Another ceiling? Can males compete for traditionally female jobs? – includes appendix
Leanne E. Atwater
One of the dramatic changes in the work force has been the increasing proportion of women working in what were once thought to be traditionally male jobs. Historically, management positions have been included among the traditionally male jobs, though the proportion of women filling lower and mid-level management positions in recent years has steadily increased (Dalton & Kesner, 1993). The proportion of women in top management positions, however, has remained very low. This lack of equal representation of women in top-management positions has been attributed to a “glass ceiling,” an invisible barrier that keeps women from entering top-level management positions (Morrison, White & Van Velsor, 1987).
A number of reasons have been suggested for the existence of the glass ceiling. First, some contend that males or male characteristics are preferred for management positions (Heilman, Block, Martell, & Simon, 1989). Others suggest that perhaps there have been too few women qualified for or interested in top-level management positions. Another reason may be tradition: management may have been seen as a male domain because, until recently, most managers and the majority of the work force in most jobs were male. Wharton and Baron (1987) offered a different explanation: People prefer to work with others of the same gender, hence men would be more suitable to supervise men. If the “same-sex preference” holds (that is, if males and females prefer and rate those of their own gender more highly) we would expect females to be considered more suitable to manage females. However, little research has directly addressed the potential for profemale bias in personnel decisions concerning traditionally female jobs, while a substantial body of evidence has documented that selection decisions for traditionally male positions have been biased against females (Deaux, 1984; Dipboye, Fromkin & Wiback, 1975; Rosen & Jerdee, 1974; Terborg & Ilgen, 1975). Nilson (1976) did find that male nurses were rated lower in prestige than female nurses. Additionally, males applying for secretarial, receptionist, or office clerk jobs have been clearly discriminated against in hiring decisions (Levinson, 1975; Business Week, July 31, 1995).
The present study investigated the extent to which males and females showed profemale bias in personnel decisions regarding managers of traditionally female jobs. Are females preferred over more qualified males as managers in female jobs? Are females preferred if those making decisions know the subordinates are predominantly females?
Also investigated were personnel decisions regarding managers in lower-versus higher-level management positions. For female jobs, are females considered more suitable for both mid-level and top-level management positions?
Background
Rater Gender Effects
The extent to which a rater sees himself or herself as similar to a ratee has been shown to influence ratings. Specifically, the rater/ratee congruence hypothesis (Pazy, 1986; Tsui & O’Reilly, 1989) suggests that raters who see ratees as more similar to themselves will rate them more positively. Therefore, women would be expected to rate other women more highly than would men. Pazy (1986) demonstrated that when the task was gender neutral, males tended to rate hypothetical male employees higher than females given identical performance information. When the job is not gender neutral, but rather male gender-typed, a number of field studies have suggested that bias against females is more pronounced for male raters than for female raters. Jacobson, Antonelli, Winning and O’Peil (1977), for instance, found that women in positions of authority over men were evaluated negatively by those men, but were not evaluated negatively by women subordinates. Similarly, Stevens and DeNisi (1980) and Van Fleet and Saurage (1984) found women had more positive attitudes toward women managers than men did. Whether gender of evaluator affects ratings of candidates in a female gender-typed job has not been tested. Given the differences described above in male and female ratings of males and females, we hypothesized that females would be more likely than males to select a female applicant as manager in a traditionally female job.
H1: Female evaluators will be more likely to display profemale bias in their applicant ranking decisions than will males.
Gender Bias in Selection Decisions
In the 1970s, researchers demonstrated that certain jobs were seen as “male” or “female” and that both males and females suffered when applying for sex-atypical jobs (cf. Cohen & Bunker, 1975; Cash, Gillen & Burns, 1977). Additionally, in 1978 Rose and Andiappan found that both males and females were biased in their selection decisions in favor of managers whose gender matched the majority of their subordinates. Thus, a predominantly male work force created obstacles for women applying for management jobs.
However, more recent evidence suggests that this trend may be changing. Women are having greater access to traditionally male jobs than they have had in the past. For example, Pazy (1992) actually found profemale bias among subjects asked to rate applicants for promotion to a management position in a gender-neutral job. Powell and Butterfield (1994) found no gender bias in selection decisions for senior-level executive positions in federal civil service (traditionally thought to be the domain of males). Haemmerlie and Montgomery (1991) disconfirmed Goldberg’s (1968) classic finding that both males and females placed a higher value on work done by a male. In the Haemmerlie and Montgomery study, the opposite was found. In engineering and law (occupations traditionally dominated by males), work was evaluated more positively when done by females than by males. Additionally, from 1981 to 1991 the percentage of women managers in the workplace increased 14 percent (from 27 to 41 percent) (Dalton & Kesner, 1993). It appears that females are now being accepted as viable management candidates in many arenas and may even be the preferred candidates in some jobs. Stover (1994), for instance, found that in universities, women were likely to be found managing departments with a high proportion of women in them. Reskin and Ross (1992) similarly found that women were typically placed in positions responsible for females.
Perhaps the shift toward more diversity in the work force and the great deal of time and money that have been spent on sensitizing employees and students about issues of diversity and equal opportunity have sensitized individuals to gender equality. If this is the case, we would expect personnel decisions to be based on qualifications, with no consideration of who is the “appropriate” gender.
Or, alternatively, gender bias may shift from bias against females to bias against males as more women enter the work force and more jobs are considered “female territory.” There is some evidence that females are more likely to be entering and advancing in traditionally male occupations than are males in traditionally female fields (Jacobs & Lim, 1992; Beller, 1985). The female share of male-dominated occupations is also increasing more than the male share of female-dominated jobs (Cotter, DeFiore, Hermsen, Kowalewski & Vanneman, 1995). If managing women is a woman’s job, management positions in traditionally female jobs may be seen as best filled by women.
Heilman’s “lack of fit” model (Heilman, 1983) suggests that “…expectations about how successful or unsuccessful an individual will be when working in a particular job are determined by the fit between the perception of the job’s requirements in terms of skills and abilities.” (p.278). In a gender-typed job, individuals of the “proper gender” will be seen as fitting the job whereas those of the inappropriate gender will not be seen as fitting.
Consistency theory (Feather & Simon, 1975) would also predict that males would be seen as unsuitable for female jobs and vice versa, because the type of work is inconsistent with the masculine or feminine image. Williams (1992), for example, cites four jobs that have been identified with “women’s work” throughout the 20th century – nursing, librarianship, elementary school teaching and social work. In these female gender-typed jobs, male characteristics would be seen as “unfit;” similarly, women would be perceived as unfit for construction or fire-fighting.
