Penalties and premiums: The impact of gender,marriage,and parenthood on faculty salaries in science,engineering and mathematics (SEM) and non-SEM fields

Abstract

The prevalence of gender wage gaps in academic work is well documented, but patterns of advantage or disadvantage linked to marital, motherhood, and fatherhood statuses have been less explored among college and university faculty. Drawing from a nationally representative sample of faculty in the US, we explore how the combined effects of marriage, children, and gender affect faculty salaries in science, engineering and mathematics (SEM) and non-SEM fields. We examine whether faculty members’ productivity moderates these relationships and whether these effects vary between SEM and non-SEM faculty. Among SEM faculty, we also consider whether placement in specific disciplinary groups affects relationships between gender, marital and parental status, and salary. Our results show stronger support for fatherhood premiums than for consistent motherhood penalties. Although earnings are reduced for women in all fields relative to married fathers, disadvantages for married mothers in SEM disappear when controls for productivity are introduced. In contrast to patterns of motherhood penalties in the labor market overall, single childless women suffer the greatest penalties in pay in both SEM and non-SEM fields. Our results point to complex effects of family statuses on the maintenance of gender wage disparities in SEM and non-SEM disciplines, but married mothers do not emerge as the most disadvantaged group.

Keywords

academic careers fatherhood premium gender wage gap marriage motherhood wage penalty parenthood postsecondary faculty SEM

Among workers in the United States and in many other nations, gender affects the impact of marriage and parental status on salaries. For male workers, marriage typically has no effect, or a beneficial impact, on wages. Fathers often receive what has been termed a fatherhood bonus in wages, in comparison to childless men (Coltrane, 2004; Glauber, 2007, 2008; Hodges and Budig, 2010). Simultaneously, marriage has inconsistent effects on women’s salaries, with different studies showing negative, no, or positive effects (Budig and England, 2001; Budig and Hodges, 2010; Sax et al., 2002). However, among US workers generally, women who are parents experience a motherhood penalty in pay (Budig and England, 2001; Glass, 2004). Budig and England (2001: 213) estimate that this penalty averages approximately 7% annually per child.

The gender-specific effects of marital and parental status on the maintenance of gender wage gaps have been attributed both to supply-side and demand-side factors (Boulis and Jacobs, 2007; Glauber, 2007; Lincoln, 2008). On the supply side, personal choices constrained by gender norms and household arrangements that allocate larger shares of housework and childcare to women diminish women’s availability for market work in comparison to men (Anderson et al., 2003; Blair-Loy, 2003; Gupta, 1999; Sokoloff, 1980). Because women who are married and/or parents typically spend more time doing housework and childcare than their male counterparts (Boulis and Jacobs, 2007), they face stronger supply-side constraints that interfere with job commitment and performance (Blair-Loy, 2003). Such choices, however, are less frequently made by women in professional positions that require years of preparation to enter (Ridgeway, 2011), and motherhood wage penalties are smaller for professional women (Budig and Hodges, 2010).

On the demand side, employers’ gendered expectations about employees’ work and family commitments lead them to devalue women workers, especially mothers (Blair-Loy, 2003; Coltrane, 2004; Correll et al., 2007; Glauber, 2007, 2008; Jacobs and Winslow, 2004; Kennelly, 1999; Kmec, 2011; Ridgeway, 2011). Employers may believe that all women will place personal life commitments above work commitments. For men, marriage and fatherhood do not conflict with images of the ‘ideal worker’, and may actually enhance men’s value as workers if employers assume that married men and fathers have greater financial responsibility and are therefore more serious about work (Correll et al., 2007; Glauber, 2007). Fatherhood bonuses are larger for men in professional jobs than in other types of occupations (Hodges and Budig, 2010). Even when women and men demonstrate similar accomplishments, workplace dynamics in male-dominated jobs typically result in greater visibility of men’s achievements and greater visibility of women’s domestic roles (Blair-Loy, 2003; Kalev, 2009; Ridgeway, 2011; Sokoloff, 1980). Although separable for heuristic purposes, supply-side and demand-side factors are hard to disentangle. Women who press for promotion less assertively may do so because of prior socialization or personal preference (Babcock and Laschever, 2003), but in a work environment where women are rarely promoted, they may regard it as futile effort.

Gender, marital, and parental statuses, and careers in SEM

Gender wage disadvantages also occur in academic work (see, for example, Barbezat, 2002; Bellas, 1994, 1997; Ginther and Kahn, 2004; Porter et al., 2008; Renzulli et al., 2006; Toutkoushian and Bellas, 2003; Toutkoushian et al., 2007; Umbach, 2007; Winslow, 2010). Wage gaps may vary among faculty according to discipline, with technology, engineering, and mathematics fields demonstrating the largest gaps (Umbach, 2007). Such gaps cannot be wholly accounted for by education and training, productivity, or institutional locale (Xie and Shauman, 2003). Patterns of advantage or disadvantage in salaries linked to marital, motherhood, and fatherhood statuses have been less explored among college faculty than among workers generally. When researchers have analyzed the effects of marital status and parenthood on faculty careers, their focus typically has been on publication productivity and/or visibility (Armenti, 2004; Cole and Zuckerman, 1987; Fox, 2005; Hunter and Leahey, 2010; Kennelly and Spalter-Roth, 2007; Long, 1990; Stack, 2004), perceptions of conflict between work and personal life (Amelink and Creamer, 2007; Fox et al., 2011), or career progressions, such as getting jobs, acquiring tenure, advancing to full professor or attaining administrative posts (Hunter and Leahey, 2010; National Research Council, 2010; Winslow, 2010; Wolfinger et al., 2008).

Studies of the impact of personal life on women academics are timely. Recruitment and retention of women into natural and physical sciences, mathematics, and engineering is an important national priority, as women are seriously under-represented in these fields and their participation is critical to meeting projected future personnel needs (National Research Council, 2010; National Science Foundation, 2010). Nevertheless, researchers suggest that impact of personal-life statuses may be greater for women in science-related fields than in other disciplines. Not only do scientific fields have cultures demanding unwavering commitment and loyalty (Fox, 2001; Fox et al., 2011), careers in those fields typically are organized in ways that make interruptions or temporary leaves difficult. The historical movement toward ‘big science’ (de Solla Price, 1986), wherein scientific research relies on large collaborative teams, a steady flow of external funding, and team-based research in technology-intensive labs, may raise demands on scientists to relinquish competing commitments.

Although most science and engineering fields require that researchers seek external funding, disciplines vary in the extent to which they depend on collaborative laboratory work. In highly interdependent collaborations, interruptions of work by key members, or even suspicions that such interruptions might occur, can risk not only the careers of individual scientists, but also the careers of more-junior team members who may rely on them for support. The masculinist cultures in which scientific work is conducted in some fields may be less family friendly and less responsive to the needs of mothers than are the cultures of other fields (Mason and Goulden, 2002; Monosson, 2008; Rosser, 1997; Rosser and Taylor, 2009; Settles et al., 2006; Williams, 2002; see Wolfinger et al., 2008 for a contrasting view). Although these factors all are linked to salaries, prior studies do not directly test the contribution of marital and parental statuses to wage gaps between men and women.

Despite the lack of focus on the effects of family statuses and gender upon salary, there is considerable research on the impact of marriage and parenthood upon other faculty outcomes, particularly among women scientists (Etzkowitz et al., 1994; Long, 2001; Mason and Goulden, 2002; National Research Council, 2010; Stewart et al., 2007). Women in tenure track and tenured positions are less likely than their male counterparts to have children (Perna, 2005), especially at elite institutions and in science-related fields. Women scientists report explicit warnings about career risks associated with childbearing (Rosser, 2005), and women academics fear that motherhood imperils academic careers in traditional science disciplines (Grant et al., 2000; Mason and Ekman, 2007; Mason and Goulden, 2002; Monosson, 2008). Studies of the effects of parenting on academic careers focus primarily on mothers. This research shows inconsistent results, with some studies, such as Cole and Zuckerman (1987) and Leahey (2006), showing that having children has no effect on women’s productivity, while others reveal short-term or long-term negative impacts (Hunter and Leahey, 2010; Kyvik, 1990; Long, 1990; Mason and Goulden, 2002; Wolfinger et al. 2008). Stack (2004) finds less productivity among women scientists with young children, while Hunter and Leahey (2010) report that although having children temporarily reduces productivity, it does not affect the disciplinary visibility (measured as citation counts) of academic mothers compared with academic fathers. Wolfinger et al. (2008) discover that having pre-school age children decreases the odds for academic women to gain tenure track positions, while older children increase their odds of gaining tenure. A frequently cited longitudinal study by Mason and Goulden (2002) posits that academic mothers suffer disadvantages, with cumulative effects as their careers progress. Few institutions offer support for work–family balance, and even when such support is available, academics are loath to take advantage of it, lest they signal lesser commitment to work (Mason and Ekman, 2007; Mayer and Tikka, 2008; Rosser, 2005; Spalter-Roth and VanVooren, 2008).

