Faculty Time Allocations and Research Productivity: Gender, Race and Family Effects
This study, drawn from data about 14,614 full-time faculty, examines total faculty work hours, research productivity, and allocation of work time among teaching, research, and service. Variation in time expenditures and research output are influenced by gender, race/ethnicity, and family (marital/parental) status, but findings are also sensitive to definitions of total work hours and research productivity. These findings have important implications for how administrators and faculty define productivity and for the status of underrepresented groups within the academy.
Relationships between individual characteristics and the career achievements of faculty are well-established. Among full-time instructional faculty, 35.9% of male faculty held the rank of full professor in 1993, compared to only 20.1% of female faculty (U.S. Department of Education, 1997). Male faculty also earn higher salaries than female faculty, even after controlling for [End Page 367] conventional salary predictors (Bellas, 1994). In addition, male faculty who are married occupy higher-level positions and earn more than their unmarried counterparts, a relationship that does not appear to hold for women (Bellas, 1992). Whites are more likely to occupy the upper ranks than faculty of color. Approximately 30.3% of White, full-time instructional faculty were full professors in 1993, compared to 27.8, 19.8, and 17.6% of Asians, Latinos, and Blacks respectively. Salary patterns across racial and ethnic groups are less clear cut and appear to interact with gender (Toutkoushian, 1998).
Differences in human capital and structural barriers contribute to intergroup differences in achievement. For example, all groups may not share equally in the resources required for research, leading to differences among groups in research output. Women publish less than men on average (Long & Fox, 1995), and unmarried men publish less than married men (Bellas, 1992), although married women appear to publish more than unmarried women (Long 1990; Astin & Davis 1985). Toutkoushian (1998) found that Blacks had the lowest level and Asians the highest level of research output among various racial/ethnic groups. Given the critical role of research productivity in tenure and promotion decisions and salary increases, these differences have obvious consequences for the status of underrepresented groups within the academy.
Our research asks (a) whether faculty differ across gender, racial/ethnic, and family status groups in how they spend their time, and (b) the extent to which any differences help explain intergroup variation in faculty research productivity. The literature on faculty frequently refers to women’s heavier teaching and service loads relative to men’s, with negative consequences for research time and productivity (Park, 1996; Menges & Exum, 1983). Parallel statements appear in the literature on minority faculty (Garza, 1993; Moses, 1989; Banks, 1984). However, these assessments typically rely on limited descriptive studies of faculty time allocations, studies that do not take into account many of the individual and institutional characteristics that may partly explain intergroup variation. In addition, comparisons of the percentages of time that faculty devote to teaching, research, and service may be misleading if not considered in the context of total hours worked. A faculty member who devotes 50% of his or her time to teaching may actually spend a greater number of hours teaching than one who devotes 60% of his or her time to teaching, if the former works substantially more hours per week.
While we are interested in assessing intergroup variation in time expenditures and how time expenditures may affect research output, our study is also motivated by the possibility that definitions of work and research output influence conclusions about intergroup differences in research productivity. Should work include only paid activities at the employing institution, [End Page 368] or should it also include unpaid services to one’s institution and one’s profession? Moreover, intergroup differences may result from the definition of research productivity used. Should research productivity include only journal articles, scholarly books, and book chapters, or should it be construed more broadly to include conference presentations and publications in nonjuried media? Despite calls for institutions of higher education to recognize and reward a broader range of scholarship (Boyer, 1990; Creamer, 1998; Park, 1996), publication counts remain the dominant method of evaluating most faculty, with traditional measures of output being valued more highly than others.
These issues and questions are significant. Faculty time expenditures—both in various activities and in total hours worked—have obvious ramifications for faculty research output, which in turn affects retention, promotion, compensation, and peer recognition. Institutions that define work activities and research productivity narrowly may disadvantage some groups while advantaging others. As Creamer (1998) observes:
The profile of faculty across this country has remained so stubbornly homogeneous because of the reluctance to relinquish traditional measures of faculty productivity. A narrow definition of what constitutes a contribution to knowledge represents only a fragment of academic discourse, and it awards the privilege of an authoritative voice to only a few scholars. Expanding these definitions will benefit minority, female, and male academics alike.(p. 4)
While we believe that conceptualizations of productivity should include teaching and service activities, we also recognize that under the current reward structure, “published research is the common currency of academic achievement” (Derek Bok, qtd. in Rau & Baker, 1989, p. 163). Consequently, we consider alternative definitions of research productivity as we examine the extent to which gender, race/ethnicity, and family status affect the time expenditures and research output of faculty.
