Positive stereotyping: Influence tactic for prejudice reduction?

Abstract

We explored the effect of making salient a positive stereotype toward an out‐group member. Combining personalization theory and contact theory, we predicted increased acceptance of an out‐group only when: (a) positive information is presented about an individual out‐group member, rather than his or her social category, and (b) that information confirms a stereotype of that out‐group. Confirming this prediction, in two 2×2 factorial experiments, each with an additional no information control condition, participants liked an out‐group more when presented with positive stereotypical information about an out‐group member than when presented with equally positive non‐stereotypical information about that out‐group member, or when presented with either type of information about the out‐group as a whole.

With the advent of modern technology members of different social groups are interacting at a much greater rate than ever before in history. Unfortunately, such interactions continue to be marked by negative evaluations of the out‐group (prejudice) as well as negative feelings toward them, which hinder intergroup relations and at the extreme lead to discrimination. Efforts aimed at boosting positive interaction with the out‐group have focused almost exclusively on reducing stereotyping and/or eliminating prejudices, with limited success. Instead of ridding people of their negative stereotypes, as has been the subject of much research, we seek to demonstrate the positive effects of making a pre‐existing positive stereotypical trait salient.

Whilst researchers have uncovered many techniques that help reduce prejudice, including the jigsaw classroom (Aronson, Blaney, Stephin, Sikes, & Snapp, 1978) and the “robbers cave” intergroup dependency approach (Sherif, Harvey, White, Hood, William, & Sherif, 1954/1961), it is not always practical to create the necessary conditions prescribed by them, especially those that involve social interaction. So often in the modern world, intergroup contact is relegated to fleeting interactions that cannot be manipulated to ensure interactions occur in a context of equal status, norms that support the reduction of prejudice, superordinate goals, and/or a climate of cooperation. Unlike many other methods of social influence aimed at reducing prejudice, the social influence technique proposed herein does not require social interaction. Similar to the prejudice reduction tactics of creating a shared superordinate identity (e.g., Gaertner & Dovidio, 1999) or promoting empathy (e.g., Batson, Chang, Orr, & Rowland, 2002), positive stereotyping does not require the interaction of two or more individuals under a constrained circumstance. Neither does it require that observers be persuaded to change previous ideas, an often difficult task. Thus, it makes a powerful addition to the list of prejudice reduction techniques.

The introduction first discusses issues that bear on positive stereotyping, considering in addition the circumstances under which out‐group generalization is likely to occur. The next section considers the level at which out‐group information is provided, arguing that information at the individual level is likely to have greater impact than that presented at the group level. Then we elaborate the notion that when positive information about an out‐group is made salient, there is a special advantage to constraining that information to be stereotype consistent. Drawing from our discussion of these issues, we anticipate that a favorable perception of an out‐group category is more likely when there is a convergence of two circumstances: (a) positive information is presented about an individual out‐group member rather than the out‐group category; and (b) the positive information is considered to be stereotypical of the out‐group, rather than irrelevant to the stereotype of that group.

POSITIVE STEREOTYPING

Stereotype valence appears to have a strong effect on how people behave. For instance, positive stereotyping has been shown to induce positive behavior in both perceiver and actor (Snyder, Tanke, & Berscheid, 1977) and augment the actor's academic performance (Shih, Ambady, Richeson, Fujita, & Gray, 2002). According to stereotype‐as‐heuristic theory, an activated stereotype biases information processing in a stereotype‐consistent manner. Thus, the mere activation of negative stereotypic information can lead to disparaging evaluations of out‐group members (Bodenhausen, 1990; Devine, 1989; Lepore & Brown, 1997). The opposite may be true as well. Specifically, an activated positive stereotype should result in a more positive view of the referent of that information. In line with this expectation, Johnston, Locke, Giles, and Rattray (1997) showed that the activation of a positive stereotype can extrematize perceptions of an out‐group target's positive stereotypical traits. Here, we extend this research by investigating whether more general evaluations of the out‐group can be improved by making a positive stereotypical attribute salient.

INDIVIDUAL VS GROUP TARGET

When it comes to influencing group perceptions, it may seem logical that general information about a group would affect perceptions of that group more than information about a single group member. However, the exemplar‐based model of stereotyping (Smith & Zarate, 1992) posits that stereotypes are mere representations of specific individuals, implying that contact with a single individual can alter generalizations about an out‐group (Allport, 1954). Several avenues of research now suggest that individuating information, in comparison to group‐level information, may be perceived and weighed differently even if the content of the information is largely the same. Individuating information has also been shown to increase compliance to the target of a request (Pratkanis, in press) suggesting that individual‐level information may be more persuasive than information received at a group level. This idea that individuating information may have greater impact on social judgments than information presented at the group level is supported by research on the use of heuristics which demonstrates that base rate and/or consensus information is often ignored or under‐weighed in favor of individuating information (Ginossar & Tropes, 1987; Kahneman & Tversky, 1972; Lockesley, Hepburn, & Ortiz; 1982; Zukier & Pepitone, 1984). Additionally, representativeness is often over‐weighted at the expense of outcome probability (Kahneman & Tversky, 1973). Research on eyewitness testimony further supports this principle. Information about a specific person exerts more influence on jury decisions than sample‐based general information about people similar to the target (Maass, Brigham, & West, 1985). Research on the better‐than‐average effect (Alicke, Klotz, Breitenbecher, Yurak, & Vredenburg, 1995) found greater reductions in the exhibition of this bias on the part of participants who compared themselves to an individuated target by comparison with those who compared themselves with an average college student, leading these authors to suggest that the abstract nature of the group information provoked less thought than individuating information. Indeed, Nisbett and Borgida (1975) suggest that because base rate information is abstract, it is often neglected in favor of more “vivid” individuating information.

