The Chivalry Script and Stereotypes about Women

In my research, I examine a cultural script that I have labeled "chivalry," characterized by themes of men protecting and providing for women. Chivalry as a cultural phenomenon is unusual in that it is possible to pinpoint its origins with considerable accuracy. In the 11^th century, the medieval knights who had once protected Europe against Viking invasion jeopardized the social order through continuous warfare among themselves. In an effort to control the knights, the Roman Catholic Church reconstructed the knight’s role to include the protection of women, the weak, and the oppressed. During the Crusades, the role of knight (who literally is chivalry, a word derived from the same root as cavalry) was transformed from brutal mercenary into, in the words of Pope Urban II, "the true soldiery of Christ." Over the centuries, the military and equestrian aspects of chivalry faded and gave way to a script describing gallantry, bravery, and deference to women. Chivalry’s survival from the 11^th to the 20^th century can be seen in its employment as an explanation for sex differences in aggression (Eagly & Crowley, 1986), helping (Eagly & Steffen, 1986), and the sentences and conviction rates of criminal defendants (Steffensmeier & Kramer, 1982). Despite its invocation as a post-hoc explanation of sex differences, little research has been done to explore the chivalry script or to assess, quantitatively, its influence on behavior.

Chivalry: Discrimination or Respect?

The chivalry script leads men to behave toward women in a way that is different from the way they would treat other men – a pattern of behavior that fits the definition of discrimination. But does this discrimination have negative consequences for women? Perhaps chivalry is a gesture of respect that acknowledges the differences between men and women and affirms the positive traits associated with women. On the other hand, if chivalry is interpreted as assistance, then perhaps chivalry helps to perpetuate the stereotype that women are less competent and independent than men are.

Research that I have conducted indicates that the chivalry script is related to both of these beliefs, which I separate into two stereotypes about women, one negative and the other positive: the belief that women are less agentic (less able to effectively achieve goals and wield authority) than men; and the belief that women are more "virtuous" (morally responsible) than men. The negative stereotype might explain why chivalrous behavior is required: men should take care of women because women are not able to take care of themselves. The positive stereotype might explain why chivalrous behavior is deserved: men should take care of women because women are angelic creatures who deserve to be put on a pedestal.

Using a variety of research methods, I have developed a program of research demonstrating the relation between chivalry and stereotypes about women. This program of research is designed to answer three questions: 1) do chivalrous beliefs co-occur with these two stereotypical beliefs, 2) do chivalrous individuals discriminate in favor of women who conform to these stereotypes, and 3) do observers of chivalrous behavior attribute it to factors within the recipient that are consistent with these stereotypes, that is, do the recipients of chivalry appear less independent than non-recipients do.

Do chivalrous beliefs co-occur with beliefs in women's virtue and lack of agency?

To answer the first question, I used survey methods to develop a measure of the degree to which an individual endorses the chivalry script. This measure, which I dubbed the Chivalry Scale, is composed of ten items such as "If there is a dangerous job to be done, it is better for a man to do it than a woman" and "A man should give up his seat to a woman if the bus is crowded." Using the responses of 410 participants, scores on the Chivalry Scale were found to have strong, positive, and significant correlations with attitude scales measuring beliefs in women’s relative incompetence, weakness, moral discipline, and sexual conservatism (Altermatt, Johnson, & Cohen, 1999). The answer to the first question is, therefore, that chivalrous beliefs do appear to co-occur with the four stereotypical beliefs.

Do chivalrous individuals discriminate in favor of women who appear to be high in virtue and low in agency?

This question moves beyond the connection between chivalry and descriptive stereotypes and considers the possibility that chivalry may be related to prescriptive stereotypes about women. That is, stereotypes that not only describe the way that women are but that also prescribe the way that women should be. Violation of prescriptive stereotypes is likely to result in negative evaluations and a loss of the privileged status that chivalrous men afford to women. To answer the question of whether chivalry is related to prescriptive stereotypes concerning women's virtue and agency, I designed a series of three studies.

