the naive perception of masculinity-femininity on the basis of expressive cues

14
JOURNAL OF RESEARCH IN PERSONALITY 12, l-14 (1978) The Naive Perception of Masculinity-Femininity on the Basis of Expressive Cues RICHARD LIPPA California State University at Fullerton Groups of naive judges rated I8 videotaped stimulus persons on masculinity. femininity, “dominance, assertiveness.” and “compassion, sensitivity to others.‘* Stimulus persons were broken down by sex and sex-typing-half were male, half female-and within sexes one third were classified as masculine, feminine, and androgynous on the basis of their scores on the Bern Sex Role Inventory. Two experiments are reported in which groups of judges rate stimulus persons on the basis of such different expressive information as videotaped pictures and recorded voices, videotaped pictures alone, videotaped bodies, videotaped heads, recorded voices, and still photos. The results showed: (I) Judges reliably rated masculinity-femininity from largely expressive cues: (2) judgments of masculinity-femininity were not predominantly determined by judgments of sex role-related traits: (3) the naive judgment of masculinity- femininity significantly corresponded to stimulus subjects’ assessed sex roles: (4) stimulus subjects (particulary males) showed a consistent display of masculinity- femininity across expressive channels: and (5) judges used different expressive cues in judging masculinity-femininity in males and females. These results are related to broader questions concerning the relation between expressive behavior and personality. As naive psychologists, most of us hold that expressive styles and personality are related. Firm handshakes are taken to indicate forceful personalities, while weak grips signal just the opposite. Loud clear voices show confidence, while soft, tremulous ones show diffidence. Slow. placid gestures reflect calm personalities. Twitches and stutters reveal anxiety. And shifty eyes proclaim dishonesty. Among professional psychologists, perhaps Gordon Allport offered the strongest support for this common-sense view that there exists a direct and necessary link between expressive behavior and personality. In Per- A shorter version of this paper was presented under the title, “The Perceived Masculinity-Femininity of Sex-Typed and Androgynous Persons,” at the Western Psychological Association Convention in Los Angeles, California. on April 9, 1976. Re- quests for reprints should he sent to the author at the Department of Psychology. California State University at Fullerton, Fullerton, CA 92634. 0092-6566/78/0121-0001$02.00/0 Copyright @ 1978 by Academic Press. Inc. All rights of reproduction in any form reserved.

Upload: richard-lippa

Post on 25-Aug-2016

213 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: The naive perception of masculinity-femininity on the basis of expressive cues

JOURNAL OF RESEARCH IN PERSONALITY 12, l-14 (1978)

The Naive Perception of Masculinity-Femininity on the Basis of Expressive Cues

RICHARD LIPPA

California State University at Fullerton

Groups of naive judges rated I8 videotaped stimulus persons on masculinity. femininity, “dominance, assertiveness.” and “compassion, sensitivity to others.‘* Stimulus persons were broken down by sex and sex-typing-half were male, half female-and within sexes one third were classified as masculine, feminine, and androgynous on the basis of their scores on the Bern Sex Role Inventory. Two experiments are reported in which groups of judges rate stimulus persons on the basis of such different expressive information as videotaped pictures and recorded voices, videotaped pictures alone, videotaped bodies, videotaped heads, recorded voices, and still photos. The results showed: (I) Judges reliably rated masculinity-femininity from largely expressive cues: (2) judgments of masculinity-femininity were not predominantly determined by judgments of sex role-related traits: (3) the naive judgment of masculinity- femininity significantly corresponded to stimulus subjects’ assessed sex roles: (4) stimulus subjects (particulary males) showed a consistent display of masculinity- femininity across expressive channels: and (5) judges used different expressive cues in judging masculinity-femininity in males and females. These results are related to broader questions concerning the relation between expressive behavior and personality.

As naive psychologists, most of us hold that expressive styles and personality are related. Firm handshakes are taken to indicate forceful personalities, while weak grips signal just the opposite. Loud clear voices show confidence, while soft, tremulous ones show diffidence. Slow. placid gestures reflect calm personalities. Twitches and stutters reveal anxiety. And shifty eyes proclaim dishonesty.