When information is presented about an individual, rather than that individual's group, it may induce more thought and hence more attitude change (Brewer & Miller, 1984) simply because it is not ignored or discounted, as is likely with group information. Furthermore, even the illusion of receiving individual information over group‐level information can lead to differences in decision making (Yzerbyt, Schadron, Leyens, & Rocher, 1994). Thus, information about representative individuals is expected to have greater impact on perceptions of the group than general information about that group.

TYPICALITY AND GENERALIZATION

The contact hypothesis suggests that negative stereotyping and prejudice can be reduced, and positive out‐group perceptions increased, by inducing contact between members of two different groups (Hewstone, 1996). Confirming evidence for this hypothesis abounds (e.g., meta‐analyses: Johnson & Johnson, 1989; Miller & Davidson‐Podgorny, 1987; Pettigrew & Tropp, 2000). Even the presentation of information about group members, in the absence of actual contact, has been shown to reduce negative bias toward the out‐group category in general (e.g., Bettencourt, Brewer, Croak, & Miller, 1992; Brewer & Miller 1984; Ensari & Miller, 2002; Urban & Miller 1998; Worchel 1986). Similarly, drawing attention to positive exemplars has been found to improve out‐group evaluations (Schwarz & Bless, 1992). However, support for the contact hypothesis is not unequivocal. Hamburger (1994) points to the dearth of evidence demonstrating generalization from a single positive contact to other members of the out‐group as a critical weakness of the contact hypothesis. Specifically, he laments that often the beneficial effects of contact are often limited to the participants in that contact situation. One goal of our research is to respond to Hamburger's (1994) observation by providing evidence that knowledge about a single out‐group member can increase acceptance of that out‐group category, if certain conditions are met.

In category‐based contact, an in‐group member tends to respond to out‐group members as interchangeable representatives of an undifferentiated category (Brewer & Miller, 1984). This implies that when category membership is cued, information about a single out‐group member might routinely be applied to the out‐group as a whole. However, the exemplar‐based hypothesis of attitude change suggests that contact with an out‐group member may not always activate stereotypes or beliefs about the group (Hilton & Von Hippel, 1996). Thus, Hewstone and Brown's (1986) Mutual Intergroup Differentiation Model argues that group membership must somehow be made salient for any generalization to the out‐group to occur. Yet individuals with stereotype‐inconsistent characteristics may be subcategorized as special cases and therefore not seen as part of a general out‐group (Lui & Brewer, 1983; Weber & Crocker, 1983). Indeed, there is much consensus that with respect to altering negative stereotypes or changing group perceptions, the typicality of individuals who exhibit stereotype‐disconfirming behavior is important (Bodenhasuen, Schwarz, Bless, & Wanke, 1995; Johnston & Hewstone, 1992; Rothbart & Lewis, 1988; Rothbart, Sririam, & Davis‐Stitt, 1995). Generalization is more likely to occur when participants lack a basis for subtyping an individual member of a social category (Kunda & Oleson, 1995). Thus, by decreasing the likelihood of subtyping, typicality promotes generalization to the out‐group category (Ensari & Miller, 2002; Hewstone & Lord, 1998; Stapel & Koomen, 1998). In contrast, stereotype inconsistent information is often ignored and/or forgotten, whereas stereotype‐consistent information garners more attention and thought and is better remembered (Bodenhausen & Lichenstein, 1987; Bodenhausen & Wyer, 1985; Cameron & Trope, 2004). In accord with these ideas, those who met a friendly, pleasant out‐group member who was typical of his group subsequently rated the entire out‐group more favorably than did those who had the same type of positive contact with an atypical out‐group member (Wilder, 1984).

THE PRESENT STUDIES

Currently, little research examines the effect of positive stereotype‐consistent information. Nevertheless, the research cited above does suggest that positive trait‐related information, when exhibited by an individual who is seen as typical of his group, is likely to affect perceptions of other members of that social category. However, no study has tested this notion by comparing the effect of presenting positively valenced information about an individual out‐group member with that induced by presenting that same information about the out‐group category as a whole, while simultaneously manipulating whether or not the positive information is stereotype consistent or stereotype irrelevant.