1. Manipulating virtue and agency in hypothetical vignettes.
In the first study, participants read a series of short vignettes designed to manipulate their impressions of a character's virtue and agency. Each vignette described a scenario in which participants encountered the character and had an opportunity to respond to them. The vignettes differed in the type of response that was appropriate. In the "defending" vignettes, the character was in danger of being assaulted by a third party and participants indicated the degree to which they would intervene or escape. In the "helping" vignettes, a failure to intervene would result in some inconvenience to the character but no physical danger. After reading and responding to each vignette, participants completed the chivalry scale. High-chivalry males were significantly more likely than low-chivalry males to indicate that they would intervene to assist the female characters in the defending and helping vignettes. In addition to this main effect for chivalry, there was also an interaction between chivalry and the characteristics of the female characters. Differences between high-chivalry and low-chivalry participants only emerged when the characters were low in agency or high in virtue (consistent with the hypothesized relation between chivalry and stereotypes about women). High-chivalry participants were significantly more likely than low-chivalry participants to defend female characters who were low in agency, and significantly more likely to help female characters who were high in virtue. When the female characters were high in agency or low in virtue, responses of high-chivalry males were almost identical to the responses of low-chivalry males.
2. Manipulating virtue and agency in a laboratory experiment.
Encouraged by results of "hypothetical situations" experiment, I designed an experiment to investigate whether high-chivalry and low-chivalry participants differed in their treatment of flesh-and-blood women. In each session, three participants and three experimental confederates met outside the laboratory. During this informal meeting, the first manipulation of the experiment took place - part of the virtue manipulation involved confederates' dress (either very conservative or "bar apparel") and choice of reading material ("Skin Art" or "Cannabis Culture," both designed to convey impressions of low virtue). After being introduced to the experiment, participants and confederates were paired up seemingly at random but always so that one participant was paired with one confederate. Each pair retired to a smaller room where they took turns asking and answering a series of scripted "icebreaker" questions. Answers to these questions were designed to further manipulate impressions and included major (Pre-Med vs. undecided, an agency manipulation) and most embarrassing photograph (falling down at sister's wedding vs. flashing a crowd at Marti Gras, a virtue manipulation). The final impression manipulation was a trivia game in which players listed examples of a category for 30 seconds. Low-agency confederates received easy categories (recent U.S. Vice Presidents) and did poorly, where high-agency confederates received difficult categories (cities in Africa) and did well. After the trivia game, participants completed a "first impressions" questionnaire, which verified that high-agency confederates were perceived as significantly more agentic (e.g., intelligent, assertive) and high-virtue confederates as significantly more virtuous (e.g., promiscuous-reversed) than their low-agency or low-virtue counterparts. The rest of the experiment was devoted to providing participants with a series of opportunities for courteous and helpful behavior. For example, each pair moved from one room to another to complete a separate task, and confederates lingered behind to see if their partner would hold the door for them. Upon reaching the second room, a precariously perched box of pencils under a desk was knocked over by confederates, who then observed how much help they received in picking up the pencils. A total of four behaviors were combined to form an index of helpful and courteous behavior. The question of whether chivalrous men discriminate among women based on their level of virtue or agency was assessed by testing whether participants' chivalry scores (assessed by the Chivalry Scale at the end of the experiment) significantly interacted with the agency or virtue of confederates. Chivalry did not significantly interact with the agency of confederates, most likely due to the unanticipated confounding of agency with friendliness. High-agency confederates were rated as significantly more friendly than low-agency confederates, and were liked by all participants. However, a significant interaction was observed between chivalry and the virtue of confederates. Whereas low-chivalry participants did not discriminate among the female confederates based on their level of virtue, high-chivalry participants did. High-chivalry participants exhibited significantly more courteous and helpful behavior toward the high-virtue confederates than toward the low-virtue confederates. Courteous behavior did not significantly interact with the other individual-difference measures administered, including the Attitudes Toward Women Scale, the Ambivalent Sexism Inventory, or its two subscales.
3. Reactions to female subtypes varying in virtue and agency.
Research on sex stereotypes has consistently identified three subtypes of women: traditional homemaker, career woman, and sexually permissive woman. In the third study, modeled after Glick, Diebold, Bailey-Werner, and Zhu (1997), participants evaluated these three subtypes of women by indicating their global positive or negative evaluation of subtype members, how much members of the subtype possessed positive and negative traits, and how much subtype members elicited positive or negative emotional reactions from them. These measures were highly intercorrelated and were combined to form an index of positive or negative evaluation. As before, participants completed a Chivalry Scale after they evaluated each subtype. Evaluations of the three subtypes varied significantly by the degree to which participants endorsed the chivalry script. Repeated-measures analysis of variance indicated that high-chivalry participants evaluated the traditional subtype significantly more positively than both the permissive subtype and the career woman subtype. The differences between the evaluations of low- and high-chivalry participants were also significant: for males, these differences were significant for the homemaker and permissive subtype; for females, differences were significant for the career woman and permissive subtype. Repeated-measures analysis of variance was conducted using individual difference measures other than chivalry, including the Attitudes Toward Women Scale and the Ambivalent Sexism Scale. Benevolent Sexism was associated with a significantly lower evaluation of career women than homemakers, but no effect was observed for permissiveness. In addition, scores on the Attitudes Toward Women Scale were associated with positive evaluations of homemakers and negative evaluations of career women.