Existing studies of faculty time expenditures have been limited in sample size and scope, and/or by statistical methods. Nonetheless, they provide a useful starting point for an analysis of the potential effects of gender, race/ethnicity, and family status on faculty time allocations and subsequent research productivity. Yuker’s (1984) summary, which conveys the state of research in the mid-1980s, indicates that gender had not yet been considered in studies of faculty time expenditures, except in a few institutional studies. He did not mention race/ethnicity or marital/parental status, indicating the lack of research on their potential effects as well. [End Page 369]
Aside from analyses of single institutions or state systems, few studies of faculty time expenditures appeared after Yuker’s (1984) summary until the National Center for Education Statistics conducted its first National Survey of Postsecondary Faculty in 1988 (NSOPF-88). Analyses of these data revealed that men and Whites worked more total hours per week than women and racial/ethnic minorities respectively, although some differences disappeared when comparisons were made by rank and institutional type (Russell et al., 1991). Analyses did not examine whether other variables, such as experience and academic field, also affected total hours worked.
The NSOPF-88 data also showed that on average men spent a higher percentage of time in research activities, while women spent a higher percentage of time in teaching and service activities (Russell et al., 1991, p. 152). After taking rank and institutional type into account, Russell et al. (1991) found most gender differences no longer statistically significant. Several smaller studies offered similar conclusions (Blackburn & Lawrence, 1995; Stark et al., 1986). Russell et al. (1991) also reported “no appreciable differences in how non-minorities and minorities allocated their time across major types of activities” (pp. 162–163), a conclusion consistent with an earlier study by Elmore and Blackburn (1983). It is important to note, however, that not adjusting for gender or racial/ethnic differences in total hours worked may understate comparisons of the time these groups allocate to various work activities. In addition, samples that contain relatively few faculty of color necessitate aggregating racial/ethnic categories or omitting race/ethnicity from the analysis, strategies that may mask significant intergroup differences.
Singell, Lillydahl, and Singell (1996) applied multinomial logit analysis to the NSOPF-88 data but restricted their sample to 1,409 full-time arts and sciences faculty employed at four-year institutions. They reported that only at research universities did women spend more time than men in teaching—at liberal arts colleges women actually spent less time in teaching. They suggested that this may be due to differences in the distribution of women and men across academic disciplines, variables they did not include in their analyses. Singell et al. found that, except at doctoral institutions, minority faculty (all groups combined) spent more time in activities outside the institution than Whites and less time in teaching, research, or service, depending on type of institution. They also found that married faculty spent more time on research at doctoral and comprehensive universities but less time on research at liberal arts institutions.
Finkelstein, Seal, and Schuster (1998) examined more recent data from the 1993 National Survey of Postsecondary Faculty (NSOPF-93). Their analyses compared faculty with seven or fewer years of full-time teaching experience to their more senior colleagues. While the study introduced some [End Page 370] controls, such as rank and broad disciplinary groupings, it too is descriptive in nature. This limits conclusions that can be drawn about the reasons for intergroup differences in faculty time allocations.
In sum, previous studies of faculty time expenditures have been limited by sample characteristics and statistical techniques. Although some recent studies have included analyses by gender and race, they typically aggregate racial and ethnic groups or exclude minorities from their analyses altogether. With the exception of Singell et al. (1996), no studies have examined the potential effects of marital status on faculty time expenditures or the potential effects of parental status. In addition to addressing these methodological issues, our study extends previous work by examining the potential effects on research productivity of intergroup variation in time expenditures, as well as the possibility that different conceptualizations of work and research output may affect the apparent productivity of groups. Traditional measures of work and research output may bias results against historically marginalized groups.
Data and Analytic Strategy
We analyzed data from the 1993 National Survey of Postsecondary Faculty (NSOPF-93) sponsored by the National Center for Education Statistics, the most recent national data available. Data collection involved a two-staged stratified random sampling process to first select a sample of 974 institutions (stratified by institutional type) from a population of 3,256 public and private postsecondary institutions, and then to select 31,354 faculty from within broad disciplinary groups. Eight hundred seventeen institutions and 25,780 individuals had participated in the study in the fall of 1992. Omitting 1,590 ineligible faculty yielded a response rate of 86.6% (National Center for Education Statistics, 1994). For this study, we restricted the sample to full-time faculty employed at two- and four-year institutions who held the rank of lecturer/instructor, assistant, associate, or full professor. These restrictions produced a sample of 14,614 faculty members. 1 The Appendix describes all variables included in our analyses and their methods of construction.
Our sample was 59% male and 41% female. Eighty percent were non-Latino Whites and 9% were non-Latino Blacks. Latinos and Asians each comprised 5% of the sample. Seventy-three percent of respondents were married; the average number of dependents was 1.38. One-fourth of the faculty were employed by two-year institutions, and one-third by comprehensive [End Page 371] universities. The rest were evenly divided between doctoral and research universities according to Carnegie Council (1994) classifications.
We first calculated average differences in time allocations among teaching, research, and service; total hours worked weekly; and research output by gender, race/ethnicity, and marital status. We then estimated regression equations predicting the percentage of time faculty allocated to teaching, research, and service activities, employing extensive controls for individual, job, and institutional characteristics. We did the same for total number of hours worked per week and for research output, using three measures of work and three measures of research output.