Brewer and Miller's (1984) personalization model argues that individual information can often carry more weight than group information. Hewstone and Brown's (1986) model of contact suggests that only situations that keep category cues salient (stereotype‐consistent conditions) will promote generalization. Thus, a merging of the two models suggests a more beneficial effect when an individual out‐group member is alleged to possess a positive stereotypical trait, as opposed to an equally positive non‐stereotypical trait. By ensuring that the positive trait attributed to the individual out‐group member is stereotype consistent, it not only maintains the salience of the social category of its referent, but additionally, it precludes subtyping of that individual. Thus, the effect of making a stereotypical trait salient will be compared with that seen when an equally positive non‐stereotypical trait is made salient. Additionally, however, as previously argued, benefit is more likely when the positive stereotypical trait information that is made salient refers to an individual out‐group member as opposed to that out‐group social category.

EXPERIMENT 1

Experiment 1 was conducted in Spain. We manipulated stereotypicality of the trait (stereotypical versus non‐stereotypical of the out‐group) and the referent of the stereotypical information (individual versus group). Based jointly on the personalization model of beneficial contact effects (Brewer & Miller, 1984), the exemplar‐based model of attitude change (Smith & Zarate, 1992), and Hewstone and Brown's (1986) Mutual Intergroup Differentiation Model, we predicted that participants exposed to information emphasizing a positive stereotypical trait in an individual out‐group member would exhibit greater liking toward that out‐group category when compared to participants: (1) for whom that same information was alleged to characterize the out‐group in general, (2) who were exposed to positive non‐stereotypical information about the out‐group individual, (3) who were exposed to non‐stereotypical positive information about that out‐group category, and (4) who received no information about the out‐group.

Method

Overview

In a field study, we used a 2 (referent of information: individual or group)×2 (type of positive trait: stereotypical or non‐stereotypical) between‐subjects factorial design with an additional no information control condition. Except for the control condition, each participant was presented with a newspaper article to read. Afterwards, all participants were given a set of dependent measures that contained manipulation checks and assessed liking of the out‐group.

Pretests

A total of 40 Spaniards were individually presented with a list of several national/ethnic out‐group categories and asked to generate adjectives describing members of those groups. Our purpose was to assess liking for these social groups and to identify traits that are stereotypically associated with them. From the list, Germans were chosen as the out‐group because there were strong consensual positive and negative stereotypes associated with them. Generally, the Spanish participants reported that they neither especially liked nor disliked Germans. For the question “How much do you like Germans,” on a 7‐point scale with 1 indicating “not at all” and 7 indicating “very much,” the mean did not reliably differ from the midpoint of the scale (M = 3.45).

A separate group of Spaniards (N = 37) was individually presented with lists of 31 positive dispositional traits partially generated from the first pretest (stereotypical adjectives) and partially selected by the experimenter (non‐stereotypical adjectives). They were asked to rate each adjective's valence on a 7‐point scale (with 0 labeled as most negative and 6 as most positive). From this measure, two traits were selected as being approximately equal in positive valence (p>.05): one that was frequently found to be stereotypical of Germans (hardworking, M = 4.8) and one that was not (creative, M = 4.6).

A third pretest conducted on another sample from this same population further assessed the degree to which traits were typical of Germans. Participants rated the typicality (or irrelevancy) of each trait with respect to Germans in general using 5‐point scales ranging from −2 (very atypical) to +2 (very typical). Ratings not significantly different from 0 were considered stereotype irrelevant (e.g., the positive traits creative, M = .3, and generous, M = −.3, fell into this domain). By contrast, hardworking had a mean rating of 1.7, giving further indication that this trait is widely considered to be stereotypical of Germans.

Participants

The 65 participants for the main study were all citizens of Spain who agreed to serve in a 15‐minute study on “memory for social perception.” They were randomly assigned to one of five experimental conditions: an individual/hardworking condition, an individual/creative condition, a group/hardworking condition, a group/creative condition, and a control condition (in which they saw no newspaper article). Participants ranged in age from 18 to 31, with a mean age of 22.49. No information on gender was recorded.

Stimulus materials

There were four different alleged newspaper articles (one for each experimental condition). Each article, in Spanish, was written by the experimenters and then pasted into a local newspaper before photocopying the news page (see Appendices for the English versions of these articles). Thus, the article appeared to be written by a fellow Spaniard. One article described a German friend1 of the article's author who was portrayed as hardworking (individual/stereotypical trait condition). A second article included a description of the German friend, but this time he was depicted as creative (individual/non‐stereotypical trait condition). The other two articles presented statistics that described Germans in general as either hardworking (group/stereotypical trait condition) or creative (group/non‐stereotypical trait condition). Aside from the manipulation of the referent of the information and stereotypicality of the trait (hardworking: stereotype‐consistent vs creative: stereotype‐irrelevant), the four articles were highly similar in content.2