Thus, both consistency theory and the “lack of fit” model would predict that males would be seen as unsuitable for female jobs, and therefore likely to suffer from gender discrimination similar to that faced by women who have attempted to enter male gender-typed jobs. If this is the case, as women enter the work force and predominate in certain jobs, we could expect a new ceiling to emerge for males. Similar to the glass ceiling females have faced, males may be considered ill-suited to manage in female domains.
Two related, more specific questions are also relevant here. First, is the level of management an issue when males seek “female” management jobs? In other words, if women are now being seen as viable applicants for mid-level positions but less likely applicants for top-level positions, are men suffering a similar ceiling when the jobs are traditionally female? History would suggest that when it comes to top executive positions, men are filling them regardless of which gender dominates the job; but this may be because until recently the majority of jobs were seen as male gender-typed. Thus, male managers have been seen as a better gender fit. Consistency theory and the “lack of fit” model would suggest that perhaps females will be seen as suitable top managers when the job is female gender-typed.
The second question addresses the ratio of female-to-male subordinates to be supervised. Kanter (1977), known for her theory of tokenism, argues that when any group represents less than 15% of an organization, its members will be subject to predictable forms of discrimination. That is, they will suffer from their minority status. If evaluators know that the majority of job incumbents to be supervised are female, they may be more likely to select a female manager. As mentioned earlier, bias against males has been demonstrated when the majority of subordinates to be supervised were female (Rose & Andiappan, 1978). However, the study by Rose and Andiappan was done a number of years ago and the bias may no longer exist.
Because the “males are managers” stereotype appears to be changing and the bias against women decreasing, we believed the gender-type of the job (i.e., Social Work) would have more influence on selection decisions than the managerial nature of the job. Consistency theory and “lack of fit” both would suggest that females would be seen as more suitable for traditionally female jobs. However, because the gender type of the top-management job might be unclear, (i.e., Social Work is female but top managers are male), we expected less bias in decisions favoring females for a top-level female gender-typed job than in decisions for a mid-level management job.
H2A: Male and female evaluators will show profemale bias in applicant rankings for top-level management positions in traditionally female jobs.
H2B: Male and female evaluators will show profemale bias in applicant rankings for mid-level management positions in traditionally female jobs.
H2C: Male and female evaluators will show more profemale bias in applicant rankings for mid-level than top-level management positions in a traditionally female job.
H3: In applicant ranking decisions for managers of traditionally female jobs, profemale bias will be more pronounced when it is clear that the majority of the subordinates are females than when the number of female subordinates is unknown.
Rater Experience
Familiarity with supervisor-subordinate interaction as it relates to gender could also affect the rater’s rankings of applicants. For example, raters who had prior management experience may have witnessed situations where matching manager and subordinate genders resulted in better performance, or less conflict. Consistent with the rater/ratee congruence hypothesis, prior research has demonstrated that subordinates tend to rate their supervisors more positively if the supervisor and subordinate are of the same gender (cf. Jacobson, Antonelli, Winning & O’Peil, 1977; Van Fleet & Saurage, 1984). Observation of better working relationships between same sex supervisor/subordinate dyads would also support the “lack of fit” model and promote stereotypes that females should supervise females. Also, those with prior management experience have likely been exposed to more training in diversity and affirmative action, and thus may be more sensitized to the value of selecting females for management jobs. As a result, we hypothesized the following.
H4: Raters with management experience will show more profemale bias than raters with no management experience.
Salary Allocation Decisions
While selection decisions for management positions are one avenue to use in assessing potential gender bias, selection/promotion decisions represent only one type of potential bias: negative access discrimination (Rosenstein & Hitt, 1986). A second type of bias, unequal treatment discrimination (Baker, Slonaker & Wendt, 1994), refers to decisions such as allocating starting salaries or salary increases. The evidence concerning promale bias in terms of pay is convincing.
Morrison and Von Glinow (1990) found that women who have advanced into management often find reward differentials when comparing themselves to men. Drazin and Auster (1987) found that women earn substantially less than men in management. Blau and Ferber (1987) pointed out that employers with discriminatory biases hire women only at a wage discount large enough to compensate for the loss of utility or level of discomfort associated with employing them.
That these pay differentials are attributable to women being less qualified or less experienced has not been supported (O’Neill, 1992). To the contrary, in the 1970’s two studies found that males were likely to be offered higher initial salaries than comparably qualified females (Dipboye, Arvey, & Terpstra, 1977; Terborg & Ilgen, 1975). Both males and females perceived wage discrimination favoring males (Rosen, Rynes, & Mahoney, 1983; Lewis, 1991). Nearly one-third of the women questioned by Lewis felt they had been discriminated against in a promotion or salary increase because they were female. Similarly, Nelson-Horchler (1991) found that a large proportion of females (74%) as well as 26% of the males felt that women received less pay for equal work. Controlling for institutional, individual, and positional characteristics, Pfeffer (1990) found that women’s salaries were lower than men’s among 20,000 high-level administrative positions in 821 colleges and universities.
There has been much speculation as to the reasons for these salary differences between males and females. For example, wage expectations have been suggested as a contributing factor. Jackson and Grabski (1988) found that women were more likely than men to believe that less pay is fair. Nivea and Gutek (1981) and Major and Konar (1989) concluded that women value pay less than men do. Thus, in terms of pay, women may be earning less because they expect to earn less, or value pay less.
If females are being discriminated against in salary decisions in traditionally male jobs, it may also be because they are viewed as less desirable and, therefore, of less worth. Similar logic may hold for males in female jobs. Males may be seen as less desirable for traditionally female jobs due to the “lack of fit” (Heilman, 1980), and thus offered lower salaries. This seems unlikely for the following reasons. First “male” characteristics are seen as more valuable than “female” characteristics, particularly for managerial jobs (Gutek & Cohen, 1987; Stover, 1994). Second, the status of a job has been shown to increase as the number of males entering the job goes up and to decrease as the number of females in the job increases. For example, Pfeffer and Davis-Blake (1987) found in a study of college administrators that the average salaries went down as the proportion of women in the group went up. Freedman and Phillips (1988) describe this as support for “institutionization,” which in essence says that at the point where work is defined as “women’s work,” wages (and, therefore, status) will decrease. As such, males may be offered even higher salaries than usual if they apply for lower-status “female jobs” in order to attract them. Third, the bulk of evidence on salary disparities between men and women still suggests that males are paid more than females even when education, job tenure, and experience are accounted for (cf. Taber, 1992). Thus, while women may be seen as more suitable for female gender-typed jobs, this will not alter the perception that work done by women is, nevertheless, worth less in monetary value than work done by men. Additionally, if salary inequities are due in part to different salary expectations on the part of males and females, we might expect the salaries offered to job applicants by males and females to differ. Two additional hypotheses follow from the above discussion:
H5A: Males will be allocated higher salaries than females when they are selected as the first-choice applicant for a top-level management position in a traditionally female job.
