gender-based influences: the promotion recommendation

16
Sex Roles, Vol. 25, Nos. 5/6, 1991 Gender-Based Influences: The Promotion Recommendation Sandra J. Hartman 1 University of New Orleans Rodger W. Griffeth George Mason University Michael D. Crino Clemson University O. Jeff Harris Northeast Louisiana University This study used a between-subjects analysis of variance design (2 x 2 x 2 x 2) to estimate the effects of rater gender, ratee gender, the gender stereotype of the job, and the gender stereotype of the ratee's personal characteristics on a promotion decision. The results indicated that female employees with masculine characteristics were evaluated as most promotable regardless of the gender stereotype of the job or the gender of the rater. We also investigated the personal characteristics raters reported using in making their decisions. We asked raters to select ratee personal characteristics that most influenced their promotion decisions. Discriminant analyses showed that male and female raters approached the rating process differently and that it is predominantly the gender stereotype of the ratee's personal characteristics rather than the ratee's gender that influences the promotion process. Considerable research associated with selection and performance evalu- ation decision processes focuses on whether ratee and rater gender have 1To whom correspondence should be addressed at Department of Management, University of New Orleans, New Orleans, LA 70148. 285 0360-0025191/0900-0285506.50/0 © 1991 PlenumPublishing Corporation

Upload: ohio

Post on 15-May-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Sex Roles, Vol. 25, Nos. 5/6, 1991

Gender-Based Influences: The Promotion Recommendation

Sandra J. Hartman 1

University of New Orleans

Rodger W. Griffeth

George Mason University

Michael D. Crino

Clemson University

O. Jeff Harris

Northeast Louisiana University

This study used a between-subjects analysis of variance design (2 x 2 x 2 x 2) to estimate the effects of rater gender, ratee gender, the gender stereotype of the job, and the gender stereotype of the ratee's personal characteristics on a promotion decision. The results indicated that female employees with masculine characteristics were evaluated as most promotable regardless of the gender stereotype of the job or the gender of the rater. We also investigated the personal characteristics raters reported using in making their decisions. We asked raters to select ratee personal characteristics that most influenced their promotion decisions. Discriminant analyses showed that male and female raters approached the rating process differently and that it is predominantly the gender stereotype of the ratee's personal characteristics rather than the ratee's gender that influences the promotion process.

Considerable research associated with selection and performance evalu- ation decision processes focuses on whether ratee and rater gender have

1To whom correspondence should be addressed at Department of Management, University of New Orleans, New Orleans, LA 70148.

285

0360-0025191/0900-0285506.50/0 © 1991 Plenum Publishing Corporation

286 I lartman et al.

a systematic impact on what should be an unbiased and rational decision- making process. Many of these studies have reported gender effects, with females typically receiving lower subjective evaluation ratings (e.g., Bartol & Butterfield, 1976; Hamner, Kim, Baird, & Bigoness, 1974; Izraeli & Izraeli 1985; Lord, Phillips, & Rush, 1980; Rice, Instone, & Adams, 1984). We propose that some of the effects that have been noted may be the result of factors other than the direct influence of bias. Specifically, in this study we consider whether differences in ratings of females may come about through raters' attempts to achieve match between ratee gender, ratee per- sonal characteristics, and stereotypical ideas of job requirements. In keep- ing with recent interest in cognitions and their relationship to behavioral intentions (e.g., Lesser & Mehdizadeh, 1988; Sudman & Schwartz, 1989), we consider how rater cognitions contribute to this process.

Several factors have led us to propose that bias may operate indi- rectly, through a matching process. First, we considered studies that have reported "main-effect" bias: differences among male and female raters, or in the rating of males and females. Among these, even the question of whether male and female raters differ in the way they evaluate a given job applicant remains unanswered. On the one hand, Sackett and Wilson (1982) and Hinricks (1978) suggest that all raters share common stereo- types of those characteristics required for job success. However, Oliphant and Alexander (1982) report evidence that, in comparison to males, females are stricter in their judgments, while others (Lord et al., 1980; Rose & Andiappan, 1978) find them more lenient.