Results and Discussion
Average Group Differences in Time Expenditures
We begin by reporting the average percentages of time faculty spent in teaching, service, and research activities by gender, four racial/ethnicity categories, and marital status. (See Table 1.) Because these percentages are most meaningful in the context of the total number of hours worked weekly, Table 1 also provides these figures. We considered three definitions of work in an attempt to determine whether intergroup differences in time expenditures are sensitive to various definitions. The first definition includes only hours of paid work at the employing institution (e.g., teaching, research, service, administration). The second definition adds hours of unpaid work at the employing institution, and the third adds unpaid professional service outside the institution. Thus, the total number of hours worked increases as the definition of work expands.
Means and Standard Deviations for Selected Variables by Gender, Race/Ethnicity, and Marital Status
|% Time in Teaching||57.9 (28.0)||
||62.1 (27.8)||58.1 (28.2)||56.6 (27.5)||58.4 (26.6)||56.0 (26.8)||
|% Time in Service||6.3 (12.1)||
||6.8 (12.9)||6.1 (12.0)||7.9 (12.6)||7.3 (13.4)||5.8 (12.1)||6.2 (11.9)||6.6 (12.6)|
|% Time in Research||15.5 (19.9)||
||11.8 (17.4)||15.1 (20.0)||12.6 (15.4)||17.3 (20.3)||23.5 (23.5)||
|Hours Worke d-1||41.2 (15.0)||
||39.8 (14.9)||41.9 (14.5)||36.7 (17.3)||38.8 (16.1)||40.8 (15.6)||
|Hours Worke d-2||46.6 (15.4)||
||45.6 (15.5)||47.0 (14.8)||43.4 (18.8)||45.6 (16.9)||47.5 (16.3)||46.7 (15.1)||46.6 (16.2)|
|Hours Worke d-3||48.7 (16.0)||
||47.7 (16.0)||49.0 (15.2)||46.6 (20.0)||47.7 (17.6)||49.5 (16.8)||48.7 (15.6)||48.8 (16.7)|
|Research Output -4||1.8 (3.5)||
||1.2 (2.6)||1.7 (3.5)||1.2 (2.8)||2.2 (3.7)||2.9 (4.9)||
|Research Output -5||2.9 (9.4)||
||2.0 (6.4)||2.9 (9.4)||2.6 (10.3)||3.9 (11.3)||3.4 (6.1)||
|Research Output -6||8.0 (14.5)||
||6.3 (11.3)||8.0 (14.5)||7.2 (14.7)||9.5 (16.6)||8.5 (13.1)||
Averages for percentage time spent in teaching, service, and research do not add to 100 because time spent in professional growth, administration, and outside consulting are excluded. These activities are included in measures of total hours worked as appropriate. We used t-tests to compare means, except for racial/ethnic groups for which we used ANOVAs (all were significant except time spent in teaching).
Hours Worked-1 is average hours spent per week in all paid activities at the institution.
Hours worked-2 adds unpaid activities at the institution.
Hours worked-3 adds unpaid professional service outside the institution.
Research output-1 includes articles in refereed journals, book chapters, other books, monographs, patents, copyrights.
Research output-2 adds works in juried media, exhibitions/performances in fine or applied arts.
Research output-3 adds articles in nonrefereed journals, works in nonjuried media, reviews, textbooks, research or technical reports, conference presentations, and software.
The group averages reported in Table 1 reveal a number of significant differences in the time expenditures of faculty by gender, race/ethnicity, and marital status. 2 On average, men devoted about 6% more work time than women to research, mainly at the expense of teaching. Similar differences are evident by racial/ethnic group, with Asians and Latinos spending greater average proportions of their work time on research than other groups. While married faculty spent more time on research than nonmarried faculty, the percentage differences are relatively small.
Regardless of which definition of work is considered, men devoted significantly more hours per week to total work activities than women, on average, while Asians and Whites tended to work more total hours than faculty in other racial/ethnic categories. On average, married faculty did not seem to work fewer hours than unmarried faculty; in fact, married faculty [End Page 372] worked significantly more hours per week using the most restrictive definition of work.
Average Group Differences in Research Output
These differences in the time expenditures of faculty across gender, racial/ethnic, and marital groups may affect research productivity. Table 1 reports average number of research products across groups for three definitions of research output. The first measure (Group 1) includes number of refereed articles, chapters, books, monographs, and patents or copyrights produced in the two years preceding the survey. 3 This category includes standard measures of productivity—those most highly regarded by institutions and many faculty. Since the first definition disadvantages faculty in some disciplines, Group 2 adds creative works in juried media and exhibitions or performances in the fine or applied arts. Group 3 broadens the measure of research output even further, to include nonjuried outlets, book reviews, textbooks, technical reports, conference presentations, workshops, and computer software.
Comparing research output across groups verifies previous research findings that, on average, male faculty have higher levels of research output than female faculty. We also found significant differences in research output by racial/ethnic group, with Asians and Latino faculty having the greatest research output. Married faculty also had higher levels of research output than unmarried faculty. These results suggest that at the aggregate level important differences exist in how faculty in selected categories allocate their time, both among competing work activities and between work and other activities, with consequences for research output. Not surprisingly, those groups of faculty who devoted more time to research and/or more time to work overall tended to have higher levels of research output.
