When Groups Are Not Created Equal: Effects of Group Status on the Formation of Intergroup Attitudes in Children

Similar documents
Effects of Physical Atypicality on. Children s Social Issues and Intergroup

In this chapter we discuss validity issues for quantitative research and for qualitative research.

Defining Psychology Behaviorism: Social Psychology: Milgram s Obedience Studies Bystander Non-intervention Cognitive Psychology:

EVALUATE SOCIAL IDENTITY THEORY. Pages Social Identity 4:22

The Role of Modeling and Feedback in. Task Performance and the Development of Self-Efficacy. Skidmore College

CHAPTER LEARNING OUTCOMES

Supplementary Study A: Do the exemplars that represent a category influence IAT effects?

Chapter 11 Nonexperimental Quantitative Research Steps in Nonexperimental Research

Does Stereotype Threat Require Stereotypes? Sarah LeStourgeon & David Phelps. Hanover College

Intelligence. Exam 3. Conceptual Difficulties. What is Intelligence? Chapter 11. Intelligence: Ability or Abilities? Controversies About Intelligence

Breaking the Bias Habit. Jennifer Sheridan, Ph.D. Executive & Research Director Women in Science & Engineering Leadership Institute

Organizational Behaviour

Risky Choice Decisions from a Tri-Reference Point Perspective

Intelligence. PSYCHOLOGY (8th Edition) David Myers. Intelligence. Chapter 11. What is Intelligence?

Perceptual and Motor Skills, 2010, 111, 3, Perceptual and Motor Skills 2010 KAZUO MORI HIDEKO MORI

Intelligence. Exam 3. iclicker. My Brilliant Brain. What is Intelligence? Conceptual Difficulties. Chapter 10

PERSON PERCEPTION September 25th, 2009 : Lecture 5

2. Which of the following is not an element of McGuire s chain of cognitive responses model? a. Attention b. Comprehension c. Yielding d.

Encoding of Elements and Relations of Object Arrangements by Young Children

11/29/2017. Gender is salient and, not surprisingly, important. Boys and girls have play preferences: By 4 years of age:

PERCEIVED BARRIERS AND ENDORSEMENT OF STEREOTYPES AMONG ADOLESCENT GIRLS OF COLOR IN STEM. Allison Scott, Ph.D. Alexis Martin, Ph.D.

The Effects of Societal Versus Professor Stereotype Threats on Female Math Performance

My Notebook. A space for your private thoughts.

Black 1 White 5 Black

Self-Handicapping Variables and Students' Performance

Introduction to Psychology Social Psychology Quiz

Title: The Relationship between Locus of Control and Academic Level and Sex of Secondary School Students

Measuring and Assessing Study Quality

Why Does Similarity Correlate With Inductive Strength?

Chapter 13. Social Psychology

Explain the formation of stereotypes and their effect on behaviour. Done by: Lynn and Daeun

CURRENT RESEARCH IN SOCIAL PSYCHOLOGY

INSTITUTIONAL LEVEL STRATEGIES IN FOSTERING INTERCULTURAL COMPETENCE DEVELOPMENT

(TEST BANK for Organizational Behavior Emerging Knowledge Global Reality 7th Edition by Steven McShane, Mary Von Glinow)

Evolutionary Psychology. The Inescapable Mental Residue of Homo Categoricus. Book Review

Desirability Bias: Do Desires Influence Expectations? It Depends on How You Ask.

Running Head: SENSITIVITY TO TAXNOMIC ASSOCIATIONS

Illusory correlation and attitude-based vested interest

Cultural threat and perceived realistic group conflict as dual predictors of prejudice

Chapter 11. Experimental Design: One-Way Independent Samples Design

Running Head: VISUAL SCHEDULES FOR STUDENTS WITH AUTISM SPECTRUM DISORDER

Influence of Implicit Beliefs and Visual Working Memory on Label Use

= Refers to Implications in instructional design Some authors

Reviewing Applicants. Research on Bias and Assumptions

Fukuoka University of Education

FEEDBACK TUTORIAL LETTER

Several studies have researched the effects of framing on opinion formation and decision

Reviewing Applicants

Everyday Problem Solving and Instrumental Activities of Daily Living: Support for Domain Specificity

Regression Discontinuity Analysis

Unconscious Knowledge Assessment

Healing Otherness: Neuroscience, Bias, and Messaging

Piers Harris Children s Self-Concept Scale, Second Edition (Piers-Harris 2)

Reducing Children s False Identification Rates in Lineup Procedures

How Does Analysis of Competing Hypotheses (ACH) Improve Intelligence Analysis?