Among professional psychologists, perhaps Gordon Allport offered the strongest support for this common-sense view that there exists a direct and necessary link between expressive behavior and personality. In Per-

A shorter version of this paper was presented under the title, “The Perceived Masculinity-Femininity of Sex-Typed and Androgynous Persons,” at the Western Psychological Association Convention in Los Angeles, California. on April 9, 1976. Re- quests for reprints should he sent to the author at the Department of Psychology. California State University at Fullerton, Fullerton, CA 92634.

0092-6566/78/0121-0001$02.00/0 Copyright @ 1978 by Academic Press. Inc.

All rights of reproduction in any form reserved.

Page 2: The naive perception of masculinity-femininity on the basis of expressive cues

2 RICHARD LJPPA

sonality, A psychological interpretation (1937), Allport asserted, “the only secure interpretation of expression seems to be in terms of stable personal dispositions. Expressive behavior, in brief, is the external aspect of inner structural consistency.”

Nowhere is the assumed agreement between expressive styles and personality more intuitively apparent than in the domain of masculinity and femininity. On an anecdotal level, Birdwhistle (1970) has listed a number of expressive cues used in American culture to display masculin- ity and femininity. These include the angle at which the upper legs are held apart when standing, the angle at which the arms are held away from the upper torso, the angle at which the pelvis is held either forward or backward, and the “flutteryness” of eyelid movements. On an even more anecdotal level, it seems clear that many persons in our culture hold strong stereotypes about the effeminacy of limp-wristedness and lisping speech in men and the mannishness of a large, loping stride and gruff speech in women. Intuitively, then, people seem to attend to a wide range of expressive cues in judging the masculinity and femininity of others.

Because of the abundance of expressive cues of masculinity and femi- ninity, this expressive domain would seem to be a fruitful area to study the relationship among expressive behavior, personality, and person per- ception. The research to be reported in this paper focuses on three main questions about expressive masculinity-femininity:

(1) Do persons judging others on masculinity and femininity attend to purely expressive cues such as gestures, body carriage, and voice quality, or do they require a wider sample of behavior to make such judgments? That is, are expressive cues sufficient to permit judges to make reliable assessments of masculinity and femininity? If so, what cues do judges use?

(2) Do perceived differences in masculinity and femininity necessarily accompany perceived differences in such sex role-related traits as domi- nance, assertiveness, tenderness, and interpersonal sensitivity, or, on the other hand, is perceived masculinity-femininity largely an expressive dimension unrelated to other trait attributions?

And finally (3) are naive judgments of a person’s masculinity and femininity “accurate’‘-that is, do they correspond to the stimulus per- son’s independently assessed sex role? In the domain of masculinity and femininity, how strong is the correspondence between expressive be- havior and personality?

In assessing masculinity-femininity, psychological researchers have traditionally relied upon personality scales developed to distinguish be- tween criterion groups of males and females or between heterosexuals and homosexuals (See Constantinople, 1973, for a recent review.). The better-known inventories include the Terman-Miles Attitude-Interest

Page 3: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTION OF MASCULINITY-FEMININITY 3

Analysis Test and the masculinity-femininity scales contained in the MMPI, the Strong Vocational Interest Blank, and the California Psychological Inventory.

In recent years, researchers have questioned the assumptions under- pinning such traditional masculinity-femininity scales. After an extensive review of the literature, Constantinople (1973) concluded that research did not support the notion that masculinity and femininity formed a single, bipolar dimension, and, in keeping with this analysis, Bern (1974) and Spence, Helmreich, and Stapp (1974) developed inventories that assess masculinity and femininity as separate and independent dimensions.

Despite such an abundance of instruments attempting to measure mas- culinity and femininity, there has been surprisingly little research carried out investigating naive judgments of masculinity and femininity and their relation to more sophisticated psychometric assessments.

Sappenfield (1965) carried out virtually the only empirical test of the correspondence between judged and assessed masculinity-femininity, and his interest was not in masculinity-femininity per se, but rather in testing the assumptions underlying the Szondi Personality Test. Sap- penfield found that naive judges could not accurately judge from facial photographs the independently assessed masculinity-femininity of 60 male stimulus subjects. It is difficult to generalize from Sappenfield’s study, however, due to the limited nature of the stimulus materials em- ployed.