Results examining bias against female ratees have also been incon- clusive. Early research generally reported bias against women (e.g., Day & Stogdill, 1972; Fidell, 1970; Rosen & Jerdee, 1974; Terborg & Ilgen, 1975). More recently, studies have found no differences (e.g., Graves & Powell, 1988; Lewis, 1986; Rice, Instone, & Adams, 1984). Other recent studies have reported differences as occurring, but only under certain circum- stances, such as where minimal information is available (e.g., Dobbins, Cardy, & Truxillo, 1988; Heilman, 1984; Heilman & Stopeck, 1985). Other research has continued to report differences but has attributed them to measurement problems (Neumark, 1988). These findings lead us to believe if a rater or ratee gender effect on evaluations could be isolated in a con- trolled environment, this effect would be minimal.

Our review of the literature leads us to discredit earlier reports of bias occurring as a main effect where, for example, preference is given to males regardless of the situation. However, there appears evidence that biases exist, but that they operate in a more complex manner than was assumed in the earlier literature. If this is the case, then study is needed to isolate the circumstances under which bias occurs. In this research, we

Gender-Based Influences 287

examine what may happen as raters attempt to match personal charac- teristics, gender, and perceived job requirements.

Personal Characteristics

One opportunity for bias may occur through the processes raters use to achieve a suitable match between ratee personal characteristics and ratee gender, for example. Note that this may happen even though male and female raters do not differ in leniency or stringency. We find that some studies suggest male and female raters share common stereotypes (Bartol & Butterfield, 1979; Schein, 1973, 1975), and ratee gender and/or personal characteristics influence both. Some research suggests that a contrast effect may exist, where females with stereotypically masculine characteristics would be overselected, while those with stereotypically female charac- teristics would be underselected (Abramson, Goldberg, Greenberg, & Abramson, 1978; Goktepe & Schneier, 1989; Heilman & Stopeck, 1985; Lee & Alvarez, 1977; Nieva & Gutek, 1980).

However, there is evidence for the opposite effect: females with ster- eotypically feminine traits have been rated more favorably than their more masculine counterparts (Bartol & Butterfield, 1976; Rice et al., 1984; Riger & Galligan, 1980). Overall, this research suggests that stereotypical male and female personal characteristics interact with ratee gender to influence evaluation decisions. However, the inconsistency of the reported results suggests that other factors besides ratee gender and personal characteristics influence decisions. One of these factors may be stereotypes about job char- acteristics.

Perceived Job Requirements and Gender Stereotype of Jobs

Rosenstein and Hitt (1986), pointing to the importance of gender stereotyping of jobs, conclude that female applicants face negative access discrimination when the job is male stereotyped and positive access dis- crimination when it is female stereotyped. Similarly, Jackson, Peacock, and Holden (1982) found that raters agreed on the characteristics required for success in job types such as carpenter or librarian, although different traits were believed necessary for success in each job. Extending this idea some- what, Cannings (1988) found evidence for higher promotion rates for fe- males where job type and background were both "appropriate" (i.e., female with training in social work--a traditionally female-dominated occupa- t i on -who was working in the field of social work).

288 llartman et aL

What may be in operation is a matching process suggested by Dipboye (1985), who pointed out that raters may attribute to an individual ratee char- acteristics consistent with their stereotype of persons similar to the ratee and attribute to a particular position requirements consistent with their stereotype of successful occupants of the position: "For instance, some jobs are consid- ered 'man's work' requiring masculine characteristics . . . " (p. 117).

Match of Ratee Gendel, Personality Characteristics, and Job

This reasoning suggests that raters may have a "mental image" or "set" of the personal characteristics associated with the kinds of jobs in which a male or female ratee is likely to succeed. An unanswered question is specifically how the ratee's personal characteristics, gender, and gender stereotyping of the job may interact to influence the selection decision. The limited evidence available suggests that raters attempt to match ratee gen- der and personal characteristics to perceived job requirements. Where the three factors match, the rater will be more likely to promote the ratee. Conversely, where one or more factors do not match, promotion will be less likely. Following from this discussion, our first hypothesis is as follows:

H1. Raters will be more likely to promote females with female personal characteristics where the job demands are perceived as stereotypically female. Conversely, they will be more likely to promote males with male personal characteristics where the job is stereotypically male. If there is lack of match among any of the factors, promotion will be less likely. This effect will occur regardless of rater gender.