These three studies provide evidence that the chivalry script is not for all women, but only for women who conform to expectations of high virtue and low agency. This finding is meaningful because it provides additional support for the hypothesis that chivalry is related to these two stereotypes about women and because it suggests that chivalry is not quite as "nice" as it is often made out to be. Chivalry is not indiscriminate politeness; rather, it is shown only to women who are "virtuous" enough to receive it and who do not exceed expectations for female agency.

3. What are the consequences of chivalry on the impressions of an observer?

My most recent study was designed to examine chivalry from the perspective of an observer of chivalrous behavior. This time, I tested whether merely observing chivalrous behavior leads to attributions about the female recipient that are consistent with the stereotypes mentioned above. Research on helping behavior has found that schoolchildren as young as five years old infer low ability from the receipt of help (Graham & Barker, 1990). In my experiment, participants viewed one of two types of videotapes. In one, a male student acts very chivalrously toward a female student (e.g., helps her carry books, buys a softdrink for her, etc.). In the second, the male student is not chivalrous – he interacts with the female student in all the same contexts but does not offer to carry books, buy a softdrink, or take advantage of any of the other scripted opportunities for chivalrous behavior. After participants viewed the videotape, they were asked to report their impressions of the characters using scales designed to measure independence, ambition, respect, purity, status, etc.

The male actors were perceived as significantly more ambitious, assertive, respectable, kind-hearted, and physically attractive when they were chivalrous as compared to when they were not. Female actors were perceived as significantly less independent when they received chivalry as compared to when they did not. This finding has important implications for the endorsement of chivalry as a healthy script for male-female interactions. If simply observing males offering chivalrous assistance to females leads to the perception that females are less independent, then chivalrous behavior helps to perpetuate these stereotypes. At the same time, the considerable rewards gained by male providers of chivalry suggest one reason for its perpetuation. [Poster from the 2001 meeting of the Society for Personality and Social Psychology]

Convergent and Discriminant Validity

Correlations between chivalry and constructs related to chivalry were assessed in both the videotape experiment and the subtype study. These correlations suggest that chivalry is strongly related to the concept of Benevolent Sexism proposed by Glick and Fiske (1996), but its relation to Hostile Sexism is less clear. Among female participants, endorsement of the Chivalry Scale is moderately correlated with Hostile Sexism (r's of .45, .35, and .24), but male participants show correlations (r's of .24, .09, and -.03) with a (Fisher-transformed and weighted) average value of only .09. Chivalry is also strongly correlated with the Attitudes Toward Women Scale, an indicator of the endorsement of traditional sex roles.

Future Directions

In my future research on chivalry, I will study the reactions of the recipients of chivalrous behavior. Like the decreased performance elicited by stereotype threat in Claude Steele’s research and by self-objectification in Barbara Fredrickson’s research, I expect female recipients of chivalrous behavior to experience reactions consistent with the stereotype that women are less competent and assertive than men. In addition, I will look beyond chivalry’s male-female relations aspect, examining the "gentleman" role promoted by chivalry and its relation to beliefs about fair competition and personal sacrifice.

Conference Posters

A news article on my research