The Gender Gap in Workplace Authority: Variation across Types of Authority Positions

T he ﬁnding that men are disproportionately represented in positions of workplace authority to an extent that cannot be explained by human capital attributes and location within the structure of the economy is well documented. Arguing that different authority positions are differentially gendered, this article tests a more reﬁned hypothesis: that the gender gap in authority is larger in positions that are relatively seen as more suitable for men. We distinguish authority positions by the amount of authority and the gender-typed control over resources they involve, hypothesizing a larger gender gap in positions with larger proportions of supervisory, as opposed to nonsupervisory tasks, and in positions with control over organizational, as opposed to human, resources. Our ﬁndings, based on Dutch linked survey and administrative data from about 32,000 employees who have started their careers between 1999 and 2016, largely support these ideas. We ﬁnd an overall larger gender gap in positions with largely supervisory tasks as opposed to positions with only some supervisory tasks. Additionally, the authority gender gap is the largest in largely supervisory positions with control over organizational resources and smallest in positions with control over human resources. We discuss both theoretical and practical implications of our ﬁndings.


Introduction
The existence of a gender gap in workplace authority across industries, including those numerically dominated by women (Grant Thornton International 2019; European Commission 2018), is well documented. Finding that men's disproportionate representation in positions of workplace authority is not explained by superior qualifications among men and men's location within the structure of the economy (Smith 2002;Longarela 2017), scholars have hypothesized that the cultural view on men as more status-worthy and competent than women (Ridgeway 2011(Ridgeway , 2014, held whether on the demand or supply side of labor markets, drives much of the gender gap in workplace authority. Indeed, the idea that men are seen as more suitable for authority positions than women is supported by ample empirical evidence (Koenig et al. 2011;Hentschel et al. 2019).
Against the backdrop of the general view on men as more suitable for authority than women, this article tests a more refined hypothesis: that the gender gap in workplace authority is larger in positions that are relatively seen as more suitable for men. Given that compared to women, men are generally seen as more competent and status-worthy (Ridgeway 2001;Fiske et al. 2002), it is likely that the more authority a position entails, the more it is seen as suitable for men. By the same token, authority positions that involve tasks stereotypically seen as women's tasks, such as people-oriented tasks-also associated with less status-may be seen as relatively more suitable for women. Provided that these views in part determine the extent to which men are overrepresented in a certain type of position, the gender gap in authority should be larger in the subset of jobs entailing more authority and tasks that are culturally seen as more appropriate for men. Some support for these expectations can be found in early work on gender and management (e.g., Kanter 1977), but tests of these ideas in contemporary samples of employees, net of employees' job-relevant characteristics, are rare.
We evaluate these ideas using linked survey and administrative data from the Netherlands. To get at the amount of authority a position entails, we differentiate positions on the basis of the proportion of supervisory, as opposed to nonsupervisory tasks they entail. To study the role of gender-typed tasks, we make a distinction between authority positions with no control over resources, control over human resources, and control over organizational resources. The distinction between control over human resources, involving people-oriented tasks, and control over organizational resources, entailing things-oriented tasks, is informed by a theoretical understanding of the former as seen as more suitable for women and the latter as more suitable for men.
The idea that authority gender gaps are relatively larger in positions seen as more suitable for men has potential practical implications for gender inequality in power at the workplace and job quality. For example, positions with higher amounts of authority and status likely accrue more power and higher job rewards to their occupants. People-oriented job tasks are in addition a source of strain associated with emotional labor and negative interpersonal experiences, overall resulting in poorer job quality (Chan and Anteby 2015). The concentration of women in positions with less authority also has consequences for their potential to influence broader inequality at work. Women's representation in authority positions has been associated with lower levels of segregation and smaller gender earning gaps at the workplace (Cohen and Huffman 2007;Stainback, Kleiner, and Skaggs 2016;Stojmenovska 2019).