The Role of Cognitions

In the preceding discussion we have suggested that raters may strive for an appropriate match between ratee gender, ratee personal charac- teristics, and job demands, and have emphasized that the process is a cog- nitive one. Underlying this argument is the renewed interest in cognitive processes in the recent literature, and specifically, in examining the rela- tionship between cognitive processes and behavioral intentions of various kinds (see, for example, Homer & Kahle, 1988; Ravlin & Meglino, 1987). In this research, we make an initial attempt to examine one area where the perceptions that raters bring to the rating process may influence the promotion decision. We are interested, therefore, in examining not only whether raters do, in fact, successfully make the hypothesized match, but

Gender-Based Influences 289

also, in terms of their cognitive processes, how such matching may take place. Specifically, a second research objective of this study is to examine rater perceptions about their use of ratee personal characteristics in the rating process.

There are several possible questions to be asked. Do male raters re- port that they use different personal characteristics than do female raters? Are different personal characteristics judged important for female ratees when compared to males? Are different personal characteristics judged to be important for success in stereotypically male jobs when compared to stereotypically female jobs? What is the correspondence between cogni- tively based judgments and behaviors? In line with our first hypothesis, as- suming that a matching process is in effect, when cognitions are examined, raters should report that they use information differently in different cir- cumstances. For example, the rated importance of any given personal char- acteristic should be different depending on ratee gender and job type. Based on these ideas, our second hypothesis is as follows:

H2. Gender of ratee and gender stereotype of the job will influence cognitions, as indicated by raters' reports of their use of ratee personal characteristics. There will be differences in which characteristics are judged most important in the decision-making process depending on whether the ratee is male or female, and on whether the job is stereotypically male or female.

METHODOLOGY

Pretest Session

An earlier experiment served as a pretest for two of the measures used in this study (Hartman, Crino, Harris, & Griffeth, 1985). Student raters (n = 33) from business administration classes at a large Southern university were presented with sets of personal attributes, ostensibly of can- didates for promotion. The attributes were taken from assessment center materials used by AT&T and the associated telephone companies. The at- tributes themselves were initially developed by Bray and his associates in their pioneering management progress study, and shown to be related to success in managerial positions over a 15-year period (Bray, Campbell, & Grant, 1974). Four different combinations of the attributes (in terms of high and low values) were developed to be either masculine (two sets) or feminine (two sets) stimuli. This strategy was based on research that iden-

290 llartman et aL

Table 1. Personal Attributes Used in Candidate Profiles

Level a

Attribute Male Female

Scholastic aptitude A AA Forcefulness A BA Perception H H Adaptability H AA Need approval of peers BA AA Managerial identification AA A Range of interests L A Organizing and planning H H Leadership skill H BA Oral communication skill H A Likability BA AA Written communication skill L H Self-objectivity A BA Need approval of superiors BA H Inner work standards BA H Resistance to stress AA L Energy AA L Decision making BA BA

aL: low; BA: below average; A: average; AA: above average; H: high.

tiffed certain attributes as masculine and others as feminine (Bern, 1974; Ritchie & Moses, 1983; Schein, 1973, 1975).

No names were attached to the resulting four stimuli. They were pre- sented to the student raters who were asked to judge whether each was a male or a female, and how confident they were of their judgment. Chi- square analyses of the resulting classifications indicated that raters consis- tently classified two of the stimuli as males and two as females (:~2, p < .001), and that the classifications were consistent with their hypothesized masculinity or femininity. Rater confidence was in the 51-75% range. Table I provides the two stimuli--one with masculine and one with feminine char- acteristics-selected for use in this study.

Subjects were also given job descriptions of four entry-level telephone company supervisory jobs. Although the jobs differed in content, they were classified at the same level and paid at the same rate (installation super- visor, word processing supervisor, assignment supervisor, and commercial supervisor). The jobs differed along the dimensions that Tienda, Smith, & Ortiz (1987) provide as female-typed vs. male-typed jobs (i.e., female-typed jobs deal with clerical/technical concerns, while male-typed jobs may involve crafts or construction). Subjects were asked to assign each of the four stim-

Gender-Based Influences 291

uli to a different job, in terms of their judgments about which sets of at- tributes best suited each job. They were also asked to indicate the charac- teristics of an ideal candidate for the job, using the Bern Sex Role Inventory (Bern, 1974). Two of the jobs, installation supervisor and word processing supervisor, were consistently assigned either masculine or feminine profiles respectively (Z2, p < .05), respectively. Furthermore, they were consistently described as masculine or feminine based upon responses to the Bern in- strument [one-way analysis of variance (ANOVA) of inventory scores, F(7,256), p < .001].