Procedure and dependent measures

Participants were approached on streets in the Basque country of northern Spain and asked to participate in a short study testing social memory. They were told they would read a newspaper article and be tested on their memory for it after they had completed a distraction task. Participants in the four experimental conditions were then given one of the four newspaper articles to read as part of a memory task. Afterwards, they responded to a written questionnaire, which was presented as a “distraction task.” It included several questions assessing liking for Germans. The following items were rated on 7‐point linear scales with described endpoints: “How likable are Germans?” (1 = not at all, 7 = very much), “To what degree does each of these characteristics describe Germans (nice, cold, and violent)?” (1 = not at all, 7 = very much), and “What is your general impression of Germans?” (1 = very negative; 7 = very positive). Afterwards, each participant in the experimental conditions completed a short memory test, which included manipulations checks. The first item was a free response question assessing memory for the target's nationality: “In the article a person [group] was described; what was his [their] ethnicity?” A second free response item determined memory for the descriptive trait associated with the target(s): “In referring to his [their] personality, how was this person [group] described?” Finally, the perceived typicality of stereotypical and non‐stereotypical descriptive traits was assessed in two items: “Do you consider hardworking [creative] as a characteristic trait of Germans in general?” (1 = not characteristic, 7 = very characteristic). Participants in the control condition did not read a newspaper article, but instead merely filled out the written questionnaire.

Results and discussion

Manipulation checks

Participants were asked to recall who was described in the newspaper to assess whether they identified the group membership of the target out‐group person correctly. Every participant correctly remembered the target's nationality. All participants also correctly identified the descriptive trait associated with the target according to their experimental condition. Confirming the pretest data, “hardworking” (M = 5.58) was perceived to be more typical of Germans than “creative” (M = 3.40), t(47) = 5.74, p<.001.

Liking for Germans

The key dependent measures were the ratings of five items: nice, cold, violent, liking, and general impression of Germans. They were combined to create a composite measure of general liking toward Germans (in the composite measure, cold and violent were reverse coded). Cronbach's alpha for all five items was .71.

A 2×2 ANOVA that did not incorporate the control condition indicated a main effect for trait. The stereotypical trait increased liking for Germans more than the non‐stereotypical trait, F(1, 48) = 4.06, p = .05, η² = .083. The individual target marginally increased liking for Germans more than the group target, F(1, 48) = 2.82, p = .10, η² = .059. More important, however, there was also an interaction between trait and target, F(3, 48) = 4.14, p = .048, η² = .084. Inspection of the means, as presented in table 1, shows that the interaction, as well as the trait and target type main effects, are solely attributable to the individual/hardworking cell. Further confirming these effects, a one‐way ANOVA of the five conditions indicated a difference in liking for the German out‐group among the five conditions, F(4, 60) = 2.84, p = .032, η² = .16. As predicted, the individual/hardworking condition produced increased liking for Germans in general (M = 4.68). LSD post hoc comparisons revealed that only participants in the individual/stereotypical condition showed an increase in liking toward Germans as compared to the control condition, p = .002. The level of liking for the out‐group produced by the individual/stereotypical condition was also greater than that found in all three of the other experimental conditions (ps<.024). In contrast, there were no differences among each of the latter three experimental conditions and the control condition on the composite measure of liking toward Germans, ps>.48.

Table 1. Mean evaluations of Germans in Experiment 1

	Hardworking	Creative
Individual
M	4.68	3.88
SD	0.43	0.98
n	12	12
Group
M	3.95	3.95
SD	0.55	0.68
n	12	13
Control			3.75
M			0.98
SD			16
The dependent variable was liking toward Germans (average composite of 5 items: nice, likable, cold, violent, and general impression of Germans).

These results support the notion that positive stereotypical information about an individual can induce perceivers to adopt more positive feelings about an out‐group. Equally positive information about an individual out‐group member that was not stereotype‐consistent failed to alter attitudes about the out‐group category, perhaps because possession of a non‐stereotypical attribute causes that individual to be subtyped. We showed that positive stereotypical information about an individual can increase liking for the group in general. However, when this same descriptive information was alleged to characterize the social category of that individual, no such benefit was obtained. These results suggest that the daunting task of eliminating negative stereotypes is not necessary in order to ameliorate intergroup perceptions. Instead, one can utilize already‐existing positive stereotypes to boost positive intergroup feelings.

EXPERIMENT 2

Experiment 2 further explores the effect of making a positive stereotypical trait of an individual salient on liking for the out‐group in general. It also examines the perceived variability of members of the relevant out‐group as a possible mediator of this effect.

