|
|
| |||||||||||||||||||||||||||||||||
Background Although the numbers of women in training and in entry-level academic positions in medicine have increased substantially in recent years, the proportion of women in senior faculty positions has not changed. We conducted a study to determine the contributions of background and training, academic productivity, distribution of work time, institutional support, career attitudes, and family responsibilities to sex differences in academic rank and salary among faculty members of academic pediatric departments.
Methods We conducted a cross-sectional survey of all salaried physicians in 126 academic departments of pediatrics in the United States in January 1992. Of the 6441 questionnaires distributed, 4285 (67 percent) were returned. The sample was representative of U.S. pediatric faculty members. Multivariate models were used to relate academic rank and salary to 16 independent variables.
Results Significantly fewer women than men achieved the rank of associate professor or higher. For both men and women, higher salaries and ranks were related to greater academic productivity (more publications and grants), more hours worked, more institutional support of research, greater overall career satisfaction, and fewer career problems. Less time spent in teaching and patient care was related to greater academic productivity for both sexes. Women in the low ranks were less academically productive and spent significantly more time in teaching and patient care than men in those ranks. Adjustment for all independent variables eliminated sex differences in academic rank but not in salary.
Conclusions Lower rates of academic productivity, more time spent in teaching and patient care and less time spent in research, less institutional support for research, and lower rates of specialization in highly paid subspecialties contributed to the lower ranks and salaries of female faculty members in pediatrics.
Listed in descending level of empirical support, the following hypotheses have been proposed for the disparities in academic advancement and salaries between women and men: (1) women publish fewer papers and secure fewer grants1,5,6,9,15,16,17; (2) women work fewer hours and spend their time in teaching and patient care rather than in research1,6,10,18,19,20; (3) women have inadequate mentorship and lack female role models3,6,7,9,16,21,22,23,24,25,26; (4) women produce less because they shoulder the burden of family responsibilities (although multiple studies have failed to support this hypothesis)2,27,28,29,30; (5) women are demoralized and less productive because of subtle or overt sex discrimination or harassment (or both)21,23,31,32; (6) women have less motivation to achieve than men,11,25 set lower career goals, and devalue the "pure research" thought necessary for traditional academic advancement; and (7) women have historically been underrepresented in entry-level positions, but with the increasing numbers of women in such positions, sex differences in the senior ranks will diminish.33
Studies of sex differences in academic advancement have been compromised by small samples, failure to include both sexes, samples of trainees and house staff rather than senior faculty, use of measures with unknown psychometric properties, or consideration of only a few variables or institutions. Few studies of academia in general,2,12,20 or of academic medicine1,11,27,28 have attempted to explore several hypotheses simultaneously in a sample of men and women. No study has assessed the collective contribution of all the hypotheses listed above to academic achievement.
To address these conceptual and methodologic issues, we conducted a survey of all academic pediatricians on the faculties of all departments of pediatrics in the United States. We present here the results of the study, known as the Career Advancement in Academic Pediatrics (CAAP) Project.
Methods
Study Design and Sample
The CAAP study was a cross-sectional survey of all physicians who were salaried faculty members of the 126 pediatric departments in the United States from November 1991 through January 1992. Of the 6441 self-administered questionnaires distributed by the offices of the department chairs, 4285 (67 percent) were returned; 1512 of the respondents (35.3 percent) were women, and 2773 (64.7 percent) were men. The proportions of male and female nonrespondents were similar. For this study, we restricted the sample to the 3999 respondents who were full-time faculty members. The mean (±SD) age of the sample was 44.3±9.01 years; 39.0 percent of the respondents were 40 years of age or younger, and 7.9 percent were 60 or older. Fifty percent of the sample had graduated from medical school after 1975. The majority of the respondents (88.3 percent) were white, 1.7 percent were black, 2.8 percent were Hispanic, 6.3 percent were Asian, and 0.9 percent were members of other races or ethnic groups.
A total of 25.9 percent of the respondents were full professors (82.9 percent were male and 17.1 percent were female), 27.4 percent were associate professors (70.6 percent male and 29.4 percent female), 40.2 percent were assistant professors (54.4 percent male and 45.6 percent female), and 6.6 percent were instructors (42.4 percent male and 57.6 percent female). With respect to geographic region, 33.2 percent of the respondents came from the Northeast, 25.1 percent from the Midwest or north-central region, 14.7 percent from the West, and 27.0 percent from the South. There were consistent parallels between the profiles for pediatrics of the Faculty Roster System of the Association of American Medical Colleges34 and the survey respondents in terms of age, sex, race or ethnic group, rank, and region. We therefore concluded that any nonresponse bias was negligible.