H5B: Males will be allocated higher salaries than females when they are selected as the first-choice applicant for a mid-level management position in a traditionally female job.
H6: Male evaluators will allocate higher salaries than will female evaluators to job applicants selected for managerial positions in traditionally female jobs.
Method
This methodology section is organized as follows: First, we present a description of the pre-test completed to identify female gender-typed jobs. Next, a description of the main study is presented. Third, the following are described: subjects; the experimental task; the selection scenarios created; the three experimental conditions; the information provided about hypothetical job applicants; data collection procedures; and the ranking and salary allocation tasks that subjects completed. Finally, the data collection design is summarized.
Identifying Gender-Typed Jobs
The determination of gender-typing of jobs was assessed empirically. Ninety-nine undergraduate business’ students from three universities were asked to review a list of seven academic disciplines and indicate whether they believed a male or a female would be more successful as the Dean (top-level manager) in that academic discipline. The job of Dean was selected as a top-level management position based on the work of Sandler and Hall (1986). The same students were also asked to review a similar list of disciplines and indicate whether they thought a male or a female would be more successful as a “Coordinator” (mid-level manager) in that discipline.
Surprisingly, virtually no students felt that this request was inappropriate, or that there was no such thing as a job more suitable to a man or a woman. Overwhelmingly, the students were able to assign either a male or a female as more likely to be successful in each of the seven disciplines rated for top-level managers: Engineering; Physical Education and Sports Medicine; Business; Architecture and Environmental Design; Social Sciences; Social Work; and Nursing. The disciplines rated for mid-level manager were indicative of a specialty within each area. For example, instead of Engineering, the Coordinator managed a group within Electrical Engineering. As can be seen from the data presented in Table 1, both male and female students saw Engineering at both levels as a “male” job, followed closely by Physical Education and Sports Medicine. Nursing and Social Work were seen by both males and females as “female jobs.” Female respondents [TABULAR DATA FOR TABLE 1 OMITTED] tended to see Business as a job suitable for males and females, while males saw Business as a “male” job. Both Architecture and Social Science were seen as rather neutral jobs by both males and females.
Because we were interested in potential gender bias in a female gender-typed job, Social Work was selected as the “female” gender-typed job for the purposes of this study. Work by Beggs and Doolittle (1993) suggested that while some occupations had become more gender-neutral since the 1970’s, Social Work has become more female gender-typed. Nursing was not selected because earlier work (cf. Cohen & Bunker, 1975) had suggested that it was so sex-bound that individuals might not be able to remove biases associated with this occupation (cf. Williams, 1992), despite the fact that a large number of male nurses perform exceptionally well in the field.
Subjects
Subjects were 191 business students from a medium-sized southwestern university. Roughly half of the students were male and half were female, and approximately half held a management position either now or in the past. 1 While there is some controversy in the literature as to whether students’ decisions are comparable to those of “real managers” in the field, few studies have revealed differences between students and managers when sex-role stereotypes are studied (Bernstein, Hakel, & Harlan, 1975; Dipboye, Fromkin & Wiback, 1975; Rosen & Jerdee, 1974). Mullins (1982) has also supported the external validity of using students as decision-makers in employment decision studies. Some critics of lab studies of potential bias suggested that bias would be greater when the raters did not know personally the individuals being rated. As such, they hypothesized that actual decisions made in work settings would reveal less bias than contrived decision situations. Contrary to this belief, Ilgen and Youtz (1986) found that bias against racial minorities was greater in the field than in the lab.
Additionally, the students who served as subjects in this study were not young, full-time students. Most were employed, over half had management experience, and their average age was 29 years. We believed this was a useful sample in which to study gender bias in that these individuals represent the work force, and likely the managerial ranks, of the future. Understanding their biases, if any, may give us insight about what to expect of decision-makers in the 21st century.
Subject Profile. Subjects completed an information sheet about themselves, indicating their gender, their age, and whether they presently (or previously) held a management position. To reduce the likelihood of subjects discerning the purpose of the study, the subject information sheet was always provided as the last sheet in the set of materials.
Procedure
The Task. A simulated selection decision experiment was designed using “paper people” to assess potential profemale bias. Paper people were desirable for this study as we wanted to ensure that the target male applicant in each decision scenario was noticeably more qualified than the target female applicant, though both would be considered qualified. This type of scenario would have been difficult, if not impossible, to create in a real personnel selection setting. A set of test materials was designed to create a scenario for a hypothetical management selection/salary allocation decision. These materials were reviewed by colleague experts and were judged to be face valid; that is, they appeared to be legitimate job applicant materials.
Subjects were asked to review two packets of materials. Each packet contained information about a large university and the managerial job for which four hypothetical applicants were being evaluated. One packet included materials describing job-relevant characteristics about two male and two female applicants for a top-level management position (Dean of Social Work). The second packet contained materials about four hypothetical applicants for a mid-level management position (Coordinator of Activities within a Social Work department).
Top and Mid-level Management Scenarios. In the scenario presented for the Dean’s job, the Dean managed a faculty of 85 and a staff of 20, had an annual budget of over 4 million dollars, and was paid a starting salary of $90,000, which was described as “somewhat below average.” The Coordinator’s job was considered a mid-level management job in which the individual was described as supervising 4 people and managing a budget of $250,000. The $45,000 salary for this position was also described as “somewhat below average.” In one condition, the ratios of male and female subordinates to be supervised by the Dean and Coordinator were given.
Experimental Conditions. Subjects ranked one set of applicants for the top-level job and a different set of applicants for the mid-level job in one of three conditions. In all three conditions, the descriptions of the applicants’ qualifications were identical. In Condition 1, no names or genders of applicants were provided. Rather, applicants were identified by numbers only. Condition 1 served as a manipulation check and a comparison group. In Condition 2, applicants were identified by names that clearly indicated whether they were male or female. In Condition 3, applicants’ names were identified as in Condition 2, and the raters were also told the number of males and females the manager would supervise. All four employees to be supervised in the mid-level, Coordinator position were female, and 100 of the 105 total in the top-level (Dean) position were female.(2)
Hypothetical Job Applicants. Each packet contained a description of four hypothetical job applicants – two males and two females. Included in the description was a subject identifier, which consisted of either a number [in Condition 1] or a name [Conditions 2 and 3]. The names were clearly indicative of gender: Jeff, Cheryl, Greg, and Carol for the Dean; Steve, Sara, Laura and Tim for the Coordinator. The individual’s current position was also identified (e.g., Associate Dean for the Dean’s job; Social Worker for the Coordinator’s job).