Intentional and Incidental Classification Learning in Category Use

Chapter 15 Learning Objectives with SubQuestions PERSON PERCEPTION: FORMING IMPRESSIONS OF OTHERS

Chapter 3 Perceiving Ourselves and Others in Organizations

Student Performance Q&A:

Incorporating quantitative information into a linear ordering" GEORGE R. POTTS Dartmouth College, Hanover, New Hampshire 03755

33 Multiple choice questions

Supplemental Materials: Facing One s Implicit Biases: From Awareness to Acknowledgment

Social and Pragmatic Language in Autistic Children

Supplementary Materials. Worksite Wellness Programs: Sticks (Not Carrots) Send Stigmatizing Signals

DEAF STUDENTS CULTURAL IDENTITY AND PARTICIPATION IN A DEAF STUDIES COURSE

Cue Saliency and Age as Factors Affecting Performance in a Card Sorting Task

How Racism Correlates with Perceptions and Attributions of Healthcare Disparities

Running Head: STEREOTYPE THREAT AND THE RACIAL ACHIEVEMENT GAP 1

Preventing Unintentional Harm: Understanding Implicit Bias in Juvenile Justice

The Costs of Accepting Gender Differences: The Role of Stereotype Endorsement in Women s Experience in the Math Domain

A Contextual Approach to Stereotype Content Model: Stereotype Contents in Context

VERDIN MANUSCRIPT REVIEW HISTORY REVISION NOTES FROM AUTHORS (ROUND 2)

Types of Group Comparison Research. Stephen E. Brock, Ph.D., NCSP EDS 250. Causal-Comparative Research 1

In our multicultural society, interracial interactions are. The Antecedents and Implications of Interracial Anxiety

Lesson 12. Understanding and Managing Individual Behavior

USE OF AN ESP COVER STORY FACILITATES REINFORCEMENT WITHOUT AWARENESS. LEWIS A. BIZO and NICOLA SWEENEY University of Southampton

Lecture Outline Schemas Part 1. Bottom-Up Processing. Schemas. Top-Down Processing. Bottom up vs. Top Down Processing. Schemas

Lecture Outline Schemas Part 1. Bottom-Up Processing

ENTREPRENEURSHIP AND STEREOTYPES: ARE ENTREPRENEURS FROM MARS OR FROM VENUS?

CHAPTER 2: PERCEPTION, SELF, AND COMMUNICATION

VIRGINIA LAW REVIEW IN BRIEF

Experimental Research. Types of Group Comparison Research. Types of Group Comparison Research. Stephen E. Brock, Ph.D.

PLS 506 Mark T. Imperial, Ph.D. Lecture Notes: Reliability & Validity

1983, NUMBER 4 (WINTER 1983) THOMAS H. OLLENDICK, DONNA DAILEY, AND EDWARD S. SHAPIRO

The Heart Wants What It Wants: Effects of Desirability and Body Part Salience on Distance Perceptions

Impression order effects as a function of the personal relevance of the object of description*

The Youth Experience Survey 2.0: Instrument Revisions and Validity Testing* David M. Hansen 1 University of Illinois, Urbana-Champaign

Module 2/3 Research Strategies: How Psychologists Ask and Answer Questions

Cultural Differences in Cognitive Processing Style: Evidence from Eye Movements During Scene Processing

Hidden Bias Implicit Bias, Prejudice and Stereotypes

ORIGINS AND DISCUSSION OF EMERGENETICS RESEARCH

CHAPTER 10 Educational Psychology: Motivating Students to Learn

On the purpose of testing:

Cognitive dissonance in children: Justification of effort or contrast?

CAN PROMOTING CULTURAL DIVERSITY BACKFIRE? A LOOK AT THE IMPLICIT EFFECTS OF PSYCHOLOGICAL REACTANCE. Jeffrey Adam Gruen

Perceived discrimination and academic resilience: A study of Albanian immigrant adolescents in Greece

Help-Giving as a Factor in Perceived Group Helpfulness and Member Satisfaction in Small Counseling Groups

Audio: In this lecture we are going to address psychology as a science. Slide #2

PERCEIVED TRUSTWORTHINESS OF KNOWLEDGE SOURCES: THE MODERATING IMPACT OF RELATIONSHIP LENGTH

Asking and answering research questions. What s it about?