Before establishing a causal link between time allocation and research productivity, we must control for other factors that could contribute to intergroup differences, such as the concentration of faculty in certain fields, differences in age and experience, and the types of institutions by which faculty are employed. For example, women’s greater time expenditures in teaching may reflect their disproportionate representation at teaching-oriented institutions, within the lower ranks, and in the humanities—all factors that increase time spent in teaching activities. To control for these factors, [End Page 373] we estimated regression models predicting both the percentage of time faculty devoted to teaching, service, and research activities, and total hours spent on work activities. These models, shown in Table 2, include the variables of primary interest (gender, race/ethnicity, marital status, and number of dependents), and an array of control variables: years of experience, age, type of institutional affiliation, highest degree, length of appointment, rank, whether the faculty member was a department chair, academic field, and the percentage of women in these fields. To simplify our presentation, we do not report or discuss all coefficients (see notes, Table 2); instead, we highlight major findings.
Multiple Regression Models Explaining Percentage of Time Spent in Teaching, Service, and Research, and Total Hours Worked (N = 14,614)
|Independent Variable||Dependent Variables||Hours-1||Hours-2||Hours-3|
|% Teaching||% Service||% Research|
|Asian||0.30 (0.92)||−0.15 (0.45)||
|Other race||−0.91 (1.64)||−0.26 (0.80)||
||−0.05 (1.02)||0.32 (1.06)|
||0.37 (0.35)||0.06 (0.29)||−0.41 (0.31)||−0.50 (0.32)|
|Research I or II||
|Doctoral I or II||
|Comprehensive I or II||
|Liberal Arts I or II||
||0.59 (.39)||0.62 (0.41)||
||−0.20 (0.29)||0.73 (0.43)||−0.23 (.36)||0.14 (0.37)||0.29 (0.39)|
Notes: Table reports unstandardized coefficients (standard errors in parentheses). Models also control for highest degree, years of experience, years of experience squared, age, age squared, appointment length, whether a faculty member is a chairperson, academic field (43 dummy variables), and the percentage of women in these fields.
**. p< .01
* p< .05
+. p< .10 (two-tailed tests)
Hours-1 is average weekly hours spent in paid activities at the institution.
Hours-2 is weekly hours spent in paid and unpaid activities at the institution.
Hours-3 is weekly hours spent in all activities at the institution plus unpaid professional service.
Regressions Predicting Percentage of Time Spent in Teaching, Service, and Research, and Total Hours Worked
Gender. Controlling for other factors, we found that women spent significantly more time in teaching than men and less time in research, but women did not differ from men in the percentage of time devoted to service activities. Note, however, that assistant professors spent more time in service activities than other faculty. Because women (and faculty of color) are disproportionately represented among assistant professors, controlling for rank may underestimate the effect of being a woman and of being a person of color on time spent in service. 4 The reasons for these gender difference in time expenditures are not clear. Women may be assigned heavier teaching loads than their male colleagues (e.g., more courses or more students per course), or women may spend more time in course preparation than men. Women may also use more labor-intensive evaluation methods or simply have greater interest in teaching—differences reported by Astin, Korn, and Dey (1991, pp. 57, 63, 77, 83), though without controls for institutional type or other relevant factors. Table 2 also shows that women worked about one hour less per week than men, regardless of the definition of work.
Race/ethnicity. Table 2 indicates that, compared to their White counterparts, Black faculty spent less time in teaching and that both Black and Latino faculty spent more time in service. Blacks’ lower time in teaching is offset somewhat by their greater time allocated to service activities. This finding supports concerns that, because Black faculty are few, they face excessive demands on their time from Black students, administrators, and community groups (Banks, 1984; Exum, 1983; Moses, 1989). Although Blacks and Whites, and Latinos and Whites did not differ in time devoted to research, Asian and “other” (race unspecified) faculty devoted a higher percentage of time to research than Whites. [End Page 376]
Differences in total hours worked weekly are also evident across racial/ethnic groups, after controlling for relevant factors. (See Table 2.) Blacks, Latinos, Asians, and faculty of other races spent significantly less time in paid activities than Whites (ranging from two and one-half to four and one-half hours per week). Broadening the definition of work to include unpaid activities at the institution and then unpaid professional service, reduces and, in some cases, eliminates the difference between White faculty and other racial/ethnic groups. However, the difference between Whites and Blacks remains statistically significant and the difference between Whites and Latinos marginally significant. Thus, after controlling for relevant predictors of time use, Whites spent more time in paid activities than other faculty, but faculty of color compensated either fully (Asians and “others”) or partially (Blacks and Latinos) by doing more unpaid activities at the institution and rendering more professional service. While the NSOPF-93 survey does not specify the nature of these activities, faculty of color may be more responsive than Whites to requests for unpaid work both within and outside their institutions. This willingness may pose problems if academic reward structures value narrower definitions of work activities and their outcomes. Faculty who spend less time in paid activities (particularly research) and more time in nonpaid activities may produce fewer standard research products, with negative consequences for career success.
