Determining public 2-year college faculty's intent to leave: an empirical model.
Among public 2-year college faculty, who represented in fall 1998 18% of all full-time and 44% of all part-time faculty (National Center for Education Statistics [NCES], 2001, p. iv), this topic is particularly important because almost half are expected to retire within the next 10 to 15 years (Fugate & Amey, 2000; Rifkin, 2000). Those who remain are subject to job opportunities at other institutions, both within and outside academe (Burnett, 2003). For example, 4-year colleges and universities seeking to fill non-tenure track positions devoted primarily to teaching rather than to research may find community college faculty highly appropriate for these positions. Another competitor is private industry, particularly for vocational-technical faculty (Fugate & Amey, 2000).
Recruiting and replacing faculty is a major concern of community college leaders. In a national study of work-related factors affecting the stress level of community college deans, Wild (2002) found that 41% of the 251 deans who responded to her queries about future challenges identified "hiring, finding, replacing, and retiring" faculty. In another national study that looked at retirement of community college faculty, the researchers found that 51% of the Chief Academic Officers in the study thought there would be "difficulty recruiting fully prepared faculty members" (Berry, Hammons, & Denny, 2001, p. 133). Thus, retaining current faculty who are not at retirement age would seem to be an important task for community colleges.
What factors are related to community college faculty's likelihood or intent to leave a position, either for another academic institution or for a career outside of academe? Since previous theoretical work has shown that the intent to stay or leave a position is a good proxy for actual turnover (Bluedorn, 1982; Lee & Mowday, 1987; Mobley, 1982; Steers & Mowday, 1981), answers to this question can alert community college administrators and faculty to problematic factors so that these factors can be alleviated. If some identified factors affecting intent to leave are beyond the control of administrators, then those administrators can accept the inevitability of some faculty departures.
Faculty turnover has been studied for decades, but primarily in the 4-year sector, and especially with faculty in research universities. Studies examining turnover initially focused on individuals and their motivations for leaving (e.g., Caplow & McGee, 1958; Flowers & Hughes, 1973), while more recent studies, influenced by research on turnover in nonacademic organizations (e.g., Bluedorn, 1982), have focused on organizational and structural variables manifested in faculty worklife and viewed as influencing attitudes toward work or job satisfaction (e.g., Matier, 1990; Weiler, 1985). Job satisfaction is seen as one of several intermediate social psychological variables that affect intent to leave or actual turnover (Johnsrud & Rosser, 2002).
Demographic characteristics (e.g., age, sex, race/ ethnicity, highest degree held, rank, tenure status, union membership status, academic discipline) and family circumstances (e.g., having children) have been studied for their relationship to faculty's degree of job satisfaction. For example, in their analysis of NSOPF-93 data, Finkelstein, Seal, and Schuster (1998) found that women faculty were less satisfied than were men on a number of factors, excluding "benefits and spousal employment" (p. 60). Hutton and Jobe (1985) also found that sex was a variable in job satisfaction. Women community college faculty members were more satisfied "overall" (p. 321) than were men community college faculty, although there were differences by sex on particular aspects of job satisfaction such as "availability of audio-visual materials and equipment" (p. 321). On the other hand, Milosheff (1990) did not find any significance of gender, department affiliation, or highest degree held to perceptions of community college faculty members' job satisfaction.
Individual characteristics have been shown to intersect with various organizational or structural characteristics in one's work life to affect job satisfaction and thus intent to leave (e.g., Barnes, Agago, & Coombs, 1998; Hagedorn, 1996; Johnsrud & Heck, 1994; Matier, 1990; Smart, 1990; Weiler, 1985). Organizational factors thought to affect faculty worklife include availability of technology and technological support, reward structures such as salary and opportunity for tenure, and support of professional development. Structural variables (1) include institutional type and discipline with consequent expectations regarding faculty roles and worklife.
To examine faculty members' morale and their intent to leave, Johnsrud and Rosser (2002) developed a model of faculty worklife that included the constructs of professional priorities and rewards, administrative relations and support, and quality of benefits and service. Professional priorities and rewards encompassed "what is important to faculty members in terms of their work and the rewards they receive for that work" (p. 522). More specifically, this included survey items on collegial relations, students, rewards and evaluation, and professional worklife. Administrative relations and support was designed to "represent the confidence faculty members have in their institutional leadership, in the extent of advocacy for faculty interests, and in their own systems of governance" (p. 523). Therefore, it addressed advocacy for faculty, confidence in leadership, and faculty governance. Quality of benefits and services included support services and the standard of living possible through salary and benefits.
Johnsrud and Rosser's (2002) data came from a survey sent in fall 1998 to faculty members in a large public system of higher education, with the system including both 4-year and 2-year colleges. Sent by the systemwide faculty senate, the survey was designed "to measure the quality of faculty worklife throughout the system" (p. 528). Although 2-year faculty were included in the survey, it is likely that survey items included issues more germane for 4-year than for 2-year faculty in that 2-year college faculty's professional worklife differs on some dimensions from that of 4-year faculty. For example, 2-year college faculty members are not expected to conduct research, although participation in out-reach/service activities and institutional governance is expected. The primary aspect of community college faculty's worklife is teaching, with the related components of advising and curriculum development (Cohen & Brawer, 2003; Huber, 1998).