In each scenario, one of the males and one of the females was clearly under-qualified relative to the other two. Descriptions of the under-qualified individuals were included as distractions from the purpose of the study. The other (qualified) male and female applicants for each job currently held the same jobs, with the male slightly, but noticeably, more qualified.(3)
The reason for presenting the male as the more qualified applicant for the traditionally female job was to test whether the gender stereotype of the job would take precedence over qualifications in decisions. In other words, we purposely biased the qualifications in favor of the gender opposite to the gender type of the job. A male was the more qualified applicant for both the top-level and mid-level management jobs. In this way, if a female were selected or given a higher salary, it would be clear that the decision was not based on an interpretation of qualifications, but rather in spite of them. In other words, we wanted to test whether a female would be seen as more suited to manage in a stereotypically female job even if she was less qualified than a male.
Also included was a listing of 11 categories of job-relevant information, with a hand-written comment beside each job category about the applicant’s competence in that area. Subjects were led to believe that these comments had been compiled by a search committee attempting to hire an individual for each position. The categories of information included: Current position, education, general ability, decision-making capability, number of years in current position, effort usually put into work, number of years of experience, willingness to sacrifice personal time to perform needed tasks, leadership ability, creativity in approaching new problems, willingness to perform tasks not liked by giving up personal preferences, and a “comments” section. To clearly distinguish the most qualified individual from the person next most qualified, the information provided in the three categories was different. The most qualified individual for the Dean’s job (a male in Conditions 2 and 3) had 5 years in his current position, 9-1/2 years academic experience, and excellent leadership ratings, while the lesser qualified individual (a female in Conditions 2 and 3) had 4-1/2 years in her current position, 9 years academic experience “very good” ratings in leadership ability. For the Coordinator’s position, in Conditions 2 and 3 the more qualified male applicant had 5 years in his current position compared to 3-1/2 years for the female, “excellent” leadership ability as compared to “very good”, and 5 years as compared to 3-1/2 years of job experience. The information provided about applicants was identical in each of the three conditions (no names, names, names plus number of female subordinates).
Terborg and Ilgen (1975) suggested that the bias effect of stereotyping women is most potent when little is known about the female being evaluated. Other more recent work has suggested that gender role stereotypes can be overridden if sufficient job-relevant information about male and female applicants is provided (Tosi & Einbender, 1985). Similarly, Heilman (1984) found that highly job-relevant information produced less differential treatment of female and male job applicants than information low in job relevance. Pazy (1992) found that job experience, not gender, was the primary consideration in the promotion decisions for mid-level management decisions. In this study, the number of pieces of task-relevant information, plus an indication that the gender-incongruent applicant was more qualified in three areas, was intended to provide the decision-maker with adequate reasons to select the male for the mid- and top-level management jobs. Thus, if the female were selected or given a higher salary, the decision would not be based on qualifications, but rather on gender congruence with a traditionally female job.
To confirm that the individual we had identified as most qualified was indeed seen that way, one set of materials was prepared with numbers substituted for names to identify the four applicants (Condition 1). This provided a standard against which to assess results in the other two conditions. Subjects evaluated materials in one of the three conditions and were asked to make selection and other decisions twice – once for the Dean’s position and again for the Coordinator. To reduce potential order effects for the two jobs, the two packets were presented in alternating order, i.e., the decisions for Dean applicants were made first by half of the subjects, and the Coordinator decisions were made first by the remaining half of the subjects. Also, the four applicants’ materials were sequenced in the packet in two different ways so that order effects could be tested. In one order, the more qualified male’s materials appeared first, the lesser qualified female’s materials appeared third, and unqualified applicants appeared second and fourth. In the second order, the applicants’ materials were reversed, the more qualified male appeared last and the qualified female appeared second. When the data were analyzed, there were no order effects; e.g., the qualified male was no more (or less) likely to be selected if his materials appeared first than if they appeared third.
Data Collection Procedures. Data were collected from students in 12 different management classes of varying size (e.g., 15-40) over a period of three weeks. The decision about which classes completed materials in each of the three conditions was random, and therefore groups of subjects were randomly assigned to conditions. Materials were handed out and subjects were told that we were interested in the information individuals use in making selection decisions. Subjects were then asked to review the materials and answer the questions. Subjects were thanked for their voluntary participation but were not debriefed at the time about our hypotheses, because we did not want to bias responses of future subjects. Once all data were collected and analyzed, subjects were informed about the purpose and results of the study via the school newspaper.
Rankings and Salary Allocations. For each job, subjects were asked to select their first and second choices for filling the job, and then to indicate the salary they would offer their first- and second-ranked applicants, keeping in mind that the present salaries were identified as “somewhat below average.” An open-ended question asked subjects why they ranked each of their first-choice applicants higher than their second-choice applicants.
Subjects were asked to rank their choices, rather than to provide ratings because earlier research has shown that more differentiation is likely to be observed when applicants are ranked rather than rated, and consequently bias is more likely to be seen if it exists (Dipboye, Arvey & Terpstra, 1977; Dipboye, Fromkin & Wiback, 1975; Pazy, 1986; Pazy, 1992). Ranking is also more realistic in selection decisions and less subject to distributional errors like leniency (Bernardin & Beatty, 1984).
Data Collection Design
A summary of the variables considered in the study, then, includes the following: rater genders (male or female), rater’s prior or current management experience (yes or no), condition (no names, names only, names plus number of female subordinates), and level of management position considered (mid or top). The level of management variable was repeated. The dependent variables of interest were the applicant selected first (male versus female) and the salary allocated to the selected applicant.
Table 2 graphically presents the data collection design. As can be seen from Table 2, there are 12 possible conditions from which data were collected (S1-12). Subjects in each of those 12 conditions, however, could select either the most qualified applicant or another applicant. In the first four conditions (no name), the expectation was that only the most qualified applicant would be selected, hence the number or proportion of subjects selecting an incorrect applicant would be [TABULAR DATA FOR TABLE 2 OMITTED] zero, making those four conditions null conditions. The design for data analysis, then, cannot follow the data collection design since it is not possible to perform computations for all interactions involving the null conditions that would be involved in a full 3X2X2X2 design. The data confirmed that these are null conditions (the actual observations obtained did not differ significantly from zero for these conditions); however, since the actual observations were slightly greater than zero, the actual values were used in the analyses as a conservative approach.
Results
Selection Decisions
The primary question of interest in this study was the extent to which profemale bias exists in selection decisions for traditionally female jobs. To test this effect, we examined the number of times the less-qualified applicant (Cheryl, Carol, or Applicant #3) was selected over the more-qualified applicant (Jeff, Greg or Applicant #1). The two treatment conditions (Applicant Gender Known and Applicant and Subordinates’ Genders Known) were compared to the gender unknown condition to test whether applicant’s or subordinates’ genders were relevant.