Family Status. Marital status and number of dependents did not affect time devoted to teaching or research activities. Interestingly, both factors influenced time devoted to service activities, but in opposite ways. Married faculty spent less time in service than comparable unmarried faculty, presumably because of family demands, though this group did not spend less time working overall than their unmarried counterparts. Faculty with more dependents spent more time in service, leading us to speculate that faculty who live in larger households may simply be more accommodating of others’ requests. Faculty with more dependents worked fewer hours per week than those with fewer dependents, consistent with the interpretation that family demands constrain the time that faculty can allocate to work activities.
Regressions Predicting Research Output
Table 3 shows the results of regressions predicting three alternative measures of research output. Recall that the first measure (Group 1) includes journal articles, chapters, books (excluding textbooks), and patents. The second measure (Group 2) adds juried exhibitions and creative works, and the third measure (Group 3) also adds other categories such as nonrefereed publications. (See the Appendix for additional details about these categories.) Our regression models include the variables of primary interest (gender, race/ethnicity, and marital and parental status), all of the individual, [End Page 378] job, and institutional variables included in our previous regression models (Table 2), and some additional variables, including percentage of time spent in teaching and service, and total hours worked weekly (broadest definition). As in Table 2, we do not report or discuss all coefficients included in the analysis. (See notes, Table 3.)
Multiple Regression Models Explaining Alternative Measures of Faculty Research Output (N = 14,614)
|Dependent Variable: “Output” during Last Two Years|
|Independent Variable||Group 1a||Group 2b||Group 3c|
||−0.022 (0.259)||−0.434 (0.396)|
||0.269 (0.339)||−0.472 (0.518)|
||−0.093 (0.605)||0.700 (0.925)|
|% Time in Teaching||
|% Time in Service||
Hours worked per week
|Research I or II||
|Doctoral I or II||
|Comprehensive I or II||0.118 (0.078)||
|Liberal arts I or II||0.057 (0.105)||−0.031 (0.303)||0.411 (0.464)|
||0.203 (0.222)||0.543 (0.339)|
Notes: Table reports unstandardized coefficients (standard errors in parentheses). Models include additional controls for highest degree, years of experience, years of experience squared, age, age squared, appointment length, whether a faculty member is a chairperson, academic field (43 dummy variables), and the percentage of women in these fields.
**. p < .01
* p < .05
+. p < .10 (two-tailed tests).
a. Includes articles in refereed professional or trade journals, chapters in edited volumes, other books, monographs, and patents or copyrights.
b. Adds creative works in juried media and exhibitions or performances in fine or applied arts.
c. Adds articles in nonrefereed professional or trade journals, creative works in nonjuried media or in-house newsletters, reviews of books, articles, or creative works, textbooks, research or technical reports disseminated internally or to clients, presentations at conferences and workshops, and computer software.
d. Includes paid and unpaid work within the institution and unpaid professional service.
Faculty who spent more time in teaching and service activities produced fewer research products, consistent with the view that teaching, service, and research tend to be mutually exclusive activities and compete for faculty members’ time and attention (Fairweather, 1993; Fox, 1992). Total hours worked also influenced research output, with greater total time expenditures leading to greater research output. As one would expect, faculty at research and doctoral institutions reported higher output than those at two-year institutions (the reference category) for all definitions of output. Faculty at comprehensive institutions differed from those at two-year institutions only for broader measures of research output. Full professors produced more research output than assistant professors (the reference category), while lecturers/instructors produced less. Associate professors produced more research output than assistant professors, but only for the narrowest measure of output.
Gender. Although women produced less than men by all three measures of research output, more than half to three-fourths of the gender difference is explained by other variables in the regression models. The gender difference is smallest for the most restrictive definition of research output, with women publishing only one-fifth fewer articles, books, chapters, and patents than men in the two years prior to the survey. While this difference is small, it is nonetheless notable because it remains after controlling for percentages of time spent in teaching and service and total hours worked weekly, as well as an array of other control variables. Thus, time allocation and total hours worked weekly do not account for all of the observed gender difference in research output.
This finding suggests that women may work at a slightly slower pace than men. Sonnert and Holton (1996) contend that women may take more care in the research process because their work is scrutinized more closely than men’s. As a result, women may spend more time in each research product relative to men. Women may also feel less confident in their research abilities than men (Schoen & Winocur, 1988), a factor that would also contribute to a more meticulous research style. It is important to recognize that, like other studies of faculty productivity, the NSOPF-93 data do not take into account research quality. The gender gap in publishing is also consistent with the existence of external barriers to publishing for women. Exum (1983) notes that women (and minority faculty) are likely to be excluded from networks that contribute to research and publishing success. They may also face greater obstacles in publishing in traditional outlets because [End Page 379] their research may challenge existing paradigms. Additionally, women and men may differ on average in how they spend their summer months, something that the NSOPF-93 survey did not assess. Women may be more likely to stay home with children during the summer while men work on campus, a difference that could contribute to gender differences in research output but which would not be reflected in the time allocation measures constructed here.
