & Hanneman, 1984). The literature presents mixed results, however, as to impact of an educational institution's size on student outcomes. Several studies at the secondary and postsecondary level uncovered positive relationships between size and outcomes, citing the economies of scale which allow larger institutions to offer more programs and degrees than smaller institutions (for example, Kuo, 1999); some demonstrate an inverse relationship (such as Anderson, 1984; Puyear, 1989; Huffman & Schneiderman, 1997); still other studies find no association between the size of organizations and student outcomes (for example, Todd & Baker, 1998; Antley, 1999).

Employment of Faculty. The second community college-level variable that could potentially influence student learning experiences, and ultimately student outcomes, concerns the employment of faculty at community colleges. Of all institutions of higher education, community colleges employ part-time faculty more frequently than other institutions of higher education (Wilson, 1994). A tendency during the 1980s and 1990s to hire faculty on a part-time contingent basis has concerned several researchers (McCabe, 1988; Cain 1999). In 1977, the number of part-time faculty in community colleges exceeded the number of full-time faculty members for the first time (Pollock & Breuder, 1982). By 1994, 65% of community college faculty members worked part-time at their institutions (American Association of Community Colleges, 1996).

The several advantages to hiring part-time faculty include the ability to respond better to community needs; budgetary flexibility; and the abiity to explore new program offerings on a short+term experimental basis, thus bringing more diverse and current curriculum into the classroom (McGuire, 1993; Lankard, 1993; Avakian, 1995). Antony and Valadez (1998) found part-time workers ranked higher on overall job satisfaction and were more committed to an academic career than their full-time counterparts. Willet (1980) found few differences between part-time and full-time faculty in terms of student ratings of teaching, course retention ratios, and student achievement in advanced courses, a finding supported by Grubb's (1999) more recent qualitative study of teaching in community colleges.

However, the literature also criticizes the use of part-time faculty and emphasizes the importance of a full-time faculty for achieving better outcomes for students (Eaton, 1994). Several studies present evidence that part-time faculty are less certain about their place in the institution, are viewed as less prepared to teach, less committed to the institution, and are not as well integrated into the local community (Cottingham, Newman, & Sims, 1982; McGuire, 1993; Griffith & Connor, 1994). It is also hypothesized that part-time faculty have little voice in institutional governance and are not encouraged to develop their skills to become better teachers and advisors to students (Rouche, Rouche, & Milliron, 1995; Little Hoover Commission, 2000).

In regard to school-to-work issues, part-time faculty are often hired because of their "real world" experience. However, few links have been made between using part-time faculty and its effect on training of students who plan to enter the workforce after leaving the community college. Beman (1980) notes that students may become increasingly skeptical of instructors who are unable to relate subjects of study to the world of work. Evidence suggests that "part-time instructors are generally less well connected to the local labor market than are full time instructors, partly because of time constraints" (Grubb, 1991, p.333). Other studies find that some aspects of part-time instruction could be the causes of student attrition, which in turn affects the eventual transition into the workplace. These factors include the alleged lower teaching quality of part-time faculty and the limited interaction between students and parttime faculty members (Albert & Watson, 1980).

Vocational Emphasis. This study aims to illuminate whether and how a vocational emphasis might influence student outcomes by isolating whether and how transfer rates influence student outcomes. In an American College Testing Program study (1969) community colleges designated as having a "vocational emphasis" were those that placed a priority on practical vocational training, with little attempt to model their curriculum to encourage transfer to four-year institutions (p. 23). Those schools that emphasized transfer curricula were designated as having a "transfer emphasis." It is recognized, however, that just because a particular institution has a high transfer rate does not mean that it does not also have a high quality vocational and occupational training program. For the purposes of this study, however, lower transfer rates serve as a proxy for a higher vocational emphasis (Dougherty, 1991).

Availability of Student Services. Several researchers have supported the notion that one of the essential elements of a community college system is providing career counseling to students (Stern, et al., 1995; Paneitz, 1997). As community colleges assume a greater role in providing vocational training, career counseling and job placement services have become increasingly important (Howard, 1998). Some features of student-support services at community colleges include assessment aids, career resource materials, counseling, follow-up studies, job listings, job search workshops, transfer assistance, co-op work experience, on-campus interviews, and career days (Stern, et. all, 1995).

Despite the fact that this school-based component is a vital part of the school-to-work legislation of the 1990s, little research has been done on the impact of student services on student outcomes (Creamer, 1994). Several sources have described vocational counseling at community colleges as in need of improvement (Lach, Kohl, & Wellman, 1979; Hammond & Porter, 1984). For example, deVries' (1998) study of community colleges in Wisconsin found that significant improvements were needed in student career services, including collaboration with area businesses and promotion of career services. Valadez's (1996) ethnographic study supports these findings, with a specific emphasis on how such limited services impact minority students and students from lower social classes. In the end then, community college students attempting to move into the workforce often bear the burden of making successful transitions (U.S. Department of Education, 1994).

In summary, this study uses information about four community college characteristics--enrollment percentage of full-time instructional faculty, percentage of students who enter a four-year program, and availability of career counseling and job placement services--to investigate a community college's impact on student outcomes.