Z-tests for differences in proportions were performed to compare the proportion of times the most qualified female identified by name (Cheryl for the Dean’s job and Carol for the Coordinator’s job) was selected compared to the percent of the time Applicant #3 was selected (all tests z-tests are shown in Appendix A). For the Dean’s job, the test proportion, or the proportion of times the less qualified applicant was selected when the gender was unknown was five percent (or two individuals). For the Coordinator’s job, this proportion was three percent or one individual.(4) The information about Applicant #3 was identical to that for Cheryl in the Dean’s scenario or Carol in the Coordinator’s scenario, except that the applicant’s name and number of female subordinates were not revealed. Applicant #1’s materials were also the same as that for Jeff (Dean) or Greg (Coordinator), with name and subordinates omitted.
The first hypothesis we tested, H1, concerned rater gender effects. Selections of first-choice applicants for each job were compared for male and female raters and tested with three binomial proportion tests, one comparing males and females selecting the female first for each condition. No rater gender effects were found. Consequently, male and female raters were combined into one rater group for testing H2, H3, and H4.
The results of the proportion tests of differences in the number of times the male vs. female applicant was selected in each of the three conditions are presented in Table 3. As can be seen from the table, when the applicant’s gender was known, 29 percent of the raters selected the less-qualified female for the Dean’s job. This was a significant increase over the proportion selected when the applicant’s gender was unknown (5%), thus supporting H2A.
For the Coordinator’s job, 26 percent of the raters selected the less-qualified female for the job when the applicant’s gender was known. This was a significant [TABULAR DATA FOR TABLE 3 OMITTED] increase over the proportion who selected the comparable applicant when gender was unknown, thus supporting H2B.
H2C proposed that more profemale bias would be shown in the mid-level management position than in the top-level position. This hypothesis was not supported. The proportions of raters selecting the female when applicant gender was known and when both applicant’s and subordinates’ genders were known were not significantly different for top- and mid-level jobs.
Regarding H3, there was no significant increase in the number of times the female applicant was selected for the Dean’s job when the gender was known for both applicant and subordinates rather than for applicant only. In this case, 32 percent of the raters selected the female for the Dean’s job. Thus, H3 was not supported. For the Coordinator’s job, 24 percent selected the female when both the applicant’s and the subordinates’ genders were known (see Table 3), again not a significant difference from when only the applicant’s gender was known. In both gender known and gender and subordinates known conditions, however, profemale bias existed – a significant proportion of the raters selected the female for a traditionally female job, even when she was less qualified than a male applicant?
While profemale bias was not more pronounced when the manager was supervising predominantly females, the comments that raters provided about why they selected their first choice applicant were more likely to indicate that the female was chosen because of her gender when the number of females to be supervised was given. When the gender of the applicant was unknown, the more qualified individual was almost always selected and reasons given for that selection never included attributions about gender. Overwhelmingly, leadership skills and/or ability and experience were given as the reasons for the selection. These were the factors we used to present the male as the more qualified applicant, thus supporting the validity of this manipulation. When gender was known, for the Dean’s position, 16% of the reasons given by those who selected the female indicated that they did so specifically because of her gender. In the condition where subordinates’ genders were known, 44% of the reasons given by those selecting the female indicated that they selected the female for the Dean’s job because a female is more suited to supervise females. Statements made by respondents included “a female is better suited to supervise females,” “a female might respond to employees better,” and “supervising more females.” When subordinates’ genders were known, 40% of the reasons given by those selecting the female indicated that being female was their reason for selecting a female for the Coordinator’s job. These comments provide additional support for our contention that raters were, in fact, selecting individuals for the job based on gender rather than qualifications, particularly where predominantly women would be supervised.
To test H4, we compared the selections made by subjects who currently held, or had held, a management position with those who had no management experience. In support of H4, a z-test of differences between proportions indicated that those who currently held or who had held a management position were more likely to select the less-qualified female for the Dean’s job (61%) than were those who had not held a management position (39%) (p [less than] .05).(6) In this case, those with management experience were more likely to select the less-qualified female than those with no experience. There was not a significant difference in selection decisions as a function of management experience for the Coordinator’s job.
While we did not hypothesize any interactions, we felt it was prudent, given the main effects, to test for interactions.(7)
We compared the gender known condition with the gender unknown condition. (The gender and subordinates condition was not included as it was not significantly different from the gender known condition and would have added needlessly to the number of tests performed.) Z-tests for differences in proportions of females selected were computed to test the two-way interactions (i.e., gender by management experience, gender by condition, and condition by management experience for both the Dean and Coordinator’s jobs). There were no significant interactions for the Dean’s job. There was one significant interaction between gender and management experience in ranking decisions for the Coordinator’s job (p [less than] .01). Females with management experience were significantly more likely to select a female for the Coordinator job than were males with or without experience or females with no management experience. We also tested for an interaction between condition (names versus no names) and position (Dean versus Coordinator). This interaction was not significant.
Summary of Selection Decision Results
To summarize the selection decision results, profemale bias existed for selection decisions made in traditionally female jobs. The profemale bias existed for both male and female raters, and for both top-level and mid-level jobs. Specifying the number of female subordinates to be supervised did not significantly alter the profemale bias in decisions, though raters were more likely to say gender was the reason for selecting a female if they knew females would be supervised. Additionally, those subjects with management experience were more likely to make decisions favoring the less-qualified female than were those without management experience, and females with management experience were more likely to rank the female applicant first than males or females without management experience.
Table 4. Analyses of Variance for Salary Decisions for Top-Level
and Mid-Level Management Jobs
A. Top-level Manager (Dean)
Main Effects and Interactions df f
Male vs Female Selected (MF) 1 .79
Condition (C) 1 .01
Gender of Rater (G) 1 4.17(*)
MF X C 1 .95
MF X G 1 1.98
C X G 1 .42
MF X C X G 1 .02
B. Mid-level Manager (Coordinator)
Male vs Female Selected 1 2.13
Condition 1 .01
Gender of Rater 1 .00
MF X C 1 .15
MF X G 1 1.43
C X G 1 .98
MF X C X G 1 3.24
* p [less than] .05
Salary Allocation Decisions
Differences in salary allocations as a function of condition, rater gender, and applicant gender were tested. Only the two conditions where gender was known were tested, as we were interested in differences allocated to male and female applicants. Two 2X2X2 analyses of variance (2 conditions by male versus female selected first by gender of rater) were performed; one for Deans and one for Coordinators. The results of these analyses are presented in Table 4. Means, standard deviations and ranges for males and females selected first for each job, as well as a breakdown of average salaries offered to males and females and by males and females, are presented in Table 5. A breakdown by condition was omitted as effects for this variable were near zero.
