Cross Cultural Research on Stereotypes

Although there are several small-scale studies of sex stereotypes in other countries (e.g., Lii & Wong, 1982;

Sunar, 1982), to date a large number of studies have been conducted by an international group of researchers cooperating in a 32-country project (Williams & Best, 1990a, 1990b). These studies have been integrated by Williams and Best and will be discussed in detail here with highlights on methodological issues and findings.

Williams and Best's Sex Stereotype Study. In their study, Williams and Best examined gender differences in trait ascription in both the USA and 30 other countries, with data from almost 9,000 children and adults. Adult participants identified stereotyped traits in their own culture, but they were not asked whether they approved of the assignment of different characteristic to men and women or if they believed that the items were self-descriptive.

Measure. In their stereotype study, Williams and Best used the 300-item Adjective Checklist (ACL) (Gough & Heilbrun, 1980). They chose this methodology so that they would have a large diverse item pool descriptive of human personality, not just stereotypes. They included both favorable and unfavorable traits in the pool and did not assume the oppositeness of men and women. Items permitted the assessment of androgyny and interfaced with existing personality research.

When translations were not already available for the ACL, translations by groups of bilinguals and backtranslation procedures (e.g., translating from English to the second language, then back to English to check translation fidelity) were used. Because individual items may not be comparable across languages, comparisons of individual item scores between countries may not have score equivalence, or similar quantitative values. Hence, Williams and Best only analyzed male and female stereotype differences within the same country—the relative gender differences within a country—rather than comparing masculinity between countries or femininity between countries.

Using the 300 ACL items, college students in each country made relative judgments by identifying the adjectives more frequently associated with men or more frequently associated with women. They were permitted to leave out items that were not associated with either gender group. This method "extracts" differences in the views of men and women rather than focusing on similarities. For example, "coarse" is infrequently used to describe either men or women, but research participants associated this adjective with men more frequently than with women.

Study Participants. Williams and Best used university students as study participants, asking them to be "cultural reporters." College students are not representative of their respective populations, but they represent narrow well-matched samples which are functionally equivalent in each country, and they are certainly products of their respective cultures.

The countries in Williams and Best's stereotype study, shown in Table 1, are not representative of all the nations of the world. The sample has a high proportion of English-speaking countries and economically developed countries. Unfortunately, these biases represent the world of cooperative research in academic psychology.

Analyses and Findings. With approximately 100 participants in each country responding to the 300 items of the ACL, the analysis began with over 750,000 "bits" of data. This required a meaningful way to reduce the data. Four scoring systems were used to summarize findings: analyses of individual items, affective meaning, psychological needs, and transactional analysis (TA) ego states. The last two are part of the standard ACL scoring procedure and will not be discussed here (see Williams & Best, 1990a).

For item analyses a simple index was devised to reflect the degree of male association or female association of a particular item in a given country. Male association is represented by an M% score computed for each item by

Table 1. Countries in Williams and Best's Study



South America




























South America

New Zealand

North America Canada United States

Zimbabwe (Rhodesia)

calculating the male association frequency and dividing it by the sum of the male plus female frequencies and discarding the decimal. Thus a high M% score indicates that an item is more frequently associated with men than with women. It does not indicate that a particular adjective would be used frequently to characterize a large portion of men who were being described with ACL items. Similarly, a low M% indicates that an item is more frequently associated with women than with men, not necessarily that the item would be used to describe a majority of women. The method teases out relative differences between men and women.

When the male-associated and female-associated items were identified in each country, a standard degree of association across all countries was used to represent the focused stereotypes, with the number of items varying from country to country. In each country, items were included in the stereotype for a particular sex if they were associated with that sex at least twice as often as with the other sex. Thus items with M% scores of 67% or greater were identified as male stereotype items, and female stereotype items were those with M% scores of 33% and below (F% score of 67% and above). Items that fell into the male-associated and female-associated groups in three quarters of the countries are shown in Table 2. The figures in parentheses beside the adjectives indicate the number of countries out of the original 25 in which the item was in the indicated group. Only three items were female-associated in all 25 countries: sentimental, submissive, and superstitious. On the other hand, six items were male-associated in all countries: adventurous, dominant, forceful, independent, masculine, and strong.