Research Hypotheses

the study aims to illuminate the multilevel nature of the school-towork transition process by considering community college structural variables that could potentially influence outcomes, while controlling for a number of student background characteristics.

The following research hypotheses guided this research: Hypothesis I, those students who attend community colleges with lower enrollments; a higher percentage of full-time faculty; a lower transfer rate; and the availability of career counseling and job placement services will experience better outcomes on average in terms of wages and use of training on the job. The strength of a multi-level analysis is its ability to analyze how higher levels of variables mediate the relationship between lower levels of variables. Thus, Hypothesis II is as follows: the more "supportive" a community college is toward vocational education (along the lines described above for Hypothesis I), the smaller the race, gender, and SES gap in wages and use of training on the job.

Data Analysis

Two models were developed and tested to express within-school and between-school relationships. The first group of models aimed to investigate the relationship between community college-level variables and student outcomes. A separate model was developed for each of the social background characteristics of students (race, gender, and SES). The second aim was to determine if the characteristics of community colleges moderated the individual-level relationships between race, gender, and SES and the dependent variables.

Basic hierarchical models consist of a two-step process that analyzes the relationships within and between the different levels (student-level and community college-level). See Arnold, 1995 for a similar example that uses the same approach. Each of these steps, while carried out simultaneously during the HLM calculations, are explained separately below.

Level-1 models

This level of analysis attempted to model the relationships among various student background characteristics and student outcomes. In this regard, a separate regression equation was estimated for each college attended by the national sample of students selected. In the first step, the within-school models were estimated using ordinary least squares (OLS) regression analysis. Using data on the selected students, the dependent variables were represented as a function of student characteristics. In the level-1 equation, the first within-school model estimated wages (and use of training) as a function of the social background of the students, including gender, race, and SES.

These models resulted in an equation for each community college that consisted of regression coefficients, called Betas in HLM, that estimated how the independent variables influenced student outcomes. The equation for each community college was as follows:

[y.sub.ij] = [B.sub.0j] + [B.sub.1j] [X.sub.1ij] + [B.sub.2j] [X.sub.2ij] + [B.sub.3j] [X.sub.3ij] + [B.sub.4j] [X.sub.4ij] + [B.sub.5j] [X.sub.5ij] + [B.sub.6j] [X.sub.6ij] + [B.sub.7j] [X.sub.7ij] + [r.sub.ij] where: i represents the [i.sup.th] student;

j represents the [j.sup.th] community college;

[y.sub.ij] represents outcomes of the [i.sup.th] student in the [j.sup.th] community college;

[B.sub.0j] is the intercept, or the average outcomes, in each community college;

[B.sub.1j] is the Beta coefficient for gender in the [j.sup.th] community college;

[B.sub.2j] is the Beta coefficient for race-ethnicity in the [j.sup.th] community college;

[B.sub.3j] is the Beta coefficient for SES in the [j.sup.th] community college;

[B.sub.4j] is the Beta coefficient for GPA, in each community college;

[B.sub.5j] is the Beta coefficient for expectations in the [j.sup.th] community college;

[B.sub.6j] is the Beta coefficient for highest degree in the [j.sup.th] community college;

[B.sub.7j] is the Beta coefficient for vocational concentration in the [j.sup.th] community college;

[X.sub.1ij] represents the gender of the [i.sup.th] student in the [j.sup.th] community college;

[X.sub.2ij] represents the race-ethnicity of the [i.sup.th] student in the [j.sup.th] community college;

[X.sub.3ij] represents the SES of the i[th] student in the [j.sup.th] community college;

[X.sub.4ij] represents the GPA of the [i.sup.th] student in the [j.sup.th] community college;

[X.sub.5ij] represents the expectations of the [i.sup.th] student in the [j.sup.th] community college;

[X.sub.6ij] represents the highest degree of the [i.sup.th] student in the [j.sup.th] community college;

[X.sub.7ij] represents the vocational concentration of the [i.sup.th] student in the [j.sup.th] community college;

[r.sub.ij] is random error in the [j.sup.th] community college.

Under this model, the within-school parameters of interest modeled at level-2 were race, gender, and SES. The other student-level variables (GPA, expectations, highest degree, and vocational concentration) were treated as control variables.

Level-2 models

During the second stage of HLM, the intercept and the regression coefficients from the level-1 model were used as outcome measures (dependent variables) in the level-2 model. This part of the analysis was designed to determine whether community college-level factors have an influence on student outcomes, when relevant background charac-teristics are controlled. These between-community college equations resulted in coefficients (called Gammas) that estimated the association of each community college-level characteristic with the average outcomes and the differences in outcomes by race, gender, and SES.