The two job descriptions and two stimuli that resulted from the first pretest were used in a second pretest, also involving student raters (n = 15). Again, there was no indication of gender on the stimulus materials, and raters were asked to classify the paper people as males or females and to assign them to jobs. The jobs were again classified with the Bern inven- tory as well. All results were consistent with the first pretest, and the job descriptions and stimuli were used in the experimental session.

Experimental Session

Sample. Upper division undergraduate business students at the same Southern metropolitan university served as raters. A total of 320 subjects, 158 men and 162 women, participated. The sample's average age was 24.0 years (range = 17-43), and nearly 72% were employed at the time of the study. Of this number, 79.5% worked in nonmanagerial positions while the remainder worked in managerial positions.

Stimulus Materials, Experimental Procedures, and the Dependent Measure. Using random assignment, subjects received one of the two stimu- lus sheets and detailed written information describing the 18 attributes. Gender of the stimulus person was manipulated by assigning the names of either Mary or Tom to the attribute sheet received by the subject. Finally, subjects received a detailed job description of either the word processing supervisor (female job) or installation supervisor (male job). Each subject was asked to make a promotion decision using the following scale: (4) more than acceptable for promotion, (3) acceptable, (2) questionable, (1) not ac- ceptable. After subjects made their promotion decisions, we asked them to list the three attributes that were most important to them in making the promotion decision.

Analyses. We used ANOVA to test the first hypothesis, that a pro- motion recommendation would be more likely to occur when ratee gender, personal characteristics, and job gender stereotype match. Although exam- ined, rater gender was not expected to enter into the recommendation.

292 l l a r t m a n et aL

The result was a 2 x 2 × 2 × 2 between-subjects factorial design (rater gender, ratee gender, personal characteristics profile, and gender stereotype of the job).

We used discriminant function analysis (DFA) to test the second hy- pothesis, dealing with the relationship of rater cognitions to the rating pro- cess. We investigated whether male and female raters reported that they used information differently, or whether they reported that they used in- formation differently depending on whether a ratee is male or female. To conduct the discriminant analysis (DA), we created an 18 Characteristics x 320 Subjects data matrix. For any given subject, a row of variables con- sisted of 15 zeros (characteristic not listed as important) and 3 ones (char- acteristic listed as important).

The rationale for the use of DFA was that if a discriminant function successfully predicted group membership (for example, predicted male vs. female raters on basis of their use of attributes), significant differences in the use of information were being reported. We interpreted the structure correlations because they are more valid than discriminant weights (Hair, Anderson, & Tatham, 1987).

We used a stepwise entry method that selects variables for entry into the analysis on the bias of their discriminating power. We believe not all variables would be equally useful in discriminating among groups and that multicollinearity among variables was likely. Green recommends the step- wise procedure for such cases (Green, 1978). The entry method used was Rao's Vwith a probability of F for entry or removal of .10, as recommended by Green (1978). Approximately two-thirds of the subjects (n = 204) were randomly selected to develop the functions and the balance (n = 106) served as the holdout sample.

RESULTS

In H1, we predicted a triple-order interaction among ratee gender, personal characteristics profile, and gender stereotype of the job. We did not find this interaction, however, indicating that the matching process did not take place as expected. We did, however, find several significant effects that are related to our previous discussion and to H1.

As expected, main effects for gender of rater and gender of ratee were not significant, indicating that male and female raters in this study did not differ in harshness or leniency (i.e., males and females assigned roughly equal ratings to ratees as a whole), and that neither males nor females were preferred. We found significant main effects for personality characteristics [male M = 2.67, female M = 2.06; F(1,318 = 70.871, p <

Gender-Based Influences 293

.01] and gender stereotype of job [male M = 2.48, female M = 2.27; F(1,318) = 9.472, p < .01]. Thus, applicants for the male job were more likely to receive the promotion recommendation than those who applied for the female job. Applicants with masculine personality characteristics (regardless of their gender) were more likely to be recommended; however, this finding should be interpreted by considering a Ratee Gender x Per- sonality Characteristics interaction described next.