As can be seen from Table 4, there were no main effects for male versus female selected. That is, contrary to H5A and H5B, males were not allocated higher salaries than females for either the top- or mid-level job. H6 was supported, however, for the top-level (Dean’s) job in that males allocated higher salaries overall than did females (M for males = $100,196 vs. M for females = $96,115). Males did not allocate higher salaries for the mid-level job.
Table 5. Means, Standard Deviations and Ranges for Salaries
Allocated to Males and Females Selected as Dean and Coordinator
Salary Allocated in Dollars
Standard
Mean Deviation Minimum Maximum
Male Selected $97,622 9,375 80.000 150,000
First as Dean
Female Selected 100,153 12,115 75,000 130,000
First as Dean
Male Selected 50,985 8,458 30,000 100,000
First as Coordinator
Female Selected 48,083 7,180 27,000 60,000
First as Coordinator
Average Salary Average Salary
Offered to Female Offered to Male
Dean Coordinator Dean Coordinator
Average Salary $95,454 46,769 96,293 51,453
Offered by
Female Raters
Average Salary 103,600 49,636 98,951 50,557
Offered by
Male Raters
In looking at the ranges and standard deviations presented in Table 5, large variances in salary allocations occurred, indicating that salary allocation decisions seemed to vary a great deal across raters. For example, even though the present salary was given, the salaries allocated for the Dean’s job when the male was selected ranged from $60,000 to $150,000, a wide range. This may be an indication of error variance and could have contributed to the lack of effects found.
Discussion
Clearly, the most interesting result of this study was the tendency for some raters to select for managerial positions a less-qualified female rather than a more-qualified male when the job to be managed was traditionally female. This effect was demonstrated for both male and female raters, at both mid- and top-levels of management. The open-ended comments confirmed that in a number of cases, raters were clearly selecting the less-qualified applicant because she was female and was expected to be more successful in a “female” job. It seems that the lack of fit or inconsistency between gender type of the job and gender of the applicant is a basis for discrimination against males in female jobs.
Clearly, specifying that the manager would supervise predominantly females had little impact on the degree of selection bias; however, it did increase the number of times subjects indicated in the open-ended comments that gender of the applicant influenced their selection decision. The lack of differences was surprising between Condition 2 (only applicant gender was known) and Condition 3 (both applicant and subordinates’ genders were known), but it indicates that if the job is female gender-typed, the profemale bias demonstrated when the gender of the job is known is not magnified by the knowledge that the subordinates are also female.
There was no evidence that evaluators would be more likely to select a female for a mid-level job than for a top-level job. In fact, the trend (though not significant) was in the opposite direction. Contrary to our expectations, bias in favor of females in a top-level job was clearly as prevalent as bias in a mid-level job.
It was also interesting that females with management experience were more likely than males or inexperienced females to rank the female applicant first for the lower-level management job. Perhaps these individuals were sensitive to the difficulties women have faced in advancing into management, or perhaps they believed there were advantages to selecting females into traditionally female jobs. This effect, however, did not hold for the top-level job.
In general, the findings that less-qualified females will be selected over more-qualified males in traditionally female jobs are somewhat disconcerting given the attention paid recently to diversity and equal employment opportunity. It is possible that previous attention has been so heavily devoted to discrimination against women and minorities that the issue of equal opportunities has not yet generalized to include males. It is also possible that, while we may be making some progress in moving males and females into their opposite-sex domains, those areas that are clearly sex-typed are still not equally accessible.
On the brighter side, the majority of raters did select the most qualified individual even though that individual was male and the job was clearly female. Perhaps this is an indication that we are moving toward more equality. While we might have anticipated that evaluators with management experience would be more sensitive to basing decisions on objective qualifications, there was some evidence that the opposite was true. Consistent with the findings of Ilgen and Youtz (1986), which suggested that decisions made in artificial settings may be less biased than decisions made in field settings, evaluators with management experience tended to display more profemale bias in decisions for top managers than those without such experience. Perhaps these managers had experienced more successful placements of managers when women were selected for supervisory positions in traditionally female jobs.
The lack of salary differences was somewhat surprising, particularly given that the female applicant was less qualified than the male. However, these results support those of Hitt and Barr (1989), which suggested that managers use different variables in selection decisions and salary decisions. That a less-qualified female would be allocated a salary equivalent to a more-qualified male is inconsistent with salary differences documented in real work settings (cf. Drazin & Auster, 1987). Perhaps being the more appropriate gender for the job increased the worth of the female applicant. Our results certainly showed no evidence that women would receive lower salaries when selected as a manager in a traditionally female job, even when a lower salary could have been justified on the basis of qualifications. The salary results (or lack thereof) should be interpreted with caution, however, as the ranges were quite large given that the current salaries were specified. This may be an indication of error variance.
In terms of male and female raters, there were no differences found in selection or salary allocation decisions as a function of rater gender. While this was contrary to our predictions and to some of the earlier research, it may be an indication that males and females view traditionally female jobs similarly. There was, however, a tendency for males to allocate higher salaries in general than females, though these higher salaries did not favor male or female applicants. This may reflect their higher salary expectations (cf. Jackson & Grabski, 1988).
The implications from this study suggest that we are, in fact, faced with the possibility of “another ceiling” or bias that may keep males from entering management positions in female arenas. Similar to females applying for “male” jobs, males also seem to suffer somewhat when applying for “female” jobs due to perceptions of a “lack of fit” because of their gender. Perhaps one remedy for this is training. Currently, most of the EEO and diversity training focuses directly or indirectly on historical biases that have been levied against women and minorities. If the results of this study generalize to the selection decisions made in the management ranks, males may also confront gender bias in decision making. And the frequency of this bias may increase as more women enter the work force and begin to predominate in numbers in more occupations. Like the complaints we have heard from women and minorities, males also may be undervalued due to their gender. The degree to which this bias operates would be of interest to test in future research. For example, would evaluators be even more likely to select a less-qualified female over a qualified male for a female job if the position were non-managerial? These and other questions pertinent to profemale bias could be the subjects of future studies.
Acknowledgment: The authors thank Susan Pyle and Kathryn Blocher for their assistance in conducting this study, and David Atwater, Cheri Ostroff, Joseph Ryan, David Waldman, and Ella Van Fleet for their helpful comments on this manuscript.
Appendix
Summary of Z-Tests for Testing Hypotheses 1 through 4
Z-Test Sig.
Hypothesis 1.
1. Dependent Variable – Proportion of subjects
selecting female first when Applicant Gender
Unknown
Contrast – Male versus Female Raters n.s.
