Child care responsibilities may not fully measure the effect of family on the careers of female faculty. Because women have traditionally been the caregivers in our society, we evaluated the extent of other dependent care responsibilities, such as elder care, that might negatively and disproportionately affect women's academic careers.

We surveyed female and male faculty in all academic departments at 24 randomly selected medical schools across the United States. We asked detailed questions about the nature of child care and other dependent responsibilities as well as aspirations, goals, rate of progress, academic productivity, and career satisfaction to determine the association between family responsibilities and these features of an academic career.

Methods

Top Methods Results Discussion Author & Article Info References

In 1995, we used a two-stage process to select a stratified random sample of full-time salaried faculty of U.S. medical schools. In the first stage, we selected the 24 medical schools from which faculty would be sampled. Of the 126 medical schools listed by the Association of American Medical Colleges (AAMC), we excluded the 6 schools outside of the contiguous United States because the AAMC considers them to be significantly different from mainland schools. To obtain reasonable numbers of female and minority faculty within each institution, we excluded 14 schools that had fewer than 200 total faculty, fewer than 50 female faculty, or fewer than 10 minority faculty. We randomly selected 24 schools from the remaining 106 eligible medical schools. Faculty size at the 24 sampled schools was similar to that in the 106 eligible schools. This sampling strategy resulted in a balanced regional diversity among the four AAMC regions of the United States and in an even division of public and private institutions.

In the second stage of the process, we selected full-time salaried faculty from each sampled school. We used the AAMC Faculty Roster System to identify approximately 4000 faculty stratified on sex (female or male), race or ethnicity (under-represented minority or other), and medical specialty (basic science, surgical specialties, medical and other specialties, or generalist medicine) within three graduation cohorts. These cohorts were defined by years of experience: those who completed their doctoral degree (MD or PhD) before 1970, those who completed it between 1970 and 1980, and those who completed it after 1980.

The AAMC listed 17 434 faculty at the 24 selected schools. Exclusion of 720 faculty who were in unique departments that did not exist at other medical schools left 16 714 faculty, of which 4156 were women. Within each school, we sampled faculty by using a 4 x 3 x 2 factorial design with equal numbers of faculty in the four academic department groupings, the three graduation cohorts, and the two sexes. Within each cell (school x department grouping x graduation cohort x sex), we sought six faculty. The most senior graduation cohort cells were filled first and were back filled, if necessary, from the more junior levels. In addition, we included all minority and generalist faculty at these 24 schools, all women in surgical specialties that had very few women, and all women who had had more than 15 years of experience since receiving their doctoral degree. The balanced factorial design yielded fewer women than men (because not all cells contained at least 6 women), as did the 100% minority and generalist sampling. This imbalance was offset by the addition of senior and surgical female faculty, resulting in a sample with nearly equal numbers of men and women.

We mailed surveys to the faculty's professional addresses. One thousand seventy-three persons were ineligible because they had left their institutions, were not full-time faculty, or had died. Nonrespondents among the remaining 3332 persons were reminded by postcard to participate and, if necessary, received a follow-up telephone call and mailing.

The self-administered survey asked 177 questions about participants' demographic characteristics; professional goals; work situation; current academic environment and rank; mentoring relationships; experiences of bias, discrimination, and harassment; academic productivity; family responsibilities; faculty compensation; and career satisfaction. Thirty questions used in the survey are the property of Drs. Linda Fried and Clair Francomano and were developed at the Johns Hopkins Medical Institutions. They were used in their original form or were modified with permission.