Correlation coefficients were computed for M% scores between pairs of countries to examine the comparability of stereotypes across countries. Across all 300 items, correlations ranged from 0.35 for Pakistan versus Venezuela to 0.94 for Australia versus England. The mean common variance across all 25 countries was 42%, indicating a substantial degree of agreement about the psychological characteristics differentially associated with men and with women.

What about exceptions to the "rules?" How often did an item which was usually in the high M% group fall into the low M% (female) category? For the male-associated items in the table, arrogant, lazy, robust, and rude were associated with women in Nigeria; assertive, humorous, and ingenious were associated with women in Malaysia; boastful, disorderly, and obnoxious were associated with

Table 2. Items Associated with Males and Females in at least 19 of 25 Countries

Male-associated items (N = 49) Female-associated items

Table 2. Items Associated with Males and Females in at least 19 of 25 Countries

Male-associated items (N = 49) Female-associated items

Active (23)

Ingenious (19)

Affected (20)

Adventurous (25)

Initiative (21)

Affectionate (24)

Aggressive (24)

Inventive (22)

Anxious (19)

Ambitious (22)


Attractive (23)

Arrogant (20)

Logical (22)

Charming (20)

Assertive (20)

Loud (21)

Curious (21)

Autocratic (24)

Masculine (25)

Dependent (23)

Boastful (19)

Obnoxious (19)

Dreamy (24)

Clear-thinking (21)

Opportunistic (20)

Emotional (23)

Coarse (21)

Progressive (23)

Fearful (23)

Confident (19)

Rational (20)

Feminine (24)

Courageous (23)

Realistic (20)

Gentle (21)

Cruel (21)

Reckless (20)

Kind (19)

Daring (24)

Robust (24)

Meek (19)

Determined (21)

Rude (23)

Mild (21)

Disorderly (21)

Self-confident (21)

Pleasant (19)

Dominant (25)

Serious (20)

Sensitive (24)

Egotistical (21)


Sentimental (25)

Energetic (22)

Stern (24)

Sexy (22)

Enterprising (24)

Stolid (20)

Shy (19)

Forceful (25)

Strong (25)

Softhearted (23)

Hardheaded (21)

Unemotional (23)

Submissive (25)

Hardhearted (21)

Unkind (19)

Superstitious (25)

Humorous (19)

Wise (23)

Talkative (20)

Independent (25)

Weak (23)

women in Japan; and lazy was associated with women in Pakistan. The exceptions for the female-associated items were even fewer: sympathetic was associated with men in France and Italy, and affected was associated with men in Germany. Impressionistically grouping the items in the table, there is some suggestion of oppositeness for the items associated with men and women (e.g., men— aggressive, dominant, women—submissive; men—stern, severe, women—sentimental, soft-hearted, affectionate). Even though these lists represent considerable cross-cultural and cross-linguistic agreement, this level of analysis is most affected by translation problems. It is perhaps remarkable there is so much similarity in the stereotypes across countries.

Williams and Best's secondly scoring system is an affective meaning analysis derived from the research of Osgood and his associates (Osgood et al., 1957). Based on his extensive research in the United States and in 23 language-culture groups (Osgood, May, & Miron, 1975), Osgood concluded that the principle components of affective meaning—evaluation (good/bad), potency (strong/weak), and activity (active/passive)—were general and could be found in all languages studied. Based on Osgood's findings, Williams and Best had separate groups of American university students use 5-point scales to rate the favorability, strength, and activity of each ACL item, without reference to gender. Standard scores for these ratings were computed by setting the overall mean equal to 500 and the standard deviation equal to 100. Thus scores above 500 indicate ratings that are more favorable, stronger, and more active, while scores below 500 indicate unfavorability, weakness, and inactivity (e.g., Aggressive = favorability 504, neutral; strength 713, very strong; activity 712, very active; Gentle = favorability 635, very good; strength 492, neutral; activity 362, very passive).