Numerous studies indicate that out‐groups generally tend to be seen as more homogenous than the in‐group (e.g., Baron, Kerr, & Miller, 1992; Judd & Park, 1988; Mullen & Hu, 1989; Park & Judd, 1990, Park & Rothbart, 1982). A focus on an individual out‐group target, as opposed to the out‐group category, may encourage thoughts about individual differences and thereby make the out‐group as a whole seem more variable in their attributes. Alternatively, or additionally, positive stereotype‐consistent information may increase the accessibility of thoughts about other positive attributes of members of the out‐group category, thereby increasing perceptions of group variability (Werth & Lord, 1992). By contrast, if stereotype‐inconsistent information is indeed seen as uncharacteristic of the group, it may also be seen as irrelevant and thus fail to elicit further thoughts about individual differences. This is in line with the research of Paolini, Hewstone, Rubin, and Pay (2004), who found that exposure to information about a single group member increased perceptions of group dispersion only when stereotype‐relevant information was present.3 To explore these ideas, we added a variability measure to Experiment 2. We expected to find that perceptions of out‐group variability would increase when stereotype‐relevant information is presented about an individual target and that this increase will mediate prejudice toward the out‐group.

Method

Overview

A field study was conducted to replicate and extend the effects of Experiment 1 with a different sample population. Participants were approached by one of two experimenters and individually asked to participate in a study on social perception. There were five conditions, one control and the four cells of a 2 (target: group or individual)×2 (trait: stereotypical or non‐stereotypical) between‐subjects design. After reading about the target and completing manipulation checks and evaluations of the target person or group, participants were asked to rate how variable they saw the out‐group.

Participants

One of two experimenters4 approached individuals in several places within southern California (e.g., at the mall, on the campus of Pasadena Community College) and asked them to participate in a short experiment. Of the 50 participants recruited in this manner, the data from 46 participants were included in the final data analysis. Four participants were excluded for suspiciousness or failure to fill out all the study measures. Participants were between the ages of 18 and 72 with a mean age of 30.63. Approximately 54.3% were male and 45.7 % were female. There were no significant interactions between gender and any of the experimental manipulations.

Stimulus materials

The materials used in this experiment were English translations of those used in Experiment 1. There were four different newspaper articles, one for each experimental condition. The articles were all prepared by the experimenter and then pasted into the USC school newspaper in order to create the illusion that a student had written the article. Pretests confirmed that, for this population, the word “hardworking” was considered typical of Germans (M = 1.32), t(36) = 9.84, p<.001, while “creative” was not (M = .26), p>.05 (−2 = very atypical, +2 = very typical). A separate pretest conducted on this population indicated that the words “hardworking” (M = 6.20) and “creative” (M = 5.98) did not differ in positivity, p>.05 (1 = very negative and 9 = very positive). As in Experiment 1, participants in the control condition saw no newspaper article.

Dependent measures

Experiment 1 used dispositional attributions and an overall impression item as a measure of liking. To see if the effects of positive stereotyping also influenced behavior tendencies with respect to the out‐group, we changed our dependent measures for Experiment 2. We asked participants, “How much would you like to be stranded on an island with someone from each of these groups?” and “How much would you like to go to a movie with someone from each of these groups?” (Germans were included in the list; 1 = not at all, 7 = very much). We also included the measure of overall impression of Germans from Experiment 1 as a general measure of prejudice reduction. These three items were combined into a single “liking” index (α = .87). In addition, a measure of variability was added so that we could test for mediation. On a 7‐point linear scale, participants were asked: “To what degree are members of each of the following groups all alike or all different?” (1 = all alike, 7 = all different) (Stangor, Jonas, Stroebe, & Hewstone, 1996).

Procedure

As in the previous experiment, participants were asked to participate in a short experiment on “memory for social perception.” Immediately after recruitment and their signing of a consent form, they were asked to read an article about which they would later be tested for memory. As in Experiment 1, the main dependent variables were located in the questionnaire that was presented as a distraction task. After completion of the questionnaire containing the main dependent variable, participants completed a memory test. Under this guise, they were given the same manipulation check items used in Experiment 1 (two open‐ended questions asking participants to recall the nationality of the target described in the newspaper article and the descriptive trait used to describe him/them, and two items rating the perceived stereotypicality of the descriptive traits: “hardworking” and “creative” in regard to Germans). Finally, participants were thanked, verbally debriefed as to the true nature of the study, and given an opportunity to exclude their data before they signed a form indicating they had in fact been debriefed. No participants chose to withhold their data.

Results and discussion

Manipulations checks

All participants indicated that the target was German. All participants also correctly identified the descriptive trait used to describe their target according to their assigned condition. Participants in the hardworking trait (stereotypical) condition (M = 4.95) considered that trait as more characteristic of Germans in general than participants in the creative trait (non‐stereotypical) condition (M = 4.18), t(33) = −2.46, p = .019.