The only other significant result was the significant Ratee Gender x Personality Characteristics interaction [F(1,318) = 7.985, p < .01]. This interaction is of particular interest considering the nonsignificant ratee gen- der main effect. Group means suggest that while raters assign the same average ratings to males and females with masculine personality charac- teristics (Ms = 2.59 and 2.79, respectively), they evaluated female ratees with masculine personality characteristics more favorably (M = 2.79) than females having female personality characteristics (M = 1.92). Based on the 4-point promotion scale used, raters recommended females with masculine characteristics for promotion but not those with feminine characteristics. Moreover, Scheff6 post hoc tests showed this later difference was significant (p < .01). Thus, while neither males nor females are preferred in overall terms, females with male profiles are overrecommended relative to females with female profiles. No such effect appears to occur for males.

Regarding H1, this effect suggests that raters attempt to match gender to characteristics for female but not for male ratees. Moreover, they do not extend the process to include job characteristics. As a result, the match is inappropriate, in that raters preferred females with masculine charac- teristics, regardless of situation.

Our second hypothesis dealt with cognitions. We suggested that rater reports about their use of information should parallel their observed deci- sions. Using rater reports of most important variables, we computed DFA for three of the main effects examined in the ANOVA: gender of rater, ratee gender, and gender stereotyping of job. Recall that, in the ANOVA, main effects for gender of ratee and gender of rater were not significant, suggesting no overall difference in preference for male vs. female ratees, or in rater harshness or leniency. However, it is still possible that raters attended to different information in the evaluation process. Therefore, our discriminant analysis, involving raters' perceptions about their use of infor- mation, was designed to investigate this possibility.

The function for ratee gender was not significant. We do not report the DFA for the profile of personality characteristics because the discriminant function for profile asks whether raters used characteristics differently under different circumstances. Because weighting of attributes was, in fact, different for the two stimulus profiles, we regard the discriminant analysis, while sig-

294 l l a r tman et al.

Table II. Discriminant Analyses for Rater Gender and Gender Stereotyping of the Job a

Male vs. female raters

Cannonical Function Eigenvalue correlation 2 df Male Female

1 .03 .17 5.9 b 2 .18 -.17

Structure Standardized Variable matrix discriminant weights

Written ~ommunication ska'll .73 .84 Decision making -.56 -.69

Male vs. female gender-stereotyped job

Group centroids c (means)

Cannonical Function Eigenvalue correlation Z 2 df Male Female

1 .04 .19 7.8 b 2 .20 - . 18

Structure Standardized Variable matrix discriminant weights

Organizing and planning .71 .72 Leadership -.70 -.71

Group centroids c (means)

aOnly variables with structure coefficients > .30 are presented. bp < .05. CCoded 1 = male, 2 = female.

nificant (p < .01), as a manipulation check. Table II presents the DFA results for rater gender and gender of job, both of which were significant.

The result reported in the top half of Table II presents the DFA for male vs. female raters, showing that the function was significant. The struc- ture coefficients for written communication skill (.73) and decision making (-.56) define this function. Using structure coefficients and group centroids as interpretive tools, we see that decision making is associated with male raters (group centroid = .18) and written communication skill with female raters (group centroid = -.17).

The result reported in the bottom half of Table II is the DFA for male vs. female gender-stereotyped jobs. The structural coefficients pre- sented in the table show that the organizing and planning (.71) and lead- ership (-.70) characteristics define this function. Taken together, using the structure coefficients and group centroids as interpretive tools, what is in- dicated is that raters used organizing and planning when selecting for the female gender-stereotyped job, and leadership for the male job.

Finally, discriminant analysis permitted further examination of the sig- nificant Gender of Ratee × Profile interaction found in the ANOVA. We

Gender-Based Influences

Table III. Four-Group Discriminant Analysis: Male Candidates with Masculine and Feminine Profiles; Female Candidates with Masculine

and Feminine Profiles

295

Cannonical Function Eigenvalue correlation 2~2 df

1 .63 .62 114.43 b 21

Standardized Structure discriminant

Variable matrix weights

Oral communication skill .38 .77 Written communication skill .37 .57 Likability .38 .62

Group Group centroids (means)

1 (Male, male profile) 0.73 2 (Male, female profile) -0.63 3 (Female, male profile) 0.70 4 (Female, female profile) -1.05

aOnly variables with structure coefficients < .30 are presented. bp < .01,

created a four-group discriminant analysis from the 2 x 2 classification (male vs. female profile and male vs. female ratee). We report the results in Table III.