In studying gender authority gaps across types of authority positions, our study differs from the vast majority of studies on the gender gap in workplace authority over the past decades that do not distinguish different authority positions (Bridges and Miller 1981;Huffman 1995;Hultin 1998;Rosenfeld, Van Buren, and Kalleberg 1998;Mitra 2003;Huffman and Cohen 2004;Yaish and Stier 2009;Mintz and Krymkowski 2010;Bygren and Gähler 2012;Dämmrich and Blossfeld 2016). We analyze data from a large sample of about 32,000 employees who have started their careers between 1999 and 2016, thereby drawing conclusions about gender and authority among relatively recent cohorts of employees. Linking survey to administrative data from the Dutch longitudinal population registers enables us to trace the survey respondents back in time and measure some authority-associated individual qualifications at the relevant point in time-at entry into the position.

Gender Status Beliefs, Gendered Job Tasks, and Authority
Status beliefs are widely held cultural beliefs about social worthiness and competence of individuals viewed as categorically distinct on the basis of gender, race, class, and other axes of categorization (Ridgeway 2011(Ridgeway , 2014. Attributions of status and competence are relevant for views on who is suitable for workplace authority because authority positions are associated with higher status than positions that do not entail authority (Wolf and Fligstein 1979). Given that gender status beliefs deem men as more status-worthy and competent than women, 1 men are in general likely to be seen as more suitable for authority positions than women. In addition, the discourse on positions of authority requiring the highest level of competence itself is gendered. As Kanter (1977) recounts in her account on the development of managerialism as a function of corporate capitalism, early images of competent managers are imbued with a "masculine ethic" emphasizing rationality, efficiency, and ability to set aside emotions, characteristics ascribed and prescribed to men (Eagly and Karau 2002). Even though the emphasis on this masculine ethic has decreased over time, men remain being seen as more suitable for authority than women (Koenig et al. 2011;Hentschel, Heilman, and Peus 2019).
Prior research has suggested that these beliefs distinctively influence women's and men's work experiences in various ways. For example, women managers are evaluated less positively than men managers (Eagly, Makhijani, and Klonsky 1992) even though they are not less motivated to manage or less effective at managing than men (Eagly, Karau, and Makhijani 1995;Eagly et al. 1994). In a study of client noncompliance in a microfinance bank, Doering and Thébaud (2017) find that clients are less likely to comply with women loan managers and managers in positions that were previously occupied by women, as opposed to men managers and managers in positions previously occupied by men. In a similar vein, Chakraborty and Serra (2020) find that workers get angrier at women managers and are more likely to question their decisions than those of men managers. These findings of greater resistance toward women managers, coupled with the findings that the gender gap in workplace authority is not explained by superior human capital attributes among men, men's location within the structure of the economy, and the gendered transition to parenthood (Smith 2002;Stojmenovska and England 2021), point to the view on men as more suitable for authority than women as pertinent to men's disproportionate representation in workplace authority.
If the view on men as more suitable for authority positions drives some or even much of the gender gap in workplace authority, is the gender authority gap larger in positions relatively seen as more suitable for men? We answer this question by looking at the amount of authority and the gender-typed tasks positions involve. Positions with higher amounts of authority are likely associated with higher status and job rewards. Given the general association of men with higher status, we expect men to be seen as relatively more suitable for authority positions that entail more authority. Consequently, they may be more likely than women to be hired especially for positions that entail more authority, or-to the same effectmay be given higher amounts of authority when hired or promoted into an authority position. Relatedly, scholars have argued that due to political pressures to have women represented in authority, women are often given positions with few authority tasks that are managerial only in name, a proposition known as the resegregation or glorified secretary hypothesis (Jacobs 1992).
Next to the amount of authority, the gendering of the type of tasks involved is also likely relevant for the extent to which a subset of authority positions is seen as suitable for men. In line with the people-things dimension of gender stereotypes (Prediger 1982), which entails an association of women with people and men with things, women tend to be employed in peopleoriented and men in things-oriented occupations (Lippa, Preston, and Penner 2014). People-oriented job tasks are associated with lower status than thingsoriented tasks, a social fact that can be interpreted using Walby's (1989) account on the connection between what she calls the "private-" and "public patriarchy." The undervalued type of work women often perform and are expected to perform at the workplace can be seen as an extension of the type of undervalued (care and service) work they do in their homes. Gender task segregation has also been found to happen within jobs. For example, women academics are more likely than men academics to perform "academic housework" -undervalued activities such as those related to teaching (Heijstra et al. 2017).