Transcription:

Child Development, July/August 2001, Volume 72, Number 4, Pages 1151 1162 When Groups Are Not Created Equal: Effects of Group Status on the Formation of Intergroup Attitudes in Children Rebecca S. Bigler, Christia Spears Brown, and Marc Markell This study was designed to examine whether the presence of implicit links between social groups and high versus low status attributes affects the formation of intergroup attitudes. Elementary school children aged 7 to 12 years (N 91) were given measures of classification skill and self-esteem, and assigned to one of three types of summer school classrooms in which teachers made (1) functional use of novel ( blue and yellow ) social groups that were depicted via posters as varying in status, (2) no explicit use of novel social groups that were, nonetheless, depicted as varying in status, or (3) functional use of novel social groups in the absence of information about status. After 6 weeks, children completed measures of intergroup attitudes. Results indicated that children s intergroup attitudes were affected by the status manipulation when teachers made functional use of the novel groups. Children who were members of high-status (but not low-status) groups developed ingroup biased attitudes. INTRODUCTION Research based on intergroup theories of stereotyping and prejudice (e.g., Allport, 1954; Sherif, Harvey, White, Hood, & Sherif, 1961; Tajfel, Flament, Billig, & Bundy, 1971) has generated a good deal of knowledge about conditions that facilitate the formation of negative intergroup attitudes (e.g., see Brewer, 1979; Messick & Mackie, 1989). Much of this research has employed minimal group manipulations, in which social groups are based on meaningless or random criteria (e.g., Billig & Tajfel, 1973; Tajfel & Billig, 1974). This research suggests that children and adults readily develop biases toward their own group. Children appear to form intergroup biases most readily when social groups are perceptually salient (e.g., Bigler, 1995; Kowalski & Lo, 1999) and used by authority figures to label and organize the social environment. This is referred to as functional use (Bem, 1983; Bigler, Jones, & Lobliner, 1997). It is important to note, however, that social stereotyping and prejudice in the United States often involve groups of unequal status. For example, European Americans (as a group) have higher levels of income, educational attainment, and occupational prestige than do African Americans. Males are similarly advantaged relative to females. Thus, children develop stereotypes about these groups in contexts that are quite dissimilar from most experimental contexts (e.g., Bigler et al., 1997; Brewer & Silver, 1978; Gagnon & Morasse, 1995; Moghaddam & Stringer, 1986; Tajfel, 1970). The primary purpose of the present study was to examine the effects of group status on the formation of children s intergroup attitudes (i.e., stereotypes, evaluations, and preferences). Several theoretical and empirical lines of work suggest that group status affects the formation of intergroup attitudes. While minimal group paradigms typically produce in-group biases, research on gender and racial attitudes indicates that children do not always develop attitudes that favor their own group. For example, although boys typically develop gender attitudes that favor their own group, girls typically develop more egalitarian (i.e., nonbiased) gender beliefs (Signorella, Bigler, & Liben, 1993). In a parallel manner, most European American children develop stereotypic (pro-white) attitudes about race, whereas African American children more often develop nonbiased, and sometimes out-group biased, attitudes (see Spencer & Markstrom-Adams, 1990). Some theorists have argued that the relative status of males versus females and European Americans versus African Americans in the United States produces these differing degrees of intergroup stereotyping and prejudice (e.g., Signorella et al., 1993; Spencer & Markstrom- Adams, 1990). The argument that group status affects intergroup attitudes is consistent with theories of stereotyping and prejudice offered by Allport (1954) and Tajfel and Turner (1979). These theorists suggest that children learn the links between social groups and status via models in the environment and, subsequently, internalize the evaluative information about groups. Thus, children come to associate higher status groups even those of which they are not members with more positive affect than lower status groups. Allport 2001 by the Society for Research in Child Development, Inc. All rights reserved. 0009-3920/2001/7204-0013