Ideally, participants in each country should have scaled each ACL item for favorability, strength, and activity, but this was not possible. However, the Osgood system has sufficient cross-cultural applicability even though particular ratings for individual items may vary by country. Indeed, in making item-by-item translations, affective meaning may determine whether one particular synonym is chosen over another.

In each country the male and female stereotype items were identified and mean favorability, strength, and activity scores for these groups of items were calculated. The ranges of the mean scores across the 25 countries is shown in Figure 1. There is considerable variation among the countries in the favorability associated with the male and female stereotypes, but the ranges of the two stereotypes overlap. In about half the countries the male stereotype was rated more favorably than the female, and the reverse was true in the remaining countries. Moreover, there was no cross-cultural tendency for one stereotype to be more favorable than the other. Frequent objections to the female stereotype are not associated with differential favorability of the adjectives attributed to men and women, but may be related to activity and strength differences.

Looking at these two dimensions, the means for all the male stereotypes are on the active and strong sides of the scales, and the female stereotypes are on the passive and weak sides of the scales, with no overlap between the distributions. Pan-culturally, male-associated items carry connotations of activity and strength, and female items carry connotations of passivity and weakness. It is likely that the differences in activity and strength, rather than a




Figure 1. Ranges of mean affective meaning scores (favorability, activity, strength) for male (M) and female (F) stereotypes across 25 countries.

differences in favorability, account for the general disfavor attributed to the female stereotype items in comparison with the male items.

In view of the variation in the stereotype scores across countries, the question arises as to how these differences may relate to cultural differences. Williams and Best (1990a) examined the relationship between stereotype scores and a number of cultural comparison variables. They used 17 demographic indices (e.g., economic/social development—GNP; education—literacy; status of women—percentage in university, percentage working outside home; religion, general demographics— population, latitude, urban/rural) and four indices of national work-related values from Hofstede's (1980) research (Power Distance, Uncertainty Avoidance, Individualism, Masculinity). They correlated these indices with the stereotype scores. Surprisingly, they found that their stereotype scores were generally unrelated to indices of economic and social development or to work-related values.

The only demographic variable that showed consistent relationships with the stereotype scores was religious affiliation. In countries with higher percentages of Catholics, the greater the relative favorability of the female stereotype and the lower the relative strength of the male stereotype. This may be related to a more significant role for women in the Catholic tradition, perhaps due to the virtue and power associated with the Virgin Mary.

Another interesting religious comparison was between the Muslim and Hindu traditions (Williams, Best, Haque, Pandey, & Verma, 1982). In Muslim theology, significant figures are male and religious practice is controlled exclusively by men, as is society. Women are expected to remain secluded in their homes and are depersonalized by traditional dress. The status of women in Hindu tradition contrasts sharply with that just described. Though most Indian women are homemakers, they also participate actively in commerce, government, religious activities, and education.

In Pakistan, a predominantly Muslim country, the traits associated with women were less favorable than those associated with men, but in India the reverse was true. While the male stereotype in each country was stronger and more active than the corresponding female stereotype, the differences were much smaller in India than in Pakistan.

Looking at male-female stereotype differentiation within each country, differences were largest in The Netherlands, Finland, Norway, and Germany, and smallest in Scotland, Bolivia, and Venezuela. The stereotypes of men and women showed greater differences in more developed countries, and in countries where Hofstede's male work-related values (Hofstede, 1980, 2001) were relatively high in Individualism. The strength and activity differences between the male and female stereotypes were greater in socio-economically less developed countries, in countries where literacy was low, and in countries where the percentage of women attending the university was low. Perhaps economic and educational advancement are accompanied by a reduction in the tendency to view men as stronger and more active than women. However, those effects were merely reduced, not eliminated, by cultural and economic factors.

Your Perfect Right

Your Perfect Right

Always wondered if you could use assertiveness and equality in your relationships and in your life? Here are some great information on how to be more assertive! Do you want to improve your career and the amount of money that you bring home? Do you want to break all the sales records in your office? Do you want to bring home more money? Do you feel as though you are just short of reaching all of your goals?

Get My Free Ebook

Post a comment