Race/ethnicity. By all three measures, Latino faculty produced significantly more research output than Whites, after controlling for other variables in the model. At the same time, the research output of the remaining racial/ethnic groups differed from Whites only for the narrowest definition of research output (Group 1). This finding demonstrates that many of the observed aggregate-level differences in research output by race/ethnicity are sensitive to how research output is measured. Within the narrowest definition of research output, we found that Blacks showed somewhat lower output than Whites. This finding raises two questions: Why do Black faculty invest their research time in less traditional and less highly rewarded research outlets? And with what consequence? As previously discussed, Blacks may face greater service demands than Whites. However, the difference between Whites and Blacks in research output persists after controlling for time spent in service and teaching, for total work hours, and for individual, job, and institutional variables. As suggested for women, Blacks may face greater difficulties publishing in traditional research outlets. Also, as suggested for women, at least part of these intergroup differences may stem from how faculty spend their summers. Blacks may have greater financial burdens than Whites of comparable income levels (Oliver & Shapiro, 1995), leading to greater pressure to spend time in (paid) summer teaching rather than (unpaid) research.
Many of the same issues are raised in the literature on Latino faculty as well (Garza, 1993; Luz Reyes & Halcón, 1988). In contrast to Blacks, however, [End Page 381] Latinos had higher levels of research output than Whites by all definitions of output, despite spending fewer total hours working weekly. Yet it is important to recognize that relatively high productivity does not ensure that Latinos (or other faculty of color, or women) will be treated fairly in the tenure and promotion process. Data from a national survey of Latino faculty found that those who had been denied tenure did not differ significantly in their research productivity from those who received tenure. However, Latinos who were denied tenure were more likely to conduct research on their own ethnic group, suggesting that such research is devalued (Garza, 1993). As Nakanishi (1993) explains, “Ethnic and gender research that frequently confronts and challenges prevailing analytical perspectives and explores sensitive issues of racism and intergroup relations has yet to be fully accepted and embraced as important, relevant, or exciting subjects of study by many faculty members” (p. 54). Clearly, closer examination of the experiences of Latino faculty is needed.
Asian faculty are often perceived as being more productive than other faculty; and, indeed, this group spent an extremely high percentage of time in research, spent more total time in paid activities at the institution, and produced higher outputs as measured by traditional means than other groups. However, Asians’ advantage over Whites disappears when the definition of research output expands to include less traditional products.
Family status. Married faculty had higher research output than unmarried faculty for all three definitions of output. Although the positive effect of marriage on men’s research output is consistent with previous findings (Bellas, 1992; Long, 1990), studies have not consistently found differences by marital or parental status for women (e.g., Cole & Zuckerman, 1987). Long (1990) did find that marriage correlated positively to research productivity for women scientists, which he attributes to higher collaboration rates among those who are married. Similarly, analyzing data from a sample of 9,500 female faculty at 98 institutions, Astin and Davis (1985) found that marriage positively affected women’s research productivity. They speculated that single women are excluded from male networks to a larger extent than married women. In other words, single women hold two deviant statuses from the normative faculty member—gender and marital status. Women of color have a third deviant status that may interfere with their ability to use networks advantageously in producing traditional measures of research output (Boice, 1993). In addition, because single women are likely to face fewer demands on their time outside of work, they may have larger blocks of time to devote to book manuscripts, which would reduce the time spent writing journal articles.
Despite their lower total work hours, faculty with more dependents produced more research output than faculty with fewer dependents (productivity measures 1 and 3). The positive effect of number of dependents on [End Page 382] research output is consistent with the explanation that men view themselves as the primary breadwinner and that publishing is an important route to higher salaries (Toutkoushian, 1998; Bellas, 1994). The greater proportion of men in the sample may overshadow any negative effect of dependents on research output for women (Long, 1990). 5
Conclusions and Policy Implications
Using data for 14,614 faculty from the NSOPF-93 survey, our analyses show considerable variation by gender, race/ethnicity, and family status in time devoted to teaching, service, and research activities, and in total hours worked weekly. Importantly, findings suggest that this variation contributes to intergroup differences in research output, but differences are sensitive to how narrowly or broadly research output is defined. Given the current emphasis of academic reward structures on research output, these findings have important implications for the status of underrepresented groups within the academy. Because research and publishing tend to be more heavily rewarded than teaching and service, faculty who devote more time to research and less to other activities will have a greater likelihood of success.
Results indicate that focusing on traditional measures of research output skews productivity ratings toward White and Asian men, and away from women and Blacks, raising the question of how the academy determines the value of different kinds of output measures (Park, 1996). Remedies include two options: (a) revise the current academic reward structure to increase rewards for teaching, service, and less traditional measures of research output, and broaden conceptualizations of work; or (2) encourage women and Blacks to spend more time in research and focus on more traditional types of research output, and, in conjunction, work to remove structural barriers to publishing.