In this level of the analysis, the community college-level characteristics were added to each model to explain the variation in each of the race, gender, and SES parameter estimates. Thus, the purpose of the gender variable was to investigate the differences in outcomes between males and females; the race variable was used to compare the achievement of minorities to non-minorities; the SES was used to examine the influence of class on student outcomes. The equation to predict outcomes is as follows:

[B.sub.0j] = [Y.sub.00] + [Y.sub.01] [W.sub.01j] + [Y.sub.02] [W.sub.02j] + ... + [Y.sub.0m] [W.sub.0mj] + [U.sub.0j] (model for intercept coefficient)

[B.sub.1j] = [Y.sub.10] + [Y.sub.11] [W.sub.11j] + [Y.sub.12] [W.sub.12j] + ... + [Y.sub.1m] [W.sub.1mj] + [u.sub.1j] (model for gender coefficient)

[B.sub.2j] = [Y.sub.20] + [Y.sub.21] [W.sub.21j] + [Y.sub.22] [W.sub.22j] + ... + [Y.sub.2m] [W.sub.2mj] + [u.sub.2j] (model for race coefficient)

[B.sub.3j] = [Y.sub.30] + [Y.sub.31] [W.sub.31j] + [Y.sub.32] [W.sub.32j] + ... + [Y.sub.3m] [W.sub.3mj] + [u.sub.3j] (model for SES coefficient) where [Y.sub.p1] is the Gamma coefficient for the first community college-level variable in the [p.sup.th] equation;

[Y.sub.p2] is the Gamma coefficient for the second community college-level variable in the [p.sup.th] equation;

[Y.sub.pm] is the Gamma coefficient for the [m.sub.th] community college-level variable in the [p.sup.th] equation;

[W.sub.p1j] represents the first value of the first community college-level variable in the [j.sup.th] college in the [p.sup.th] equation;

[W.sub.p2j] represents the second value of the first community college-level variable in the [j.sup.th] college in the [p.sup.th] equation;

[W.sub.pmj] represents the value of the [m.sup.th] community college-level variable in the [j.sup.th] college in the [p.sup.th] equation; m is the number of community college-level parameters.

Results

The following section presents the HLM results for the dependent variables wages and use of training. The coefficients for HLM are interpreted somewhat differently than for the coefficients in traditional multiple regression analyses. As indicated in Table 1, gender and race were dummy variables (with females = 1, males = 0; minorities = 1, nonminorities = 0). thus, for the level-2 (student level) models of the gender, race, and class effects in wages and use of training, a positive coefficient indicates females faring better than males and minorities faring better than nonminorities, all else being constant. If the coefficient is negative, that is an indication that females and minorities are doing worse in terms of outcomes, as compared to males and nonminorities. Because SES is a continuous variable (see Table 1 for values), the interpretation of these coefficients is somewhat different. In this case, a positive coefficient indicates a larger difference (gap) between students of high and low SES; that is, the variable is positively related to the social class slope. On the other hand, a negative coefficient would indicate a smaller difference in wages between students of high and low SES; thus the gap between social classes would be smaller (and the effect of social class would be smaller).

Wages

the results for the equation predicting average wages indicate that one organizational variable was significantly related to student outcomes (see Table 2). Average annual wages were higher in community colleges with higher enrollments. Although this difference in annual wages was significant (p [is less than or equals] .01), the difference was small (as indicated by the gamma coefficient). None of the other community college variables were significantly related to average wages.

Table 2
Effects of Community College-Level Variables on

Predictors of Student Outcomes (WAGES)

Effect                     Gamma Coeff.      S.E.       T-Value

On Intercept (Avg. wages)

Intercept                      19435.34       441.29   44.04(**)
Enrollments                       15.80         0.05    3.09(***)
% Faculty FT                      -2.85        21.15   -0.14
Transfer Rate                    -13.23        21.75   -0.61
Career Couns.                    454.92      2010.37    0.23
Placemnt. Ser.                  -913.04      1276.21   -0.72

On Race Slope Coefficient

Intercept                      -1799.83      1790.93   -1.01

Enrollments                       -0.13         0.17    -0.80
% Faculty FT                       6.71        83.41     0.08
Transfer Rate                     27.09        83.24     0.33
Career Couns.                  -7441.19      1257.01    -0.69
Placemnt. Ser.                 -4478.99      4231.57    -1.06

On Gender Slope Coefficient

Intercept                      -4781.71      1342.05   -3.56(***)
Enrollments                       -0.19         0.14   -1.38
% Faculty FT                     -74.98        64.11   -1.17
Transfer Rate                   -115.06        65.04   -1.77(*)
Career Couns.                   9879.29       647.18    1.53
Placemnt. Ser.                  -792.16      3562.99   -0.22

On SES Slope Coefficient

Intercept                        106.10       366.40    0.29

Enrollments                        0.05         0.04    1.24
% Faculty FT                       3.93        17.79    0.22
Transfer Rate                    -35.31        21.17   -1.67(*)
Career Couns.                  -1926.11      1567.17   -1.23
Placemnt. Ser.                  3410.54      1453.81    2.34(**)

(*) p [is less than or equal to] .10

(**) p [is less than or equal to] .05

(***) p [is less than or equal to] .01

The race-ethnicity slope equations reveal that, on average, no variables predicted the variation in the race-ethnicity effect in wages. Additionally, after controlling for the between-college variables, there continued to be no difference, on average, between minorities and nonminorities in annual wages, as indicated by the nonsignificant intercept.