In the context of workplace authority, the notion that women are more suitable for dealing with people leads to the expectation that women will be seen as relatively more suitable for authority positions that have to do with managing people compared to positions of managing things. Prior work based on old and selective samples (e.g., single workplaces or samples restricted to fulltime employees) largely from the United States provides some support for this idea. In her classic work of gender and work at a large industrial corporation, Kanter (1977) has indeed found that when in managerial positions, women tended to have personnel jobs, whereas men held positions in which they made decisions about the organization's budget and strategy. A few studies using data from the 1980 Class Structure and Class Consciousness project (Wright et al. 1982;Wright 1989) have found a larger gender gap in positions that involve control over organizational resources (among others, making decisions regarding the budget and distribution of funds) than control over human resources, net of individual and job characteristics (Jaffee 1989;Adler 1994;Hopcroft 1996;Ishida 1995). 2 Evidence from more recent cohorts and contexts other than the United States is scarce. Whereas Wright, Baxter, and Birkelund (1995; Australia, UK, Canada, Sweden, Norway, Japan) find smaller gaps in making decisions about the organization than pay and hiring of personnel, Ishida (1995;Japan, Britain), also using Wright's data, finds, somewhat conflictingly, larger gender gaps in control over organizational than human resources. Based on data from the Israeli Mobility Survey 1991-1992, Kraus and Yonay (2000) find smaller gaps in authority to promote and reward others than make organizational decisions in women-dominated and gender-integrated occupations, and the opposite in men-dominated occupations.
While past work studying gender authority gaps across types of positions is limited, based on the literature on the relationship between gender status beliefs and authority, as well as the broader literature on gendered job tasks, we expect to find an overall larger gender in positions with larger, as opposed to smaller, amounts of authority. Additionally, we expect a larger gender gap in authority positions with things-oriented as opposed to peopleoriented tasks. We differentiate positions with different amounts of authority by taking into account the proportion of supervisory, as opposed to nonsupervisory, tasks of the position, assuming that positions with larger proportions of supervisory tasks entail more authority. Prior work has used the number of employees supervised as a measure of the amount of authority, concluding that women have lower amounts of authority than men because they tend to supervise fewer employees than men (Hultin 1998;Abendroth, Maas, and van der Lippe 2011;Blommaert, Meuleman, et al. 2019;Blommaert, van der Lippe, et al. 2019). Our measure of proportion of supervisory tasks is a more direct measure of task composition and is less conflated with workplace size than measures of the number of employees supervised. To differentiate authority positions with differently gendered tasks, we consider the type of control over resources a position entails, distinguishing between control over human resources and control over organizational resources, the former involving people-oriented authority tasks and the latter things-oriented tasks.
We test ideas about differences in gender gaps across types of authority positions in a large sample of employees in the Netherlands, controlling for a number of job-relevant individual attributes and job and workplace characteristics. By using a more detailed measure for the task composition of authority positions, our study adds more nuance to extant research studying the gendered distribution of workplace authority mostly by using a binary measure of whether one supervises others on the job (Bridges and Miller 1981;Huffman 1995;Hultin 1998;Rosenfeld, Van Buren, and Kalleberg 1998;Mitra 2003

Data and Sample
We use a sample of about 32,000 respondents from three rounds (2014, 2015, and 2016) of the Dutch National Survey of Working Conditions (Nationale Enquête Arbeidsomstandigheden [NEA]), a large-scale annually repeated crosssectional survey representative of employees in the Netherlands, excluding the self-employed. 4 Using unique respondent identifiers, we link the survey data with administrative data on respondents' employment histories based on the Dutch population registers. Since the latter is available only for respondents who have started their career in the year 1999 or later, we focus on this group of respondents. This is also a substantively interesting population of relatively recent employee cohorts. 5 Of the initial sample of 33,415 respondents whose careers started in 1999 or later, and whose main activity at the time of the survey was working, we exclude 203 individuals in armed forces occupations, 942 individuals with missing information on authority status, and 86 individuals with missing information on workplace size. Information based on data from the population registers is complete for all respondents. In total, after listwise deletion, we drop only about 4 percent of the respondents. This leaves us with a final analytic sample of 32,184 individuals (51 percent women). Women are slightly overrepresented in our sample (in 2016, women comprised 48 percent of all employed individuals aged 25-35 in the Netherlands).