1152 Child Development (1954, p. 54) and other theorists have used the term self-hatred to describe the situation in which an individual mentally identifies himself with the practices, outlook, and prejudices of the dominant group. More recent perspectives have called into question, however, the degree to which intergroup biases affect the self-system (Cross, 1991; Spencer & Markstrom- Adams, 1990). It appears that negative attitudes toward in-groups (and positive attitudes toward out-groups) often develop as object information about social groups in the environment, leaving self-identity and self-esteem unaffected (Powell, 1985; Spencer, 1985). In one of the few developmental studies to examine group status and intergroup attitudes, Yee and Brown (1992) gave preschool and elementary schoolage children false feedback about their team s performance in an athletic contest. Children were later asked to evaluate their in-group (i.e., own team) and out-group (i.e., other team). Findings indicated that group status did not affect children s affective ratings of the in-group and out-group. Group status did, however, affect children s performance evaluations of the groups. Children placed on fast teams rated their own team as faster than the other team, whereas children placed on the slow team rated their own team as slower than the other team. This finding is intriguing in that it suggests that children as young as 3 years of age are attentive to group comparisons. Several methodological factors limit the conclusions that can be drawn from this study, however. Most important, participants were told that members of the highand low-status (i.e., fast and slow) groups were selected for membership in a particular group on the basis of their performance on the athletic task. Thus, when children in the low-status group rated their in-group as slower than the out-group, this could be construed as an accurate rating of performance (rather than an out-group bias). In contrast, the present study looked at whether children develop in-group (or out-group) biases when the individual members of high- and low-status groups are not actually different on the dimensions rated by participants. In other words, this research explored whether status information leads to the development of stereotypes of individuals. Additionally, this study looked at whether children might develop beliefs about the status of social groups from implicit (rather than explicit, as in Yee & Brown, 1992) information about social groups. Today, the relative status of males versus females, and European Americans versus African Americans, is communicated most often to children indirectly. In many school settings, for example, children are presented with a disproportionate number of male versus female, and European American versus African American, models who are in high- versus low-status positions (e.g., school principal, famous inventor, President of the United States). Consistent with previous theories (e.g., Allport, 1954), it was hypothesized that children would use such implicit information about group status in developing intergroup attitudes. For the status of social groups to affect intergroup attitudes, children must be able to detect a covariation between a particular social group and traits or behaviors that convey status. Research with adults suggests that conscious and nonconscious detection of covariation is quite good in the absence of initial preconceptions about relations (e.g., Alloy & Tabachnik, 1984). Furthermore, once an initial correlation has been detected (even nonconsciously), people continue to act in accordance with such a correlation (Hill, Lewicki, Czyzewska, & Schuller, 1990). Fewer studies have examined whether, and under what conditions, children detect covariations between social groups and attributes, but several recent studies suggest that such processes may be within the capabilities of even young children (Levy & Haaf, 1994; Younger, 1992). Thus, it seems likely that elementary school-age children will notice the positive and negative attributes associated with novel social groups, and that these associations will, in turn, affect the formation of intergroup attitudes. A second important aspect of the detection of group-to-attribute links concerns the complexity of the task. Many laboratory tasks present information (i.e., experimental stimuli) in a constrained context that might actually promote attention to the group-toattribute relation of interest. Stimuli are often made less complex, for example, by constraining the dimensions along which individuals vary (e.g., only a single gender), or constructing social group-to-attribute links that are correlated perfectly (e.g., Levy & Haaf, 1994). The present study examined whether the implicit presentation of links between novel social groups and attributes that convey status would affect intergroup attitudes in a naturalistic (i.e., field) setting, in which a great deal of individuating information about group members was available, and individuals did not always exhibit the attributes linked to their groups. To examine these issues, an intergroup research paradigm similar to that used by Bigler and her colleagues (Bigler, 1995; Bigler et al., 1997) was used. Specifically, elementary school-age children enrolled in a summer school program were randomly assigned to a novel social group, denoted by colored T-shirts ( yellow or blue ), in their classroom. In the experimental conditions, children s classrooms