Job satisfaction is affected by a wide variety of factors. While not based upon research in higher education organizations, Herzberg's Two Factor Theory (1966) of job satisfaction has heavily influenced studies of faculty job satisfaction, including that of community college faculty (e.g., Diener, 1985; Hardy & Laanan, 2003; Hill, 1986-1987; Truell, Price, & Joyner, 1998). Herzberg (1966) theorized that job satisfaction was influenced by "intrinsic factors" or "motivators" relating to actual job content or "what the person does" (p. 74) and by "extrinsic factors" or "hygienes" associated with the work environment or "the situation in which [the person] does" (p. 75) the work. Examples of intrinsic motivator factors for academe would be the work itself, such as teaching and research, and recognition of that work through achievement of tenure and promotion. Examples of extrinsic factors would be working conditions such as levels of compensation, administrative support, and working conditions. As factors external to the work itself, extrinsic hygiene factors do not contribute to job satisfaction but rather to job dissatisfaction. Alternatively, the presence of motivators lead to job satisfaction but their absence does not lead to job dissatisfaction. Although Herzberg's work is not always cited as the theoretical framework, the concept of intrinsic and extrinsic aspects of job satisfaction has been used in studies of faculty job satisfaction (e.g., Olsen, 1993).
Few studies have directly examined community college faculty's intent or propensity to leave their current position except as a factor in job satisfaction studies. In some of these studies, the issue is raised indirectly through questions about intent to stay, either at the current institution or at community colleges in general. Thus, in their study of the demographics and job satisfaction of full-time faculty in Florida's community colleges, Benoit and Smith (1980) asked respondents if they "expected to remain in the community college field until retirement" (p. 267). In addition, in his study of job satisfaction of community college faculty in one southeastern state, Diener (1985) found that 92% "would not change jobs or were not eager to do so" (p. 351). Neither Deiner nor Benoit and Smith used statistical analysis to relate intent to stay with any aspects of the job or faculty demographics. In contrast, Locke, Fitzpatrick, and White (1983) used factor analysis in their examination of the job satisfaction of faculty at one community college and found that job satisfaction was related to one's "intended tenure" (p. 351). Similarly, to examine the relationship between job satisfaction and "organizational commitment" (Hill, 1986, p. 4), Hill (1986) conducted a statewide study of New York developmental education faculty. Using stepwise regression, he found that faculty members who were more satisfied with their jobs were more "committed to their employing institutions" (p. 5). Hill also found relationships between faculty's rank, highest degree held, age, sex, and departmental affiliation and their level of job satisfaction.
One quantitative study that focused directly on community college faculty's intent to leave as opposed to intent to stay found a relationship between it and job satisfaction. McBride, Munday, and Tunnell (1992) examined propensity to leave in their study of the job satisfaction of community college faculty in the 11 states in the Southern Association of Colleges and Schools regional accrediting body. Using stepwise multiple regression, the authors found that certain job satisfaction factors influenced, through direct effects only, the propensity to leave (e.g., dissatisfaction with the work itself, salary, policy and administration). The demographic variables of gender and highest degree held were not influential, but age was: Older faculty members were less likely to leave.
Only one study seems to have been conducted about part-time community college faculty's job satisfaction (Valadez & Antony, 2001). Using 1992-1993 NSOPF data that included almost 7, 000 part-time 2-year college faculty, the authors used factor analysis to develop three dimensions of satisfaction (autonomy, students, and demands and rewards) and found some job satisfaction but also some concerns. The authors did not focus upon intent to leave or stay.
What is missing in research on community college faculty's job satisfaction and intent to leave are efforts to understand how demographic variables, professional and institutional worklife issues, and job satisfaction simultaneously interact to explain faculty intentions to leave. While previous research has provided some knowledge about the intended departure of community college faculty, its method of analysis has been a limitation. For example, regression analysis does not represent a complete test of any theory since one is not able to lay out more complex models with direct and indirect effects of latent and demographic variables and to account for more sources of measurement error, thus improving the accuracy of the parameter estimates.
Purpose of Study
This study proposes to use structural equation modeling to develop a model that simultaneously defines multidimensional constructs such as worklife, job satisfaction, and intent to leave (i.e., through confirmatory factor analysis), and tests the direct effects of workplace variables and satisfaction on community college faculty's intent to leave. Figure 1 represents this proposed conceptual model, which draws upon Herzberg's (1966) work to create its constructs. In addition to the demographic variables, the effect of worklife is hypothesized to directly affect job satisfaction, and job satisfaction is hypothesized to directly affect intent to leave. Community college faculty members' worklife is hypothesized to directly or indirectly influence their intent to leave. Structural equation modeling (SEM) is an appropriate procedure for examining this type of research hypothesis because it (a) corrects for measurement error, which provides more accurate estimates of structural parameters of interests, and (b) permits the study of direct and indirect effects (e.g., worklife) simultaneously. In other words, SEM provides a more complete test of our proposed hypothesis.
[FIGURE 1 OMITTED]
To investigate individual-level perceptions of community college faculty members' worklife and satisfaction on their intent to leave, this study uses data from the 1999 National Survey of Postsecondary Faculty (NSOPF-99), sponsored by the National Center for Educational Statistics and the National Science Foundation, to measure the quality of faculty members' professional and institutional worklife throughout the United States. The NSOPF-99 database is a nationally representative sample of faculty members in nonprofit higher education institutions. Over 3, 390 postsecondary institutions were included, with an 83% response rate among faculty (N = 18, 043).
Using the NSOPF-99 restricted database, we created a subset retaining only those faculty (both full- and part-time) who were in 286 public 2-year postsecondary institutions (i.e., community and associates colleges) and designated instruction as their primary responsibility. Thus for this individual-level study, 968 community college faculty members were selected as the subset from the national representative sample. Demographic characteristics of the respondents or control variables included (a) sex, (b) race/ethnicity, (c) age, (d) highest degree held, (e) rank, (f) tenure status, (g) union membership status, (h) full or part time status, (i) administrative service as department chair, (j) length of time in current position, (k) average years of teaching in higher education, and (l) prior employment at a 4-year institution.