Workplace Authority
Our measure of workplace authority combines two dimensions of task composition of the position: the proportion of tasks that is supervisory, and information on whether the position entails control over human and organizational resources. We distinguish three types of positions based on the proportion of supervisory tasks they involve: positions without any supervisory tasks (no authority), positions with some supervisory tasks and mostly nonsupervisory tasks (some authority), and positions with mostly supervisory tasks (largely supervisory positions). This classification is based on three survey questions: "Do you supervise other employees in your position?" (yes/no), "Are your tasks exclusively supervisory?" (yes/no), and "What kind of tasks does most of your work consist of?" (supervisory/nonsupervisory).
Respondents with largely supervisory positions were additionally asked whether their position involves authority to make (1) decisions about personnel matters such as hiring and pay raises (yes/no) and (2) decisions regarding the organization's financial or strategic policies, such as the budget or multiannual strategic plan (yes/no). Using these two questions, we further split the largely supervisory category into three types of positions, involving: no control over resources ("no" to both questions), control over human resources ("yes" to the first and "no" to the second question), and control over organizational resources ("yes" to the second question). In total, thus, the outcome variable contains five categories: no authority, some supervisory tasks, largely supervisory with no control over resources, largely supervisory with control over human resources, and largely supervisory with control over organizational resources. This measure, based on respondents' direct reporting of authority tasks, captures authority better than occupation-based measures. In our data, only 17 percent of employees with authority tasks work in an executive occupation. Table 1 shows descriptive statistics separately for women and men. Nineteen percent of women and thirty-four percent of men have positions with at least some supervisory tasks. Among supervisory positions, positions with only some supervisory tasks (and mostly nonsupervisory tasks) are the most common: 13 percent among women and 22 percent among men. For individuals with largely supervisory positions, 2 percent of women and 5 percent of men have no control over resources, 2 percent of women and 3 percent of men have control over human resources, and 2 percent of women and 4 percent of men have control over organizational resources.

Analytic Strategy
To model the gender gap in workplace authority, we use multinomial logistic regression analyses. 6 Our discussion of results focuses on predicted probabilities as a more intuitive way (as opposed to relying on regression coefficients) of presenting and understanding findings based on nonlinear probability models (Mize 2019). We begin our analyses with a baseline model regressing the authority outcome only on gender (women are the reference category). Based on this model, we calculate and plot predicted probabilities of having each type of authority positions for women and men. We make conclusions about the gender gaps in the different subsets of authority positions by discussing relative differences in predicted probabilities of having each type of position between men and women.
In a subsequent model, we adjust for gender differences in authorityassociated individual characteristics to show how the authority gender gap looks for women and men with similar job-relevant individual characteristics. In this model, we control for highest completed education (coded as lower secondary education, higher secondary education, Bachelor, and Master), work experience (cumulative number of years in employment at entry into position), and tenure with current employer (cumulative number of years working for the current employer at entry into position) as authority-associated attributes on which women and men differ on average. Building on the baseline model, we regress the authority outcome on gender, education, work experience, and tenure with the current employer and predict probabilities of having each type of authority position for women and men.
We measure work experience and tenure with the current employer at entry into the position, rather than at the time of the survey. This allows us to take into account the qualifications of respondents at the relevant point in time-when they were hired or promoted into the authority position. The measure of work experience is based on data from the Dutch population registers, which contain complete information on all (changes in) activities of respondents over time.