Bigler, Brown, and Markell 1153 contained posters that depicted unfamiliar blue and yellow group members as having higher versus lower status attributes (e.g., good academic, athletic, and leadership abilities). Because previous research suggests that teachers functional use of novel social groups affects the formation of intergroup bias (e.g., Bigler, 1995; Bigler et al., 1997), two experimental conditions were used. In the first experimental condition, teachers made functional use of the color groups in their classrooms (posters plus functional use condition). That is, teachers made extensive use of physical and procedural groupings by color groups (see Bigler, 1995). In the second experimental condition, teachers ignored the presence of the color groups, referring either to classrooms as a single unit, or to the individual classroom members (posters only condition). In the control condition, teachers made functional use of the color groups, but posters were not included in the classrooms (no posters condition). This condition allowed for a test of whether the posters contributed to the formation of intergroup bias above and beyond the effects of the functional use of groups. Based on previous research, it was predicted that children in the control (i.e., no posters) classrooms would develop intergroup biases that favored their own color group. It was also expected that children in the experimental (i.e., posters plus functional use and posters only) classrooms would detect the covariation between the color groups and status, and form stereotypes based on this information. Thus, children in high-status groups were expected to show higher levels of intergroup bias than children in low-status groups. Although it was expected that the presence of implicit messages about status would affect children s attitudes at the group level, it was also expected that some children might be more susceptible than others to the experimental manipulations. Several lines of research suggest that there are important individual differences among children and adults in their tendency to endorse social stereotypes (e.g., Aboud, 1988; Levy & Dweck, 1999; Levy, Stroessner, & Dweck, 1998; Signorella, 1987). Based on previous theorizing and research, it was hypothesized that children s selfesteem (Tajfel & Turner, 1979) and classification skill (Bigler, 1995; Spencer, 1985) would moderate the effects of condition on children s intergroup attitudes. Tajfel and Turner s (1979) self-identity theory postulated that self-esteem serves as a motivator of intergroup bias. Specifically, they argued that individuals with low self-esteem might attain increased self-esteem via devaluation of the out-group. Research examining links between self-esteem and intergroup bias suggests, however, that higher (rather than lower) self-esteem is associated with in-group bias in children (Bigler et al., 1997; Gagnon & Morasse, 1995) and adults (Abrams & Hogg, 1988). The mechanism by which self-esteem affects intergroup attitudes is not well understood, however. The manipulation of group status may inform the understanding of this mechanism. It is possible, for example, that intergroup manipulations produce a threat to self-esteem that individuals with higher, but not lower, selfesteem are motivated to overcome via intergroup bias (Lemyre & Smith, 1985). In this study, children in the low-status (but not high-status) condition would be expected to experience a particular potent threat to their self-esteem. High self-esteem children in this condition might, therefore, show higher levels of intergroup bias than low self-esteem children. In addition to examining self-esteem as a predictor of intergroup bias, it is also interesting to consider self-esteem as an outcome variable. Considerable research shows that children who are members of low-status social groups do not typically have lower self-esteem than children who are members of highstatus groups (see Crocker & Major, 1989). For example, girls do not have lower global self-esteem than boys (Major, Barr, Zubek, & Babey, 1999), and African American children do not have lower global selfesteem than European American children (Pallas, Entwisle, Alexander, & Weinstein, 1990). Thus, it was expected that, at the group level, children s selfesteem would be unaffected by the group status manipulations used here. It seems possible that children s self-esteem is unaffected by group status because young elementary school-age children lack the cognitive skills to coordinate their conceptions of self and their social group (Spencer, 1985). To better understand how the relation between self-esteem and group membership might be affected by cognitive development, a number of cognitive skills that seemed related to the ability to integrate self and social group conceptions were assessed, including multiple classification and class inclusion skills (see Kofsky, 1966; Piaget, 1970). It was predicted that among children placed in low-status groups, only those who had advanced classification skills would show lower self-esteem as a result of the experimental procedures. METHOD Participants Participants were 91 children (43 girls, 48 boys) attending a summer school program in the Midwest.