While many studies show that marital status has no impact on academic productivity (Zuckerman and Cole, 1975), more recent research by Fox (2005) suggests that marital status has more complex effects that depend, in part, upon the order of the marriage (first or subsequent) and the employment status of the spouse. Other studies suggest that faculty members benefit from marriage, regardless of gender, but the marriage premium is more stable and larger for men (Toutkoushian, 1998; Toutkoushian et al., 2007).

Finally, prior research on faculty in a broad range of academic fields lends support to the idea that there are asymmetrical effects of specific combinations of personal-life statuses on salaries and other career outcomes (Fox et al., 2011; Toutkoushian et al., 2007). Long (1990) argues that marriage and parenthood have opposite effects for women academics, with marriage enhancing their careers while parenthood detracts from careers. Cole and Zuckerman (1987) find that neither marriage nor parental status affects the productivity of scientists employed in PhD institutions, whether women or men. Since much of this research focuses on individuals who have survived in academia long enough to attain tenured or tenure-track positions, these patterns may reflect selectivity: women who survive to this point may be unusually talented, motivated, or well supported in managing potential work-family conflicts (see Rayman and Brett, 1995). A recent study by Fox, Fonseca and Bao (2011) found that both women and men who are parents experience moderate family-to-work and work-to-family conflicts, but the magnitude of these conflicts is greater for women than for men, especially for perceived interference of work with family life. That study focused on subjective perceptions of conflicts but did not measure the impact of conflicts on productivity or salary.

Although prior research finds that research productivity is a primary predictor of salary for academic women and men, it also reveals that productivity differences cannot wholly account for gender wage gaps (Bellas, 1994; Renzulli et al., 2006). Productivity differences – a supply-side factor – appear in some, but not all, disciplines and institutions (Renzulli et al., 2006). There is evidence that in science and engineering, women fail to receive equal benefits, especially advancement in rank, for their productivity (Long et al., 1993; Long and Fox, 1995). Moreover, marital and parental premiums and penalties may be inconsistent across disciplines, particularly where women are scarce and work–family policies are weak, or where normative patterns of work make interruptions for personal life difficult. As Fox (2001) observes, careers in science and engineering are hierarchical, with substantial inequality in rewards, and core values such as rationality and order are more closely linked to cultural conceptions of masculinity than of femininity. Women thus may be alienated from cultures of academic science, making their achievement less visible and less well rewarded. Thus, research on how gender affects productivity, career advancement, and rewards such as salary has not produced consistent findings.

In addition to inconsistent findings, studies of the impact of marital and parental status in the sciences, engineering, mathematics and other fields are limited in other important ways. Although some studies include members of many different fields and types of institutions (e.g. Fox et al., 2011), others focus only on single institutions (e.g. Amelink and Creamer, 2007), on scientists in elite institutions only (e.g. Cole and Zuckerman, 1987), or on scientists in a single discipline (e.g. Long, 1990). Many studies took place when women constituted only a small proportion in their fields, and before recruitment and retention of women into academic science and engineering became a national priority (National Research Council, 2010).

Research on scientists and engineers usually does not provide contrasts with faculty in other fields. Many studies cannot disentangle the possibly separate and opposing effects of marital and parental status on academic careers and ignore categories such as single parents. Finally, more studies focus on the impact of personal-life statuses on research productivity or other career outcomes, rather than directly addressing salaries or the gender wage gap. Only a few studies have compared outcomes across specific disciplines within science and engineering. Thus, there is a need for a comprehensive study of the effects of marital and parental statuses on gender wage gaps that compares faculty in scientific, mathematics, and engineering disciplines with those in other fields, and which also considers the potential impact of specific disciplinary groups within SEM on salaries.

Focus of this study

This study explores motherhood penalties and fatherhood premiums for higher-education faculty members in the USA. It compares the impact of gender, marital status, and parental status, in combination, on salaries of faculty in natural and physical science, engineering, and mathematics (SEM) fields with those in non-SEM fields, controlling for institutional locale, professional age, occupancy of an administrative post, academic rank, and productivity. What constitutes a scientific, engineering, or mathematics field has not been consistently defined in prior research. The National Science Foundation includes social scientists in this category, but many researchers do not. In this study we limit SEM to natural and physical scientists, engineers and mathematicians and classify social scientists with non-SEM academics. Women make up a larger share of social scientists in comparison to faculty in SEM fields (West and Curtis, 2006), and salary ranges for women and men in the social sciences are lower and more similar to trends in the humanities. Academic social science research does not always require the collaborative teams and technology-intensive labs prevalent in many of the sciences. Furthermore, masculinist cultures tend be more prevalent in the natural, and physical sciences and in engineering (see for example Fox, 2001; National Research Council, 2010).

We base our analyses on the 2004 release of the National Study of Postsecondary Faculty (NSOPF:04), a nationally representative sample of US college and university faculty in all academic fields (National Center for Education Statistics, 2011). These data are more comprehensive and representative of US college and university faculty than data used in much prior research.

Research strategy

Our first task is to explore whether the combined effects of gender, marital status, and parental status on salaries reported in previous studies with more limited samples appear in the NSOPF:04 data. We first address the following question:

1. How do the combined effects of gender, marital status and parental status affect faculty salaries in SEM and non-SEM disciplines?

To determine whether differences in gender, marital status and parental status appear after controlling for measures of scholarly productivity, we then address the following question:

2. How does productivity change the effects of gender, marital status, and parental status on salary in SEM and non-SEM disciplines?

If structural aspects of work differ across SEM fields, it is possible that combined effects of gender, marital status and parental status also will differ across subfields. To test for this possibility, we address question three:

3. How does location in specific sets of SEM disciplines affect the impact of gender, marital status, and parental status on salary, after controlling for productivity?

If there are combined effects of gender, marital status and parental status on faculty pay, the patterns may not be parallel across SEM and non-SEM fields. To test for this possibility, we address question four:

4. How do the combined effects of gender, marital status, and parental status on salary differ between faculty in SEM disciplines and non-SEM disciplines?

To determine whether these measures having differing effects on SEM versus non-SEM faculty, we compare coefficients using z-test scores across models to determine whether SEM and non-SEM faculty experience significant differences in the determinants of salary outcomes.

Data

As noted earlier, we use data from the 2003–2004 wave of NSOPF:04, a nationally representative dataset of faculty in the US conducted at approximately 5-year intervals by the National Center for Education Statistics. NSOPF surveys represent the most comprehensive data source for higher education faculty in the USA, covering 920 public and private, not-for-profit postsecondary institutions, including doctoral, masters, baccalaureate, associate-degree, and special focus institutions. It includes information on research accomplishments, teaching workloads, marital and parental status, discipline, rank, degrees and salary for 26,600 individual faculty members (National Center for Education Statistics, 2011).¹ We analyze the effects of parental and marital status upon salary of full-time, tenured, and tenure-track faculty within SEM and non-SEM academic disciplines located in either doctoral-degree granting institutions or institutions that primarily offer baccalaureate or master’s degrees.^2,3

The final sample contains 8350 individuals representing 490 institutions, representing a minimum of one individual per institution and a maximum of 100, with an average of 17 per institution. Of this sample, 4900 faculty work at doctoral institutions and 3440 at masters/bachelors institutions. SEM disciplines account for 2550 faculty, of whom 480 (19%) are women. Non-SEM disciplines account for 5800 faculty, 2400 (41%) of whom are women. A total of 50 institutions contain non-SEM faculty exclusively, while another 10 contain SEM faculty only. The remaining 430 institutions include both SEM and non-SEM faculty. See Tables 1 –4 for additional descriptive statistics.