Other studies looking at naive conceptions of masculinity and feminin- ity have asked subjects to rate the characteristics of ideal “masculine” and “feminine” persons and to rate the masculinity and femininity of groups of acquaintances. Using such techniques, Jenkin and Vroegh (1969) found that subjects did not place masculinity and femininity at opposite ends of a single dimension. For example, they found subjects’ semantic differential ratings of the concepts “most feminine person you can imagine” and “most masculine person you can imagine” to be posi- tively correlated. While Jenkin and Vroegh’s study represents a success- ful effort to begin to understand people’s “implicit personality theories” about masculinity and femininity, it has little to say about the cues used in judging masculinity and femininity in direct person perception.

The two experiments to be described in this paper were carried out to investigate empirically how naive judges do in fact judge the masculinity and femininity of others. Subjects were asked to judge videotaped stimu- lus persons, broken down by sex and sex-typing, on a number of per- sonality characteristics including masculinity and femininity.

In the first experiment to be reported, judges were asked to rate stimu- lus persons on the basis of “complete“ information-that is, judges viewed videotapes showing the subjects’ full bodies and also heard their

Page 4: The naive perception of masculinity-femininity on the basis of expressive cues

4 RJCHARD LJPPA

recorded voices. With the data from this study, we hoped to answer two major questions: (I) Do perceived differences in masculinity and feminin- ity necessarily accompany perceived differences in sex role-related traits (such as dominance and sensitivity to others)? And (2) do judges’ ratings of masculinity and femininity correspond to the assessed sex-typing of the stimulus persons?

In the second experiment to be described, different groups of judges were asked to rate the same stimulus persons used in the first experiment on the basis of various types of “limited” information. For example, one group of judges viewed just the videotaped pictures of the stimulus per- sons without the accompanying sound, another group viewed just the videotaped bodies of the stimulus persons, and a third group viewed just their videotaped faces. Another two groups of judges rated just the re- corded voices of stimulus subjects and just still photographs of the sub- jects. With these ratings based on different expressive information, we hoped to investigate what cues were used in the judgment of masculinity and femininity.

METHOD

Stimulus subjects were selected from a population of 46 subjects (25 males and 22 females) who had been videotaped during previous research as they roleplayed being junior high school math teachers. These subjects had been videotaped as they stood next to an easel, drew triangles on a pad of paper, and explained them to an imagined class. Thus the task in which the subjects were engaged and the content of their speech were fairly constant across subjects.

Of the original 46 subjects, 9 males and 9 females were selected on the basis of their scores on the Bern Sex Role Inventory (BSRI) (Bern, 1974). The BSRI asks an individual to rate himself on a number of personality characteristics, some of which have greater social desirability for men and others for women. Thus the BSRI has both a masculinity scale and a fenininity scale, and a subject’s “androgyny” score is defined as Student’s t ratio for the difference between his masculinity and femininity scores. If this ratio is statistically sig- nificant, then the subject is classified as sex-typed: if the ratio is between - 1 and +I, he is classified as androgynous.

From among the male and female subjects respectively, 3 subjects were selected who were the most extremely “masculine sex-typed” and another 3 subjects were selected who were the most extremely “feminine sex-typed.” Thus 12 subjects were chosen: 3 masculine males, 3 feminine males, 3 masculine females, and 3 feminine females.

In addition, 3 androgynous males and 3 androgynous females were randomly selected from all those males and females classified as androgynous. Thus, in total, 18 stimulus subjects were selected-half male and half female-and one-third within each sex was classified as masculine sex-typed, feminine sex-typed, and androgynous.

Procedure

EXPERIMENT 1

Forty-eight introductory psychology students (half male, half female) viewed 45sec. segments of 21 stimulus subjects (3 “practice” subjects

Page 5: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTION OF MASCULINITY-FEMININITY 5

followed by the 18 stimulus subjects described earlier). These naive judges heard the accompanying sound track of stimulus persons’ voices as well as viewed their full videotaped pictures. After viewing each stimulus person, the judges were asked to rate the person on eight personality characteristics: “calm, relaxed,” “masculine,” “compassionate, sensi- tive to others,” “outgoing, extraverted,” “anxious, nervous,” “domi- nant, assertive,” “ reserved, introverted,” and “feminine.” The judges rated each stimulus person on each personality characteristic using a seven-point scale that ranged from “l-not at all” through “4- moderately” up to “7-extremely.”