Only periods in which the main activity of the respondent was working count as work experience. Education is measured at the time of the survey because it is insufficiently documented in the register. Women in our sample have higher levels of educational attainment and fewer years of work experience and tenure with the current employer than men at entry into the position (see table 1). In a final model, we additionally include part-time factor (at entry into position, coded as 0-40 percent, 41-60 percent, 61-80 percent, and 81-100 percent), industry (13 categories), and workplace size (number of employees at the workplace), as three job characteristics that are differentially distributed among women and men and that are relevant for the authority opportunities a job offers. Women in our sample are more likely to work part-time and at smaller workplaces than men and are concentrated in certain industries, such as the health industry (see table 1). These are relevant differences because the presence and extent of authority structures differ across industry and workplace size (McGuire and Reskin 1993). Working part-time is additionally seen as less compatible with having authority given that work-devotion narratives are particularly strong in high-status positions such as authority positions (Blair-Loy 2005; Cha and Weeden 2014). The predictions based on the models adjusting for individual and job authority-associated characteristics assume the sample distribution of controls. Figure 1 shows the results from the multinomial regression analyses of the gender gap in different types of authority positions. The figure plots women's and men's average predicted probabilities of having authority positions with only some supervisory tasks, largely supervisory positions with no control over resources, largely supervisory positions with control over human resources, and largely supervisory positions with control over organizational resources. For each subset of positions, we compare women's and men's predicted probabilities based on the three successive models we described earlier: the model without controls; the model adjusting for education, work experience, and tenure with the current employer; and the final model, additionally controlling for part-time factor, industry, and workplace size. The respective coefficients of interest are shown in table 2. The coefficients for the control variables can be found in, see the online supplementary material for, table S1.

Results
A first look at figure 1 reveals that men have higher probabilities of having workplace authority than women, regardless of the type of position we look at. Focusing on the controls-unadjusted gender gaps (first set of bars), men's probability of having some supervisory tasks (21.9 percent) is 8.5 percentage points higher than that of women (13.3 percent). For largely supervisory positions with no control over resources, men's probability of having these positions is 2.8 percentage points higher than women's (4.9 as opposed to 2.1 percent). Men's probability of having largely supervisory positions with control over human resources is 1 percentage point higher than that of women (2.6 as opposed to 1.6 percent) and that for largely supervisory positions with control over organizational resources is 2.6 percentage points higher (4.2 percent among men, 1.6 percent among women). These differences in probabilities between women and men are both substantively and statistically significant. 7 However, the size of the relative gender gap in authority differs across the different subsets of positions. We observe the largest gender gap for largely supervisory positions with control over organizational resources, and the smallest gap for largely supervisory positions with control over human resources. Men's average probability of having authority positions with control over organizational resources is 2.57 times higher than that of women. In contrast, their probability of having authority positions with control only over human resources is 1.61 times higher than that of women. The gender gap in largely supervisory positions with control over human resources is comparable to that in positions with only some supervisory (and mostly nonsupervisory tasks): men's average probability of having these positions is 1.64 times higher than women's. The gender gap in largely supervisory positions with no control over resources seems to fall somewhere in between, with men being 2.37 times as likely as women to have such a position. Overall, thus, these results suggest that the gender gap in authority is relatively smaller in positions with only some supervisory tasks-which presumably entail lower amounts of authority than largely supervisory positions-and when they do involve largely supervisory tasks, positions of managing human resources.
Does the fact that women and men on average differ on authority-associated individual characteristics explain this pattern? The second set of bars in figure 1 show, for each type of position, men's and women's probabilities of having these positions when controlling for highest completed education, work experience at entry into position, and tenure with the current employment at entry into position, assuming the sample distribution of these covariates. For all positions, comparing the first to the second bars for women and men, we see a small increase in the predicted probabilities for women and a small decrease in the predicted probabilities for men, altogether making the absolute differences in predicted probabilities between men and women slightly smaller (8.2 percentage points for positions with some supervisory tasks, 2.6 percentage points for largely supervisory positions with no control over resources, 0.8 percentage points for largely supervisory positions with control over human resources, and 2.3 percentage points for largely supervisory positions with control over organizational resources).