Table 1.

Descriptive data

	Mean	Standard deviation	Minimum	Maximum
Dependent variable
Total institutional income	77568.44	37131.90	40.00	300000.00
Log of total institutional income	11.16	0.46	3.69	12.61
Controls
Associate professor	0.31	0.46	0	1
Full professor	0.40	0.49	0	1
Professional age	18.14	11.33	1	50
Administrator status	0.10	0.30	0	1
Doctoral institution	0.59	0.49	0	1
Masters or Bachelors institution	0.41	0.49	0	1
Marital and parental status by gender
Married man no kids	0.22	0.41	0	1
Married woman no kids	0.11	0.31	0	1
Married father	0.34	0.47	0	1
Married mother	0.12	0.33	0	1
Single father	0.02	0.15	0	1
Single mother	0.03	0.16	0	1
Single man without children	0.07	0.26	0	1
Single woman without children	0.09	0.28	0	1
Productivity
Total hours worked per week	51.18	11.69	1	100
Number of classes taught	2.33	1.61	0	20
Any funded scholarly activity	0.43	0.50	0	1
Number of career refereed articles	21.98	34.25	0	200
Number of career patents	0.40	1.66	0	15
Number of career reviews	5.80	11.42	0	80
Number of career presentations	42.12	61.72	0	500
Discipline
Non-SEM discipline	0.70	0.46	0	1
Engineering	0.05	0.22	0	1
Biological Sciences	0.09	0.28	0	1
Computer Science	0.02	0.14	0	1
Physics and Astronomy	0.03	0.16	0	1
Geological, Atmospheric and Environmental Sciences	0.03	0.16	0	1
Mathematics/Statistics	0.04	0.21	0	1
Chemistry	0.03	0.17	0	1
Agriculture and Related Sciences	0.02	0.13	0	1

N=8350.

Table 2.

Women as a percentage of total faculty in SEM and non-SEM disciplines

Discipline	%	Total N
All SEM fields	19	480 of 2550 faculty
Engineering	9	40 of 440 faculty
Biological Sciences	27	200 of 730 faculty
Computer Science	18	30 of 170 faculty
Physics and Astronomy	10	20 of 210 faculty
Geological, Atmospheric and Environmental Sciences	19	40 of 210 faculty
Mathematics/Statistics	19	70 of 370 faculty
Chemistry	20	50 of 250 faculty
Agricultural Sciences	20	30 of 150 faculty
Non-SEM Disciplines	41	2400 of 5800 faculty

Table 3.

Percentages of SEM and non-SEM faculty by gender holding specific marital status and parental status combinations

Discipline type	SEM (%)		Non-SEM (%)
Gender	Women	Men	Women	Men
Married	70	88	67	84
Parents	47	59	42	54
Single parents	8	3	7	4
Married with children	40	56	35	50
Married without children	30	32	32	34
Single without children	22	8	26	12
	n=480	n=2070	n=2400	n=3400

Table 4.

Average SEM and non-SEM faculty salaries by gender, marital status, and parental status combinations

Discipline type	SEM ($)	Non-SEM ($)
All men	86,214	81,982
All women	71,845	65,037
Married men without children	90,861	83,296
Married women without children	71,346	66,619
Married fathers	84,988	84,385
Married mothers	75,377	66,195
Single fathers	89,493	81,363
Single mothers	65,562	65,787
Single men without children	75,011	68,633
Single women without children	68,331	61,253
	n=2550	n=5800

Values of n are rounded to the nearest 10.

Key measures

Natural log of salary

The dependent variable is the individual’s total annual institutional salary (excluding income from contracts, consulting, royalties, or other income besides salary from the employing academic institution). Salary includes supplemental compensation for overload and summer teaching, grant funding, monetary awards, and administrative duties. We take the natural log of this variable so that changes attributed to the independent variables can be expressed as a proportional change in total base income. We do this, because the effect sizes of the independent variables in absolute dollar amounts are likely to differ across institutions – for example, the salary difference between regular faculty and administrators across all institutions is unlikely to be exactly $6000. It is more likely to be a consistent, roughly equivalent proportional change, such as a 10 percent increase in base salary for promotion to an administrative position. Table 1 provides further information on the measurement of variables.⁴

Marital status × parental status × gender

To examine the impact of marital and parental statuses, as they intersect with gender, we include eight binary variables, each measuring a unique and mutually exclusive set of statuses: married men with no children; married women with no children; married fathers; married mothers; single fathers; single mothers; single men with no children; and single women with no children. Married fathers, the group most likely to receive wage premiums in the US labor market generally, are the contrast category. Both married individuals and those living with a partner or significant other are coded as married, while those who have never married, or are separated, divorced, or widowed are coded as single. These data do not distinguish between heterosexual and same-sex couples, but no respondents are excluded based on sexual orientation or legal marital status. Respondents living with dependent children are coded as parents, while faculty with no children or no dependent children are coded as childless. Although it would be worthwhile to examine differences between parents with dependent children, parents with adult children, and those who have never had children, the NSOPF:04 measures do not permit such fine-grained analyses. Our measures are limited to marital and parental status, as NSOPF:04 contains no measures of parenting behaviors, such as time allocated to childcare or specific childcare activities performed.

Using sets of dummy variables in this way is comparable to running three-way interactions for marital status, parental status, and gender, but the results are easier to interpret (see also see Renzulli et al., 2011 and Reynolds, 2004, for other studies that use this approach). It also allows us to detect nuanced differences that occur at the intersections of family and gender.

Scholarly productivity

To distinguish the effects of scholarly productivity from the effects of marital and parental status, we incorporate several measures of productivity, including total hours worked per week, the number of classes taught per semester, presence of an external funding award, number of career refereed articles published, number of career patents, number of career reviews (including book reviews and edited book chapters), and number of career presentations.⁵ All productivity variables are coded as cumulative count variables, with the exception of funding, which is coded 1 if the individual has current grant funding and 0 if not.

Discipline

The NSOPF:04 survey identifies faculty by primary teaching discipline. Any faculty members who are not currently teaching are classified by their primary research discipline. We divide their disciplines into one of two categories: SEM and non-SEM. We further subdivide the SEM category into specific discipline groupings by using eight dummy- coded variables: engineering; biological sciences; computer sciences; physics and astronomy; geological, atmospheric and environmental sciences; mathematics and statistics; chemistry; and agricultural and related sciences.⁶ Engineering is the contrast category. The SEM and non-SEM categories and the eight subdisciplinary groups are mutually exclusive. These groupings are common across colleges and universities, and there are similarities in research activities within each grouping. For SEM faculty, there is a minimum of 150 members per group and a maximum of 730, with an average per group of 320. For more information on these variables, see Table 2.

Control variables

We control for professional age, measured as the time since each academic began his or her first postsecondary position, rounded to the nearest full year. We also control for administrative appointments, defined as holding positions such as department chair, associate dean, dean, or higher-level administrators. Administrators are coded 1, and non-administrators are coded 0 (the contrast category). Prior research demonstrates gaps in salaries between faculty employed in research-intensive universities, typically those offering doctoral degrees in their fields, and other types of institutions. Our preliminary analyses with this dataset reveal that salaries of full-time, tenure track faculty do not differ significantly between master’s and bachelor’s institutions, but that location in a doctoral-granting institution increases the average salary for all categories of faculty. We therefore differentiate between doctoral institutions (coded 1) and masters/bachelors institutions (coded 0 and the contrast category).