To simplify the analyses, we divided the respondents into three groups according to specialty: neonatology, intensive care, or cardiology (28.6 percent), all other subspecialties (44.4 percent), and general pediatrics (28.8 percent).
Measures
We developed multi-item scales for each of the following variables: institutional support of research, distribution of work time, family responsibilities, and career attitudes and perceived barriers to advancement. Multi-item scales were created with the use of cluster scoring. Reliability coefficients for internal consistency were computed for these scales, all of which met or exceeded the standards for group comparisons (Cronbach's alpha, >0.70). Composite measures were transformed into scales ranging from 0 to 100 for ease of interpretation. We calculated the ratio of time spent in teaching and patient care to time spent in research, adjusted for the total number of hours worked each week. We also measured gratification from teaching, patient care, research, and administration, using five-point scales ranging from "extremely unrewarding" to "extremely rewarding." Because of multicollinearity, we created four summary scales of career attitudes and barriers. We used these summary scales in multivariate models.
We asked 10 questions about academic productivity, including the numbers of unsolicited articles for which the respondent was the first, last, or only author; solicited articles; book chapters; book reviews; named lectureships; and positions as principal investigator, co-principal investigator, or consultant on grants and contracts. Two productivity measures were derived from principal-components analysis with a varimax (orthogonal) rotation of these 10 items. Standardized composite-factor scores were developed and used in multivariate analyses.
Statistical Analysis
Descriptive statistics were used to assess sex differences for all independent variables and covariates. In separate multiple regression models for academic rank and salary, seven blocks of variables were entered: background and training, academic productivity, distribution of work time, institutional support of research, family responsibilities, career attitudes, and perceived barriers to advancement. Since academic rank was ordinal, both multiple linear regression and discriminant analysis were used. The results of the two analyses were similar; we report the multiple regression results. Because of substantive sex differences in all the modeled independent variables, separate models were developed for men and women for both academic rank and salary. Tolerances and variance-inflation factors were investigated to assess multicollinearity.35 Mallow's Cp coefficients were analyzed for variable selection and overspecification in the multiple regression models.35
Results
Academic Rank, Salary, and Background and Training
At every rank, except that of instructor, there were proportionately more men than women. This disparity was most pronounced in the senior ranks. Unadjusted mean annual salaries reported by male faculty members were substantially higher than those reported by female faculty members at each academic rank (e.g., $128,520 vs. $110,497 among full professors and $71,496 vs. $66,212 among instructors). For both male and female faculty members, salaries were higher in the Northeast than in all the other regions (for men, $134,152 vs. an average of $126,306 in all other regions; and for women, $113,137 vs. an average of $109,375 in all other regions).
The male respondents were older than the female respondents (45.4 vs. 42.1 years, P<0.001), a higher proportion of the men were white (89.6 percent, vs. 85.8 percent of the women; P<0.001), and a higher proportion were married (89.1 percent vs. 72.1 percent, P<0.001). More men than women specialized in intensive care, neonatology, or cardiology (28.8 percent vs. 23.4 percent, P<0.001), and more women than men were general pediatricians (35.9 percent vs. 24.8 percent, P<0.001). A significantly higher proportion of male respondents had completed fellowships and received postgraduate degrees.
Work Time, Institutional Support, Family Responsibilities, and Career Attitudes
Male faculty members reported more hours worked per week than female faculty members (64.4 vs. 60.5 hours, P<0.001). Female respondents reported spending proportionately more time in teaching and patient care than male respondents (Table 1). This difference was most pronounced among instructors and assistant professors, with the men reporting significantly fewer hours per week in teaching and patient care than the women (34.9 vs. 40.1 hours, P<0.001) and more hours in research (20.4 vs. 15.0 hours, P<0.001). Since the men at these ranks reported working more total hours per week than the women (64.7 vs. 58.2 hours, P<0.001), these differences reflected a substantial disparity in the distribution of work time (Figure 1).
|
|
Men took significantly less time off for the birth or adoption of children than did women (approximately 3 vs. 90 days), but men perceived significantly greater delays in their careers because of family responsibilities.