The conclusions regarding the differences in the gender gaps across the types of positions however remain unchanged. Adjusted for gender differences in highest educational attainment, work experience, and tenure with the current employer, the gender gap in authority is largest in largely supervisory positions with control over organizational resources and smallest in largely supervisory positions with control only over human resources. Men's average probability of having authority positions with control over organizational resources (4 percent) is 2.31 times higher than that of women (1.7 percent). In contrast to this, men are 1.50 times as likely as women to have positions with control over human resources (2.5 as opposed to 1.7 percent). This is a comparable relative difference to the gender difference in having some supervisory tasks: 1.60 times higher probability among men (21.7 as opposed to 13.5 percent). For largely supervisory positions with no control over resources men are 2.21 times as likely as women to have these positions (4.7 as opposed to 2.1 percent). All differences remain substantively and statistically significant.
Finally, how are the gender gaps we observe shaped by the fact that the opportunities for having workplace authority (and potentially the type of workplace authority) vary across job characteristics that covary with gender? The third set of bars in figure 1 is based on the models that next to education, work experience, and tenure with the current employer additionally control for part-time factor, industry, and workplace size. The first observation here is that adding job and workplace characteristics makes a more substantive difference to the gender gap in authority than controlling for individual authority-associated attributes. The predicted probabilities for women and men come closer together than when adding only individual characteristics. These differences however remain statistically significant and for the most part also substantively meaningful (7.2 percentage points for positions with some supervisory tasks, 1.6 percentage points for largely supervisory positions with no control over resources, 0.4 percentage points for largely supervisory positions and control over human resources, and 1.7 percentage points for largely supervisory positions with control over organizational resources).
Net of individual and job characteristics, the gender gap in authority remains largest for positions with mostly supervisory tasks and control over organizational resources. Men's average probability of having these positions (3.7 percent) is 1.89 times larger than that of women (1.9 percent). This difference is followed by the difference in largely supervisory positions with no control over resources (1.65 times higher among men; 4.1 as opposed to 2.5 percent among women). Next in size is the gender gap in positions with only some, and mostly nonsupervisory, tasks where men are 1.52 times as likely as women to have such a position (21.1 as opposed to 13.9 percent). Lastly, the gender gap in largely supervisory positions with control over human resources is the smallest: men (2.3 percent) are 1.21 times as likely as women (1.9 percent) to have such a position. This is also a substantively small absolute difference of 0.4 percentage points, albeit statistically significant. Altogether, the discussion of the results based on the three models suggests that the pattern we observe in the controls-unadjusted model is robust to the inclusion of individual-and job-level controls. Our finding of substantial gender gaps in authority net of job-relevant individual attributes and job and workplace characteristics is in line with existing evidence of gender authority gaps net of individuals' qualifications and location within the structure of the economy (Smith 2002;Longarela 2017). The differences in gender gaps across types of authority positions we observe provide support for our theoretical expectations.
So far we have discussed the difference between the different subsets of authority positions on conceptual grounds. One way to study the place of these different positions on the organizational hierarchy is to look at the earnings of occupants of each type of position. In figure 2, we show average predicted earnings for women and men in the five types of positions we examine: positions with no authority, positions with some supervisory authority, largely supervisory positions with no control over resources, largely supervisory positions with control over human resources, and largely supervisory positions with control over organizational resources. These predictions are based on OLS regression models, regressing earnings first only on the interaction between gender and our authority outcome (first set of bars), and then in addition on a number of controls-education, work experience, tenure with current employer, parttime factor, industry, and workplace size (second set of bars). Respondents' individual earnings are obtained from the Dutch register using unique respondent identifiers. We use a measure of hourly nominal earnings, calculated by dividing the total gross earnings in the year of the survey by the total number of hours paid in income in that year. We log-transform the variable to address skew and adjust 2014 and 2015 earnings to 2016 earnings according to the Consumer Price Index. The earnings in the Dutch register are not top-coded. Figure 2 shows a clear income hierarchy between occupants of the different types of authority positions. Occupants of largely supervisory positions with control over organizational resources earn the highest hourly earnings (about 29 euros per hour among men and 27 euros per hour among women in these positions), followed by occupants of largely supervisory positions with control over human resources (23 euros among men, 20 euros among women). The figure additionally shows that women earn less than men for the same positions (all differences are substantively meaningful and statistically significant). The patterns we observe largely hold also net of education, work experience, tenure with current employer, part-time factor, industry, and workplace size. Adjusted for these, men in largely supervisory positions with control over organizational resources earn on average 24 euros per hour; women in these positions earn 21 euros per hour. Occupants of largely supervisory positions with control over human resources follow, with about 20 euros per hour on average among men and 19 euros among women. The difference in earnings between occupants of positions with control over organizational and those with control over human resources is both substantively meaningful and statistically significant. The difference in earnings between women and men with largely supervisory positions with no control over resources and respectively women and men with only some supervisory tasks is no longer statistically significant. Women without authority earn about 17 and men about 18 euros per hour. 8

Discussion
The idea that not all authority positions are the same has long been acknowledged (Adler 1994). This article aimed to contribute to a more comprehensive view of the gender gap in workplace authority and associated gender inequality at work by studying whether the gender gap in authority is larger in positions seen as more suitable for men. We approached the latter by differentiating authority positions by the proportion of supervisory, as opposed to nonsupervisory tasks of the job, and the type of control over resources they involve. The former aimed to capture the amount of authority, and the latter the gender-typed tasks surrounding control over resources positions entail.