There is considerable disagreement in the research literature about whether academic rank should be included in the estimation of gender equity in faculty salaries. Although salary is logically linked to rank, many researchers argue that including rank as a predictor may create a downward bias in estimates of gender inequality. If promotions for women faculty members are delayed in comparison with those for similarly accomplished men, including rank may obscure the effects of other variables that are linked to salary, for example a woman faculty member at the ranks of associate professor may have a salary roughly equal to that of her men counterparts, who are at the associate rank, but lower than those of similarly accomplished men who are full professors. Research suggests that gender-linked promotion bias may affect promotion rates of scientists (Long et al., 1993). Given equal productivity levels, men are more likely to be promoted to full professor, while women remain in the associate professor rank. The salaries of these women may be high for associate professors but lower than those for full professors with similar records. Rank, therefore, can be a confounding factor if there is gender bias in the time to promotion from associate to full professor (Barrett and Sansonetti, 1988; Boudreau et al., 1997). However, other researchers argue that progression through the typical faculty ranks in the USA of assistant, associate, and full professor is a lockstep process, with little variation, and that rank is thus a key predictor of salary (see Becker and Toutkoushian (2003) for a review). According to these researchers, excluding measures of rank leads to omitted variable bias. We acknowledge this debate and provide models both with and without measures of rank.⁷

In the analyses in which we include rank, we use two dummy variables to indicate whether an individual in our sample is a full or associate professor (with assistant the contrast category). Inclusion of rank made a substantive difference only in the analyses for SEM faculty. Thus, we report results of both equations (with and without rank) for SEM faculty. Since the inclusion or exclusion of rank made no substantive difference in results for non-SEM faculty, we report only the equations estimated without rank for them.⁸

Analytic strategy

Because we anticipated differential effects of marital status, parental status and gender for SEM and non-SEM faculty salaries, we ran separate regression models for the two groups. Our cases are not independent, because multiple faculty members in our sample work at the same institution; thus we used the cluster command in Stata with robust standard errors to correct for this non-independence without sacrificing degrees of freedom.⁹ Values of the dependent variable are independent between groups of faculty at the same institution, but within each faculty–institution group these values are non-independent. The Stata cluster command allows us to specify which institution each faculty member is nested within and corrects for any correlation resulting from this non-independence. Our first goal is to determine salary differences linked to family statuses and gender among SEM and non-SEM faculty. To answer our first research question, our initial models for each discipline group examine the direct, combined effects of marital status, parental status, and gender on logged salary.

The next set of models for each discipline category includes the research and productivity variables, in order to see if those variables affect the relationship between gender, marital status and parental status, and faculty salaries. These models thus address our second research question. The third set of models adds specific discipline groupings to the SEM analyses to address our third question about whether the effects of marital status and parental status for women and men SEM faculty are similar or dissimilar across disciplinary groups within SEM.¹⁰

The last set of models analyze the issue of rank by including measures for associate and full professors (assistant professors are the reference category), controlling for family status and productivity, for the SEM faculty, and their disciplinary group within SEM. Here we explore whether rank exerts an independent influence upon faculty salaries after we have accounted for the effects of other relevant predictors.

Our fourth research question asks whether salary penalties and premiums attached to specific marital statuses and parental statuses for women and for men are consistent across SEM and non-SEM fields. Given the normal career patterns in different disciplines, we anticipate that marital and parental statuses might have differential impact on SEM and non-SEM faculty. We therefore examine the coefficients in our separate discipline category models to see if this is the case. By testing separate models, we are essentially testing an interaction between our independent variables and our discipline categories. We tested coefficients using a z-score test across these two models to determine whether women and men in SEM and non-SEM fields experience significant differences based on marital and/or parental statuses in the determinants of salary outcomes. Models 1 and 4 show the results of these analyses, when only controls, family statuses, and gender are included, while Models 2 and 5 show the results of coefficient tests for models including productivity measures. Coefficients for variables that show statistically significant differences for SEM and non-SEM faculty are highlighted in bold print.

Results

Representation and salaries of women and men

Despite influxes of women faculty into SEM disciplines, they remain a distinctive minority in most of those disciplines (only 19% in our sample), and their placement across fields remains uneven (Rosser and Taylor, 2009). Although women constitute nearly 27 percent of faculty in the biological sciences, they make up only 9 percent in engineering and 10 percent in astronomy and physics (Table 2). Gender is connected systematically to marital and parental statuses (Table 3). In both SEM and non-SEM fields, a larger proportion of men are married and/or parents. Larger proportions of men than women are married with children, though roughly the same proportion of men and women are married without children across both discipline types. Strikingly, across both groups, proportionately more women than men are single, and more women are single parents (Table 3).

Finally, mean salaries for faculty members are related to gender and to location in both SEM and non-SEM disciplines. Overall men in SEM disciplines have average annual salaries of $86,214, while women in those disciplines earn only $71,845. Men in non-SEM fields have average annual salaries of $81,982, while women average $65,037. Even though SEM faculty members earn higher salaries than those in non-SEM fields, women in both groups earn less than the men. In both SEM and non-SEM fields, the average salary for women faculty is roughly 80 percent of the average salary for men faculty, although women in SEM fields are slightly closer to their men counterparts (Table 4). These patterns parallel those for academic scientists reported in prior research (see for example Etzkowitz et al., 1994; Perna, 2005). Mean salaries for women and men are also affected by marital and parental statuses (Table 4).

Patterns of premiums and penalties in SEM and non-SEM disciplinary groups

Our first set of analyses address how combinations of gender, marital status, and parental status affect faculty salaries in SEM and non-SEM disciplines (Table 5, Model 1 and Table 6, Model 5). We find a strong effect of gender on salaries across all combinations of family status. In SEM, married fathers have the highest mean salaries, but earnings of married men without children and single fathers are not significantly different from those of married fathers. However, women in all categories earn less than married fathers (Table 5, Model 1). Furthermore, the gap between their salaries and married fathers is relatively small (5.4 percent) for single men without children, the only category of men in SEM disciplines with salaries significantly lower than those of the reference group of married fathers. The gap for married mothers is also small (2.4 percent). The three remaining women faculty categories in SEM have proportionately larger wage deficits from the reference group of married fathers, ranging from 10.3 percent to 14.4 percent. Thus, men with any type of family responsibilities (spouse and/or children) earn equivalent salaries to the reference group, while all women earn proportionately less, regardless of family statuses.

Table 5.

Income of women and men SEM faculty by family status, controlling for professional age, institution type, administrator status, and rank

	Model 1		Model 2		Model 3		Model 4
	B	Change (%)	B	Change (%)	B	Change (%)	B	Change (%)
Constant	10.84	***	10.79	***	10.89	***	10.84	***
Controls
Associate professor							0.13	13.77***
Full professor							0.27	31.13***
Professional age	0.01	1.01***	0.01	0.78 ***	0.01	0.82***	0.00	0.25**
Administrator status	0.28	32.58 ***	0.24	27.00 ***	0.24	26.62***	0.19	20.56***
Doctoral institution	0.33	39.51 ***	0.16	17.35***	0.15	15.95***	0.15	16.30***
Marital and parental status by gender (married father is the reference category)
Married man without children	–0.01	–1.01	–0.02	–1.50	–0.01	–1.16	0.01	1.19
Married woman without children	–0.11	–10.33***	–0.07	–6.99 **	–0.06	–6.15*	–0.03	–3.05
Married mother	–0.02	–2.36*	–0.01	–0.64	0.00	–0.34	0.00	0.32
Single father	0.00	0.00	–0.01	–1.10	–0.01	–0.71	–0.01	–1.22
Single mother	–0.13	–12.28*	–0.11	–10.60*	–0.11	–10.15*	–0.07	–6.31
Single man without children	–0.06	–5.36 *	–0.01	–1.22	0.00	–0.25	0.02	2.21
Single woman without children	–0.16	–14.44 ***	–0.10	–9.15 ***	–0.08	–7.91**	–0.05	–5.34*
Productivity
Total hours worked per week			0.00	0.17 **	0.00	0.15*	0.00	0.17*
Number of classes taught			–0.03	–2.59***	–0.03	–2.66***	–0.02	–2.44***
Any funded scholarly activity			0.09	8.88***	0.09	9.29***	0.08	8.47***
Number of career refereed articles			0.00	0.24***	0.00	0.23***	0.00	0.18***
Number of career patents			0.01	1.14***	0.01	0.69*	0.01	0.66*
Number of career reviews			0.00	0.23 ***	0.00	0.27***	0.00	0.24***
Number of career presentations			0.00	0.01	0.00	0.02	0.00	0.01
Discipline (engineering is the reference category)
Biological Sciences					–0.09	–8.19	–0.09	–8.26***
Computer Science					0.02	2.18	0.03	3.44
Physics and Astronomy					–0.10	–9.88***	–0.11	–9.97***
Geological, Atmospheric and Environmental					–0.17	–15.97***	–0.18	–16.05***
Mathematics/Statistics					–0.14	–13.06***	–0.14	–12.72***
Chemistry					–0.15	–14.19***	–0.16	–14.44***
Agriculture and Related Sciences					–0.20	–18.21***	–0.20	–17.80***
R²	0.30		0.42		0.44		0.47
F change	84.09		107.28		81.24		83.09
Significance of F change	0.00		0.00		0.00		0.00

Dependent variable: log total institutional income.