With regard to the first option, growing pressures from state legislatures for greater attention to undergraduate teaching and public service may attenuate the emphasis on traditional measures of research output, particularly at public universities. These efforts to increase faculty accountability, driven largely by growing constraints on state budgets and rising tuition costs, are not likely to diminish in the near future. In addition to external pressures for professors to spend more time in teaching and service, voices from within the academy have called for greater recognition of these activities [End Page 383] (e.g., Creamer 1998; Park 1996; Boyer 1990). Despite these appeals, radical changes in the current reward structure seem unlikely since faculty who have succeeded and gained status under existing structures will resist altering it (Rau & Baker, 1989; Banks 1984). Academic institutions also have a stake in maintaining the current system since they too gain status from the reputations of their faculty. Moreover, the current shift away from tenured positions and toward the use of part-time and adjunct faculty will likely cause criteria for tenure and tenure-track jobs to become more rather than less restrictive.
If we assume that the academic reward structure will not change dramatically in the near future (or that it will become more restrictive rather than more inclusive), it is important to ensure that all groups have equal opportunities for success. Teaching and service loads should be equitably distributed within and among departments. Equity encompasses not only number of courses, but number of course preparations, frequency of new courses, number of students, and support from graduate assistants. In addition, the level of preparation required of courses can vary (graduate versus undergraduate; course content that changes very little versus that which changes almost daily because it deals with current issues). And, of course, the labor intensity of different types of evaluation methods varies tremendously. Similarly, equity in service goes beyond merely counting the number of committees on which faculty serve (though this is an important initial step). Some committees are far more demanding than others, and administrators should take this factor into account when making committee assignments. Working toward fairness in task distribution demands that faculty have a thorough knowledge of the “rules of the game” at their particular institution—something that is currently not the case (Tierney & Bensimon, 1996). Administrators and colleagues have a responsibility to provide guidance in negotiating the tenure and promotion system and to make the reward structure explicit. What is required for tenure and promotion? Are book chapters valued or should faculty devote their time only to refereed journal articles? How are coauthored publications viewed relative to sole-authored? Should untenured faculty spend time pursuing external grant funds and, if so, how will any resulting gaps in the vita be interpreted? How much weight is given to teaching and service, relative to publishing? Is it advisable to decline requests for service, or will this be viewed negatively? Answers to questions like these are crucial in successfully negotiating tenure and promotion systems, yet these questions are not asked or are left unanswered far too often. While informal mentoring systems may provide guidance for some, women and faculty of color may be less likely to be mentored or to benefit from informal and formal networks (Johnsrud, 1993; Reid, 1990). Consequently, explicit criteria for tenure and promotion, and formal mentoring systems are crucial to the success of these groups. Such [End Page 384] mechanisms may also help reduce structural barriers to publishing if exclusion from social and professional networks is detrimental to publishing success, as some researchers suggest (Sonnert & Holton, 1996; Fox, 1991; Astin & Davis, 1985).
We caution that simply reducing teaching or service loads may not necessarily lead to higher productivity if faculty spend their time on other activities (Yuker, 1984). For example, Hesseldenz (1976) found a trade-off for faculty at one university between time spent in teaching and time spent in institutional and professional service, rather than between time spent in teaching and time spent in research. Yet our findings show that higher levels of research output are associated with lower time expenditures in teaching and service, indicating that, for this large national sample, there is indeed a trade-off between these activities. However faculty members’ orientations to research and teaching are important predictors of time spent in research (Blackburn & Lawrence, 1995; Fox, 1992; Stark, Lowther, & Hagerty, 1986). Publication success (or its lack) can, of course, influence one’s orientation to research and teaching by creating a feedback mechanism.
The two options we have outlined represent two different approaches to addressing the underrepresentation of women and faculty of color in the academy. The first calls for a change in the academic reward structure, while the second calls primarily for changing individual behavior to better accommodate the existing reward structure, although such an adaptation may also involve structural change. Our own view is that both types of change are needed. We advocate a broadening of the academic reward structure to include less-traditional measures of research output and measures of teaching and service productivity. However, we also recognize that change in this direction is likely to be slow, if it occurs at all. Thus, the reality is that individuals must meet the expectations of the current reward system (or at least know what the expectations are so that they can make informed decisions about the consequences if they do not meet them), while also working to change it. If we are to increase the promotion and retention rates of women and faculty of color (and attract them to the profession initially), we must both reevaluate the existing reward structure with an eye toward removing gender and racial bias and also help individuals meet institutional expectations.
Marcia L. Bellas is an assistant professor of sociology at the University of Cincinnati. Her research includes stratification in labor markets and households, with a focus on academia and the work of professors.
Robert K. Toutkoushian is the Executive Director of the Office of Policy Analysis for the University System of New Hampshire. His research interests include issues relating to faculty compensation, gender and racial equity, state funding for higher education, student demand for postsecondary education, graduate program ratings, and higher education costs. An earlier version of this paper was presented at the annual meetings of the Southern Economic Association, November 1997, in Atlanta, GA. The authors thank Marianne Ferber, Georganne Rundblad, Paula Stephan, and two anonymous referees for their helpful comments on earlier drafts of this paper. Address inquiries to Marcia L. Bellas, Department of Sociology, University of Cincinnati, P.O. Box 210378, Cincinnati, OH 45221-0378; telephone: (513) 556-4700; fax (513) 556-0057; e-mail: Marcia.Bellas@uc.edu
1. We first restricted our analyses to faculty at four-year institutions but found that adding other faculty did not substantively change the results; we therefore opted for the more inclusive sample.