Measures and Variable Definitions

Career progress was self-assessed by participants on a five-point Likert scale on which 1 indicates slower progress and 5 indicates faster progress. We developed a career satisfaction scale that measured two dimensions: satisfaction with work environment and satisfaction with academic advancement (Cronbach = 0.87). This instrument consisted of McGlynn's four-item scale [7] and three additional items. Faculty responsibility for child care ("What is your degree of responsibility for such childcare tasks as caring for sick children, car pooling or meal preparation for children, compared to your spouse/partner/other relative/nanny/hired help?") was measured on a five-point Likert scale on which 1 indicates no responsibility and 5 indicates sole responsibility. The effect of child care responsibilities on the ability to work on weekends ("On weekends, do your childcare responsibilities impact on your ability to do professional work?") was also measured on a five-point Likert scale on which 1 indicates "not at all" and 5 indicates "considerably." Aspiration to become a department chair or a full professor ("To what degree do you aspire to being a 1] chair of a department, 2] full professor?") was assessed on a five-point Likert scale on which 5 indicates great aspirations to these positions. The relative importance of nonprofessional goals ("My nonprofessional goals are as important to me as my professional goals") was assessed on a five-point Likert scale on which 1 indicates "strongly disagree" and 5 indicates "strongly agree."

Departments were coded as primary care (general internal medicine, family medicine, pediatrics, and geriatrics); medical and other specialties (internal medicine subspecialties, pediatric subspecialties, neurology, physical medicine, radiology, emergency medicine, anesthesia, and psychiatry); surgical specialties (general surgery and its subspecialties, including obstetrics and gynecology); and basic sciences. Race or ethnicity was defined as under-represented minority or other. The number of publications is the total number of articles published in refereed journals.

Statistical Analysis

In Table 1, simple frequency distributions and means ±SD are used to describe male and female survey respondents. In Table 2, Table 3, Table 4, and Table 5, we model outcomes separately for faculty with and those without children. In producing these tables, we used multivariate models to estimate all relations between sex and outcomes. The following independent variables appear in all models: medical school, specialty (basic science, surgical specialties, medical and other specialties, and generalist medicine), race (under-represented minority or other), seniority (years since first faculty appointment), seniority squared, age, and marital status (married or partnered compared with single). For each outcome, we report the adjusted means for men and women (using the least mean squares method) and the 95% CI and P value for the adjusted mean difference, which is the main effect of "female" in a model that uses all of the above variables as predictors. For professional outcomes (Table 4), we also include hours worked per week, total years of training, percentage of time spent on clinical work, and percentage of time spent on research. Finally, to help judge whether differences between the sexes differ for faculty with and those without children, we report P values for the interaction of parental status and sex in a model developed for all faculty.

To simplify reporting, all numbers in the tables refer to ordinary least-squares regressions, regardless of the form of the dependent variable. Findings did not differ (and thus are not separately reported) when we used logistic regression to predict dichotomous outcomes or when we analyzed the skewed, non-negative variable "number of publications" (n) by using ordinary least-squares regression to predict the natural logarithm of "n + 1." We used SAS Statistical Software, version 6.11 (Cary, North Carolina), for all analyses.

Role of Funding Source

The Robert Wood Johnson Foundation funded the study but had no role in its design, conduct, or reporting.

Results

Top Methods Results Discussion Author & Article Info References

Respondent Characteristics

The overall number of respondents was 1979, and response rates were similar, approximately 60%, for male and female faculty. Among respondents, women were slightly younger than men and were less likely to be married or to have children (Table 1). Even among faculty with children, women had fewer children. With regard to specialty, the similar distributions of female and male respondents by specialty reflects the balance built into the sampling design. Women with children averaged 5 fewer hours of professional work per week than other faculty, including men with children, men without children, and women without children (54.0 hours compared with 59.4 hours, 59.0 hours, and 58.5 hours, respectively). The allocation of time to administration, patient care, and research was similar for men and women, but women spent slightly more time teaching than men did. Women were less often full professors and were more often instructors and assistant professors than men were.

Dependent Responsibilities

Female faculty averaged more hours of dependent responsibilities per week than men (11.2 compared with 7.5 hours; P < 0.001), but most of these hours (approximately 95%) for both women and men were devoted to child care. Women with children spent more waking weekday hours on child care responsibilities and were less able to work on weekends because of such responsibilities (Table 2). Women, with or without children, were about twice as likely as men to be responsible for other dependents (elderly parents and other relatives not necessarily living at home). However, the time spent by faculty who had such responsibilities was similar for men and women. Work-related travel was more limited by family responsibilities for women with children than for men with children, but it was not a problem for men or women without children.