N=2540.

***

p>.001; **p>.01; *p>.05.

Coefficients in bold indicate that the magnitude of the effect of these variables is significantly different for SEM versus non-SEM faculty.

Table 6.

Income of women and men non-SEM faculty by family status, controlling for professional age, institutional type, administrator status, and rank

	Model 5		Model 6		Model 7
	B	Change (%)	B	Change (%)	B	Change (%)
Constant	10.88	***	10.79	***	10.77	***
Controls
Associate professor					0.13	13.54***
Full professor					0.33	38.40***
Professional age	0.01	1.05***	0.01	0.91 ***	0.00	0.19**
Administrator status	0.22	24.61 ***	0.16	17.59 ***	0.11	11.74***
Doctoral institution	0.24	27.38 ***	0.14	14.91***	0.15	15.95***
Marital and parental status by gender (married father is the reference category)
Married man without children	–0.06	–5.75 **	–0.05	–5.33 **	–0.04	–3.44*
Married woman without children	–0.18	–16.81 ***	–0.17	–15.55 ***	–0.14	–13.41***
Married mother	–0.15	–13.67 ***	–0.13	–11.75 ***	–0.09	–8.85***
Single father	–0.04	–3.70	–0.04	–4.23	–0.02	–1.75
Single mother	–0.17	–15.38***	–0.15	–14.27***	–0.13	–12.28***
Single man without children	–0.16	–14.96 ***	–0.14	–12.72 ***	–0.11	–10.51***
Single woman without children	–0.24	–20.94 ***	–0.21	–18.86 ***	–0.17	–15.21***
Productivity
Total hours worked per week			0.00	0.30 ***	0.00	0.27***
Number of classes taught			–0.03	–3.02***	–0.03	–2.78***
Any funded scholarly activity			0.07	7.39***	0.07	6.88***
Number of career refereed articles			0.00	0.24***	0.00	0.17***
Number of career patents			0.00	0.44	0.00	0.34
Number of career reviews			0.00	0.04	0.00	–0.03
Number of career presentations			0.00	0.06 ***	0.00	0.05***
R²	0.21		0.27		0.31
F change	119.38		97.74		110.6
Significance of F change	0.00		0.00		0.00

Dependent variable: log total institutional income.

N=5800.

***

p>.001; **p> .01; *p>.05.

Coefficients in bold indicate that the magnitude of the effect of these variables is significantly different for SEM versus non-SEM faculty.

The story changes somewhat for non-SEM faculty (Table 6, Model 5). With the exception of single fathers, all groups earn less than the reference group of married fathers. In contrast to SEM disciplines, married men without children also are disadvantaged. The gap between men and women with similar family statuses diverges sharply. Married mothers earn 13.7 percent less than married fathers. The penalty for married men without children is 5.8 percent, but it becomes 16.8 percent for married women without children. Single fathers earn about the same, while single mothers earn 15.4 percent less, than married fathers. Single men without children earn 15 percent less than the reference group, but single women without children earn nearly 21 percent less (Table 6, Model 5). Overall, while almost all non-SEM faculty earn less than the non-SEM married fathers, women earn proportionately less than men with identical family status.

In these and in the analyses reported below, control variables in the regression models involving SEM and non-SEM faculty operate as anticipated. Professional age, location in a doctoral institution and administrative appointments all contribute significantly to higher salaries. We performed analyses with and without the rank variable for SEM and non-SEM faculty members. Since the change made no difference in substantive results for non-SEM faculty members, we report only the results of the analysis that exclude rank for this group. Since there were some substantive differences in results for SEM faculty, we report results of both analyses for SEM academics.

Independent effects of rank

To consider the independent effects of rank upon faculty salaries, we add measures for associate and full professors to the full models (Table 5, Model 4 and Table 6, Model 7). After controlling for family status, productivity, and controls, we find that rank exerts no substantive influence on non-SEM faculty salaries. Some coefficients are slightly reduced, but there are no changes in significance levels. Including rank in these models may lead to underestimating the effects of gender on salary; thus, it is telling that including rank in the non-SEM model does nothing to diminish the general pattern of penalties faced by all groups of women faculty, and to a lesser extent, by married and single men without children. The change in the proportion of the variance explained is also modest, moving from .27 to .31, suggesting that rank is not an influential basis of salary inequities in non-SEM fields (Table 6, Model 7).

Including rank in the SEM model yields more complicated results. While most patterns remain unchanged with rank included, there are two exceptions. The coefficients for married women without children and single mothers remain negative but become nonsignificant, leaving only childless single women with significantly lower salaries than the reference category of married men with children (Table 5, Model 4). Including rank in the model obscures other gender-based effects tied to family statuses or promotion bias. Since levels of productivity are controlled, the changes occurring with the inclusion of rank are consistent with patterns one would expect if women experienced promotion delays, but unfortunately this question cannot be addressed with the NSOPF:04 data.

Premiums and penalties, with productivity controlled

Our second question addresses whether measures of scholarly productivity moderate the relationship between gender, marital status, parental status, and salary (Table 5, Model 2 and Table 6, Model 6). These analyses control for research and teaching productivity. Measures of research productivity and greater numbers of hours worked per week are generally linked to higher pay, while the number of classes taught is linked to lower earnings. There is some variation between SEM and non-SEM faculty in terms of which research productivity measures have a significant impact on pay. For SEM faculty, all measures of research productivity positively affect salary, with the exception of presentations (Table 5, Model 2). For non-SEM faculty, funding, number of peer-reviewed articles, and number of conference presentations are linked to higher salaries, while reviews and patents have no effect (Table 6, Model 5). This is not surprising, given that publications such as book reviews do not typically garner the prestige associated with peer reviewed articles, and patenting rates outside SEM disciplines are low. Different forms of productivity are differentially ranked across fields.

Our results concur with prior research showing that involvement in teaching negatively affects salary (Fairweather, 2005; Marthers and Parker, 2008). Productivity moderates the impact of marital and parental status and gender for at least some faculty groups (Table 5, Model 2 and Table 6, Model 5). For SEM faculty, with productivity measures included, most groups of men earn salaries on a par with those of the highest-paid group, married fathers. The penalties among SEM faculty for married mothers and single men without children are no longer significant, suggesting that both of these groups earn less than married fathers because they are less productive. Thus, supply side factors may be a source of disadvantages for these groups. Significant penalties for married women without children, single mothers, and single women without children remain but are reduced in magnitude. Thus, for these groups a portion of the decreases may be linked to lower productivity, but productivity differences alone cannot explain the entire differential. Likewise, adding productivity measures to the model produces a moderate but not dramatic increase in R-squared, from .30 to .42 (Table 5, Model 2).

For non-SEM faculty, including productivity measures does not markedly change overall patterns (Table 6, Model 5). All groups except single fathers continue to earn less than married fathers. The size of the discrepancy is reduced, but the reductions are small. For example, when productivity measures are not taken into account, single mothers earn 15.4 percent less than married fathers. With productivity controlled, this gap drops only slightly, to 14.3 percent. Reductions in penalties for other marital and parental status groups are similarly small. Thus, productivity accounts for only a limited proportion of the variation in non-SEM faculty salaries, while marital and parental statuses are linked more strongly to salary inequalities. This conclusion is well supported in light of the fact that no group’s salary deficits become insignificant when productivity measures are taken into account, nor does the R-squared statistic change very much (0.21 versus 0.27) (Table 6, Model 5). Supply side factors seem to have only minimal impact on salary discrepancies among non-SEM faculty.

Premiums and penalties within specific SEM disciplinary groups

Our third research question addresses whether being in specific SEM disciplinary fields affects marital and parental premiums and penalties, after productivity is taken into account (Table 5, Model 3). Although our data did not allow us to test for associations with individual disciplines, we grouped SEM fields into disciplinary groups with similar work structures. Mean salaries differ across these disciplinary groups (Table 4), but our analyses show that these groups do not moderate the effects of marital or parental status on salary. Women in all marital and parental status combinations still have lower salaries, with the exception of married mothers. Including the disciplinary group variables produces only minute changes in the overall size of penalties, for example the largest absolute change occurs for single women without children, from 9.2 percent less than married fathers to 7.9 percent. Beyond these overall effects, the inclusion of specific SEM disciplinary groups in the model produces no meaningful change in the results, and the R-squared statistic changes only from .42 to .44 after the addition of discipline groups (Table 5, Models 2 and 3). Thus, there is no evidence, as some prior studies propose, that differences in normative work structures across scientific disciplinary groups produce greater or lesser penalties for mothers or for other women in those fields.