Regarding the model itself, we aimed to reflect as much as possible the world of 2-year college faculty in our choice of variables to empirically define and measure the latent constructs of worklife and satisfaction. Thus, unlike studies of 4-year faculty's worklife, we did not include such factors as availability and quality of graduate students and research and teaching assistants, graduate courses taught or prepared, and the advising and directing of master's and doctoral work with graduate students (e.g., Rosser, 2004). Rather, for worklife we included dimensions of concern for all faculty members regardless of institutional setting: administrative support and facilities, professional development, and technology support. Illustrative items delineated within each dimension were factors germane to 2-year faculty. We also included these particular dimensions because they are extrinsic factors under administrative control and thus could be improved or strengthened should administrators wish to alleviate job dissatisfaction. For job satisfaction, we examined the intrinsic factors of faculty's decision-making authority, student advising, course preparations, and workload. We also included the extrinsic factors of salary, benefits, and job security even though, according to Herzberg's (1966) work, these factors do not contribute to job satisfaction. We believe, however, that since these areas are initially negotiated for each individual prior to employment, they are individual personal commitments associated with an individual's job satisfaction. Additionally, these extrinsic factors could also be altered by administrators, should they choose to do so. For example, administrators could increase the job security of part-time faculty by promising them a certain number of courses within a given period. Some full-time faculty could be given greater job security if they were moved to a tenure-track position, if possible.
Faculty Members' Worklife
Worklife quality is measured with a set of 14 statements regarding faculty members' professional and institutional issues. Respondents were asked to indicate from poor to excellent, statements regarding the quality of their professional and institutional worklife. The scale on these worklife issues ranged from 1 to 4, with 1 indicating a poor perspective on the worklife issue and 4 indicating an excellent perspective. Drawing upon the work of Johnsrud and Rosser (2002) and Rosser (2004) and desiring to create measures of the quality of worklife that were more global, the items were reduced to three interrelated dimensions (alpha coefficients in parentheses). Administrative support and facilities (.67) covers the areas of secretarial support, classroom space, office space, and library holdings. Professional development (.88) consists of opportunities for faculty members' professional development through availability of internal professional travel funds, internal tuition remission funds, training to improve research/teaching, release time, and sabbaticals. Technology support (.85) includes computers and local networks, centralized computer facilities, internet connections, technical support for computers, and audio-visual equipment. Using confirmatory factor analysis (described later), the multidimensional construct of faculty members' worklife was defined and measured by the three dimensions of administrative support, professional development, and technology support.
Faculty Members' Satisfaction
Satisfaction, the intervening variable in this study, was created from a set of 10 questions (alpha coefficients in parentheses). The first dimension reflected the motivator of decision making authority (.73), comprised of 3 items about the faculty members' level of satisfaction with their authority to make decisions about the content of courses and courses taught, as well as other job decisions. The second dimension consisted of 3 items that examined statements regarding their satisfaction with another intrinsic factor, advising and workload (.80), including time available to advise students and to prepare for class, and overall satisfaction with their workload. The third set of 4 questions focused on the extrinsic factor of benefits and security (.80). This factor includes the level of satisfaction faculty members have with their salary, job security, benefits, and advancement opportunity. Therefore, the multidimensional construct of satisfaction may be defined as three interrelated dimensions: decision-making authority, advising and workload, and benefits and security.
Faculty Members' Intent to Leave
The dependent variable in this study, faculty members' intent to leave, was defined as two sets of two questions. The first set of questions asked faculty members the extent to which they were likely to leave their career for another full- or part-time position within 3 years. The second set of questions asked the extent to which faculty members were likely to leave their institution for another full- or part-time postsecondary position within 3 years. These four items were measured on 3-point scales, 1 indicating that faculty members were not likely to leave their position and 3 indicating that they were very likely to leave their position (alpha = .65). These questions suggested that those faculty members with higher scores would be more likely to intend to leave.
The demographic profile of the 968 faculty members in the study was as follows: 498 (51.4%) of the faculty members were female and 470 (48.6%) were male. There were 127 (13.1%) ethnic minorities and 841 (86.7%) Caucasians. The average age was 47.72 (SD = 9.79), with 617 (63.8%) faculty members 45 or older and 350 (36.2%) 44 or younger. Of those faculty members who were a part of this sample, 463 (47.8%) held a full-time faculty position while 505 (52.2%) held a part-time position at their institutions.
For 117 (12.1%) of the faculty, the highest degree held was a doctorate, 535 (55.3%) held a master's, 30 (3.1%) held a professional degree, 200 (20.7%) held a bachelor's, and 66 (6.9%) had an associate's degree, certificate, or equivalent.
As for academic rank, 115 (11.9%) were full professors, 74 (7.6%) associate professors, 66 (6.8%) assistant professors, 550 (56.8%) were instructors, and 16 (1.7) were lecturers. The remaining 163 (16.9%) faculty members were unclassified by these rank titles or rank was not applicable. Of those faculty members who reported their tenure status, 230 (23.8%) were tenured and 738 (76.2%) were untenured or not on a tenure track.
Regarding union membership status, 689 (71.2%) were members. The remaining 279 (28.8%) indicated that unionization was not an option, they were not eligible for available unionization, or they chose not to be members. Of the total group of faculty members, 84 (8.7%) were noted as department chairs.
The average length of time these community college faculty members had been in their current position at the institution was 9.01 years (SD = 8.39), and the average years of teaching in higher education was 11.76 (SD = 9.10). There were 168 (17.4%) faculty members who had been employed at a 4-year institution at some point prior to working at a 2-year institution.