Liking measure

A 2×2 ANOVA that did not incorporate the control condition was carried out on the liking index described earlier. It showed a main effect for trait, F(1, 32) = 9.35, p = .004, η² = .226, and a marginal main effect for target, F(1, 32) = 3.77, p = .061, η² = .105. Although the omnibus interaction between trait and target was not significant (p<.39), as is recommended when testing a specific theoretically derived interaction pattern (Abelson & Prentice 1997; Rosenthal, Rosnow, & Rubin, 2000) we conducted an a priori interaction test using contrast weights +3 for the individual/stereotypical cell and −1 for the other three experimental cells, to examine the specific interaction hypothesis predicted by our theoretical analysis and supported by the results obtained in Experiment 1. The obtained reliable contrast, t(41) = 3.43, p = .001, confirms our predicted interaction. A test of residual variance that was unaccounted for by the planned contrast was not reliable, F(2, 36) = 2.50, ns. In further confirmation, a one‐way ANOVA applied to the five conditions revealed a difference in liking for the out‐group, F(4, 41) = 3.53, p = .014, η² = .26,5 which, as shown by LSD post hoc comparisons, indicated that participants in the individual target/stereotypical trait condition liked the out‐group more (see table 2) than participants in each of the other conditions (ps<.046). Moreover, when compared to the control condition, only participants who read an article about a hardworking German (stereotypical/individual target condition) exhibited increased liking toward the out‐group as a whole (mean difference = 1.55, p = .033). No other experimental condition differed significantly from the control condition or from each other, ps>.12.

Table 2. Mean evaluations of Germans in Experiment 2

	Hardworking	Creative
Individual
M	6.56	4.52
SD	1.15	1.50
n	9	8
Group
M	5.1	3.96
SD	1.24	2.14
n	10	9
Control
M			5.00
SD			1.48
n			10
The dependent variable was liking toward Germans (average composite of three items: general impression of Germans, how much you would like to be stranded on an island with a German, and how much would you like to watch a movie with a German).

Variability measure

A one‐way ANOVA on the perceived variability measure showed differences between conditions, F(4, 40) = 3.83, p = .01, η² = .28. A post hoc LSD comparison revealed that those in the individual target/stereotypical trait condition perceived the out‐group as marginally more variable (M = 6.00) than did those in the control condition (M = 4.80), p = .07. The out‐group was also seen as more variable by those in the individual target/stereotypical trait condition, than by participants in both of the non‐stereotypical trait conditions, ps<.005 (see table 3 for means). An independent samples t‐test conducted on this measure indicated that the variability of the out‐group was not affected by target type, p>.05. However, variability of the out‐group was affected by trait; when targets possessed stereotypical traits, the outgroup was perceived as more variable (M = 5.7) than when the targets possessed non‐stereotypical traits (M = 3.87), t(33) = −3.89, p = .001. Both non‐stereotypical target conditions failed to produce a significant increase in out‐group variability when compared to the control condition, ps>.17. These findings support the idea that non‐stereotypical information does not impact perceptions of out‐group variability.

Table 3. Mean perceptions of variability of Germans in Experiment 2

	Hardworking	Creative
Individual
M	6.00	3.86
SD	1.66	0.69
n	9	7
Group
M	5.50	3.89
SD	1.43	1.62
n	10	9
Control
M			4.80
SD			1.40
n			10
The dependent variable was “to what degree are Germans all alike or all different?” rated on a 7‐point scale (1 = all alike, 7 = all different).

We next explored the hypothesis that variation in perceptions of group variability resulting from reading the news articles mediated the greater liking of the out‐group seen in the individual/stereotypical cell compared to the no information control condition (Baron & Kenny, 1986). The individual/stereotypical condition was associated with greater liking for the out‐group, β = .52, p = .022 and was also marginally associated with an increased perception of out‐group variability, β = .38, p = .10. Perception of out‐group variability predicted liking for the out‐group, β = .57, p<.001. Finally, when controlling for perceived variability, the individual/stereotypical condition was no longer associated with increased liking for the out‐group, β = .27, p = .11, but variability remained a significant predictor of liking, β = .66, p<.001. Thus, with the caveat that the individual/stereotypical condition was only marginally linked to perceived variability, the four criteria suggested by Baron and Kenny (1986) for establishing mediation were fulfilled.

When this same mediational analysis was run backwards, however, it showed the mediational implication of the preceding analysis to be untenable. The individual/stereotypical condition marginally predicted perceptions of out‐group variability, β = .38, p = .10. The individual/stereotypical condition also reliably predicted liking, β = .52, p = .022. Liking predicted perceived variability of the out‐group, β = .57, p<.001. Finally, when controlling for out‐group liking, the individual/stereotypical condition no longer predicted variability for the out‐group, β = −.025, p = .89, but liking remained a significant predictor of variability, β = .78, p = .001. Thus, with the caveat that the individual/stereotypical condition was only marginally linked to perceived variability, the four criteria suggested by Baron and Kenny (1986) for establishing mediation were again fulfilled, suggesting this time that liking mediated perceived group variability. Taken together, these analyses cannot support the view that perceptions of out‐group variability are a mediating cause of increased liking for the out‐group. They are more consistent with the notion that liking and variability are covarying symptoms of the experimental manipulations.