When controlling for the various between-college variables, there was a difference between females and males in average earnings, with females earning, on average, $4782 per year less than males as indicated by the significant intercept (p [is less than or equal to] .01). the general level of the scope coefficient of gender on wages is gauged by the intercept which equals 4781.7. Given significant gender effects in wages for most (or all) of the "school" models (discussed above), this is expected. The gender coefficients shown in the within-community college equation were negatively associated with transfer rates (p [is less than or equal to] .10). that is, the negative coefficient means that females are doing worse as compared to males in schools with higher transfer rates. Although the difference in wages was quite small ($115 per year), the significance (p [is less than or equal to] .10) should be noted.

The results for the SES equation indicate that two community college structural variables are related to class differences in earnings. First, community colleges with higher transfer rates had less of an SES effect on wages than community colleges with lower transfer rates, as indicated by the negative intercept (p [is less than or equal to] .10). Nevertheless, community colleges that have placement services are associated with a more disequalizing distribution of wages along class lines, as indicated by the positive intercept (p [is less than or equal to] .05). that is, this variable, availability of placement services, is positively related to social class's effect on wages. The positive SES coefficient ($3410) is significant and indicates larger differences between students of high and low SES. Thus, having placement services is associated with a stronger relationship between SES and wages.

Use of Training

The results for the equation predicting average use of training on the job indicate that, on average, use of training was higher in those community colleges with career counseling (p [is less than or equal to] .10) (Table 3). the race-ethnicity coefficient was not significantly associated with either the average use of training on job (the intercept) or any of the community college characteristics. When looking at the gender equation, however, several variables in the community college-student model were associated with a gender gap in the use of training on the job. First, on average, gender had a differentiating impact on the use of training on the job (p [is less than or equal to] .10) with females less likely to use training on the job (as indicated by the significant intercept). Second, community colleges in which career counseling and placement services were available had a wider gender gap in the use of training (p [is less than or equal to] .10 for both variables), as indicated by the negative coefficients. Thus, these community colleges differentiated more along gender lines, with females using their training less often than males. Also, when controlling for the between-college variables, there continued to be no gap, on average, between males and females in use of training, as indicated by the nonsignificant intercept. Thus, having placement services was associated with a stronger relationship between SES and use of training.

Table 3
Effects of Community College-Level Variables on
Predictors of Student Outcomes (TRAINING)

Effect                  Gamma Coeff.   S.E.         T-value

On Intercept (Avg. Wages)

Intercept                     2.18      0.03      75.58(**)
Enrollments                  -0.000     0.000     -0.27
% Faculty FT                  0.0005    0.001      0.43
Career Couns.                 0.17      0.11       1.57(*)
Placemnt. Set.               -0.022     0.08      -0.26

On Race Slope Coefficient

Intercept                     0.08      0.11       0.67
Enrollments                   0.000     0.000      0.68
% Faculty FT                 -0.006     0.005     -1.18
Transfer Rate                -0.003     0.005     -0.73
Career Courts.                0.49      0.68       0.72
Placemnt. Ser.                0.03      0.27       0.13

On Gender Slope Coefficient

Intercept                    -0.15      0.09      -1.70(*)
Enrollments                   0.000     0.000      0.53
% Faculty FT                  0.004     0.004      0.78
Career Couns.                -0.59      0.41      -1.72(*)
Placemnt. Set.               -0.38      0.23      -1.65(*)

On SES Slope Coefficient

Intercept                     0.0004    0.03       0.02
Enrollments                   0.000     0.000     -0.43
% Faculty FT                  0.001     0.001      0.98
Transfer Rate                 0.001     0.001      1.11
Career Couns.                -0.06      0.16      -0.61
Placemnt. Set.                0.22      0.07       3.11(***)

(*) p [is less than or equal to] .10

(**) p [is less than or equal to] .05

(***) p [is less than or equal to] .01

The SES coefficient was not associated with average use of training on the job by social class. One variable, however, emerged as statistically significant in regard to use of training on the job. Colleges with placement services were associated with a more disequalizing distribution of use of training along socioeconomic class lines (p [is less than or equal to] .01).

Discussion

The results of this study provide some insight into the potential impact of community college structural variables on average student outcomes (Hypothesis 1). Table 4 provides a summary of the results. The equation modeling the intercept for average wages shows student enrollments to be statistically significant in a positive direction. That is, average wages are higher in community colleges with higher. The processes leading to this finding may be attributed to the fact that larger community colleges may be able to benefit from the economies of scale that are associated with higher student enrollments. Some community colleges are reluctant to invest in occupational training programs because of the expense required to develop, implement, and maintain these programs. Larger schools, however, may be better able to offer such programs, which may generally train for higher paying occupations. Additionally, location in an urban area versus a rural area often confers more advantages in terms of job availability and higher wages.

Table 4
Summary of Multilevel Analysis:
Disequalizing vs. Equalizing Effects

                              Gender Gap              Race Gap
Community College
Variable                  Wages       Training     Wages   Training

Enrollments                --            --         --        --
% Faculty Part-time        --            --         --        --
Transfer Rate          Disequaliz.       --         --        --
Career Counseling          --        Disequaliz.    --        --
Placement Services         --        Disequaliz.    --        --

                               SES Gap
Community College
Variable                  Wages       Training

Enrollments                --            --
% Faculty Part-time        --            --
Transfer Rate           Equaliz.         --
Career Counseling          --            --
Placement Services     Disequaliz.   Disequaliz.