3. While the regression models predict research output within the two years prior to the survey, we are aware that there is typically a lag time between completing research and achieving some outcome measure, such as a publication or an exhibition. Hence, some of the independent variables (e.g., time allocations) do not actually precede the time period measured by the dependent variable. Moreover, productivity may also predict time allocations, if less/more productive faculty choose or are required to spend more/less time in teaching. These time order issues are unavoidable given the cross-sectional nature of the data.
4. Because of our concerns about over-controlling, we first estimated equations without rank, discipline, and the sex composition of fields. Adding them did not substantively change the results so we included all variables simultaneously.
5. For the regressions reported in Tables 2 and 3, we tested for interaction effects between gender and race/ethnicity, gender and family status (marriage and number of dependents), and gender and type of employing institution. While several coefficients were statistically significant, they added almost nothing to the explained variances, so we report the more parsimonious models.
Appendix. Variable Definitions
|Variable Name||Definition and Method of Construction|
|Female||1 if female, 0 otherwise|
|Married||1 if currently married, 0 otherwise|
|Dependents||Number of dependents|
|Race||Series of dummy variables for non-Latino White, non-Latino Black, Latino, Asian, and other; non-Latino White is the reference category|
|Highest degree||Series of dummy variables for doctorate (Ph.D., Ed.D.), professional (e.g., M.D., D.D.S.), and other (e.g., B.A., M.A.); other is the reference category|
|Years of experience||Sums years of experience in three most recent jobs, including current job (part-time jobs weighted by .50)|
|Age||Year of survey (93) minus year of birth (midpoint of the age category used for respondents who did not report year of birth but reported age in a category)|
|Carnegie classification (current employer)||Series of dummy variables for Research I or II, Doctoral I or II, Comprehensive I or II, Liberal Arts I or II, and Two-year; Two-year is the reference category|
|Appointment length||Number of months employed per year in current position|
|Chairperson||1 if faculty member is a department chairperson, 0 otherwise|
|Rank||Series of dummy variables for lecturer/instructor, assistant, associate, and full professor; assistant professor is the reference category|
|Academic field||43 dummy variables representing principal field of teaching|
|Research funding||Total grant funds for research during the 1992–1993 academic year (in millions of dollars)|
|Satisfied w/RA availability||1 if respondent rated the availability of research assistance as “good” or “very good,” 0 otherwise|
|Satisfied w/personal computer||1 if respondent rated the availability of personal computers “good” or “very good,” 0 otherwise|
|Satisfied w/central computing||1 if respondent rated the availability of central computing facilities as “good” or “very good,” 0 otherwise|
|% time in teaching||Estimated percentage of total work time allocated to teaching activities in fall 1992 (includes teaching, grading papers, preparing courses, developing curricula, advising or supervising students, and working with student organizations or intramural athletics)|
|% time in research||Estimated percentage of total work time allocated to research/scholarship in fall 1992 (includes research, reviewing and preparing articles or books, attending or preparing for professional meetings or conferences, reviewing proposals, seeking outside funding, giving performances or exhibitions in the fine or applied arts, and giving speeches)|
|% time in service||Estimated percentage of total work time allocated to service/other nonteaching activities in fall 1992 (includes providing legal or medical services or psychological counseling to clients or patients, paid or unpaid community or public service, service to professional societies/associations, and other activities not considered teaching, research, professional growth, administration, and outside consulting or freelance work)|
|Hours worked-1||Average hours spent per week in all paid activities at employing institution in fall 1992|
|Hours worked-2||Average hours spent per week in all paid and unpaid activities at employing institution, in fall 1992|
|Hours worked-3||Average hours spent per week in all paid and unpaid activities at employing institution, and all unpaid (pro bono) professional service activities outside the institution in fall 1992|
|Research Output-1||Number of research products generated in the two years prior to the survey, including articles published in refereed professional or trade journals, chapters in edited volumes, other books, monographs, patents, or copyrights|
|Research Output-2||Number of research products generated in the two years prior to the survey, including articles published in refereed professional or trade journals, chapters in edited volumes, other books, monographs, patents, copyrights, creative works published in juried media, and exhibitions or performances in fine or applied arts|
|Research Output-3||Number of research products generated in the two years prior to the survey, including articles published in refereed professional or trade journals, chapters in edited volumes, other books, monographs, patents, copyrights, creative works published in juried media, exhibitions or performances in fine or applied arts, articles published in nonrefereed professional or trade journals, creative works published in nonjuried media or in-house newsletters, published reviews of books, articles, or creative works, textbooks, research or technical reports disseminated internally or to clients, presentations at conferences and workshops, and computer software|