1154 Child Development The summer school program is a laboratory school run by a midsized university in the Midwest serving a middle-class community. 1 Fourteen additional students were enrolled in the program but were not included in the final sample because of inability to obtain parental permission to give the pre- and posttest measures (3 students), extended school absences (10 students), and incomplete data (1 student). Participants ranged in age from 7 years, 0 months to 12 years, 4 months (M 8,10; SD 1,1), and most were White and from middle-class backgrounds. Children were assigned to one of nine classrooms with similar-age peers prior to the start of the study. One classroom of younger (completed kindergarten or first grade), middle (completed second grade), and older (completed third or fourth grade) children was randomly assigned to each experimental condition. Means and standard deviations (in parentheses) for age across the posters only, posters plus functional use, and no posters conditions were 8,8 (1,0), 9,0 (1,1), and 8,10 (1,2), respectively. Overview of Procedure Students enrolled in the summer program attended class from 8 a.m. until noon each weekday over a period of 6 weeks. During the first week of school, all children were seen individually by a female experimenter and given pretest measures of self-esteem and classification skill. Classrooms were then matched for overall age level and assigned to one of three conditions: a posters only, a posters plus functional use, and a no posters condition. Three classrooms of 7 to 13 children participated in each condition. In each classroom, children were given a T-shirt, called a work shirt, to wear during the school day. Half of the children in each room were given yellow shirts and half were given blue shirts. In the two experimental conditions (posters only and posters plus functional use), classrooms contained posters that implicitly linked the novel color groups to higher and lower status attributes (e.g., traits and roles). In half of these 1 It is designed to provide teacher training and research opportunities (e.g., tests of new curriculum designs), as well as to serve the academic needs of children in the community. Parents were informed of the experimental nature of the program prior to enrollment and, in addition, were sent letters asking for permission to administer the experimental measures to their child. Only those children whose parents returned consent forms, and who themselves agreed to participate, were tested. Students included in this sample were drawn from 28 different elementary schools. Only four of the 91 students were assigned to a summer school classroom containing a peer who had been a member of their classroom during the prior academic year. rooms (posters plus functional use), the teachers made functional use of the novel groups, whereas in the other half (posters only), the teachers ignored the perceptually salient color groups. The control (no posters) classrooms did not contain posters, although teachers used the novel color groups in a functional manner. After 4 weeks, children were given posttest measures of self-esteem and intergroup attitudes. During the last week of school, all children who participated in the summer program were exposed to a combined intervention and debriefing lesson designed to link children s classroom experience to their understanding of stereotyping and prejudice. 2 Pretest Measures Self-esteem. The Piers-Harris Children s Self-Concept Scale (CSCS; Piers, 1969) was used to assess self-esteem. Half of the scale s 80 items were randomly selected to serve as a pretest measure and the other half were used as a posttest measure. Thus, total pretest scores ranged from 0 to 40, with higher scores indicating higher self-esteem. Classification skill. Classification skill was assessed using procedures developed by Jones and Bigler (1996), following Inhelder and Piaget (1964). Tasks were designed to measure (1) consistent sorting, (2) resorting, (3) multiple classification, and (4) class inclusion skills. Children were first asked to sort a set of 12 pictures (bears and elephants that were gray or brown) into two groups (consistent sorting task). Children were then asked to sort the cards along a second, new dimension (resorting task). One point was given for each correct sort. Next, children were shown a 2 2 matrix and, following a demonstration sort by the experimenter, were asked to sort a new set of cards into the appropriate cells of the matrix. Children were then asked to justify their arrangements. Scoring was based on the child s sort and justification: 3 points for a correct sort and justification, 2 points for an incorrect sort but appropriate justification of the experi- 2 In this study, the potential for children to develop negative attitudes toward other groups and, perhaps more seriously, their own group presented serious ethical concerns. Two considerations were important in justifying this research. First, exposure to implicit information linking one s own social group to lower status traits and roles is a common occurrence in educational classrooms for many children in the United States, including racial and ethnic minorities, females, and individuals with physical disabilities. Thus, the potential to better understand the consequences of such exposure seemed highly important. Second, prior research using intergroup methodologies has failed to find any effects of intergroup manipulations on children s selfesteem or behavior toward in-group or out-group members (see Bigler, 1995; Bigler et al., 1997).