The Measurement Model
The first goal of the analysis was to empirically define and measure the multidimensional constructs of worklife, satisfaction, and intent to leave through confirmatory factor analysis (CFA). Because underlying constructs cannot be directly measured, they must be indirectly defined through a set of observed variables. In this study, there were three dimensions or observed variables comprising worklife, three dimensions of satisfaction, and two dimensions of intent to leave. The validity of the measurement model was examined through CFA using a maximum likelihood fitting function (2) with Mplus (Version 2.14, Muthen & Muthen, 2003). In the CFA approach to testing models, one tests the variance-co-variance matrix implied by the model against the variance-covariance matrix of the actual data (Rosser, 2004).
Several indices may be examined to determine the extent to which the data fit the hypothesized model. The chi-square coefficient for the model is 26.821 with 8 degrees of freedom (p = .001). Although the chi-square is significant, rejecting the model on this basis alone might be ill advised because of the chi-square's sensitivity to sample size. The chi-square test is centered on sample size of the data, and therefore, alternative indices of the model fit are advised when evaluating models (Hu & Bentler, 1995; Raykov & Marcoulides, 2000).
Other indices may be examined to determine the extent to which the data fit the hypothesized model. The indices most commonly used to examine the measurement model's fit are the comparative fit index (CFI), the Tucker-Lewis index (TLI), the root-mean-square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). The fit indices for the model indicate that the CFI of .990 and TLI of .964 provide indications of the amount of variances and covariances in the data accounted for by the proposed model. In general, values as close to 1.0 (i.e., above .95) indicate an acceptable fit of the model to the data (Loehlin, 1998). The RMSEA value is .030, which is nonsignificant (p = .483). The RMSEA offers a "close" test of statistical fit for the model. The close test allows for a discrepancy of fit per degree of freedom. The results of this index suggest an excellent fit of the proposed structural model to the observed data. The SRMR value for the model was .025. The SRMR describes the average magnitude of the residuals or, in this case, the variances unaccounted for in the data. Although this coefficient can depend on the scaling of variables, in most cases, an SRMR less than .08 indicates a well fitting model (Hu & Bentler, 1999). Overall, these indices suggest an excellent fit of the proposed measurement model to the observed data comprising the latent constructs of worklife, satisfaction, and intent to leave.
As illustrated in Figure 2 (i.e., the measurement model), the parameter estimates relating the observed variables to their latent constructs comprising worklife, job satisfaction, and intent to leave are summarized. Three observed variables comprising the latent construct of work-life (i.e., technical support, administrative support, professional development) contributed significantly in the definition and measurement of faculty members' quality of worklife (.65, .70, .13, respectively). Decision-making authority, advising and workload, and benefits and security contributed significantly to the definition and measurement of faculty members' satisfaction (.48, .61, .92, respectively). Intent to leave was defined by faculty intentions to leave their institution for another full- or part-time position in higher education, and to leave their academic career for another full- or part-time non-postsecondary position (.62, .79, respectively). The correlation between worklife and satisfaction was .41, the correlation between satisfaction and intent to leave was -.53, and the correlation between worklife and intent to leave was -.20. Once we have adequately defined and measured the latent variable constructs through CFA, we are then able to examine the various direct and indirect relationships among the variables in the final SEM model.
The Final Structural Equation Model
Structural equation modeling (SEM) is a statistical methodology that provides researchers with a comprehensive method to propose and subsequently test theoretical propositions about the interrelations among the constructs (Raykov & Marcoulides, 2000; Rosser, 2004). SEM also takes into account the measurement error that is widespread in most disciplines and contains latent variables. In this case, for example, SEM allows for the simultaneous examination of those demographic characteristics that may explain (either directly or indirectly) the latent variable constructs such as worklife, satisfaction, and the intent to leave. (3)
Similar to the measurement model, the first step in SEM is to assess how well the proposed model fits the data after adding the predictor variables. The validity of the proposed structural equation model was also examined by Mplus version 2.14 (Muthen & Muthen, 2003), using a maximum likelihood fitting function. (See Appendix A for the correlations between the observed variables included in the final structural equation model.) In the final SEM model, the chi-square coefficient for the model is 59.229 with 37 degrees of freedom (p = .012). The significant [X.sup.2] is due to sample size, and other indices of model fit show that the model provides an acceptable representation of the data. The CFI of .991 and TLI of .979 provide good indications of the amount of variances and covariances in the data accounted for by the proposed model. The RMSEA value is .025, which is nonsignificant (p = 1.00), and the SRMR value for the model was .026. Overall, the indices suggest a "good fit" of the proposed model to the observed data.
[FIGURE 2 OMITTED]
Figure 3 summarizes the parameter estimates relating the observed variables to their latent constructs comprising worklife, job satisfaction, and intent to leave. In the final model, the observed variables comprising the latent construct of worklife contributed significantly in the definition and measurement of faculty members' quality of worklife (.60, .79, .14, respectively), their satisfaction (.53, .68, .84, respectively), and their intent to leave (.79, .62, respectively). After the model has been well defined and measured, we are now able to examine important relationships among the variables in the full and final SEM model.
Worklife, Job Satisfaction, and Intent to Leave
As illustrated in Figure 3 (significant standardized path estimates), having a doctorate (e.g., dummy coded as 1 = doctorate, 0 = other degrees held) had a significant and negative impact on the perceptions of this faculty's worklife (-.08), but was nonsignificant on their satisfaction and intentions to leave. Age had a positive and significant impact on their worklife (.19) but not on satisfaction or intent to leave. Years faculty members were in the position had a significant and negative impact on their worklife (-20) and intent to leave (-17) but not on their level of satisfaction. Being employed at a 4-year institution prior to their current position had a significant and negative impact on faculty satisfaction (-.16) but not on worklife or intentions to leave. Being a full-time faculty member had a significant and negative impact on intention to leave (-.21) but not on faculty perceptions of worklife or satisfaction. Work-life had a significant and positive impact on satisfaction (.46) but no direct effect on intent to leave. Satisfaction had a significant and negative impact on intent to leave (-.47).