Aspirations and Goals

Having children had little effect on faculty aspirations and goals (Table 3). Faculty, both with and without children, aspired less strongly to chairing departments than to becoming full professors, and they generally agreed that nonprofessional goals were as important as professional goals. However, among faculty with children, men were significantly more professionally ambitious than women. Among faculty without children, differences in aspirations between the sexes were smaller: The sexes equally aspired to full professorships and were similar in their interest in chairing a department.

Institutional Support and Professional Outcomes

Women with children were less likely than men with children to receive research funding from their institutions (Table 4). Such support was similar, however, for women and men without children. Compared with men who had children, women who had children had less secretarial support. However, the level of secretarial support for male and female faculty without children was similar. Access to research assistant time for women, regardless of parental status, was similar to that of men.

The likelihood of having a research grant and the mean number of grants obtained within the previous 2 years were similar for male and female faculty regardless of parental status. Women with children, however, had fewer peer-reviewed publications than men with children; among faculty without children, men and women were again similar.

Career progress was rated as slower by women than by men, although this difference was less marked for faculty without children. Women with children rated their career satisfaction lower than did men with children. Men and women faculty without children were similarly satisfied with their careers.

Obstacles to Career Advancement

Meetings held before 8:00 a.m., after 5:00 p.m., or on weekends were more frequently perceived as a problem by women with children than by any other group (Table 5). Women, regardless of parental status, were more likely to be concerned about the absence of a part-time tenure track at their institution. Lack of on-site childcare, emergency child care, parental leave, and other family leave policies were more problematic for women faculty with children than for other faculty.

Discussion

Top Methods Results Discussion Author & Article Info References

In this study of academic faculty across the United States, women with children published less, had slower self-perceived career progress, and were less satisfied with their careers than were men with children. Parental status is a significant factor in these lower outcomes; few, and smaller, differences between the sexes were seen in these measures for faculty without children. Sex differences among faculty with children, however, are substantial.

Modest sex differences were seen in the aspirations and goals of faculty with children. Overall, however, these differences were surprisingly small. Compared with men, women in academic medicine had greater dependent responsibilities but had similar allocations of time to professional responsibilities and had greater obstacles to an academic career. Meetings after hours, lack of child care, lack of parental and other family leave policies, and lack of part-time tenure tracks were significantly more burdensome for female faculty with children.

Women faculty with children had less institutional support than men with children. Whether their support was similar to that of men when they started but declined because of lower academic productivity or whether their lower productivity resulted in part from less support is unknown. In a group less able to expand working hours because of dependent responsibilities, however, institutional support may be especially critical for maintaining productivity.

Among faculty with children, women had fewer peer-reviewed publications than men, even after controlling for variables such as years as a faculty member, number of hours worked per week, and hours of dependent responsibilities. Most studies have reported fewer publications for women [4-6, 8-14], but this is not a universal finding [15]. Several reasons have been put forth to explain the difference between the sexes in publications, including women's lack of interest in research [4, 8, 9], lower career motivation and aspirations [9, 12], and general demoralization due to subtle or overt sex discrimination or harassment [16-18]. Kaplan and colleagues, however, dispute this last hypothesis; they found higher morale among female faculty than among male faculty [4].

Although some studies have found that family responsibilities do not account for differences between the sexes in academic careers [4, 6], others have found these responsibilities to be significant [1-3, 19]. We focus on the outcome of publications as a quantifiable measure of productivity that is relied on heavily in promotion and salary decisions. In our previous work, we found that parental status had no effect on publication [16]. When we explored this further, however, in our present study, we found that parental status affects men and women differently. Family responsibilities consume more time for female faculty, and women are less able than men to expand their working hours. We also found evidence of less institutional support for female faculty with children. Previous studies have also suggested sex-related differences in institutional support, including personnel, space, and equipment [5, 8, 20]. Kaplan and colleagues [4] found that even the most productive female faculty had less institutional support for research than her male counterparts did. These studies did not evaluate degree of support according to parental status.