In sum, the results of Experiment 2, which used Anglo participants, replicated those of Experiment 1, which used Spanish participants. It showed that exposure to information about a stereotype‐consistent attribute of an out‐group individual produced more positive perceptions of that out‐group. We also examined the role that perceived out‐group variability plays in positive stereotyping. The added measure of variability indicated that exposure to information about a stereotype‐consistent attribute of an out‐group individual marginally increased perceptions of out‐group variability compared with those who had no such information. In contrast, exposure to either an individual or a group failed to affect perceptions of out‐group variability. Nevertheless, mediational analyses failed to support the notion that out‐group variability mediated the increase in liking for the out‐group that was found when the target was portrayed as both stereotypical and as an individual. They suggest instead that perceptions of variability and out‐group liking are covarying consequences of the experimental manipulations.

GENERAL DISCUSSION

This research investigated the effects of positive stereotyping on out‐group perceptions. Both experiments showed that liking for the out‐group can be increased by exposure to information about a single member of that out‐group who possesses a salient stereotypical trait. Exposure to the same trait information about the group as a whole did not augment positive perceptions of the out‐group. Likewise, exposure to positive stereotype‐irrelevant trait information about an individual or group failed to promote more positive perceptions of the out‐group. These findings fit with Brewer and Miller's (1984) predictions on individuating information and group perceptions and with the exemplar‐based model of stereotype change (Smith & Zarate, 1992). Information presented at the individual level had more impact than information presented at the group level only when that information cued category membership, thereby indicating typicality and simultaneously precluding subtyping.

In Experiment 2 we expected out‐group variability to increase with exposure to an individual stereotype‐consistent target, allowing us to explore whether variability mediated the prejudice reduction found under these conditions. Although there was no reliable difference in out‐group variability between target types, we did find that the provision of stereotypical trait information about an individual did tend to increase perceptions of out‐group variability, by comparison with conditions wherein no information was presented about a target at all or in conditions in which the target possessed a non‐stereotypical trait. Mediational analyses, however, provided no support for the view that positive perceptions of the out‐group were caused by a perception of increased out‐group heterogeneity. In view of these findings, future work in this area should look to other possible causes for the increased positive perceptions created by the stereotypical trait/individual target conditions. Perhaps the individual target induces more attention and hence better memory than the group target. Perhaps, too, it avoids eliciting the negative affect typically associated with out‐group. Future research in this area should therefore include mood and memory measures.

At this point, it seems prudent to acknowledge a few limitations of the current research. From our manipulations, it is impossible to state the duration of the effects of positive stereotyping. Future research should explore the long‐term effects of positive stereotyping and determine if out‐group perceptions are only boosted temporarily, while the positive information remains accessible in memory, or whether this reduction in prejudice is more long lasting. Another limitation is that the out‐group only elicited mild prejudice in these experiments. Thus, future research is needed to demonstrate whether the affects of positive stereotyping generalize to more strongly stigmatized groups. Additionally, the extent to which our findings have a more general application would be strengthened had we extended them to more target out‐groups, each of which possesses its own distinctive stereotype.

Based on research on the contact effect, some may argue that the description of the individual target as a “friend” of the author's may, by itself, boost evaluations for the target's out‐group (Pettigrew, 1997; Wright, Aaron, McLaughlin‐Volpe, & Ropp, 1997). As applied to this experiment, however, the out‐group friendship effect only predicts a main effect of target. The friendship effect does not explain our obtained interaction between target and trait or the results of two one‐way ANOVAs, wherein LSD post hoc tests showed that only the individual target/stereotypical trait condition received more favorable evaluations than the control condition and that evaluations in the individual target/stereotypical trait condition were more positive than those in all three of the other experimental conditions (which did not reliably differ from the control or each other).

Implications for intergroup relations

At the individual level, positive stereotyping has been shown to positively bias the behavior both of targets (Shih et al., 2002) and perceivers (Snyder et al., 1977). These studies suggest that if people are encouraged to focus on the positive stereotypical characteristics of individual out‐group members, they are likely to feel more positive about that out‐group. Furthermore, direct contact is not necessary for these effects to occur. Methods of social influence that present information about individual group members while at the same time emphasizing their positive stereotypical traits are powerful enough to reduce prejudice. These types of experiences may reduce feelings of competitiveness and other negative cognitions and emotions that normally arise in intergroup situations, while increasing accessibility of positive thoughts and feelings toward an out‐group and its members.

In sum, our outcomes suggest that it is not necessary to eradicate negative stereotypes, a notoriously difficult task, in order to improve perceptions of out‐groups. Positive perceptions of out‐groups can be bolstered instead by emphasizing a positive trait that is considered stereotypical of that group. This implies that prejudice can be combated by increasing contact with individual out‐group members whose positive stereotypical attributes are salient, or merely increasing access to individual‐level information about out‐group members. Since it does not require complicated social interactions specially designed to control a host of factors known to be critical to reducing prejudice (i.e., be of equal status, have a superordinate goal, etc.), this technique of social influence has the added benefit of being extremely easy to implement in a world filled with a multitude of media outlets. Positive stereotyping can capitalize on media outlets like newspaper articles or Internet web pages and blogs to induce more positive perceptions of out‐group members. Positive stereotyping may in fact be the key to offsetting poor group relations in the modern multi‐cultural world and paving the way for more positive intergroup connections to develop.