Being a female faculty member or an ethnic minority had no significant impact on their perceptions of their worklife, satisfaction, and intent to leave. Several other demographic characteristics (i.e., tenure status, union membership, department chair status, faculty rank, over the age of 44, and years of teaching) were included in the preliminary analyses, but were dropped from the final structural model because they had no impact on other variables in the model.
In short, two demographic characteristics had a negative impact on intent to leave, either for another academic position or for a career outside academe: being a full-time faculty member and having been in a position longer. No demographic characteristics were positively associated with intent to leave. There was no significant difference between full- and part-time faculty in perceptions of worklife and job satisfaction. However, as indicated, full-time faculty were less likely to leave.
Being employed at a 4-year institution prior to employment at the community college was the one demographic variable that contributed to lack of job satisfaction. As for worklife, those who held a doctorate and who had worked more years in the position were more negative in their perceptions about their worklife. However, older faculty members were more positive in their perceptions.
Using LISREL 8.54 (Joreskog & Sorbom, 2000), we also examined the various indirect and total effects in the structural equation model. Table 1 provides a recap and summary of the significant and nonsignificant direct, indirect, and total effects for the demographic variables and the latent constructs of worklife, satisfaction, and intent to leave. The significant effect of faculty members' holding a doctorate and their age on worklife and the significant effect of transferring from a 4-year institution on satisfaction were mitigated through faculty members' satisfaction (.046, -.095, .042, respectively) on intent to leave. Demographics with the largest total effects on intent to leave were from years in the position (-.115) and full-time status (-.268). Examining the direct and indirect effects of the various constructs revealed that the latent construct of job satisfaction had a significant and negative effect on intent to leave (-.473). Faculty members' worklife had a direct positive and significant effect on their satisfaction (.460) but a nonsignificant effect on their intent to leave. Although worklife did not have a significant and direct effect on intent to leave, the total effect through satisfaction remains powerful and negative on intent to leave (-.218). Given the powerful affective response of satisfaction, this large total effect of worklife on intent to leave suggests that the more faculty members are positive about their worklife, the less likely they are to leave. The demographic variables in this study explain 5% of the variance ([R.sup.2]) in faculty members' worklife. However, the model explains 25% of the variance in satisfaction and 37% of the variance in intent to leave.
[FIGURE 3 OMITTED]
Discussion and Implications
Several of the faculty demographics found in this study match fairly well with those found by Huber (1998) in her analysis of community college faculty's responses to the Carnegie Foundation for the Advancement of Teaching's National Survey of Faculty administered in 1997. In Huber's study, 47% of the faculty were female and 88% were White compared to slightly over 51% female and almost 87% White in our study. The average age in Huber's study was 50.7 compared to 47.72 in ours. As regards academic background, Huber found that 18% of the faculty had a doctorate, 64% a master's, 8% a bachelor's, and 3% an associate's degree; in our findings, 12.1% for the doctorate, 55.3% for the master's, 20.7% for the bachelor's, and 6.9% for the associate's degrees. Differences in results may reflect the higher percentage of community college part-time faculty in the NSOPF data (52.2% compared to 21% in Huber's study) and the greater number of community colleges that received the survey in our study: 329 institutions received surveys, and faculty at 298 completed surveys with an unweighted response rate of 90.6% (92.8% weighted), compared to 34 institutions in Huber's study. The Carnegie Foundation survey did not ask about tenure status, union membership, service as department chairs, or previous experience in the 4-year sector.
Because of our use of structural equation modeling, the demographic characteristics found in our study could be examined for their direct and indirect effects upon community college faculty's job intent to leave. Unlike regression analysis, structural equation modeling allows the researcher to empirically define and measure latent variables such as work-life, job satisfaction and intent to leave while simultaneously examining the direct and indirect effects of demographic characteristics on individuals' worklife experience, job satisfaction, and intent to leave. Once we had adequately defined and measured the latent variables through confirmatory factor analysis, we were then able to examine the various relationships among the variables in the final SEM model. In addition, SEM accounts for more sources of error among the observed (i.e., constructs comprising the latent variables) and latent variables in the analysis.
What SEM has enabled us to see in this national study of instructional public community college faculty's intent to leave is the impact of certain demographic characteristics (age, years in current position, prior employment at 4-year institutions, and full-time versus part-time status) on worklife, satisfaction, and intent to leave. The finding that older faculty are the most satisfied is consistent with findings from Hill's (1983) study of the job satisfaction of community college faculty in Pennsylvania. Older faculty were also found to be less likely to leave in McBride, Munday, and Tunnell's (1992) study of community college faculty at institutions within the Southern Association of Colleges and Schools regional accrediting body.
Another variable similar to age is the number of years faculty members have been in their current position. Although this profile characteristic of the faculty does not have a direct impact on their satisfaction, it does negatively affect their perceptions of worklife and their intent to leave, depending upon the length of time at the institution. Although those who have been at an institution longer were less positive about their worklife, they were also less likely to indicate they would change careers or leave for another higher education position than were faculty members who have only served in their current position for a few years. It might be that both increased age and years in the position reflect an individual's commitment to stay in a current position or within academe. Alternatively, these factors might make finding another job more difficult, so older, long-term faculty stay out of desperation or despair.