2. Dependent Variable – Proportion of subjects
selecting female first when Applicant Gender
Known
Contrast – Male versus Female Raters n.s.
3. Dependent Variable – Proportion of subjects
selecting female first when Applicant and
Subordinates’ Genders Known
Contrast – Male versus Female Raters n.s.
Hypothesis 2A
1. Dependent Variable – Proportion of subjects
selecting female first for Dean’s job
Contrast – Applicant Gender Unknown versus
Applicant Gender Known p [less than] .05
2. Dependent Variable – Proportion of subjects
selecting female first for Dean’s job
Contrast – Applicant Gender Unknown versus
both Applicant and Subordinates’ Genders Known p [less than] .05
Hypothesis 2B
1. Dependent Variable – Proportion of subjects
selecting female first for Coordinator’s job
Contrast – Applicant Gender Unknown versus
Applicant Gender Known p [less than] .05
2. Dependent Variable – Proportion of subjects
selecting female first for Coordinator’s job
Contrast – Applicant Gender Unknown versus
both Applicant and Subordinates’ Genders Known p [less than] .05
Hypothesis 2C
1. Dependent Variable – Proportion of subjects
selecting female first when Applicant Gender
is Known
Contrast – Dean versus Coordinator n.s.
2. Dependent Variable – Proportion of
subjects selecting female first when both
Applicant and Subordinate Genders Known
Contrast – Dean versus Coordinator n.s.
Hypothesis 3
1. Dependent Variable – Proportion of subjects
selecting female first for Dean’s job
Contrast – Applicant Gender Known versus both
Applicant and Subordinates’ Genders Known n.s.
2. Dependent Variable – Proportion of subjects
selecting female first for Coordinator’s job
Contrast – Applicant Gender Known versus both
Applicant and Subordinates’ Genders Known n.s
Hypothesis 4
1. Dependent Variable – Proportion of subjects
selecting female first for Dean’s job
Contrast – Rater with Management Experience
versus No Experience p [less than] .05
2. Dependent Variable – Proportion of subjects
selecting female first for Coordinator’s job.
Contrast – Rarer with Management Experience
versus No Experience n.s.
Notes
1. Twenty-eight subjects did not indicate their gender and twenty-two did not complete the question regarding management experience.
2. We obtained the male to female faculty ratios for 20 schools of Social Work in the U.S. The proportions of women ranged from 41% for faculty only to 93% when faculty and staff were considered. We, therefore, concluded that this scenario was acceptable.
3. These materials were reviewed by three independent raters and the most qualified applicants were easily identified, as were those who were under-qualified
4. Following the suggestions of Fromkin and Streufert (1976) and Fisher (1984), three subjects (less than 5%) were dropped from the analyses presented here since they selected one of the much less qualified applicants in the Applicant Gender Known or Applicant’s and Subordinates’ Genders Known conditions as either their first or second choice. These choices were used as distractions and were not pertinent to the purpose of the study. However, all analyses were conducted with and without these three cases. Their omission did not significantly affect the results or conclusions reached.
5. At the suggestion of an anonymous reviewer, we ran a follow-up study to assess whether evaluators assumed applicants were female (because the job was female), even when the gender was not given. We administered a simple post-survey which was identical to the original set of materials for the Dean and Coordinator, when no applicant names were given. Subjects were simply asked to review the materials in the packets and tell us about their assumptions of gender of applicants. Twenty-five subjects completed the Dean’s survey and 26 completed the Coordinator’s survey. For both positions, over 60 percent replied that they did not think about whether the applicants were male or female. Less than 5% for each position assumed most or all of the applicants were female. On the basis of these post-test results, we feel confident that the subjects in our study were not making female gender attributions about applicants when that information was not given.
6. We tested to assess whether the effect for management experience was an artifact of age. The sample was broken into two groups – those 30 and over and those under 30. There was no significant difference in the proportion of raters who selected the female as a function of age.
7. We thank an anonymous reviewer for suggesting that we test for interaction effects.
References
Baker, B., Slonaker, W.M. & Wendt, A.C. (1994). Employment discrimination: Another challenge for the project manager. Project Management Journal, 25(4): 22-25.
Beggs, J. & Doolittle, D. (1993). Perceptions now and then of occupational sex typing: A replication of Shinar’s 1975 study. Journal of Applied Social Psychology, 23: 1435-1453.
Beller, A.H. (1985). Changes in the sex composition of U.S. occupations, 1960-1981. Journal of Human Resources, 20(2): 235-250.
Bernardin, H. & Beatty, R. (1984). Performance appraisal: Assessing human behavior at work. Boston, MA: Kent.
Bernstein, V., Hakel, M. & Harlan, A. (1975). The college student as interviewer: A threat to generalizability? Journal of Applied Psychology, 60: 266-268.
Blau, F. & Ferber, M. (1987). Discrimination – empirical evidence from the United States. American Economic Review, 77: 316-320.
(Business Week, July 31, 1995.)
Cash, T., Gillen, B. & Burns, D. 1977). Sexism and beautyism in personnel consultant decision-making. Journal of Applied Psychology, 62: 301-311.
Cohen, S. & Bunker, K. (1975). Subtle effects of sex role stereotypes on recruiters’ hiring decisions. Journal of Applied Psychology, 60: 566-572.
Cotter, D., DeFiore, J., Hermsen, J., Kowalewski, B. & Vanneman, R. (1995). Occupational gender desegregation in the 1980s. Work and Occupations, 22(1): 3-21.
Dalton, D. & Kesner, I. (1993). Cracks in the glass: The silent competence of women. Business Horizons, 36: 6-11.
Deaux, K. (1984). From individual differences to social categories: Analysis of a decade’s research on gender. American Psychologist, 39: 105-116.
Dipboye, R., Fromkin, H. & Wiback, K. (1975). Relative importance of applicant sex, attractiveness, and scholastic standing in evaluation of job applicant resumes. Journal of Applied Psychology, 60: 39-43.
Dipboye, R., Arvey, R. & Terpstra, D. (1977). Sex and physical attractiveness of raters and applicants as determinants of resume evaluations. Journal of Applied Psychology, 62: 288-294.
Drazin. R. & Auster, E. (1987). Wage differences between men and women: Performance appraisal ratings vs. salary allocation as the locus of bias. Human Resource Management, 26: 157-168.
Feather, N. & Simon, J. (1975). Reactions to male and female success and failure in sex-linked occupations. Journal of Personality and Social Psychology, 31(1): 20-31.
Fisher, C. D. (1984). Laboratory experimentation. Pp. 169-185 in T. S. Bateman & Ferris, G. R. (Eds.), Methods & analysis in organizational research. Reston, VA: Reston Publishing Company.