-- indicates nonsignificant findings

The second dependent variable modeled was use of training. The intercept equation for this variable reveals that the presence of career counseling has a positive relationship with the average use of training on the job. This finding is consistent with theoretical expectations. The positive influence of the presence of career counseling confirms the literature regarding the importance of career counseling for young adults attempting to make the transition from school to work (Bushnell, 1978; Fadale, 1982).

The hypothesized effects of community college structural variables on differentiation in the dependent variables with regard to race-ethnicity, gender, and SES were supported in several cases and not supported in others (Hypothesis 2). The equation modeled to predict the race-ethnicity gap in earnings shows that no community college structural variables were associated with the race-ethnicity gap in wages or use of training. This relationship merits further investigation. Since minorities tend to be disproportionately represented in the lower SES groups, a multilevel model that could account for the complexities in the relationship between race-ethnicity and class and outcomes is suggested. The literature points to the differential experiences of minority community college students as compared to nonminority students. For example, Valadez (1994) found that minorities faced barriers in their goals of transferring or entering more career oriented health programs. In another discussion, Valadez (1993) cites the need for community colleges to provide specialized services to minority students, especially as college campuses become more multicultural. Wechsler (1991) also calls for improvement in the support services offered to minority students.

The gender equations reveal that differences in student outcomes by gender were somewhat more apparent than differences by race-ethnicity. One variable in the community college level model was related to the gender gap in earnings. Colleges with a higher transfer rate had wider gaps between males and females, with females earning $115 per year less than males at such community colleges. Because higher transfer rates are associated with a weaker vocational emphasis, this result does not confirm theoretical expectations regarding the school-to-work transition. That is, it was hypothesized that the gender gap would be narrower in those schools with a lower transfer rate (and thus, a stronger vocational emphasis). It is not clear why males with a community college education seem to benefit more from community colleges with higher transfer rates; this relationship merits further investigation. Perhaps those community colleges with a high transfer rate devote an inordinate amount of attention to preparing students for transfer to a four-year college, rather than preparing them for the workplace, and females benefit less than males from this allocation of resources. It is also noted that the differences in wages ($115 per year) is not substantial. Nonetheless, the significant t-value suggests a need for further research.

Variables in the community college model for use of training point to several factors associated with the gender influence. The gender influence in use of training was greater in community colleges with career counseling and placement services. That is, community colleges which offered these services had a disequalizing effect along gender lines for both of these variables. This negative association between gender and availability of services is of some concern because of the supposed importance of such services in reducing gaps in student outcomes. This finding could possibly be related to career counseling practices that encourage females to enter gender-stereotyped traditional jobs, which are usually lower paying and less satisfying. Additionally, females may receive unclear signals regarding the potential to enter the fields in which they are trained and thus take jobs unrelated to their training and education. However, this seems to contradict the findings of Ahrens and Boatwright (1997) who found that female students were more satisfied than male students with career planning and job placement services offered at community colleges.

The results for this model also show significant differences in outcomes by social class background, although the results were contrary to expectations. Community colleges that have higher transfer rates (and thus a less "supportive" environment as defined in Hypothesis I) are associated with a more equalizing distribution of wages along class lines. However, community colleges with placement services have an opposite and disequalizing effect on wages. Additionally, the availability of placement services has the same disequalizing effect for average use of training on the job. That is, the SES gap in the use of training is wider in community colleges that offer job placement services. Perhaps availability of such services favors the middle class and upper class students who may have more "informal support" (in the form of mentoring) that complements the services offered through placement offices at community colleges. As noted by Laden (1999), minority students and students from lower SES backgrounds find that support services are geared more toward the dominant culture. Thus, lower SES students may tend to use placement services less often than students from a higher class background.

Recommendations and Conclusions

This study began with a discussion about the limitations of research on community colleges and the school-to-work transition. These limitations include a lack of attention to the bridging role of community colleges, an emphasis on transfer students, and a focus on individualistic explanations of student outcomes. The present study attempted to address these limitations by using hierarchical linear modeling to determine if and how community college structure affects the school-to-work transition.

Multilevel research methods can enable educational researchers to better conceptualize the school-to-work transition. Raudenbush and Bryk (1988) attest to the importance of multi-level modeling for school effects research to remediate the gap between methods and theory. They argue that the difficulties in past research "served as indicators of a fundamental mismatch between the relatively complex, multilevel reality we have sought to study and the comparatively simplistic, single-level statistical models used to study that reality" (p. 469).

Results suggest a direction for future research in this regard. Other analyses could incorporate broader measures of the economy and labor market trends that may influence community college effectiveness and wages. Lastly, school effects research could be strengthened through longitudinal studies, possibility through multilevel modeling. Raudenbush and Bryk (1988) suggest the use of time series data and three-level hierarchical models to meet this objective.

There is also a need to consider alternative measures of school context and school effects. The literature on school effects outlines some of the challenges in isolating school context and organizational characteristics, including empirical uncertainty, insufficient measurement technology, and concerns about feasibility (Oakes, 1989). This study focused on variables that seemed to be specifically related to the school-to-work transition. Thus, future research could be directed toward formulating additional measures of community college structure that may be related to student outcomes in the labor force. A separate paper written about this project provides the results of a model using measures of state coordination and the impact of these variables on student outcomes (Mobley, forthcoming). These measures included level of state funding, years of existence of state community college system, number of local institutions, type of governance board, degree to which the state is involved in program approval, and centralization rank.