Other than the work of Valadez and Antony (2001), almost no research has been conducted on the job satisfaction of part-time community college faculty. Thus, this study adds to our knowledge of part-timers by indicating that they would be more likely than full-time community college faculty to leave their institution or academe for another position. When faculty members acquire full-time status within their institution, they have greater job security than do part-time faculty and thus might be less interested in looking for another position. In addition, being employed full-time at an institution suggests a mutual commitment by the institution and by the faculty member, which may lead to greater job satisfaction.
The finding that faculty who were employed at 4-year institutions prior to employment at the community college are not satisfied with their current job merits further exploration. Further examination of the NSOPF-99 indicated that those who had last been at doctoral granting institutions prior to employment at the community college were less satisfied than those who had been at a 4-year, non-doctoral granting institution. Since previous research indicates that some women community college faculty preferred their 2-year college faculty experience over their 4-year college experience because they disliked the pressure to publish at 4-year colleges (Townsend, 1998), it would be interesting to determine what characteristics of 2-year employment were deemed less satisfactory by this study's group of faculty.
Another important aspect of this study is its demonstration that the quality of faculty worklife is paramount to community college faculty members and thus has a strong and positive effect on their overall level of satisfaction. More importantly, this study suggests that the perceived quality of worklife can generate a positive or negative response as to whether faculty members are indeed satisfied, and therefore whether they intend to stay or leave their academic career or institution.
Of the three dimensions measuring faculty worklife, administrative support and facilities was the most important. Like 4-year faculty, 2-year faculty members want secretarial services, appropriate library services, and sufficient classroom and office space. Given the erosion in most states of funding for higher education (Arnone, 2004; Massy, 2003; Voorhees, 2001), institutions are having to take numerous steps to pare expenses, including cutting back on secretarial staff, reducing funds to libraries, increasing class size, and sometimes asking faculty to share their offices with other faculty. These changes in worklife may be one reason why faculty members who have been at an institution longer are more negative about their worklife. They remember when administrative (i.e., secretarial) support was greater, funding for library holdings increased regularly, and facilities were less crowded. In addition, in our study faculty members with a doctorate were also more critical of their worklife. Their reasons for being more critical merit further research since it is unclear from any previous research on community college faculty how their holding a doctorate relates to criticism of one's current workplace.
Technical support was the next most significant contributor to perceptions of community college faculty members' quality of worklife and professional development is the least. Community colleges have readily incorporated instructional technology, especially for distance education (Cohen & Brawer, 2003; Lever-Duffy & Lemke, 1996), so it only makes sense that technical support would be important to their faculty. Professional development may be the least important contributor to community college faculty's worklife because of most of the ways in which it is defined in this study (availability of travel funds, training to improve research/teaching, release time, and sabbaticals). While the support of faculty members' professional development and research activities through travel, release time, and sabbatical leave has been shown to be important in the retention of research faculty members (Rosser, 2004), these kinds of professional development activities are less common in the overall worklife of community college faculty (Cohen & Brawer, 2003). Rather, professional development also occurs through administrative support for "tuition reimbursement, tuition waivers, teaching excellence awards, ... instructional equipment and supplies, computer software, ... and professional books" (Miller, Finley, & Vancko, 2000, p. 70). Only one of these, tuition reimbursement, was included under the dimension of professional development.
All three dimensions of worklife (i.e., administrative support and facilities, technical support, professional development) have a positive influence on the satisfaction of this group of community college faculty members. Moreover, these findings support the previous work of Herzberg (1966) in that those intrinsic factors or motivators relating to one's job content and the extrinsic factors or hygienes relating to the situation in which one works has a positive influence on faculty members' overall satisfaction, and subsequently on their intent to stay or leave academe or their institution.
As regards job satisfaction, institutional leaders can take heart that the variable of benefits and security does not seem to be an important source of dissatisfaction among full-time, instructional community college faculty. Since benefits and security are established through contract negotiation in unionized settings (and recall that over 70% of the faculty in this study were unionized), administrators may have little control over them for individual faculty (Castro, 2000; Cohen & Brawer, 2003). Salary and job status (full- or part-time) are negotiated and established before a faculty member starts a position, so acceptance of them initially may suggest some satisfaction with them. Unless desperate for employment, individuals seeking a full-time faculty position would look elsewhere if they were really dissatisfied with an institution's salary and job status offer.
Similarly, advising, course preparation, and workload are not a major concern, nor is the decision making authority faculty members hold regarding the type of courses taught, the selection of course content, and other job-related decisions. Course workload is established prior to employment, as may also be the number of course preparations, particularly for those community college faculty working in a unionized setting (Castro, 2000; Cohen & Brawer, 2003). Again, one can assume, as with initial salary and job status, too much dissatisfaction with the initial terms of employment (e.g., course workload and number of preparations) would lead to rejection of the job offer. Advising in many community colleges is conducted by counselors rather than faculty (Grubb, 2001), so student advising may be a non-issue. Finally, the authority faculty members have over the content and the type of courses taught as well as their decision making over other job-related decisions contributed to the overall satisfaction of faculty in this study. While many community college faculty members do not benefit from a tenured system that protects their rights and control over these curriculum areas (less than 24% of the faculty in this study were tenured), community college faculty are seemingly satisfied with the authority and control they do have. Perhaps this is because collective bargaining agreements prescribe the decision-making authority of faculty, and the majority of faculty members in this study were unionized (Castro, 2000; Cohen & Brawer, 2003).