Table III shows that only one significant (p < .01) discriminant func- tion emerged. Inspection of group centroids shows that Groups 1 and 3 (Group 1 consists of males with a masculine profile and Group 3 consists of females with a masculine profile) have similar centroid scores (Group 1 = .73; Group 3 = .70). Groups 2 and 4 (males with a female profile and females with a female profile respectively), although similar in sign (nega- tive), differ in magnitude (-.63 and -1.05, respectively). These results show that oral and written communication skills and likability were used by raters in making their promotion decision of candidates with male characteristics. With Groups 2 and 4, there is evidence that gender does make a difference. Thus, the profile of personality characteristics serves to separate the groups, but there is evidence of an interaction in that only under the female profile does gender make a difference.

DISCUSSION

Recall that H1 considers that raters search for match between ratee gender, personal characteristics (profile), and job characteristics. We ex-

296 l la r tman et al.

pected that promotion would be more likely where there was such a match. Our results do not support this hypothesis. Instead, they suggest that job type does not enter into the decision-making process, and that the match between the remaining factors--personal characteristics and ratee gen- d e r - o p e r a t e s quite differently than expected.

Use of job information was not as hypothesized, and we saw no evi- dence of a search for a match between personality characteristics and job requirements. This finding is in contrast to results from the pretest, where there was evidence that raters did make an effort to match personality char- acteristics to perceived job requirements. In the pretest, however, raters were not told the gender of the applicant, while in this experiment gender was given. Apparently, knowing the gender of the candidate actually served to inhibit raters' ability or willingness to attempt to match characteristics to perceived job demands.

In this study, raters used job information only to the extent that they were more likely to recommend candidates for the masculine job for pro- motion. The literature includes no direct evidence to suggest why ratees receive higher ratings when applying for masculine jobs. Perhaps the tra- ditional managerial job is seen as masculine and raters feel more comfort- able in evaluating candidates for such jobs.

The significant Ratee Gender x Ratee Personality Characteristics Profile interaction indicates that even though, on the average, raters do not prefer either males or females, they do have different expectations of the characteristics required of male and female managers. The interaction suggests that while raters prefer the masculine profile for all ratees, this preference is especially pronounced for females. A female with a masculine profile was more strongly preferred and a female with a feminine profile more strongly rejected than her male counterpart. Apparently, raters be- lieved to be successful a woman should have masculine characteristics, while having such characteristics is not necessary for a man. This finding is consistent with the literature arguing for a contrast effect, in which a woman exhibiting out-of-character behavior will be favored (Nieva & Gutek, 1980; Schein, 1973). Considering these findings, we suggest that raters do take personal characteristics into consideration in the rating pro- cess, but that they do not use them to achieve a match with job demands. Instead, particularly for female ratees, those with male characteristics are preferred while those with female characteristics are rejected.

Discriminant analysis was used to test our second hypothesis, involv- ing the role of cognitions. H2 proposed that gender of ratee and gender stereotype of the job wilt influence raters' use of ratee personal charac- teristics. We expected differences in which characteristics were judged most important in the decision-making process depending on whether the ratee

Gender-Based Influences 297

is male or female, and on whether the job is stereotypically male or female. The discriminant analysis results are consistent with these findings in sev- eral respects.

The function for gender stereotyping of the job was significant, and can be thought of as corresponding to and providing information about the decision-making process that leads to the significant job main effect in the ANOVA. Raters appear to attend to somewhat different information in dealing with the two jobs. Only two personal characteristics entered into the function, suggesting that raters may not be making use of a sufficient number of ratee characteristics to permit them to make an adequate match to job demands. Additional research is needed here. We asked subjects to identify the three personal characteristics they used in making their deci- sions. What would be the outcome if we had asked for more than three (e.g., five or ten)? Also, what would be the outcome if we had asked sub- jects to provide their own idiosyncratic cognitive influences? This experi- ment represented an important first step in identifying the cognitive influences in the rating process.

In general, ANOVA results indicate that while raters do not differ in leniency or stringency, they prefer ratees applying for masculine jobs, and that while they do not prefer male or female ratees in overall terms, they match in the sense of preferring females with masculine profiles. H2 suggests that rater cognitions should mirror and help explain these findings.

Examination of the four-group discriminant function for masculine vs. feminine profile and male vs. female ratees provides information about the Profile x Ratee Gender interaction. While that function does offer some evidence for an interaction, the primary influence upon raters is the gender stereotype of ratee personal characteristics. Overall, evidence from this study indicates that preference for traditionally masculine personal characteristics, rather than ratee gender, underlies the decisions made.