Notes

1. The reference to the target as a friend of the author's in the individual conditions cannot account for the obtained results in view of the fact that the LSD post hoc comparisons on the one‐way ANOVAs for both studies found no reliable differences between the individual/non‐stereotypical target condition and any of the other conditions save the individual/stereotypical target condition (ps>.43).

2. The individual and group articles differed slightly in length (215 words and 150 words respectively). Thus, length is confounded with the “individual vs group” manipulation. However, it is unlikely that the difference in article length can account for the obtained effect. The one‐way ANOVAs conducted on both studies yielded a simple effect between the two individual conditions (stereotypical and non‐stereotypical) but no simple effect is found between the group conditions. Yet for the comparison within each of these simple effects, length is controlled. Thus, length per se cannot simultaneously account for (a) the effect of trait found in the two individual conditions and (b) the absence of an effect of trait in the two group conditions.

3. Paolini et al. (2004) also tested whether positive information about out‐group members would decrease prejudice toward the out‐group, but failed to find this effect. This may be attributable to the fact that target profiles contained both stereotypical and counterstereotypical traits, rather than solely the former.

4. There were no significant experimenter×treatment interactions.

5. There are relatively smaller n sizes in Experiment 2 than Experiment 1. Nevertheless, a reliable effect was found, indicating a robust phenomenon. In addition, the fact that these results conceptually replicate those of Experiment 1 further boosts our confidence in them.

APPENDIX A

Article used for the individual target/stereotypical trait condition.

Appreciating Others

Editorial: Personal thoughts on friends

By JULIE MARSHALL

I loved the article by Nancy Grove on October 26th. I think she is right when she says that all of us, without exception, are fortunate to have incredible people around us. The problem is that we often are so focused on ourselves that we forget to look outside and learn from other people, or to tell them how much we admire them and how much they teach us.

One of the persons that I admire the most is my friend Mark, a German student that I met a few months ago. I admire Mark because he is one of the most hardworking people I have ever met. Mark is one of those people who always does his job. In our study group, he never puts aside a task, and he doesn't mind staying as long as he needs to finish a project. Mark is always busy doing something productive, and he finds the time to complete what he is doing.

If there is something that defines Mark, besides being German, and all who know him would agree, it is his capacity for work, and all of his friends think that is one of his best qualities.

|see Editorial, page 15|

Appendix B

Article used for the individual target/non‐stereotypical trait condition.

Appreciating Others

Editorial: Personal thoughts on friends

By JULIE MARSHALL

I loved the article by Nancy Grove on October 26th. I think she is right when she says that all of us, without exception, are fortunate to have incredible people around us. The problem is that we often are so focused on ourselves that we forget to look outside and learn from other people, or to tell them how much we admire them and how much they teach us.

One of the persons that I admire the most is my friend Mark, a German student that I met a few months ago. I admire Mark because he is one of the most creative people I have ever met. Mark is one of those people who always has original, out of the ordinary ideas. When trying to solve a problem, he always does it with imagination and finds creative solutions to any kind of situation. Mark enjoys all kinds of creative activities and does everything in a different and innovative way and not just because he is German. He always has an original point of view about things.

Mark is definitely one of the most creative people I know, and all of his friends think that is one of his best qualities.

|see Editorial, page 15|

Appendix C

Article used for the group target/stereotypical trait condition.

Appreciating Others

Poll: National survey indicating hardworking group

By JULIE MARSHALL

Results from a recent national survey about different social groups indicate that the Germans are considered one of the most hardworking people, when compared with other groups and nationalities. In general, people concur that the Germans are very hardworking and that they always do their job.

Both those who have a personal relationship with someone from Germany and those who have a business relationship with the Germans agree: they never put aside a task, they don't mind staying as long as they need to finish their job, they are always busy, and they always find the time to complete what they are doing.

In general, most people agree that, compared with other people or nationality groups, the Germans are very hardworking and conclude that that is one of their best qualities.

|see POLL, page 15|

Appendix D

Article used for the group target/non‐stereotypical trait condition.

Appreciating Others

Poll: National survey indicating creative group

By JULIE MARSHALL

Results from a recent national survey about different social groups indicate that the Germans are considered as one of the most creative people, when compared with other groups and nationalities. In general, people conclude that the Germans are very creative and that they always have original, out of the ordinary ideas.

Both those who have a personal relationship with someone from Germany and those who have a business relationship with the Germans agree: when trying to solve a problem, they always do it with imagination, they always find creative solutions to any kind of situation, they enjoy all kinds of creative activities, and they do everything in a different and innovative way.

In general, most people agree that the Germans have an original point of view about things and conclude that that is one of their best qualities.

|see Poll, page 15|