Our study systematically examined the relation between family responsibilities and academic productivity and other aspects of an academic career among faculty in all departments at a variety of institutions across the United States. Previous work in this area has been limited to medical school faculty in one department or one institution. Our data are derived from a representative sample of medical schools in the United States, and we were able to control for many of the factors that might confound sex-related studies, including years as a faculty member, hours of work per week, institution, and marital and parental status. Our study also looks at dependent responsibilities other than child care. Elder care and other dependent responsibilities affect a minority of faculty and consume far less time than child care responsibilities.

Our work has limitations, the most important of which is that it is cross-sectional, not longitudinal. Because faculty who left academic medicine were not sampled, our findings probably understate the effect of family responsibilities on academic careers. Only a longitudinal study could fully identify these effects. A follow-up study that carefully distinguishes among factors for parents of children of different ages is needed to help identify policies that could alleviate some of the problems faced by faculty with children.

Because of the confidentiality concerns of the AAMC, we could not directly compare respondent and nonrespondent faculty. However, because questions about family responsibilities were part of a larger survey of the total experience of academic faculty, response bias is probably less pronounced than it would be in a more narrowly focused study. Our sampling design over-represents senior women and women in the surgical specialties, but we controlled for this in the multivariate analyses. Finally, our study is based on a survey instrument, and we do not know whether men and women faced with a similar reality differ systematically in how they respond to such questions. For instance, although the literature documents that family responsibilities affect the careers of women more than the careers of men [1, 3], men in the study by Kaplan and colleagues [4] reported a more negative effect of family responsibilities on their careers than women did.

In our study, family responsibilities go a long way toward explaining an overall difference between the sexes in publication. Other factors (such as networking and faculty responsibilities other than publishing) that affect the academic productivity of faculty and other ways of assessing productivity need further investigation. Women may spend their time in tasks that are less academically rewarded than research and writing. We found that men and women had similar allocations of professional time to clinical work, administration, and research but that women spent slightly more time teaching. Kaplan and colleagues [4] found that female pediatricians spent more time teaching.

Several studies of sex and publications in science have shown that although women have fewer publications, the citation rate of their publications is significantly higher than that of men [13, 21]. If women focus on fewer, more important papers than men do, simply counting numbers of publications may not be the best way to assess productivity. The fact that women with children have fewer publications may also reflect differences between the sexes in networking. Women with young children are less likely to be included in professional networks. In addition, women's networks tend to include fewer superordinates and colleagues from previous institutions [22]. This may also hinder women's academic productivity.

Family responsibilities, especially child rearing, disproportionately affect the careers of female faculty. Compared with men with children, women with children have modestly lower career aspirations, similar job descriptions, greater obstacles to an academic career, less institutional support, slower self-perceived career progress, and lower career satisfaction. Many of these obstacles can be addressed with little monetary cost to academic institutions, but doing this will require directed effort, such as scheduling departmental meetings during routine work hours and making part-time tenure tracks available to faculty. Greater attention to mentoring activities and the creation of an office for women with non-professional as well as career-related support may be important. Medical schools must make greater efforts to remove impediments, must reward teaching and clinical work with greater academic recognition, and must provide increased institutional support for the careers of faculty with children.

Drs. Ash, Friedman, and Moskowitz and Ms. Scaramucci: Boston University, 720 Harrison Avenue, Suite 1102, Boston, MA 02118.

Dr. Barnett: Murray Research Center, Radcliffe College, 10 Garden Street, Cambridge, MA 02138.

Ms. Szalacha: Graduate School of Education, Harvard University, Cambridge, MA 02138.

Dr. Palepu: Room 620-B Burrard Building, Health Research Centre, St. Paul's Hospital, 1081 Burrard Street, Vancouver, British Columbia, Canada V6Z 1Y6.