Although judges rated the stimulus persons on eight personality charac- teristics, we were primarily interested in only four of these ratings-those on masculine: feminine: compassionate, sensitive to others: and domi- nant, assertive. The last two trait dimensions were selected to represent personality characteristics stereotypically assumed to relate to masculin- ity and femininity. In fact, an item analysis carried out on the Bern Sex Role Inventory (using data from 167 introductory psychology students) showed that such items as “dominant” and “assertive” correlated most strongly with the masculinity scale, while such items as “compassionate” and “tender” loaded most highly on the femininity scale.

By comparing the four types of ratings listed above, we could analyze the relation between judges’ ratings of masculinity and femininity and their ratings on traits stereotypically assumed to relate to masculinity and femininity. If masculinity-femininity judgments were made largely on the basis of judgments on sex role-related traits, then we would expect a large correlation between ratings of masculinity-femininity and ratings of sex role-related traits. If, on the other hand, judgments of masculinity and femininity were based largely on expressive cues unrelated to sex role- related trait attributions, then we would expect low correlations between ratings of masculinity-femininity and ratings of sex role-related traits.

Results

A single rating of masculinity and femininity was calculated for each stimulus subject by averaging the ratings of all 48 judges. Similarly, a single rating on compassion, sensitivity to others and dominance, asser- tiveness was calculated for each stimulus person.

Judges’ ratings of stimulus persons’ masculinity and femininity were strongly negatively correlated-the perceived femininity of female stimu- lus subjects correlated -36 with the perceived masculinity of the same subjects (p < JOI). For male stimulus subjects the correlation was - 39 (p < .OOl). Since ratings of masculinity and femininity were so strongly related, we combined these ratings into a single rating of masculinity-

Page 6: The naive perception of masculinity-femininity on the basis of expressive cues

6 RICHARD LIPPA

femininity by subtracting the femininity rating from the masculinity rating.

To check the reliability of judges’ ratings, the group of 48 judges was split into two groups (each consisting of half males and half females) and correlations were computed between the mean ratings of each group. These reliability measures were uniformly high. For ratings of masculinity-femininity, the correlation between groups was .94 (p < .OOl> for female stimulus subjects and .95 (p < .OOl) for male stimulus subjects. For ratings on compassion, sensitivity to others, the correlations between groups were .97 (p < .OOl) for male stimulus subjects and .99 (p < .OOl) for female stimulus subjects, and for ratings on dominance, assertiveness, the correlations were .95 (p < .OOl) for male stimulus subjects and .97 (p < ,001) for female stimulus subjects.

These high reliabilities reflect the large number of judges employed, but they also indicate that there was substantial agreement among judges in rating the various personality characteristics of stimulus persons.

Our data indicated that judges’ ratings of masculinity-femininity were more strongly related to judgments of dominance, assertiveness than to ratings of compassion, sensitivity to others. Using the mean ratings of all 48 judges, the correlations between ratings of masculinity-femininity and ratings on compassion, sensitivity to others were .17 (ns.) for male stimulus subjects and -. 10 (n.s.) for female stimulus subjects. Ratings of masculinity-femininity correlated .59 (p < .05) with ratings of domi- nance, assertiveness for male stimulus subjects and .49 (p < -1) for female stimulus subjects.

Given the high reliabilities of all types of ratings, these correlations were not extremely large. Even for the highest correlation (between ratings of masculinity-femininity and dominance, assertiveness for male stimulus subjects), the ratings on one dimension accounted for only 35% of the variance of the ratings on the other dimension. Clearly, most of the variance in the ratings of masculinity-femininity was accounted for by judgments unrelated to the sex role-related traits of dominance, assertive- ness and compassion, sensitivity to others.