Future research could incorporate other institutional and cultural variables, such as the attitude toward school-to-work and policies related to career development and institutional values concerning minority students and students from lower class backgrounds (Grubb, 1999; Laden, 1999). Gonzalez and colleagues (1998) call for the inclusion of economic indicators into such analyses.

Several of the measures used in this study could benefit from additional analysis and further refinement. For example, wages were not standardized for work schedule, numbers of hours worked, or geographic location. This means that the current research did not account for the varying work schedules of community college students or the locations of the various community colleges. Additionally, future analyses could compare private community colleges with public community colleges, as the current study did not distinguish between the two types of institutions.

The fact that race was not significant in any of the equations merits further investigation, especially given the fact that race/minority status is nearly always related to educational outcomes. Future models could focus on the other variables, such as class, that interact with race to influence outcomes. Additional studies could focus on other institutional barriers and individual level variables that influence the success of minority students who see their community college degree as a means for advancing in society (Valadez, 1993; Valadez 1996).

One of this study's strengths originates in its use of a national sample of students (from the High School and Beyond Survey). Because this study used a national sample, by its very nature, it could not consider the wide variation in terms of effectiveness among individual institutions, a characteristic that may apply more to community colleges because of their more localized orientations. Research on the school-to-work transition might be better conceptualized if based on data that would allow for a qualitative comparison between community colleges with similar structures. This would allow for more insight regarding the different elements of variation and how these influence student outcomes. A more comprehensive analysis would also be better able to account for the regional and state differences in labor market and other economic forces that influence the school-to-work transition. Similarly, another factor not considered in this investigation was the distinction between rural and urban community colleges which have very different labor markets and industrial-business environments. For example, rural areas are having to contend with the disappearance of industries such as assembly plants, textile mills, and food processing plants (Valadez & Killaclcy, 1995).

In summary, while multilevel analysis does have limitations that cannot be ignored, it is an important tool for determining contextual effects. In essence, they are "one possible way of bridging the gap between macro- and micro-sociology.' Thus, they can only add to our understanding of the school-to-work transition of community college students (van den Eeden & Huttner, 1982, p. 16).

References

Ahrens, A.M., & Boatwright, M.A. (1997). Satisfaction with career planning and job placement services at two-year public colleges. Community College Journal of Research and Practice 21(7), 617-26.

Albert, L., & Watson, R. (1980). Mainstreaming part-time faculty: Is sue or imperative? In M. Parsons (Ed.), Using part-time faculty effectively (pp. 73-80). New Directions for Community Colleges, Vol. 8(2). San Francisco: Jossey-Bass.

Alfred, R., Hudgins, J., Ewell, P., & McClenney, K. (1999). Core indicators of effectiveness for community colleges. Washington, D.C.: Community College Press.

American Association of Community Colleges. (1996). A pocket profile of community colleges: Trends and statistics, 1995-1996. Washington, DC: Author.

American College Testing Program. (1969). The two-year college and its students: An empirical report. Iowa City, Iowa: Author.

Anderson, K. (1984). The effects of college type and characteristics on educational attainment. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 256 206).

Antley, H.W. (1999). The development ora predictive model for one-year freshman retention rate: A macro-approach. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 433 787).

Antony, J.S., & Valadez, J.R. (1998). An exploration of the job satisfaction of American part-time college faculty. (ERIC Document ED 427 597).

Arnold, C. (1995). Using HLM and NAEP data to explore school correlates of 1990 mathematics and geometry achievement in grades 4, 8, and 12: Methodology and results. Washington, DC: National Center for Education Statistics.

Avakian, A. (1995). Conflicting demands for adjunct faculty. Community College Journal 65(6), 34-36.

Banks, D.L. (1992). Environmental factors influencing the community college transfer function. Dissertation Abstracts (DAI 52/4196).

Beman, R. (1980). Observations of an adjunct faculty member. In M. Parsons (Ed.), Using part-time faculty effectively (pp. 81-84). New Directions for Community Colleges, Vol. 8(2). San Francisco: JosseyBass.

Berman, P., & Weiler, D. (1990). Enhancing transfer effectiveness: A model for the 1990s. Washington, DC: American Association of Community and Junior Colleges and the National Effective Transfer Consortium.

Bowles, S., & Levin H. (1968). The determinants of scholastic achievement: An appraisal of some recent evidence. Journal of Human Resources, 3, 3-24.

Bushnell, D. (1978). Cooperation in vocational education. Washington, DC: American Association of Community and Junior Colleges and the American Vocational Association (ERIC Document ED 164 052).

Cain, M. (1999). The community college in the twenty-first century: A systems approach. Lanham, MD: University Press.

Carter, E. (1975). Follow-up study of New River Community College transfers who graduated in the 1972-73 school year [and]follow-up of 1974 graduates in occupational-technical programs. Dubline, VA: New River Community College. (ERIC Reproduction Service, Document No. ED 118 171).