More male than female respondents aspired to the rank of division chief (28 percent vs. 17 percent) or department chair (13 percent vs. 3 percent). The highest rank that half the female faculty members aspired to was that of full professor. Men considered their career aspirations to be more achievable than did women. Men also reported significantly fewer career delays, higher achievement orientation, more commitment to their career paths, and greater satisfaction with the progress of their careers (Table 1). There were no statistically significant differences in the proportions of male and female respondents who found teaching and patient care "very rewarding" or "extremely rewarding"; however, a higher proportion of the men found research very or extremely rewarding (68.6 percent vs. 56.7 percent, P<0.001).
Male faculty members also reported significantly fewer barriers to advancement than female faculty members but were more demoralized about their careers. Women reported more sex and age discrimination, more problems in professional life (such as isolation from colleagues), and more negative interference from senior colleagues.
Academic Productivity
Male faculty members published significantly more articles (as first, last, or only author), book chapters, and book reviews and gave more named lectures than did female faculty members (Table 2). Male respondents reported twice as many positions as principal investigator, co-principal investigator, or consultant on research grants.
|
Variables significantly associated with academic rank were similar for male and female faculty members, with some important exceptions. Among respondents in the Northeast, among those specializing in neonatology, intensive care, or cardiology, and among those who had completed fellowships, men had higher academic positions than women. Academic productivity accounted for the most age-adjusted variation in academic rank for both male and female faculty members (Table 3). Neither the distribution of work time nor the amount of institutional support accounted for the significant variation in academic rank among the men. Male faculty members who perceived few career delays due to family responsibilities and female faculty members who reported adequate institutional support of research were likely to have higher academic positions. Among both men and women, higher overall career satisfaction and more career problems and frustrations were associated with higher academic ranks.
|
Sex Differences among the Most Academically Productive Respondents
We divided the sample into three equal groups on the basis of academic productivity. In the most productive group, there were similarities between male and female faculty members and significant differences between them and their less productive colleagues. As compared with the other two groups, the most productive male and female faculty members were in subspecialties other than neonatology, intensive care, and cardiology, had completed fellowships, spent less time in teaching and patient care than in research, worked more hours per week, reported a higher quality of mentorship, had a much greater proportion of institutional support for their research, reported higher overall career satisfaction, had a higher degree of achievement orientation, and reported greater discrimination and career problems and frustrations. Even in the advanced ranks, the ratio of time spent in teaching to time spent in research and the ratio of time spent in patient care to time spent in research were lower among the highly productive female faculty members (0.90 and 1.24, respectively) than among their less productive female colleagues (1.38 and 2.06, respectively).
In the most productive group, the men were older than the women, reported a higher quality of mentorship, received more institutional support for research, took less time off for the birth or adoption of children, reported fewer delays in their careers due to family responsibilities, expressed higher overall career satisfaction, and reported less discrimination and fewer career problems and frustrations (Table 4). At the rank of a midlevel associate professor, highly productive male faculty members earned an average annual salary of $122,172 in 1992 dollars, as compared with $102,189 for highly productive female faculty members (P<0.001).
|
To determine whether male and female faculty members with the same characteristics differed in rank and salary, we generated multivariate estimates of the expected mean salaries and ranks, using background and training characteristics as well as the variables listed in Table 1 and Table 2. The probabilities of achieving the rank of full professor, along with 95 percent confidence intervals, are shown in Figure 2A. The probabilities did not differ for male and female faculty members with identical characteristics.
|
Discussion
The proportion of women in the senior academic ranks has not changed appreciably in the past 10 years.2,8,10 To date, the hypothesized trend toward equalization in the representation of women at senior faculty levels as a result of the increasing numbers of women in training and entry-level faculty positions8 has not materialized. We conducted a multivariate analysis of factors that have been hypothesized to contribute to sex differences in academic achievement. For virtually all the independent variables and covariates that we examined, there were statistically significant differences favoring male faculty members. Of all the variables examined, the only finding favoring female faculty was less career demoralization. With some notable exceptions, however, the analyses produced remarkably similar relationships for both male and female faculty.
Although age was the variable most closely correlated with both rank and salary, since this was a cross-sectional study, it was not possible to estimate the influence of time spent at each rank on academic advancement. Some people have suggested that lifting the time restrictions on promotions will eventually resolve sex differences in academic rank. Our data, however, suggest that such differences may reflect a complex process involving productivity and differential institutional support for and use of work time by male and female faculty members. If so, lifting time constraints on promotions may not resolve existing sex differences.