Judges’ ratings of masculinity-femininity were the only ratings to dis- tinguish among the different sex-typing categories of stimulus persons. Table 1 lists the mean ratings on masculinity-femininity for stimulus subjects broken down by sex and sex-typing along with the results of a 2 x 3 ANOVA carried out on these data. Not surprisingly, there was an enormous main effect for the sex of stimulus persons (F = 760.13,~ < .OOl), with males being perceived more masculine and less feminine than females. Of more interest was the significant main effect for sex-typing (F = 11.87, p < .005). Clearly, judges did perceive differences in the masculinity-femininity of masculine sex-typed, androgynous, and

Page 7: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTION OF MASCULINITY-FEMININITY 7

TABLE I

MEAN RATINGS OF MASCULINITY-FEMININITY OF SUBJECTS

BROKEN DOWN BY SEX AND SEX-TYPING”

Masculine Androgynous Feminine

Males 3.35 2.99 2.03 Females -1.90 -2.59 -2.85

a Fax = 760.13 (p < ,001): Fsex.~y~,,,~ = Il.87 (P < .0X).

feminine sex-typed persons. The perceived differences were in the ex- pected direction: Masculine sex-typed persons were perceived as more masculine and less feminine than feminine sex-typed persons, with an- drogynous persons being perceived in between the other two groups.

A similar analysis of judges’ ratings on compassion, sensitivity to others and on dominance, assertiveness showed no significant main ef- fects for either sex or sex-typing of stimulus persons (all F’s < 1.51, n.s.).

Discussion

In general, the results of this study indicated the importance of purely expressive cues in the judgment of masculinity-femininity. Naive judges, as a group, tended to give highly reliable ratings of stimulus persons’ masculinity-femininity, and these ratings significantly corresponded to the independently assessed sex-typing of stimulus persons.

The fact that judges’ ratings of masculinity-femininity discriminated among the sex-typing categories, while ratings of such sex role-related traits as dominance and interpersonal sensitivity did not, indicates the largely expressive nature of the cues used to judge masculinity- femininity. Stimulus persons were clearly not being judged masculine or feminine solely on the basis of their perceived dominance or sensitivity. Rather, it appeared that judges were attending largely to expressive cues that had little to do with such trait attributions.

In a sense, the stimulus materials presented to judges forced them to base their judgments of personality largely on expressive cues. Stimulus subjects were viewed as they carried out a standardized, stereotyped task (teaching triangles) that provided little direct information on which to base personality judgments.

Although Experiment 1 showed that judges were “accurately” perceiv- ing masculinity-femininity, it did not provide any data indicating what cues judges used to make their ratings of masculinity-femininity. Exper- iment 2 was designed to investigate what types of cues were most impor- tant in determining judges’ ratings of masculinity-femininity.

Page 8: The naive perception of masculinity-femininity on the basis of expressive cues

RICHARD LJPPA

EXPERIMENT 2

Procedure

Using the same procedure as in Experiment 1, five additional groups of judges were asked to rate the same stimulus persons used previously after viewing or hearing limited stimulus information. One group of judges (N = 14) made their ratings after viewing the videotaped pictures of stimulus persons without the accompanying sound. Another group of judges (N = 15) rated stimulus persons after viewing videotaped segments of just the persons’ bodies-stimulus persons’ heads were blocked from sight on the TV monitor. A third group ofjudges (N = 14) viewed segments ofjust the stimulus persons’ heads, and a fourth group of judges (N = 16) rated the stimulus persons after listening to recordings of their voices. The same audio and videotaped segments used in Experiment 1 were used in Exper- iment 2.

A fifth and final group of judges (N = 18) viewed still photographs of each stimulus person and made their ratings on the basis of this informa- tion. The photographs were taken of the stimulus persons from a TV monitor-that is, the photographs were essentially “snapshots” of the subjects as they appeared in the videotaped segments. Ail subjects were photographed as they faced toward the TV camera, and each judge rated two separate photos of a given stimulus person.

Results

As in Experiment 1, the ratings from each group of raters were aver- aged to create one masculinity-femininity rating for each stimulus person. So, for a given stimulus person, there was one masculinity-femininity rating based on the video picture, one based on his videotaped body, one based on his voice, and so on.