In sum, the picture that emerges from this study is that public community college faculty members, whose primary responsibility is instruction, are well satisfied with the aspects of their worklife examined in this study, particularly those dimensions of technical support and facilities and administrative support and facilities. Given faculty's positive perceptions of worklife and their job satisfaction, this study provides only limited insight into what community college leaders at the institutional level can do to retain their current faculty and see that they are satisfied. One possibility, due to the importance of technology in classroom teaching nowadays (Groves & Zemel, 2000; Paulson, 2002), is for administrators to maintain or increase technical support. Additionally, increasing administrative support through providing more secretarial assistance, improving facilities through increasing library holdings, and lobbying for more classroom buildings is another way to enhance work-life perceptions and improve retention of faculty. If other factors pertaining to professional development were used in national faculty surveys such as the NSOPF, it may be that future research would find that opportunities for professional development contributed more to community college faculty's perceptions of their worklife.
Community college faculty members' satisfaction does matter, and it is clearly a contributing factor affecting intent to leave. Longer length of time at an institution, older age, and full-time faculty status are demographic characteristics that seem to demonstrate a commitment to a specific institution as well as to the community college professorate in general. Institutional leaders need to honor this commitment as they seek to retain their faculty.
(1) Variables that define individuals as groups by the differentiation of the organization are considered to be structural variables, and therefore they should be used in multilevel analyses. The sample size in this study limits our ability to conduct an appropriate multilevel analysis, and thus the use of only individual level profile characteristics of the respondents.
(2) Maximum Likelihood (ML) is a general estimation procedure that produces estimates for the population parameters that maximize the probability of observing the data that are actually observed, given the model. ML is the most commonly used estimation method. An advantage of the ML estimation method is that it is generally robust, and it produces estimates that are asymptotically efficient and consistent. With large samples, ML estimates are usually robust against mild violations of the assumptions, such as having non-normal errors (Hox, 2002. p. 37).
(3) The NSOPF-99 weight (WEIGHT) is used to minimize the influence of large sample sizes and to correct for the nonsimple random design on standard errors. (See Thomas & Heck, 2001, for further discussion).
Arnone, M. (2004, January 17). State spending on colleges drop for the first time in 11 years. The Chronicle of Higher Education, 50(19), 19a.
Barnes, L. B., Agago, M. O., & Coombs, W. T. (1998). Effects of job-related stress on faculty intention to leave academia. Research in Higher Education, 39(4), 457-469.
Benoit, R. J., & Smith, A. (1980). Demographics and job satisfaction of Florida community colleges. Community/Junior College Quarterly of Research and Practice, 4(3), 263-276.
Berry, L. H., Hammons, J. O., & Denny, G. S. (2001). Faculty retirement turnover in community colleges: A real or imagined problem? Community College Journal of Research and Practice, 25(2), 123-136.
Bluedorn, A. C. (1982). A unified model of turnover from organizations. Human Relations, 35(2), 135-153.
Burnett, S. (2003, November 10). High-stakes hold 'em. Community College Week, 16(7), 6-8.
Caplow, T., & McGee, R. J. (1958). The academic marketplace. New York: Basic Books.
Castro, C. R. (2000). Community college faculty satisfaction and the faculty union. In L. S. Hagedorn (Ed.), What contributes to job satisfaction among faculty and staff (New Directions for Institutional Research, No. 105, pp. 45-55). San Francisco: Jossey-Bass.
Cohen, A. F., & Brawer, F. B. (1987). The collegiate function of community colleges. San Francisco: Jossey-Bass.
Cohen, A. F., & Brawer, F. B. (2003). The American community college (4th ed.). San Francisco: Jossey-Bass.
Diener, T. (1985). Community college job satisfaction. Community/Junior College Quarterly of Research and Practice, 9(4), 347-357.
Finkelstein, M. J., Seal, R. K., & Schuster, J. H. (1998). The new academic generation: A profession in transformation. Baltimore, MD: The Johns Hopkins University Press.
Flowers, V. S., & Hughes, C. L. (1973, July/August). Why employees stay. Harvard Business Review, 51(4), 49-60.
Fugate, A. L., & Amey, M. J. (2000). Career stages of community college faculty: A qualitative analysis of their career paths, roles, and development. Community College Review, 28(1), 1-22.
Groves, M. M., & Zemel, P. C. (2000). Instructional technology adoption in higher education: An action research case study. International Journal of Instructional Media, 27(1), 57.
Grubb, W. N. (2001, June). "Getting into the world": Guidance and counseling in community colleges (Teachers College Community College Research Center. Working Paper #1) [On-line] Available: http://www.tc.columbia.edu/ccrc/PAPERS/grubb3.pdf
Hagedorn, L. S. (1996). Wage equity and female faculty job satisfaction: The role of wage differentials in a job satisfaction causal model. Research in Higher Education, 37(5), 569-598.
Hardy, D. E., & Laanan, F. S. (2003, November). Understanding community college faculty: Characteristics, opinions, and satisfaction. Paper presented at the annual meeting of the Association for the Study of Higher Education, Portland, OR.
Herzberg, F. (1966). Work and the nature of man. Cleveland, OH: World Publishing.
Hill, E. A. (1986). Job satisfaction facets as predictors of commitment to or withdrawal from the work organization among selected community college faculty in New York State. Community/Junior College Quarterly of Research and Practice, 10(1), 1-11.
Hill, M. (1983). Some factors affecting the job satisfaction of community college faculty in Pennsylvania. Community/Junior College Quarterly of Research and Practice, 7(4), 303-317.
Hill, M. D. (1986-1987). A theoretical analysis of faculty job satisfaction/dissatisfaction. Educational Research Quarterly, 10(4), 36-44.
Hox, J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Lawrence Erlbaum Associates.
Hu, L. T., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling (pp. 76-99). Thousand Oaks, CA: Sage Publications.
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55.
Huber, M. T. (1998). Community college faculty attitudes and trends, 1997. Stanford, CA: National Center for Postsecondary Improvement, Stanford University.
Hutton, J. B., & Jobe, M. E. (1985). Job satisfaction of community college faculty. Community/Junior College Quarterly of Research and Practice, 9(4), 317-324.