Freedman, S. & Phillips, J. (1988). The changing nature of research on women at work. Journal of Management, 14(2): 231-252.
Fromkin, H. L. & Streufert, S. (1976). Laboratory experimentation. Pp.415-465 in M.D. Dunnette (Ed.), Handbook of industrial and organizational psychology. Chicago: Rand McNally.
Goldberg, P. (1968). Are women prejudiced against women? Trans-Action, 5: 28-30.
Gutek, B. & Cohen, G. (1987). Sex ratios. sex role spillover, and sex at work: A comparison of men’s and women’s experiences. Human Relations, 40: 97-115.
Haemmerlie, F. & Montgomery, R. (1991). Goldberg revisited – pro-female evaluation bias and changed attitudes toward women by engineering students. Journal of Social Behavior and Personality, 6: 179-194.
Heilman, M. (1980). The impact of situational factors on personnel decisions concerning women: Varying the sex composition of the applicant pool. Organizational Behavior and Human Performance, 26: 386-395.
Heilman, M. (1983). Sex bias in work settings: The lack of fit model. Research in Organizational Behavior, 5: 269-298.
Heilman, M. (1984). Information as a deterrent against sex discrimination: The effects of applicant sex and information type on preliminary employment decisions. Organizational Behavior and Human Performance, 33: 174-186.
Heilman, M., Block, C., Martell, R. & Simon, M. (1989). Has anything changed? Current characterizations of men, women and managers. Journal of Applied Psychology, 74: 935-942.
Hitt, M. & Barr, S. (1989). Managerial selection decision models: Examination of configural cue processing. Journal of Applied Psychology, 74(1): 53-61.
Ilgen, D. & Youtz, M. (1986). Factors affecting the evaluation and development of minorities in organizations. Research in Personnel and Human Resources Management, 4: 307-337.
Jackson, L. & Grabski, S. (1988). Perceptions of fair pay and the gender wage gap. Journal of Applied Social Psychology, 18: 606-625.
Jacobs, J. & Lim, S. (1992). Trends in occupational and industrial sex segregation in 56 countries, 1960-1980. Work and Occupations, 19(4): 450-486.
Jacobson, M., Antonelli, J., Winning, P. & O’Peil, D. (1977). Women as authority figures: The use and nonuse of authority. Sex Roles, 3: 365-375.
Kanter, R. (1977). Men and women of the corporation. NY: Basic Books.
Levinson, R. (1975). Sex discrimination and employment practice: An experiment with unconventional job inquiries. Social Problems, 22: 533-543.
Lewis, S. (1991). Breaking through the glass ceiling. Planning, 57: 7-13.
Major, B. & Konar, E. (1989). Predicting career-entry pay expectations: The role of gender-based comparisons. Journal of Business and Psychology, 3: 331-340.
Morrison, A. & Von Glinow, M. (1990). Women and minorities in management. American Psychologist, 45: 200-208.
Morrison, A., White, R. & Van Velsor, E. (1987). Breaking the glass ceiling. Reading MA: Addison-Wesley.
Mullins, T. (1982). Interviewer decisions as a function of applicant race, applicant quality, and interviewer prejudice. Personnel Psychology, 35: 163-174.
Nelson-Horchler, J. (1991). The best man for the job is a man. Industry Week, 240: 50-52.
Nilson, L. (1976). The occupational and sex-related components of social standing. Sociology and Social Research, 60: 328-336.
Nivea, V. & Gutek, B. (1981). Women and work: A psychological perspective. New York: Praeger.
O’Neill, J. (1992). A flexible work force: Opportunities for women. Journal of Labor Research, 13(1): 67-72.
Pazy, A. (1986). The persistence of pro-male bias despite identical information regarding causes of success. Organizational Behavior and Human Decision Processes, 38: 366-377.
Pazy, A. (1992). Sex-linked bias in promotion decisions – the role of candidates’ career relevance and respondents prior experience. Psychology of Women Quarterly, 16: 209-228.
Pfeffer, J. (1990). Gender-based wage differences: The effects of organizational context. Work and Occupations, 17: 55-78.
Pfeffer, J. & Davis-Blake, A. (1987). The effect of the proportion of women on salaries: The case of college administrators. Administrative Science Quarterly, 31: 1-24.
Powell, G. & Butterfield, D. (1994). Investigating the “glass ceiling” phenomenon: An empirical study of actual promotions to top management. Academy of Management Journal, 37: 68-86.
Reskin, B. & Ross, C. (1992). Jobs, authority, and earnings among managers. Work and Occupations, 19: 342-365.
Rose, G. & Andiappan, P. (1978). Sex effects on managerial hiring decisions. Academy of Management Journal, 21: 104-112.
Rosen, B. & Jerdee, T. (1974). Influence of sex-role stereotypes on personnel decisions. Journal of Applied Psychology, 59: 9-14.
Rosen, B., Rynes, S. & Mahoney, T. (1983). Compensation, jobs and gender. Harvard Business Review, 61: 170-172.
Rosenstein, J. & Hitt, M. (1986). Experimental research on race and sex discrimination: The record and prospects. Journal of Occupational Behavior, 7: 215-226.
Sandler, B. & Hall, M. (1986). The campus climate revisited: chilly for women faculty, administrators, and graduate students. Washington, D.C.: Project on the Status of Education of Women, Association of American Colleges.
Stevens, G. & DeNisi, A. (1980). Women as managers: Attitudes and attributions for performance by men and women. Academy of Management Journal, 23: 355-361.
Stover, D. (1994). The horizontal distribution of female managers within organizations. Work and Occupations, 21(4): 385-402.
Taber, C. (1992). From a glass slipper to a glass house. American Advertising, Fall: 20-22.
Terborg, J. & Ilgen, D. (1975). A theoretical approach to sex discrimination in traditionally masculine occupations. Organizational Behavior and Human Performance, 13: 352-376.
Tosi, H. & Einbender, S. (1985). The effects of the type and amount of information in sex discrimination research: A meta-analysis. Academy of Management Journal, 28: 712-723.
Tsui, A. & O’Reilly, C. (1989). Beyond simple demographic effects: The importance of relational demography in superior-subordinate dyads. Academy of Management Journal, 32: 402-423.
Van Fleet, D. & Saurage, J. (1984). Recent research on women in management. Akron Business and Economic Review, Summer: 15-24.
Wharton, A. & Baron, J. (1987). So happy together?: The impact of gender segregation on men at work. American Sociological Review, 52: 574-587.
Williams, C. (1992). The glass escalator: Hidden advantages for men in the “female” professions. Social Problems, 39(3): 253-267.
COPYRIGHT 1997 JAI Press, Inc.
COPYRIGHT 2004 Gale Group