Again, reliabilities were calculated by dividing each of the groups of raters into two equal or near-equal groups and correlating the mean ratings of these two subgroups. Although the number of judges in each of our groups was substantially smaller than in the previous study, the reliabilities were, in general, quite high. For the five new groups of raters, the mean of the reliabilities for the masculinity-femininity ratings (com- puted separately for male and female stimulus persons) was .75 and the median was .84.

As a preliminary analysis, a 2 x 3 x 6 repeated-measures ANOVA (sex of stimulus person x sex-typing x type of rating) was carried out on the five new types of masculinity-femininity ratings plus the old masculinity-femininity ratings based on “complete” information from Experiment 1. The results of this ANOVA showed the same pattern in the data as previously reported-there was a strongly significant effect for sex

Page 9: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTION OF MASCULINITY-FEMININITY 9

on stimulus persons (F = 325.97,~ < .OOl) and also a significant effect for the sex-typing of stimulus persons (8’ = 3.93, p < .05).

Two similar ANOVAs were carried out on ratings of “dominance, assertiveness” and “compassion, sensitivity to others.” As before, there were no significant main effects for sex-typing (both F’s < 1): however, with the addition of judgments based on limited information, gender stereotypes begin to appear in the data.

For ratings of dominance, assertiveness there was a marginally sig- nificant main effect for sex of stimulus persons (F = 3.31, p < .09), with men being perceived as more dominant and assertive than women. And for the ratings on compassion, sensitivity to others, the main effect for sex of stimulus persons reached significance (F = 9.45,~ < .Ol). with women being perceived as more compassionate and sensitive to others than men.

To determine which of the masculinity-femininity ratings based on limited information distinguished among the three sex-typing categories of stimulus persons, ANOVAs using a 2 x 3 design (sex x sex-typing) were carried out separately on the five masculinity-femininity ratings based on limited information.

All of these ANOVAs showed significant main effects for the sex of stimulus persons: however, not all showed significant main effects for sex-typing. There were significant sex-typing main effects in the masculinity-femininity ratings based on the video picture without sound (F = 5.05, p < .05) and for ratings based on videotaped bodies (F = 4.19, p < .05), but not for ratings based on videotaped heads (F = .3 I, n.s.), voices (F = 2.05, p < .25) or still photographs (F = .77, n.s.).

Thus, at least when male and female stimulus persons were analyzed together, it appeared that the most accurate masculinity-femininity cues were visual ones, mainly body cues. The absence of a significant sex- typing main effect for ratings based on still photos was important since it tended to rule out the importance of just physical appearance or dress as cues in the judgment of masculinity-femininity. Clearly, the judges who had access to complete information were responding to more subtle expressive cues than were present in just still photographs of stimulus persons.

To analyze how the various types of ratings of masculinity-femininity were related, these ratings were correlated with one another. These intercorrelations, calculated separately for male and female stimulus per- sons, are shown in Table 2.

A number of interesting patterns emerge from this table of correlations. The first is that the stimulus persons, particularly the males, show a consistent display of masculinity-femininity across expressive channels. In the table of correlations for males, all of the correlations are positive and 6 out of 15 are significant at the .05 level. Particularly impressive in

Page 10: The naive perception of masculinity-femininity on the basis of expressive cues

10 RICHARD LIPPA

TABLE 2 CORRELATIONS AMONG MASCULINITY-FEMININITY RATINGS

BASED ON DIFFERENT INFORMATION

Voice Picture

and voice Picture

Type of rating

Body Head Photo

.s3** Male stimulus subjects (N = 9)

.60*” .41 .44 .32 .52’ ..56* .51” .I9

.25 .g,** .73”% .624=+ .23

.63*”

Female stimulus subjects (fV = 9) .65”%. -.05 -.lO -.I0 -.06

.50" .57* .43 .53* ,96”” .42 .v*

.51” .89** .73”L

Voice Picture and voice Picture Body Head Photo

Voice

Picture and voice Picture Body Head

*p 4 .I. **: p < .05.

this table is the .60 (p < .05) correlation between masculinity-femininity judged on the basis of the video pictures with that judged from recorded voices. Agreement between ratings based on stimulus males’ bodies and faces was also high (r = .62, p < .05). In other words, groups of judges having access to completely nonoverlapping samples of expressive be- havior agreed fairly strongly in their assessment of stimulus males’ degree of masculinity-femininity.