Comparing magnitudes of difference in SEM and non-SEM disciplinary groups

Our final research question considers whether the magnitude of premiums and penalties for women and men with various combinations of marital and parental status varies between SEM and non-SEM disciplines. z-test scores across models reveal two significant differences prior to controlling for productivity (Table 5, Model 1 and Table 6, Model 5).¹¹ Married men without children in SEM fields show no deficits vis-a-vis married fathers, but their non-SEM counterparts are penalized. The penalties for non-SEM faculty who are married women without children, married mothers, single men without children, and single women without children are significantly larger than they are for SEM faculty, at least relative to their respective reference categories.

With productivity measures added, married men without children, married mothers, and single men without children in SEM earn roughly the same as married fathers in those fields, while these same family status groups in non-SEM fields continue to experience penalties (Table 5, Model 2 and Table 6, Model 6). Married women without children and single women without children in both SEM and non-SEM fields earn less than married fathers, but the penalties are much larger for non-SEM faculty (15.6 versus 7 percent for married women without children and 18.9 versus 9.2 percent for single mothers). Interestingly, the penalties for single mothers are not significantly different between SEM and non-SEM faculty, although they suffer the consistently largest penalty for all faculty groups, both before and after productivity measures are included. In addition, single fathers never have salary deficits among either faculty group, suggesting that the consequence of being a single parent is gendered (Table 5, Models 1 and 2 and Table 6, Models 5 and 6). Women in both SEM and non-SEM fields appear to experience penalties linked to personal statuses, but they are not consistent across SEM and non-SEM fields. Since these outcomes persist with productivity controls, patterns are consistent with demand-side explanations. They also suggest that single mothers but not single fathers may be devalued as workers.

For women in SEM, marital status is a better predictor of outcomes than motherhood. Among women faculty in SEM, single mothers and single childless women fare the worst, with penalties of 10.6 percent and 9.2 percent, respectively. Married women without children see a smaller penalty of 7 percent and, most surprisingly, married mothers experience no penalty (Table 5, Model 2). In non-SEM fields, women without children fare the worst relative to the reference group, with marital status compounding these effects to a degree. After productivity is accounted for, single, childless women experience penalties of 18.9 percent while married, childless women are penalized 15.6 percent. Single mothers do slightly better, with penalties of 14.27 percent, and married mothers fare the best of all, with the smallest penalties at 11.8 percent (Table 6, Model 6).

Discussion

Our research provides evidence of continuing gender gaps in wages in SEM and non-SEM disciplines that is consistent with Coltrane’s (2004) assertion that gender still matters for workplace rewards, despite efforts to combat inequalities. We also find that family status, in combination with gender, influences the direction and the magnitude of wage gaps, but our findings about the impact of family status for higher education faculty is not entirely consistent with more general patterns in the US workforce (Budig and England, 2001; Budig and Hodges, 2010; Hodges and Budig, 2010). Our results are more consistent with those reported by Fox (2005), who discovered complex relationships between gender, family status, and career outcomes, and also Ridgeway’s (2011) argument that professional women who make significant investments in preparation for careers are less apt to suffer motherhood penalties than other women workers.

We find clearer evidence for fatherhood premiums than for motherhood penalties. This is particularly apparent for non-SEM faculty, where fathers earn more than all of the other categories of faculty we examined. Among SEM faculty, however, fathers are not better paid than faculty in other family categories, especially when productivity is controlled. Men with at least one family commitment, such as a spouse and/or children, generally fare better than others, but some of the differences are linked to research productivity differences. Childless married men, single fathers, and, surprisingly, married mothers have earnings that are not significantly different from those of married SEM fathers. The lower wages paid to married mothers and single childless men compared with married fathers are largely linked to lower productivity for these groups. Thus, it is the non-SEM faculty, rather than the SEM faculty, who most clearly benefit from fatherhood premiums (Glauber, 2008). These findings suggest that supply-side factors affecting productivity may be relatively more influential in SEM disciplines, while demand-side factors that result in differential reward to different gender and family status groups may exert more influence in non-SEM fields.

Women in both SEM and non-SEM fields appear to be subject to penalties tied to personal statuses, but these patterns are not parallel across SEM and non-SEM disciplines and are not the same as those generally found in the US labor market (Budig and England, 2001). In SEM disciplines, it is marital status – being unpartnered – that disadvantages women faculty the most. Married mothers, along with single men, earn less than married fathers, but these differences are largely attributable to their lower research productivity than that of married fathers. For mothers in SEM, the results are consistent with studies suggesting that mothers experience at least short-term declines in productivity when they have children (Hunter and Leahey, 2010; Stack, 2004). Whether these deficits will translate into longer-term cumulative career disadvantages akin to those discussed by Mason and Goulden (2002) remains to be seen. The productivity deficits among single men are more difficult to interpret, since these faculty most closely approximate the ‘unencumbered worker’ that is often stereotyped as the ideal scientist. These faculty members may lack the instrumental and emotional support that a spouse/partner can provide to enhance an academic career (Sokoloff, 1980). Being single is considerably rarer for men than for women in SEM fields, and their status may marginalize them from networks of collaboration and support.

Another interpretive challenge is the lack of impact of family statuses on wage penalties for married, childless women and single mothers in SEM, once rank is controlled. Such a pattern could occur if women are more likely to be promotion-delayed in comparison to colleagues of similar accomplishments. These single women may lack network ties enjoyed by married women in academia – ties that help with advancement in rank. The NSOPF data are not sufficiently detailed at the individual level to allow a test of this possibility, but it is an important question for future research.

Unlike women in SEM fields, parental status makes the most difference for women in non-SEM fields, with childless women experiencing greater salary deficits than women with children. For academic women in both groups, however, a lack of family commitments has the more powerful negative association with salary, once productivity is controlled. This pattern represents a sharp contrast to US labor markets overall, where motherhood penalties typically appear (Budig and Hodges, 2010). Furthermore, including rank in the models made little difference in the outcomes for these groups, suggesting that demand-side bias leading to differential and more negative evaluation of women workers, regardless of family statuses or productivity, remains entrenched in non-SEM fields. We did not anticipate this finding, because so much attention has been focused on gender and family status effects in SEM fields, where they have been presumed to have a greater impact on gender inequality. Our findings may reflect selectivity effects, where only the most talented and highly motivated women enter and are retained in SEM fields.

Our findings identify single mothers as a distinctively disadvantaged group, but we find no evidence that they are any more disadvantaged in SEM than in non-SEM disciplines. The fact that these disadvantages persist, even after we control for productivity, suggests that single mothers may face unique levels of demand side disadvantage. If employers generally assume that women are less committed to work, then single mothers, who bear the bulk of their financial and domestic responsibilities on their own, may be perceived as even less dedicated workers (Kennelly, 1999). The persistent disadvantages for single mothers, but not for single fathers, is a problem that future research should address.