Cohen, A., & Brawer, F. (1989). The American community college (2nd Edition). San Francisco: Jossey-Bass.

Cohen, A., & Sanchez, J. (1997). The transfer rate: A model of consistency. Community College Journal 68(2), 24-26.

Coleman, J. (1961). The adolescent society. New York: Free Press of Glencoe.

Commission on the Skills of the American Workforce. (1990). America's choice: High skills or low wages. Rochester, NY: National Center on Education and the Economy.

Cottingham, T., Newman, M., & Sims D. (1982). Is there help for parttime faculty--our new and neglected majority? Community College Review, 9(1), 12-17.

Creamer, D. (1994). Synthesis of literature related to historical and current functions of student services. In G. Baker (Ed.), A handbook on the community college in America: Its history, mission, and management (pp. 439-53). Westport, CT: Greenwood Press.

Cross, P. (1990). Transfer: Major mission of community colleges? Paper presented at the Annual International Conference on Leadership Development of the League for Innovation in Community Colleges. (ERIC Reproduction Service, Document No. ED 322 972).

deVries, J.J. (1998). School-to-work planning: Career guidance and development functions at Wyoming community colleges. Community College Journal of Research and Practice 22(1), 67-77.

Diener, T. (1986). Growth of an American invention: A documentary history oft he junior and community college movement. Contribuions to the Study of Education, No. 16. Westport, CT: Greenwood Press.

Dougherty, K. (1987). The effects of community colleges: Aid or hin drance to socioeconomic attainment? Sociology of Education 60(2), 86-103.

Dougherty, K. (1991). The community college at the crossroads: The need for structural reform. Harvard Educational Review, 61(3), 331-336.

Dougherty, K. (1994). The contradictory college: The conflicting origins, impacts and futures of the community college. Albany, NY: SUNY Press.

Eaton, J. (1994). Strengthening collegiate education in community col leges. San Francisco, CA: Jossey-Bass.

Fadale, L. (1982). Achieving sex equity in two-year college programs. Community/Junior College Quarterly of Research and Practice, 6, 187-191.

Farmer, E., & Honeycutt, F.D. (1999). Community college administrators and faculty opinions of Tech Prep. Community College Journal of Research and Practice 23(8), 717-725.

Gonzalez, A., de-la-Torre, A., & Garcia, J. (1998). Minority student achievement and workforce success in Arizona: ,4 research study. Tucson, AZ: Arizona University Mexican-American Studies and Research Center. (ERIC Reproduction Service, Document No. ED 434702).

Griffith, M., & Connor, A. (1994). Democracy's open door: The community college in America's future. Portsmouth, NH: Boyton/Cook.

Grubb, N. (1991). The decline of community college transfer rates: Evidence from national longitudinal surveys. Journal of Higher Education 62(2), 194-224.

Grubb, N. (1996). Working in the middle: Strengthening education and training for the mid-skilled labor force. San Francisco: Jossey-Bass.

Grubb, N. (1999). Honored but invisible: An inside look at teaching in community colleges. New York: Routledge.

Grubb, N., Badway, N., Bell, D., Bragg, D., & Russman, M. (1997). Workforce, economic, and community development: The changing landscape of the entrepreneurial community college. Mission Viejo, CA: League for Innovation in the Community College.

Hammond, L., & Porter, G. (1984). Follow-up study of 1981-82 students: North Carolina community college system. Raleigh, NC: Division of Planning and Research Services, North Carolina Department of Community Colleges. (ERIC Reproduction Service, Document No. ED 259 806).

Hershey, A., Silverberg, M.K., Haimson, J., Hudis, P., & Jackson, R. (1999). Expanding options for students.' Report to Congress on the national evaluation of school-to-work implementation. Princeton, NJ: Mathematica Policy Research (ERIC Reproduction Service, Document No. ED 436 641).

Hollingsworth, R., & Hanneman, R. (1984). Centralization and power in social service delivery systems. Boston: Kluwer Nijhoff Publishing.

Howard, T. (1998). Addressing the needs of the changing workplace through the School-to-Work Opportunities Act: Collaborative effort in curriculum development, teacher training, career services, and mentor training. Journal of Cooperative Education 33(2), 38-43.

Huffman, J.P., & Schneiderman, S. (1997). Size matters: The effect of institutional size on graduation rates. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 410 882).

Kuo, E.W. (1999). Analysis of ESL course offerings in community colleges. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 427 795).

Jencks, C., Smith, M., Acland, H., Bane, M., Cohen, D., Gintis, H., Heyns, B., & Michelson, S. (1972). Inequality: A reassessment of the effect of family and schooling in America. New York: Basic Books.

Kane, T., & Rouse, C. (1993). Labor market returns to two-year and four-year college: Is a credit a credit and do degrees matter? Faculty Research Working Paper R93-38. Cambridge, MA: Harvard University.

Lach, I., Kohl, P., & Wellman, F. (1979). Follow-up study of FY 1978 occupational graduates of Illinois public community colleges. Spring field, IL: Illinois Community College Board. (ERIC Reproduction Service, Document No. ED 176 823).