Female stimulus persons, on the other hand, showed consistent display of masculinity-femininity through visual channels, but the visual chan- nels did not seem to correspond with judgments based on their voices.

This sex difference in the consistency of masculinity-femininity cues across expressive channels was statistically tested by converting each type of masculinity-femininity rating to a 2 score (using the means and standard deviations of that type of rating over the 18 stimulus persons). For each stimulus person, the variance of his or her six different types of masculinity-femininity ratings was computed. This variance essentially measured the agreement, for a given subject, of his masculinity- femininity ratings based on different expressive information.

A 2 x 3 ANOVA (sex x sex-typing) was carried out on these variances.

Page 11: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTION OF MASCULINITY-FEMININITY II

The main effect for sex was significant (F = 6.50, p < .02), while that for sex-typing was not (F = 1.98, p < .18). The mean variance for male stimulus persons was .019, while that for females was ,092: Clearly the perceived masculinity-femininity of females was more variable across expressive channels than was that of males.

Another sex difference that is apparent from Table 2 is the varying importance of various types of cues in judging the masculinity-femininity of males and females. In general, the judges in our study seemed to utilize body cues and physical appearance cues more in judging the masculinity-femininity of females than in judging that of males. For example, for males the judgments of masculinity-femininity based on still photos correlated only .I9 with judgments based on complete informa- tion, while for females the correlation was .53. While the difference between these two correlations was not significant, it was suggestive.

Body cues clearly appeared to be more important in judging the masculinity-femininity of females. For example, for male stimulus per- sons the correlation between masculinity-femininity judged from their videotaped bodies with that judged from their full videotaped pictures was .25, while for females this correlation was .96. The difference between these two correlations was significant (Z = 2.93,~ < .002). Similarly, for male stimulus persons the correlation between masculinity-femininity judged from their videotaped bodies with that judged from still photos was .23, while for females this correlation was .89. Again, the difference was significant (2 = 2.06, p < .02). Thus it appears that the judges used different cues in judging the masculinity-femininity of males and females.

Not only did judges employ different cues in judging the two sexes, but the various cues seemed to be differentially valid in reflecting males and females independently assessed sex-roles. Table 3 lists the correlations of stimulus subjects’ judged masculinity-femininity with their “androgyny scores” on the Bern Sex Role Inventory.

The correlations for masculinity-femininity ratings based on complete information (video picture and voice) were very high-for male stimulus subjects, r = .90 @ < .OOl), and for female stimulus subjects, r = .79 0, < .005). These correlations indicate again that the judges, as a group. were accurately perceiving the assessed sex-typing of stimulus persons.

The correlations for masculinity-femininity ratings based on limited information suggest that various expressive cues were differentially valid for the two sexes. For example, the correlation between androgyny scores and masculinity-femininity judged from voices was significant for males (Y = .73,p < .Ol) but not for females (v = .44,p < .12). This pattern was reversed for masculinity-femininity ratings based on video pictures and bodies; the correlations of these ratings with stimulus subjects an- drogyny scores were significant for females but not for males. While none

Page 12: The naive perception of masculinity-femininity on the basis of expressive cues

12 RJCHARD LJPPA

TABLE 3 CORRELATIONS BETWEEN MASCULINITY-FEMININITY RATINGS AND ANDROGYNY SCORES FOR MALE AND FEMALE STIMULUS SUBJECTS

Type of rating Males Females

Picture and voice Voice Picture Body Head Photo

,go**** .79*** ,7-j*** .45 .39 .64** .48* .61*” .32 .15

-.05 .52*

* p < .I. **p < .05.

*** p < .Ol. **** p < ,001.

of these differences between correlations for males and females was statistically significant, given the small sample size these results seem suggestive nonetheless.

DlSCUSSlON

The research described in this paper demonstrates that masculinity- femininity can be reliably judged from largely expressive cues, and that such judgments reflect the independently assessed sex-typing of stimulus persons.