For the correspondence between the cognitive data and actual rater behaviors, DFA results detected some differences in cognitions that do not manifest themselves in the ANOVA. In contrast to the nonsignificant main effect for raters in the ANOVA, which suggested no difference in leniency, the corresponding DFA was significant. This finding suggests that there may be some differences in approach to the rating problem by males and females, although such differences apparently do not affect rater leniency.

SUMMARY

This study has examined several aspects of the rating process: whether there are common stereotypes that male and female raters share, whether

298 Hartman et al.

raters use different information for male and female ratees, and whether job information enters into the decision-making process. Evidence from this study suggests that while male and female raters arrive at equal decisions in that neither group gives evidence of leniency or stringency, they may approach the process somewhat differently. Raters appear to share com- mon stereotypes in their preferences for the male rather than the female gender-stereotyped jobs and for the male rather than the female profile. There were no differences in the ways raters use information when evalu- ating male and female ratees, but all raters appear to prefer female ratees with a masculine profile. This finding argues for a contrast effect and for use of common stereotypes by raters. Use of job information remains un- clear. There is little evidence for rater match between job demands and ratee characteristics.

We believe our experiment is useful because it has demonstrated how objective and cognitively oriented measures may be used to aid in under- standing the decision-making process. Of interest is our use of attributes and job descriptions from an ongoing organizational decision-making set- ting.

However, this research is limited in that it has dealt with only one narrowly defined aspect of the promotion recommendation process. Finally, since the focus of this study was on the universal aspects of decision making rather than on the specifics of the promotion process (we used the pro- motion recommendation as the organizational context), use of the labora- tory and college students, in the present context, is probably acceptable. However, additional research is needed to examine the promotion process in actual organizations.

REFERENCES

Abramson, P. E., Goldberg, P. A., Greenberg, J. I-I., & Abramson, U. M. (1978). The talking platypus phenomenon: competency ratings as a function of sex and professional status. Psychology of Women Quarterly, 2, 114-124.

Bartol, K. M., & Butterfield, D. A. (1976). Sex effects in evaluating leaders. Journal of Applied Psychology, 61, 446-454.

Bern, S. L. (1974). The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology, 42(2), 155-162.

Bray, D. W., Campbell, R. J., & Grant, D. L. (1974). Formative years hi business. New York: Wiley.

Cannings, K. (1988). Managerial promotion: The effects of socialization, specialization, and gender. Industrial and Labor Relations Review, 42, 77-88.

Day, D. R., & Stogdill, R. M. (1972). Leader behavior of male and female supervisors: A comparative study. Personnel Psychology, 25, 353-360.

Dipboye, R. L. (1985). Some neglected variables in research on discrimination in appraisals. Academy of Management Review, 10, 116-127.

Gender-Based Influences 299

Dobbins, H. G., Cardy, L. R., & Truxillo, M. D. (1988). The effects of purpose of appraisal and individual differences in stereotypes of women on sex differences in performance ratings: A laboratory and field study. JomTzal of Applied Psychology, 3, 331-558.

Fidell, L. S. (1970). Empirical verification of sex discrimination in hiring practices in psychology. Amel~can Psychologist, 25, 1094-1098.

Goktepe, R. J., & Schneier, E. C. (1989). Role of sex, gender roles, and attraction in predicting emergent leaders. Journal of Applied Psychology, 74, 165-167.

Graves, H. L., & Powell, N. G. (1988). An investigation of sex discrimination in recruiters' evaluations of actual applicants. Journal of Applied Psychology, 3, 20-29.

Green, P. E. (1978). Analyzing multivariate data. Hinsdale, IL: The Dryden Press. Hair, J. F., Anderson, R. E., & Tatham, R. L. (1987). Multivariate data analysis: With readhlgs.

New York: Macmillan Publishing Co. Hamner, W. C., Kim, J. S., Baird, L., & Bigoness, W. J. (1974). Race and sex as determinants

of ratings by potential employers in a simulated work sampling task. Journal of Applied Psychology, 56, 705-711.

Hartman, S. J., Crino, M., Harris, O. J., & Griffeth, R. (1985). Importance of ratee personality characteristics: Rater nominations of most important characteristics for male and female managers and for jobs of different types. Proceedings, Southern Management.