Female faculty members were less productive at all ranks than their male counterparts. Women accounted for less than 20 percent of the highly productive respondents, and these women did not differ from the highly productive men in terms of hours worked per week, ratio of time spent in teaching and patient care to time spent in research, institutional support for research, or achievement orientation. However, the highly productive women reported a poorer quality of mentorship, less-adequate institutional support of their research, more time taken off for the birth or adoption of children, and less overall career satisfaction than their male counterparts. The most productive female faculty members worked an average of six hours per week longer and spent proportionately more of their time in research than their less productive female colleagues.
According to our study and others, female faculty members, particularly those in the early phases of their careers, spend more time in teaching and patient care than their male colleagues.1,6,10,18,19,20 Carr et al. reported that women preferred to spend more time in teaching but did not differ from their male colleagues in their preferences for how they spent the remainder of their time.1 Although female faculty members found research less rewarding than male faculty members, this variable was not significantly related to productivity, rank, or salary.
Although even the most productive female faculty members reported a lower quality of mentorship, which is often cited as a barrier to academic advancement among women,3,6,7,9,16,21,22,23,24,25,26 we found no empirical support for this hypothesis. Differences in mentorship may operate more subtly to influence such factors as the distribution of work time and the allocation of institutional support for research. These subtle effects of mentorship, "protected time," and institutional support of research on academic advancement require further research.
The hypothesized inverse relation between family responsibilities and academic productivity,2,27,28,29,30 although a popular explanation for sex differences in academic advancement,16 was not supported by our findings or those of other studies.36 Paradoxically, a larger proportion of men than women in our study reported career delays because of family responsibilities. It may be that women learn to accommodate family responsibilities in the context of their professional and other social roles.37,38 Men may perceive the burden of these responsibilities as more directly in conflict with the demands of their professional roles. Whatever the underlying sociological differences, family responsibilities do not appear to account for sex differences in academic advancement in pediatrics.
Career attitudes, such as achievement orientation, have been hypothesized to explain sex differences in academic advancement,11,25 but such attitudes did not explain sex differences in rank or salary in our study.
Finally, we found that after adjustment for all observed sex differences in academic productivity, distribution of work time, institutional support for research, family responsibilities, and career attitudes, male and female faculty members attained similar academic ranks, but women did not receive similar salaries. These findings support those of a recent study of 5057 faculty members at 310 universities, suggesting that even in fields where women are more prevalent than they are in medicine, female faculty members are paid less than their male colleagues.17 These findings suggest a pervasive prejudice against women, despite similar scholarship, that is manifested throughout academia and is not unique to medicine.
The results of our study have three implications for policy making. First, time spent in teaching and patient care has a differentially negative effect on the academic productivity of female faculty members. Second, institutional support of research, in terms of protected time and access to research space, research assistants, and support staff, is substantially lower for women than for men. Finally, although work-site child care and other such arrangements have a high social value, these endeavors alone are unlikely to diminish the differences in academic advancement between male and female physicians.
Because our findings are based on associations between variables, we cannot establish causal relations. Longitudinal studies of the careers of physicians in pediatrics and other academic specialties, with the use of comprehensive measures to test multiple hypotheses, are needed. Our data do suggest, however, that greater representation of women in medical training and entry-level faculty positions will not eliminate sex differences in academic advancement over time. Female faculty members will remain at greater risk than their male colleagues for failing to meet their career aspirations and will continue to receive lower salaries. Further institutional efforts are required to eliminate the differences in rank and salary between women and men in academic medicine.
Supported by grants from the Association of Pediatric Department Chairmen, New England Medical Center, the Mead Johnson Company, the National Institute of Child Health and Human Development, and the Bureau of Maternal and Child Health.
We are indebted to Drs. Arlene Ash, Thomas Boat, Evan Charney, Russell Chesney, Sheldon Greenfield, Alan Gruskin, Sol Levine, Elizabeth MacAnarney, Frank Oski, Robert Petersdorf, George Ray, and Alvin Tarlov for their comments on the manuscript; to Mrs. Heather Kelly for her statistical and analytic efforts; and to Ms. Linda Jackson for her efforts in manuscript editing and preparation.
Source Information
From the Primary Care Outcomes Research Institute (S.H.K., L.M.S., K.A.D.) and the Department of Pediatrics, Floating Hospital for Children (J.G.S.), New England Medical Center, Boston; the College of Medicine, University of Vermont, Burlington (C.F.P.); and the College of Medicine, University of Iowa, Iowa City (R.P.K.).
Address reprint requests to Dr. Kaplan at the Primary Care Outcomes Research Institute, New England Medical Center, 750 Washington St., Box 345, Boston, MA 02111.
References
| |||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