Using linked survey and administrative data and a large sample of employees in the Netherlands who have started their careers between 1999 and 2016, we find that the gender gap in authority is largest in largely supervisory positions with control over organizational resources and smallest in largely supervisory positions with control over human resources. The gender gap in positions with control over human resources is comparable to the gender gap in positions with only some supervisory tasks that are mostly nonsupervisorypositions we suspect entail the lowest amounts of authority. Differences in levels of education, work experience, and tenure with the current employer between men and women, as well as women's and men's hours worked and location within the structure of the economy (particularly industry and the size of the workplaces at which they work), do not explain the patterns that we observe.
Our findings concerning control over resources are in line with the literature on gender task segregation along the people-things dimension, as well as early work (e.g., Kanter 1977) that suggests that women are overrepresented in positions that have to do with personnel at the workplace. As argued by Kanter (1977), not only are these considered to be less important to the organization, but also as more corresponding to what are typically seen as women's tasks. An emerging literature on the personal and health implications of having job authority has begun to reveal additional consequences of having women represented in personnel authority jobs. For example, in a longitudinal study of authority and breast cancer, Pudrovska (2013) finds that women with authority had a significantly higher risk of breast cancer diagnosis over a period of thirty years than women who are not employed or employed but have no workplace authority. Importantly, the relationship between job authority and breast cancer was conditional on having authority to hire, fire, or influence others' pay. The author theorizes that because women are not seen as suitable for authority, when in positions of control over human resources, they go through stressful interpersonal experiences that increase their risk of breast cancer.
The question of how the gender gap in workplace authority looks across types of positions has become even more timely since Kanter's (1977) case study and Wright's 1980 Class Structure and Class Consciousness project (Wright 1989), given the growing number of positions of authority and women's representation in them. In the Netherlands, women's representation in authority has grown over time-especially after the second half of the 1990s-while that of men has stayed the same (van der Lippe, van Doorne-Huiskes, and Blommaert 2014). This suggests that new authority positions (potentially occupied by women) have been created, raising the question of what type of positions these are. While our findings are based on data from the Netherlands, we believe that they are not exclusive to the Dutch context. Given the evidence of the existence of gendered authority and status beliefs across countries (Bowen et al. 2007; Maniraj Singh and Enid Kiaye 2013; Katila and Eriksson 2013), and in light of the theoretical considerations outlined above, we suspect that similar patterns can also be observed in other contexts.
While our analyses treat women as a homogenous group, gender status beliefs interact with cultural beliefs about social worthiness and competence of individuals viewed as categorically distinct on the basis of other axes of categorization, such as race, class, and migration status. The handful of studies that have looked at multiplicative effects of social meanings find larger disadvantages in authority among Black women compared to White women and Black men, relative to the disadvantages these groups experience compared to White men (Elliott and Smith 2004;Shams and Tomaskovic-Devey 2019). A more complete picture of authority gaps necessitates taking account of individuals' location at the intersection of socially meaningful categories.