Heilman, M. E. (1984). Information as a deterrent against sex discrimination: The effects of applicant sex and information type on preliminary employment decisions. 01~anizational Behavior and Human Performance, 38, 174-186.

Heilman, M. E., & Stopeck, H. M. (1985). Being attractive, advantage or disadvantage? Performance-based evaluations and recommended personnel actions as a function of appearance, sex and job type. Ot~ganizational Behavior and Human Decision Processes, 35, 202-215.

Hinricks, J. R. (1978). An eight-year follow-up of a management assessment center. Jout71al of Applied Psychology, 63, 596-601.

Homer, P. M. & Kahle, L. R. (1988). A structural equation test of the value-attitude-behavior hierarchy. Journal of Personality and Social Psychololo; 54, 638-646.

Izraeli, D. N., & Izraeli, D. (1985). ~ex effects in evaluating leaders: A replication study. Journal of Applied Psychology, 70, 340-346.

Jackson, D. N., Peacock, A. C., & Holden, R. R. (1982). Professional interviewers', trait inferential structures for diverse occupational groups. Organizational Behavior and Human Performance, 29, 1-20.

Lee, D. M., & Alvarez, K. M. (1977). Effects of sex on description and evaluation of supervisory behavior in a simulated industrial setting. Journal of Applied Psychology, 62, 405-410.

Lesser, J., & Mehdizadeh, M. (1988). Contribution of cognitive theories of psychology to economic systems. Psychology: A Journal of Haman Behavior, 25, 19-27.

Lewis, G. B. (1986). Gender and promotions. Joarnal of Human Resources, 21, 406-419. Lord, R. G., Phillips, J. S., & Rush, M. C. (1980). Effects of sex and personality on perceptions

of emergent leadership, influence, and social power. Journal of Applied Psychology, 65, 176-182.

Neumark, D. (1988). Employers' discriminatory behavior and the estimation of wage discrimination. Journal of Human Resources, 23, 279-295.

Nieva, V. F., & Gutek, B. A. (1980). Sex effects on evaluation. Academy of Management Rev&w, 5, 267-276.

Oliphant, V. N., & Alexander, E. R., Ill. (1982). Reactions to resumes as a function of resume determinateness, applicant characteristics, and sex of raters. Personnel Psychology, 35, 829-842.

Ravlin, E. C., & Meglino, B. M. (1987). Effect of values on perception and decision making: a study of alternative work values measures. Joto71al of Applied Psychology, 72, 666-673.

Rice, R. W., Instone, D., & Adams, J. (1984). Leader sex, leader success, and leadership process; two field studies. Journal of Applied Psycholo~; 69, 12-31.

Riger, S., & Galligan, P. (1980). Women in management: An exploration of competing paradigms. American Psychologist, 35, 902-910.

300 l lar tman et aL

Ritchie, R. J., & Moses, J. L. (1983). Assessment center correlates of women's advancement into middle management: A 7-year longitudinal analysis. Journal of Applied Psychology, 68, 227-231.

Rose, G. L., & Andippan, P. (1978). Sex effects on managerial hiring decisions. Academy of Management Journal, 21, 104-112.

Rosen, B., & Jerdee, T. H. (1974). Effects of applicant's sex and difficulty of job of candidates for managerial positions. Journal of Applied Psychology, 59, 511-512.

Rosenstein, J., & Hitt, A. M. (1986). Experimental research on race and sex discrimination: The record and the prospects. Journal of Occupational Behaviol, 7, 215-226.

Sackett, P. R., & Wilson, M. A. (1982). Factors affecting the consensus judgment process in managerial assessment centers. JourlTal of Applied Psychology, 67, 10-17.

Schein, V. E. (1973). Relationship between sex role stereotypes and requisite management characteristics. Journal of Applied Psychology, 57, 95-100.

Schein, V. E. (1975). Relationship between sex role stereotypes and requisite management characteristics among female managers. JoutTtal of Applied Psychology, 60, 340-344.

Sudman, S., & Schwartz, N. (1989). Contributions of cognitive psychology to advertising research. Journal of Advertising Research, 29, 43-53.

Terborg, J. R., & llgen, D. R. (1975). A theoretical approach to sex discrimination in traditionally masculine occupations. Organizational Behavior and Human Pelformance, 13, 352-376.

Tienda, M., Smith, S. A., & Ortiz, V. (1987). Industrial restructuring, gender segregation, and sex differences in earnings. American Sociological Review, 52, 195-210