The NSOPF:04 data do not contain all of the detailed measures we would like to have to determine with greater certainty why academic mothers are in a different place in the earnings queue than workers in other fields in the US. It would be valuable to have more detailed information on parenting behaviors and on relative amounts of time invested in work, childcare, or other forms of domestic labor. More nuanced productivity measures, such as citation counts and other antecedents of rank, also would allow us to more effectively untangle the effects of rank from those of gender. However, several explanations are plausible. Although we cannot test the effect directly, we may be seeing an impact of selection bias that applies to people who enter and persist in academic careers (Rayman and Brett, 1995). Women who take academic jobs are less likely to be married or to have children than men who do so – a supply-side factor. Some women may choose to forego heavy family commitments to avoid being caught between two greedy institutions. Those who combine motherhood and academic work may be extraordinarily talented and committed, may have strong personal support systems to help with family life, or may, as prior research has suggested, arrange their family lives in a way that minimizes conflict with work (Grant et al., 2000; Sonnert and Holton, 1993a,b). The selectivity factor probably applies less to the US labor market generally, and selectivity may affect all entrants into academic jobs requiring high achievement, early commitment, and years of postgraduate training

As Long (1990) has suggested, marriage and motherhood may operate in opposite directions for women scientists who are mothers, with some of the benefits of marital status offsetting potential disadvantages of motherhood (see also Budig and England, 2001). On the supply side, women may receive interpersonal support, career guidance, access to expanded networks, or greater understanding of the everyday demands of scientific careers from spouses, especially spouses working in related fields (Creamer, 1999; Schiebinger et al., 2008). Married men may benefit even more on the supply side, since wives traditionally have helped husbands directly and indirectly with academic careers than the reverse (Fowlkes, 1968). The NSOPF:04 contains data that are too limited on spousal employment to explore these questions, but they are worth exploring in the future. Single women might be more marginalized from networks of support and advice, a possibility raised by Long (1990), and single childless women may be sidelined from informal, but influential, social networks in institutions where most of their colleagues are married parents.

On the demand side, the initiatives aimed at bringing more women into academic jobs, especially in SEM fields, have focused primarily on issues affecting married and parental women (e.g. spousal placement programs, maternity leaves, daycare) (Mason and Goulden, 2003). In programs designed to recruit more women into SEM fields, married women, and particularly mothers, may have been especially valuable targets for recruitment, because they could serve as role models for younger women hoping to combine SEM careers with family life. Few other occupations, particularly professional occupations, have experienced such intensive efforts to improve workplace climates for women and mothers. Nonetheless, these efforts have had little impact on childless women or single mothers in SEM disciplines, who continue to suffer salary penalties even after accounting for productivity. Discovering why these women have been left behind deserves more research attention. Exploring why motherhood – indeed all types of gender-linked penalties – are stronger in non-SEM than in SEM disciplines is also an important question to pursue. If institutions have improved climates in SEM fields benefitting married mothers, such improvements should be identified and extended to other units within colleges and universities.

On a cautionary note, we must warn against an overly optimistic interpretation of our findings. When we control for productivity, we find no evidence of a motherhood penalty in SEM disciplines. Mothers in SEM who manage to remain as productive as their men counterparts do avoid salary penalties, but this does not mean than motherhood has no impact on research productivity. Even temporary declines in productivity for mothers can be problematic, especially if they begin a pattern of cumulative disadvantage that persists throughout a career. Furthermore, factors other than salary are important for quality of life. Qualitative research consistently shows that academic women who are married and have children endure considerable stress and time-management pressure in their everyday lives (Armenti, 2004; Jacobs and Winslow, 2004; Rosser and Taylor, 2009; Ward and Wolf-Wendel, 2004; Williams, 2002). Family-oriented men also report considerable conflict between family and career demands (Spalter-Roth and VanVooren, 2008). Institutional supports to ease these conflicts frequently are unavailable (Stewart et al., 2007), or women and men fear stigma and devaluation as uncommitted workers if they use benefits such as parental leave or reduced-duty schedules (Glass, 2004; Spalter-Roth and VanVooren, 2008). The disadvantage may be less in salary than in the personal burdens borne by women more than men for maintaining work–family balance. Thus, efforts to recruit more women into academia, especially into SEM disciplines, should persist in efforts to identify and eradicate gender biases affecting women, regardless of their family configurations.

Footnotes

Acknowledgements

The authors would like to thank two anonymous reviewers and the editor of SSS for their comments and suggestions.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

Biographical note

Kimberly Kelly is an Assistant Professor of Sociology at Mississippi State University, as well as the director of the Gender Studies Program and an affiliate of the Center for Computer Security Research at MSU. Her recent research focuses upon gender inequality in STEM fields, particularly barriers to entry and advancement in STEM fields.

Linda Grant is a professor of Sociology at the University of Georgia and a faculty affiliate of the Institute for Women’s Studies at UGA. Her most recent research focuses on gender and careers in science, medicine, and academia.

References

Amelink

Creamer

(2007) Work-life spillover and job satisfaction of married/partnered faculty members. Journal of Women and Minorities in Science and Engineering 13(4): 317–332.

Anderson

Binder

Krause

(2003) The motherhood wage penalty revisited: Experience, heterogeneity, work effort, and work schedule flexibility. Industrial and Labor Relations Review 56(2): 273–294.

Armenti

(2004) May babies and posttenure babies: Maternal decisions of women professors. Review of Higher Education 27(2): 211–231.

Babcock

Laschever

(2003) Women Don’t Ask: Negotiation and the Gender Divide. Princeton, NJ: Princeton University Press.

Barbezat

(2002) History of pay equity studies. In: Toutkoushian

(ed.) Conducting Salary-Equity Studies: Alternative Approaches to Research. San Francisco: Jossey-Bass, 9–40.

Barrett

Sansonetti

(1988) Issues concerning the use of regression analysis in salary discrimination cases. Personnel Psychology 41(3): 503–516.

Becker

Toutkoushian

(2003) Measuring gender bias in the salaries of tenured faculty members. New Directions in Institutional Research 117: 5–20.

Bellas

(1994) Comparable worth in academia: The effects on faculty salaries of the sex composition and labor-market conditions of academic disciplines. American Sociological Review 59 (6): 807–821.

Bellas

(1997) Disciplinary differences in faculty salaries: Does gender bias play a role? Journal of Higher Education 68 (3): 299–321.

10.

Blair-Loy

(2003) Competing Devotions: Career and Family among Women Executives. Harvard: Harvard University Press.

11.

Boudreau

Sullivan

Balzer

Ryan

Yonker

Thorsteinson

Hutchinson

(1997) Should faculty rank be included as a predictor variable in studies of gender equity in university faculty salaries? Research in Higher Education 38(3): 297–312.

12.

Boulis

Jacobs

(2007) The Changing Face of Medicine: Women Doctors and the Evolution of Health Care in America. Ithaca, NY: ILR Press.

13.

Budig

England

(2001) The wage penalty for motherhood. American Sociological Review 66(2): 204–225.

14.

Budig

Hodges

(2010) Differences in disadvantage: Variation in the motherhood penalty across white women’s earnings distribution. American Sociological Review 75(5): 705–728.

15.

Cole

Zuckerman

(1987) Marriage, motherhood, and research performance in science. Scientific American 255(2): 119–125.

16.

Coltrane

(2004) Elite careers and family commitment: It’s (still) about gender. Annals of the American Academy of Political and Social Science 596: 214–220.

17.

Correll

Benard

Paik

(2007) Getting a job: Is there a motherhood penalty? American Journal of Sociology 112(5): 1297–1338.

18.

Creamer

(1999) Knowledge production, publication productivity, and intimate academic partnerships. Journal of Higher Education 70(3): 261–277.

19.

De Solla Price

(1986) Little Science, Big Science … and Beyond. New York: Columbia University Press.

20.

Etzkowitz

Kemelgor

Neuschatz

Uzzi

Alonzo

(1994) The paradox of the critical mass of women in science. Science 266(1): 51–54.

21.

Fairweather

(2005) Beyond the rhetoric: Trends in the relative value of teaching and research in faculty salaries. Journal of Higher Education 76(4): 401–422.

22.

Fowlkes

(1980) Behind Every Successful Man: Wives of Medicine and Academe. New York: Columbia University Press.

23.

Fox

(2001) Women, science and academia: Graduate education and careers. Gender & Society 15(5): 654–666.

24.

Fox

(2005) Gender, family characteristics, and publication productivity among scientists. Social Studies of Science 35 (1): 131–150.

25.

Fox

Fonseca

Bao

(2011) Work and family conflict in academic science: Patterns and predictions among women and men in research universities. Social Studies of Science 41(5): 715–735.

26.

Ginther

Kahn

(2004) Women in economics: Moving up or falling off the academic career ladder? Journal of Economic Perspectives 18(3): 193–214.

27.

Glass

(2004) Blessing or curse? Work-family policies and mothers’ wage growth over time. Work and Occupations 31(3): 367–394.

28.

Glauber

(2007) Marriage and the motherhood wage penalty among African Americans, Hispanics, and Whites. Journal of Marriage and the Family 69(4): 951–961.

29.

Glauber

(2008) Race and gender in families and at work: The fatherhood wage premium. Gender & Society 22 (1): 8–30.