In a larger context, this research addresses one of the major questions in the study of expressive behavior: Does expressive behavior reflect personality, and, if so, what is the nature of the correspondence? In this study, we have shown that the expressive behavior of masculinity- femininity as rated by naive judges does indeed reflect “personality” as assessed by the sex-role inventory employed.

However, the research reported in this paper points to the problems with too naive a “sign equals trait” approach to the study of expressive behavior. Our data indicate that naive judges of masculinity-femininity used different cues in judging males and females, and this may result from the fact that males and females in this culture display masculinity and femininity somewhat differently. If this interpretation is correct, then, in the domain of masculinity-femininity at least, a given expressive cue may have a different “meaning” or validity depending on the sex of the stimulus person. That is, particular expressive signs do not equally reflect underlying dispositions for all stimulus persons.

This is intuitively obvious for the expressive cues of an underlying state or disposition such as anxiety. Some people tremble when anxious; some

Page 13: The naive perception of masculinity-femininity on the basis of expressive cues

PERCEPTlON OF MASCULINITY-FEMININITY 13

stutter. Others display nervous twitches, and still others fidget. Clearly, if a researcher attempts to correlate measures of any single expressive sign with an independent measure of anxiety, the correlation will probably be low-simply because a given expressive cue is not an equally good indica- tor of anxiety for all persons.

The strategy, then, in studying expressive behavior, must be either to separate persons into groups who display similar types of expressive signs or to use naive judges who can observe and integrate whatever cues are available. In our study we found that “naive” judges are far from naive: They showed themselves to be sophisticated and accurate observers of expressive masculinity and femininity.

Despite individual differences in the expressive display of personality, our research indicates that, at least in the domain of masculinity- femininity cues, people (particularly males) display fairly consistent ex- pressive behavior across different expressive channels. Such a finding is in keeping with Allport’s notion that expressive behavior is consistent, and that such consistency reflects a consistent personality (Allport & Vernon, 1967; Allport, 1937).

In interpreting our data, however, we break with Allport insofar as he held that expressive behavior is a “naturally” occurring, unlearned ex- pression of personality. Insofar as masculinity and femininity are cultur- ally defined and culturally enforced, we should also expect their accom- panying expressive behaviors to be so defined and enforced.

The most plausible explanation for the high correlations between per- ceived masculinity-femininity and assessed androgyny in our study would seem to be that people actively monitor aspects of their expressive behavior and enforce certain expressive styles on their behavior. The domain of masculinity-femininity is a likely area to search for expressive consistencies since (I) there seem to be specific socially defined expres- sive cues of masculinity and femininity and (2) in this culture at least, there are strong socialization pressures on people to display “appro- priate” sex-role behavior.

On an anecdotal level, Birdwhistle (1970) has observed that the expres- sive behaviors of masculinity-femininity appear at a very early age. Developmental research on the acquisition of such expressive behaviors could shed light on how expressive consistency develops and how expres- sion comes to reflect personality.

REFERENCES Allport, G. W., & Vernon, P. E. Studies in expressive morlement. New York: Hafner, 1967

(originally published 1933). Allport. G. W. Personality, A psychological interpretution. New York: Henry Holt, 1937.

Page 14: The naive perception of masculinity-femininity on the basis of expressive cues

14 RICHARD LIPPA

Bern, S. The measurement of psychological androgyny. Journal of Consulting and Clinical

Psychology, 1974, 42, 155-162. Birdwhistle, R. Masculinity and femininity as display. In Kinesics andcontexf. Philadelphia:

University of Pennsylvania Press, 1970. Constantinople, A. Masculinity-femininity: An exception to a famous dictum? Psychologi-

cal Bulletin, 1973, 80, 389-407.

Jenkin, N., & Vroegh, K. Contemporary concepts of masculinity-femininity. Psychological

Reports, 1969, 25, 679-697.

Sappenfield, B. R. Test of a Szondi assumption by means of m-f photographs. Journd of Personality, 1%5, 33, 409-417.

Spence, J., Helmreich, R., & Stapp, J. The personal attributes questionnaire: A measure of sex role stereotypes and masculinity-femininity. Journal Supplement Abstract Service

Catalog of Selected Documents in Psychology, 1974, 4, 43 (Ms. No. 617).