Our findings potentially underestimate gender authority gaps in all types of positions we look at, given that the age of our sample comes at the cost of not examining outcomes at later stages of respondents' careers. It is likely that men's advantage we observe in our sample is larger at later career stages, if gender inequalities in authority grow over the career course, as has been shown to be the case for gender inequality in earnings (Weisshaar and Cabello-Hutt 2020). However, studying authority outcomes among employees who have started their careers between 1999 and 2016, we are able to do away with concerns that the outcomes we observe are in part a result of past discrimination experienced by older employee cohorts.
More generally, our findings add to the discussion on whether authority is best measured dichotomously or in a polytomous fashion. In a review of this ongoing discussion in the authority literature, Smith (2002) points out that the conclusion depends on the type of research question. Our study suggests that for a better understanding of gender inequality in workplace authority and associated work outcomes, polytomous measures of workplace authority, viewing authority as residing in the work activities of employees, seem more appropriate.

Notes
1. In writing of women and men, we refer to gender as assigned social position, which may or may not coincide with respondents' gender identities. Genders other than women and men are not part of this study because the secondary data we use classify individuals into these two categories. We additionally recognize the analytic distinction between men and masculinity and refer to masculinity as a set of attributes and behaviors ascribed and prescribed to men. 2. In cases where authors do not make explicit comparisons but provide the data, we make these comparisons ourselves. Sometimes authors make conclusions based on absolute, rather than relative, gender authority gaps. Given that both women and men are more likely to have supervisory than decision-making authority, for example, finding a larger absolute gap for supervisory positions is not indicative of women being more strongly underrepresented in supervisory than decision-making positions. Relative differences are more informative and thus the differences we discuss. This means that our conclusions about a study's findings sometimes differ from what the original authors themselves conclude. 3. Studies have also operationalized workplace authority using occupationbased classifications of jobs as managerial and nonmanagerial (Shenhav 1992;Blum, Fields, and Goodman 1994;Powell and Butterfield 1994;Moore and Shackman 1996;Huffman 1999;Maume 1999;Baxter and Wright 2000;Reskin and McBrier 2000;Brown and Jones 2004;Maume Jr. 2004;Stainback and Tomaskovic-Devey 2009;Ng and Sears 2017;Mun and Jung 2018;Shams and Tomaskovic-Devey 2019). As noted in the Method section, occupation-based measures fall short of capturing the full spectrum of authority. 4. Pooling the 2014, 2015, and 2016 rounds of the survey ensures larger statistical power for the analyses. The data collection approach and survey questions are comparable across the three years. 5. To identify respondents who started their careers in 1999 or later, we use respondents' highest completed education (as indicated in the survey) combined with information on when they were last registered as a student from the population registers. 6. The Wald test for combining alternatives in multinomial logistic models (Long and Freese 2014) suggests that the five categories of the outcome variable are distinct. 7. The figures show 83% confidence intervals for the means for women and men to correspond to a 95% significant difference between group means (Goldstein and Healy 1995). 8. The replication package to this article is available at the corresponding author's website (http://www.stojmenovska.com). The data used in this article are nonpublic microdata from Statistics Netherlands. These, and other nonpublic microdata, are under certain conditions accessible for statistical and scientific research. For more information, contact microdata@cbs.nl.

Supplementary Material
Supplementary material is available at Social Forces online, http://sf.oxfordjou rnals.org/. Stephanie Steinmetz is an Associate Professor of Sociology at the University of Lausanne where she is a member of the LINES and LIVES centers and FORS. She is also an affiliated researcher at the University of Amsterdam where she works with the WageIndicator Foundation. Her research focuses on gender inequalities in education and the labor market from a cross-national and intersectional perspective. Her work has been published in journals such as the European Sociological Review, Work, Employment and Society, and Social Politics. Beate Volker is the Head of the Department of Human Geography and Spatial Planning at Utrecht University. Her research is on networks and network outcomes in neighborhoods as well as in other social settings, on trends in social capital and communities, and in adult friendships. She has published among others in Social Networks, European Sociological Review, and Sociological Science.