30.

Grant

Kennelly

Ward

(2000) Revisiting the gender, marriage, and productivity puzzle in scientific research. Women’s Studies Quarterly 38(1): 62–83.

31.

Gupta

(1999) The effects of transitions in marital status on men’s performance of housework. Journal of Marriage and the Family 61(3): 700–711.

32.

Hodges

Budig

(2010) Who gets the daddy bonus? Organizational hegemonic masculinity and the impact of fatherhood on earnings. Gender & Society 24(6): 717–745.

33.

Hunter

Leahey

(2010) Parenting and research productivity: New evidence and methods. Social Studies of Science 40(3): 433–451.

34.

Jacobs

Winslow

(2004) Overworked faculty: Job stresses and family demands. Annals of the American Academy of Political and Social Science 596: 104–129.

35.

Kalev

(2009) Cracking the glass cages? Restructuring and ascriptive inequality at work. American Journal of Sociology 114(6): 1591–1643.

36.

Kennelly

(1999) ‘That single mother element’: How white employers typify black women. Gender & Society 13 (2): 168–192.

37.

Kennelly

Spalter-Roth

(2007) Parents on the job market: Resources and strategies that help sociologists attain tenure track jobs. American Sociologist 37(4): 29–49.

38.

Kmec

(2011) Are motherhood penalties and fatherhood bonuses warranted? Comparing pro-work behaviors and conditions of mothers, fathers, and non-parents. Social Science Research 40(3): 444–459.

39.

Kyvik

(1990) Motherhood and scientific productivity. Science, Technology, & Human Values 21(1): 54–71.

40.

Leahey

(2006) Gender differences in productivity: Research specialization as the missing link. Gender & Society 20(6): 754–780.

41.

Lincoln

(2008) Gender, productivity, and the marital wage premium. Journal of Marriage and Family 70(3): 806–814.

42.

Long

(1990) The origins of sex differences in science. Social Forces 68(4): 1297–1316.

43.

Long

(2001) From Scarcity to Visibility: Gender Differences in Careers of Doctoral Scientists and Engineers. Washington, DC: National Academies of Science.

44.

Long

Allison

McGinnis

(1993) Rank advancement in academic careers: Sex differences and the effects of productivity. American Sociological Review 58: 703–722.

45.

Long

Fox

(1995) Scientific careers: Universalism and particularism. Annual Review of Sociology 21: 45–71.

46.

Marthers

Parker

(2008) Small colleges and new faculty pay. Academe 94(4): 45–49.

47.

Mason

Ekman

(2007) Mothers on the Fast Track: How a New Generation Can Balance Family and Careers. New York, NY: Oxford University Press.

48.

Mason

Goulden

(2002) Do babies matter? The effects of family formation on the lifelong careers of academic men and women. Academe 88(6): 21–27.

49.

Mayer

Tikka

(2008) Family friendly policies and gender bias in academia. Journal of Higher Education Policy and Management 30(4): 363–374.

50.

Monosson

(ed.) (2008) Motherhood, the Elephant in the Laboratory. Ithaca, NY: Cornell University Press.

51.

National Center for Education Statistics (2011) National Study of Postsecondary Faculty-Overview. Washington, DC: US Department of Education. Available at: nces.ed.gov/surveys/nsopf/ (accessed 14 June 2012).

52.

National Research Council (2010) Gender Differences in Critical Transitions in the Careers of Mathematics, Science, and Engineering Faculty. Washington, DC: National Academies Press.

53.

National Science Foundation (2010) ADVANCE: Increasing the participation and advancement of women in academic science and engineering careers. Arlington, VA: National Science Foundation. Available at: http://www.nsf.gov/funding/pgm_summ.jsp?pims_id=5383 (accessed 14 June 2012).

54.

Perna

(2005). Sex differences in faculty tenure and promotion: The contribution of family ties. Research in Higher Education 46(3): 277–307.

55.

Porter

Toutkoushian

Moore

III

JR (2008) Pay inequities for recently hired faculty, 1988–2004. Review of Higher Education 31(4): 465–487.

56.

Rayman

Brett

(1995) Women science majors: What makes a difference in persistence after graduation? Journal of Higher Education 66(4): 388–414.

57.

Renzulli

Grant

Kathuria

(2006) Race, gender, and the wage gap: Comparing faculty salaries in predominately white and historically black colleges and universities. Gender & Society 20(4): 491–510.

58.

Renzulli

Parrott

Beattie

(2011) Racial mismatch and school type: Teacher satisfaction and retention in charter and traditional public schools. Sociology of Education 84(1): 23–48.

59.

Reynolds

(2004) When too much is not enough: Actual and preferred work hours in the United States and abroad. Sociological Forum 19(1): 89–120.

60.

Ridgeway

(2011) Framed by Gender: How Gender Inequality Persists in the Modern World. New York: Oxford University Press.

61.

Rosser

(1997) Re-Engineering Female-Friendly Science: Applying Women’s Studies Methods and Theories to Attract Students. New York: Teachers College Press.

62.

Rosser

(2005) Through the lenses of feminist theory: Focus on women and information technology. Frontiers: A Journal of Women Studies 26(1): 1–23.

63.

Rosser

Taylor

(2009) Why are we still worried about women in science? Academe 95 (3): 7–10.

64.

Sax

Hagedorn

Arredondo

DiCrisi

III (2002) Faculty research productivity: Exploring the role of gender and family-related factors. Research in Higher Education 43(4): 424–446.

65.

Schiebinger

Henderson

Gilmartin

(2008) Dual-Career Academic Couples: What Universities Need to Know. Stanford, CA: The Michelle R. Clayman Institute for Gender Research.

66.

Settles

Cortina

Malley

Stewart

(2006) The climate for women in academic science: The good, the bad, and the changeable. Psychology of Women Quarterly 30(1): 47–58.

67.

Sokoloff

(1980) Between Money and Love: The Dialectics of Women’s Home and Market Work. Santa Barbara, CA: Praeger.

68.

Sonnert

Holton

(1993a) Gender Differences in Scientific Careers: The Project Access Study. New Brunswick, NJ: Rutgers University Press.

69.

Sonnert

Holton

(1993b) Who Succeeds in Science? The Gender Dimension. New Brunswick, NJ: Rutgers University Press.

70.

Spalter-Roth

VanVooren

(2008) PhDs at mid-career: Satisfaction with work and family. Washington, DC: America Sociological Association.

71.

Stack

(2004) Gender, children, and research productivity. Research in Higher Education 45(8): 891–920.

72.

Stewart

Malley

LaVaque-Manty

(2007) Transforming Sciences and Engineering: Advancing Academic Women. Ann Arbor: University of Michigan Press.

73.

Strathman

(2000) Consistent estimation of faculty rank effects on academic salary models. Research in Higher Education 41(2): 237–250.

74.

Toutkoushian

(1998) Racial and marital status differences in faculty pay. Journal of Higher Education 69(5): 513–541.

75.

Toutkoushian

Bellas

(2003) The effects of part-time employment and gender on faculty earnings and satisfaction: Evidence from the NSOPF:93. Journal of Higher Education 74: 172–195.

76.

Toutkoushian

Bellas

Moore

(2007) The interaction effects of gender, race, and marital status on faculty salaries. Journal of Higher Education 78(5): 572–601.

77.

Umbach

(2007) Gender equity in the academic labor market: An analysis of disciplines. Research in Higher Education 48(2): 169–193.

78.

Ward

Wolf-Wendel

(2004) Fear factor: How safe is it to make time for family? Academe 90(6): 16–19.

79.

West

Curtis

(2006) AAUP faculty gender equity indicators 2006. Washington, DC: American Association of University Professors (AAUP).

80.

Williams

(2002) How academe treats mothers. Chronicle of Higher Education, 6 June.

81.

Winslow

(2010) Gender inequality and time allocations among academic faculty. Gender & Society 24(6): 769–793.

82.

Wolfinger

Mason

Goulden

(2008) Problems in the pipeline: Gender, marriage, and fertility in the ivory tower. Journal of Higher Education 79(4): 388–405.

83.

Xie

Shauman

(2003) Women in Science: Career Processes and Outcomes. Boston, MA: Harvard University Press.

84.

Zuckerman

Cole

(1975) Women in American science. Minerva 13(1): 82–102.