Johnsrud, L. K., & Heck, R. H. (1994). A university's faculty: Predicting those who will stay and those who leave. Journal for Higher Education Management, 10(1), 71-84.
Johnsrud, L. K., & Rosser, V. J. (2002). Faculty members' morale and their intentions to leave: A multilevel explanation. The Journal of Higher Education, 71(1), 34-59.
Joreskog, K., & Sorbom, D. (2000). LISREL: User's Guide, Version 8.54. Lincolnwood, IL: Scientific Software International.
Lee, T. W., & Mowday, R. T. (1987). Voluntarily leaving an organization: An empirical investigation of Steers and Mowday's model of turnover. Academy of Management Journal, 30(4), 721-743.
Lever-Duffy, J., & Lemke, R. A. (1996). Introduction. In J. Lever-Duffy, R. A. Lemke, & L. Johnson (Eds.), Learning without limits (pp. vii-xii). League for Innovation in the Community College & Miami-Dade Community College District.
Locke, E. A., Fitzpatrick, W., & White, F. M. (1983). Job satisfaction and role clarity among university and college faculty. Review of Higher Education, 6(4), 343-365.
Loehlin, J. C. (1998). Latent variable models: An introduction to factor, path, and structural analysis (3rd ed.), Mahwah, NJ: Lawrence Erlbaum.
Massy, W. F. (2003). Honoring the trust: Quality and cost containment in higher education. Bolton, MA: Anker.
Matier, M. W. (1990). Retaining faculty: A tale of two campuses. Research in Higher Education, 31(1), 39-61.
McBride, S. A., Munday, R. G., & Tunnell, J. (1992). Community college job satisfaction and propensity to leave. Community/Junior College Quarterly of Research and Practice, 16(2), 157-165.
Miller, R. I., Finley, C, & Vancko, C. S. (2000). Evaluating, improving, and judging faculty performance in two-year colleges. Westport, CT: Bergin & Garvey.
Milosheff, E. (1990). Factors contributing to job satisfaction in the community college. Community College Review, 18(1), 13-22.
Mobley, W. H. (1982). Employee turnover: Causes, consequences, and control. Menlo Park, CA: Addison-Wesley.
Muthen, L. K., & Muthen, B. O. (2003). Mplus: The comprehensive modeling program for applied researchers user's guide, Version 2.14. Los Angeles, CA: Muthen & Muthen.
National Center for Education Statistics. (2001, September). Institutional policies and practices: Results from the 1999 National Study of Postsecondary Faculty, Institution Study. Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement. (NCES 2001-201)
National Survey of Postsecondary Faculty, 1999 (1999). Washington, DC: National Center for Educational Statistics, U.S. Department of Education.
Olsen, D. (1993). Work satisfaction and stress in the first and third year of academic appointment. Journal of Higher Education, 64(4), 453-471.
Paulson, K. (2002). Reconfiguring faculty roles for virtual settings. The Journal of Higher Education, 73(1), 123-140.
Raykov, T., & Marcoulides, G. A. (2000). A first course in structural equation modeling. Mahwah, NJ: Lawrence Erlbaum.
Rifkin, T. (2000). Public community college faculty (New Expeditions Issues Paper No. 4). Washington, DC: American Association of Community Colleges.
Rosser, V. J. (2004). Faculty members' intentions to leave: A national study on their worklife and satisfaction. Research in Higher Education, 45(3), 285-309.
Smart, J. C. (1990). A causal model of faculty turnover intentions. Research in Higher Education, 31(5), 405-424.
Steers, R. M., & Mowday, R. T. (1981). Employee turnover and post-decision accommodation processes. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 3, pp. 235-281). Greenwich, CT: JAI Press.
Thomas, S. L., & Heck, R. H. (2001). Analysis of large-scale secondary data in higher education research: Potential perils associated with complex sampling designs. Research in Higher Education, 42, 270-277.
Townsend, B. K. (1998). Women faculty satisfaction with employment in the community college. Community College Journal of Research and Practice, 22(7), 655-661.
Truell, A. D., Price, W. T., & Joyner, R. L. (1998). Job satisfaction among community college occupational-technical faculty. Community College Journal of Research and Practice, 22(2), 111-122.
Valadez, J. R., & Antony, J. S. (2001). Job satisfaction and commitment of two-year college part-time faculty. Community College Journal of Research and Practice, 25(2), 97-108.
Voorhees, R. A. (2001). Financing community colleges for a new century. In M. B. Paulsen & J. C. Smart (Eds.), The finance of higher education: Theory, research, policy & practice (pp. 480-500). New York: Agathon Press.
Weiler, W. C. (1985). Why do faculty members leave a university? Research in Higher Education, 23(2), 270-278.
Wild, L. (2002, April). Work-related stress factors affecting the community college dean. Paper presented at annual meeting of Council for the Study of Community Colleges, Seattle, WA.
An earlier version of this study was presented at the annual meeting of the American Educational Research Association in San Diego, CA, in April 2004.
Vicki J. Rosser is Assistant Professor of Higher Education in the Department of Educational Leadership & Policy Analysis at the University of Missouri-Columbia. Barbara K. Townsend is Professor of Higher Education in the Department of Educational Leadership & Policy Analysis at the University of Missouri-Columbia.
|Printer friendly Cite/link Email Feedback|
|Author:||Rosser, Vicki J.; Townsend, Barbara K.|
|Publication:||Journal of Higher Education|
|Date:||Jan 1, 2006|
|Previous Article:||The impact of appointment type on the productivity and commitment of full-time faculty in Research and Doctoral institutions.|
|Next Article:||Factors contributing to faculty incorporation of diversity-related course content.|