Laden, B.V. (1999). Celebratory socialization of culturally diverse students through academic programs and support services. In S. Shaw, J.R. Valadez, & R.A. Rhoads (Eds.), Community colleges as cultural texts: Qualitative explorations of organizational and student culture (pp. 173-94). Albany, NY: SUNY Press.

Lankard, B.A. (1993). Part-time faculty in adult and vocational education. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 363 797).

Little Hoover Commission. (2000). Open doors and open minds: Improving access and quality in California's community colleges. Sacramento, CA: Little Hoover Commission.

McCabe, R. (1988). The educational program of the American community college: A transition. In J. Eaton (Ed.), Colleges of choice: The enabling impact of community colleges (pp. 93-115). New York: MacMillan.

McCabe, R., & Day, P. (1998). Developmental Education: A Twenty-First Century Social and Economic Imperative. Mission Viejo, CA: League for Innovation in the Community College.

McGuire, J. (1993). Part4ime faculty: Partners in excellence. Leadership Abstracts 6(6), 1-3.

Mintzberg, H. (1979). The structuring of organizations. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Mobley, C. (Forthcoming). Community colleges and the school-to-work transition: A multilevel analysis. Forthcoming in Sociological Inquiry.

National Research Council. (1994). Preparing for the workplace: Charting a course f or federal postsecondary training policy. Washington, DC: National Academy Press.

Oakes, J. (1989). What educational indicators? The case for assessing the school context. Educational Evaluation and Policy Analysis 11(2), 181-99.

Orr, M. (1998). Integrating secondary schools and community colleges through school-to-work transition and educational reform. Journal of Vocational Education Research 23(2), 93-113.

Paneitz, B. (1997). Community college students' perceptions of student services provided when enrolled in telecourses. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 435 409).

Parnell, D. (1985). The neglected majority. Washington, DC: The Community College Press.

Pincus, F. (1986). Vocational education: More false promises. In S. Zwerling (Ed.), The community college and its critics (pp. 41-52). New Directions for Community Colleges, No. 54. San Francisco: Jossey-Bass.

Pollard, R. (1991). A comparison of attitudes toward tech-prep programs. Community College Review, 19(3), 34-42.

Pollock, A., & Breuder, R. (1982). The eighties and part-time faculty. Community College Review, 9(4), 58-62.

Puyear, D.E. (1989). Student retention in the Virginia Community College System, Academic years 1985-86 through 1987-88. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education (ERIC Reproduction Service, Document No. ED 304 198).

Raudenbush, S., & Bryk, A. (1988). Methodological advances in analyzing the effects of schools and classrooms on student learning. Review of Research in Education, 15, 423-75.

Rouche, J.E., Rouche, S.D., & Milliron, M.D. (1995). Strangers in their own land: Part-time faculty in American community colleges. Washington, D.C.: The Community College Press.

Schuyler, G. (1997). ERIC review: The assessment of community college economic impact on the local community or state. Community College Review 25(2), 65-80.

Shaw, K., Valadez, J., & Rhoads, R. (1999). Community colleges as cultural texts: Qualitative explorations of organizational and student culture. Albany, NY: SUNY Press.

Shults, D. (1997). The School-to-Work Opportunities Act (1994) and community college preparedness. NOVA Southeastern University. (ERIC Reproduction Service, Document No. ED 409 039).

Silverberg, M.K. (1996). Building school-to-work systems on a Tech Prep foundation: The status of school-to-work features in Tech Prep initiatives. Princeton, NJ: Mathematica Policy Research (ERIC Re production Service, Document No. ED 413 442).

Stern, D., Finkelstein, N., Stone J., Latting, J., & Dornsife, C. (1995). School-to-work: Research on programs in the United States. London: Falmer Press.

Todd, T., & Baker, G. (1998). Institutional effectiveness in two-year colleges: The Southern region of the United States. Community College Review 26(3), 57-76.

U.S. Department of Education. (1994). National assessment of vocational education - Volume 1: Summary and recommendations. Washington, DC: Author.

Valadez, J.R., & Killacky, J. (1995). Opening the shutter. New Directions for Community Colleges 23(2), 5-7.

van den Eeden, P., & Huttner, H. (1982). Multi-level research. Current Sociology, 30(3).

Wechsler, H. (1991). Meeting the transfer challenge: Five partnerships and their models. Washington, DC- American Association of Community Colleges. (ERIC Reproduction Service, Document No. ED 338 165).

Willett, L. (1980). Comparison of instructional effectiveness of full- and part-time faculty. Community/Junior College Research Quarterly, 5(1), 23-30.

William T. Grant Foundation Commission on Work, Family, and Citizenship -- Youth and America's Future. (1988). The forgotten half.' Noncollege youth in America. Washington, DC: Author.

Wilson, R. (1994). Education department study of faculty members f'inds most have full-time appointments. Fact file: A profile of full-time faculty members with teaching duties, Fall 1992. Chronicle of Higher Education (November 23), p. A16.

"Funding for this research was provided by the American Educational Research Association (AERA) and the National Science Foundation (NSF), Grant #RED9452861. The opinions expressed in this study do not necessarily reflect those of either AERA or NSF. I am grateful to these agencies for their support."

Catherine Mobley is a professor of sociology at Clemson University. [e-mail]