Bigler, Brown, and Markell 1155 menter s correct sort, and 1 point for a correct sort but inappropriate justification. Finally, children were asked a series of five questions designed to tap their understanding of hierarchical relations among categories. One point was given for each correct response and thus, classification scores ranged from 0 to 10, with higher scores indicating more advanced skill. Experimental Conditions To expose children to implicit links between the novel social groups and attributes that convey status, six large, attractive posters were designed. The posters were 55 cm 71 cm in size and included colored drawings and actual photographs of unfamiliar children and adults wearing yellow and blue shirts. The posters conveyed information about evaluative traits, roles, and behaviors, including intelligence, leadership ability, athletic prowess, classroom behavior, and occupational prestige. For example, one poster depicted the six children who won athletic events during the previous year s Sports Olympics Day. Children in yellow shirts were depicted as winning the majority (five out of six) of the athletic contests. Another poster depicted five out of the six weekly spelling bee winners as wearing yellow shirts. The posters were placed in the experimental classrooms prior to the start of the summer program. On the first day of school, classroom teachers gave a 3 min explanation of the posters in their room. They stated that the posters depicted summer school staff, participants, and activities from the previous year s summer school program. Teachers stated that some of these activities might be repeated this year, but teachers made no mention of the group membership (i.e., shirt color) of any of the individuals in the posters. In the postersonly condition, the placement (and brief description) of the posters was the only experimental manipulation. In the posters plus functional use condition, classroom teachers were instructed to make use of physical and spatial dichotomies, as well as verbal categorization, in their classrooms (see Bem, 1983; Bigler, 1995). Teachers were also instructed not to favor one group over another or to promote group competition because conditions have been shown to produce negative intergroup attitudes (e.g., Sherif et al., 1961). Informal classroom observations indicated that teachers showed high levels of compliance with all instructions. For example, teachers in each of the classrooms made creative use of color-based bulletin boards and seating arrangements. Teachers routinely asked children to perform tasks by color group (e.g., lining up at the door for recess) and frequently mentioned color group membership when addressing individual children. Formal observations of the children s own use of color groups were not conducted, and thus it is not possible to present data concerning the frequency with which children themselves used the groups in a functional manner. Anecdotal evidence suggests, however, that children readily accepted the color groups as important categories and frequently used them to organize their activities. It was common, for example, for children to request that the two color groups be allowed to compete in various activities (e.g., Let the yellow group play against the blue group! We can beat them! ). The teachers always denied such requests but their occurrence suggests that some children may have engaged in implicit or covert group competition. The control classrooms did not contain posters but the teachers used the color groups in a functional manner. Thus, classroom structure and organization centered around the color groups. Again, teachers were instructed to treat the two groups equally and to forbid group competition. Informal observations indicated that teachers complied with instructions. Posttest Measures Overview. After 4 weeks in the experimental classrooms, children were seen individually by one of four female experimenters and given a series of posttest measures. To help minimize boredom, testing was separated into two short sessions. Those measures that did not make the color group salient as a dimension for responding were given in the first session, so that the salience of the color groups would not bias responding. In the first session, children completed a posttest measure of self-esteem and sociometric ratings of their classroom peers. In the second session, children were given a trait stereotyping measure and asked a series of evaluative questions about the novel groups. Finally, children in the experimental classrooms were given a test of their memory for the information presented in the posters, which had been removed from the classrooms prior to posttesting. Trait stereotyping. Children s perceptions of the traits associated with each color group were assessed using a procedure developed by Bigler (1995). Participants rated how many of the children in each color group were characterized by five positive (friendly, good, nice, pretty, and smart) and five negative traits (dirty, mean, naughty, selfish, and lazy), taken from the Preschool Racial Attitude Measure II (Williams, Best, Boswell, Mattson, & Graves, 1975). Children responded using a 4-point scale that included all of the blue/yellow group, most of the blue/yellow group, some of the blue/yellow group, or none of the blue/yellow group. Children made two judgments

1156 Child Development for each trait, one for their in-group and one for their out-group. Evaluation of in-group versus out-group. Children s evaluations of their in-group and out-group were assessed by asking a series of seven questions: (1) whether they would like to change their group membership in the classroom, (2) which color group a new schoolmate would prefer, (3) which color group would win a hypothetical spelling contest, (4) which color group would win a hypothetical math contest, (5) which color group would have the most time-outs for bad behavior at the end of the school session, (6) which color group would have the most absences from school at the end of the school session, and (7) which color group would, when the children were adults, have the most important and highest paying jobs. Children responded by choosing their own color group, the other color group, or both groups (e.g., a tie ). Peer preferences. To assess possible color group bias in peer preferences, children were asked to rate how much they liked to play with each child in their classroom. Participants responded using a 5-point scale ranging from not too much to a lot. RESULTS Pretest Measures Self-esteem. A one-way ANOVA by condition indicated that there were no significant differences in selfesteem across the three conditions. Overall, children showed moderately high levels of self-esteem (M 31.57, SD 6.15). A 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) ANOVA indicated no significant differences across groups. Classification skill. A one-way ANOVA by condition produced a significant effect, F(2, 88) 3.33, p.05. Post hoc analyses indicated that children in the posters plus functional use condition (M 7.81, SD 1.73) showed significantly better classification skills than did children in the posters only (M 6.94, SD 2.21) and no posters (M 6.33, SD 2.55) conditions (which did not differ from each other). A 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) ANOVA indicated no significant differences across groups. Posttest Measures: Overview The first major question was whether children s intergroup attitudes differed as a function of exposure to implicit messages indicating that their social group was high or low in status, and whether possible effects were exacerbated when teachers used the social groups in a functional manner. Thus, ANOVAs were used to test for group status (i.e., high versus low) and condition (i.e., posters only versus posters plus functional use) effects on the dependent measures. Because cell sizes would have become unacceptably small if all potential between-group variables were included in the ANOVAs, effects of gender of participant and classroom within condition were evaluated via preliminary ANOVAs. Results indicated no significant main effects or interactions involving gender, and thus data were pooled across this variable. Significant effects were found, however, for classrooms (i.e., classes within conditions). Examination of the means indicated a consistent pattern of lower levels of stereotyping among those classrooms comprised of older (rather than younger) children within each condition. These effects were, therefore, interpreted as age (rather than classroom) effects. Consequently, age was included as a between-subjects variable (i.e., younger versus older children, based on a median split) in all subsequent ANOVAs. To ensure that possible differences in intergroup attitudes among groups were not due to selective memory for the poster content, children were asked to recall the shirt color of the individuals depicted in the posters (e.g., the school principal). Children received one point for each of 12 possible correct answers. A 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger versus older) ANOVA indicated no significant main effects or interactions. The failure to find differences in memory across groups suggests that any subsequent difference in intergroup attitudes across conditions, or high- and low-status groups, could be attributed to differences in attention to, or memory for, the implicit information about group status presented via the posters. The experimental design prevented the simultaneous comparison of all three classroom conditions within a single ANOVA (i.e., because the control group lacked high- and low-status groups). Therefore, comparisons of the two experimental groups are presented first, followed by comparisons of the experimental and control groups. All comparisons among means following significant ANOVAs were conducted using the Newman-Keuls method of multiple comparisons. Trait Stereotyping: Experimental Conditions Positive traits. Scores were analyzed using a 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger

Bigler, Brown, and Markell 1157 versus older) 2 (target: in-group versus out-group) ANOVA, with the latter variable as a repeated measure. The three-way interaction of condition, status, and target was significant, F(1, 50) 8.12, p.01. Means are presented in Table 1. Post hoc comparisons indicated that among children in the posters plus functional use condition, high-status (but not lowstatus) children gave more positive evaluations to the in-group than to the out-group. The two-way interaction of condition and age was also significant, F(1, 50) 7.15, p.05. Post hoc comparisons indicated that among children in the posters plus functional use (but not posters only) condition, younger children gave more positive trait ratings than did older children (M 4.93, SD.65, and M 3.85, SD 1.0, for younger and older children, respectively). Negative traits. Scores were analyzed using a 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger versus older) 2 (target: in-group versus out-group) ANOVA, with the latter variable as a repeated measure. Only the two-way interaction of condition and age was significant, F(1, 50) 6.34, p.05. Post hoc tests indicated that among children in the posters plus functional use condition, older children gave more negative trait ratings than did younger children (M 1.27, SD 1.0, and M.72, SD.65, for older and younger children, respectively). In contrast, among children in the posters only condition, younger children gave more negative trait ratings than did older children (M.82, SD.79, and M.31, SD.41, for younger and older children, respectively). groups, the high-status children in the posters plus functional use condition were compared to those in the control classrooms (no posters). Positive trait ratings were analyzed using a 2 (condition: no posters versus high-status posters plus functional use) 2 (age: younger versus older) 2 (target: in-group versus out-group) ANOVA, with the latter variable as a repeated measure. Means are presented in Table 1. The predicted interaction of condition with target failed to reach significance, F(1, 43) 2.64, p.11, indicating that children in both conditions showed ingroup biased attitudes. There was, however, a nonsignificant trend for high-status children in the posters plus functional use condition to show higher levels of in-group bias compared with children in the no posters condition. The two-way interaction of target and age was also significant, F(1, 43) 4.12, p.05, indicating that younger (but not older) children rated the in-group more favorably than the out-group (M 2.45, SD.50, and M 1.96, SD.48, for in-group and outgroup ratings, respectively. The main effect for target was also significant, F(1, 43) 17.83, p.05, indicating that children rated the in-group more favorably than the out-group (M 2.35, SD.47, and M 2.03, SD.46, for in-group and out-group ratings, respectively). Negative traits. Negative trait ratings were analyzed using a 2 (condition: no posters versus high-status posters plus functional use) 2 (age: younger versus older) 2 (target: in-group versus out-group) ANOVA, with the latter variable as a repeated measure. No significant main effects or interactions were found. Trait Stereotyping: Control Group Positive traits. To examine the effects of the posters above and beyond that of the functional use of Table 1 Positive Trait Ratings by Condition, Status, and Target In-Group Target Out-Group Condition N M SD M SD No posters 33 2.28.39 2.06.52 Posters only High status 17 2.05.57 2.18.58 Low status 14 2.37.50 2.16.52 Posters plus functional use High status 14 2.50.39 1.96.26 Low status 13 2.22.62 2.23.80 Note: Scores range from 0 to 3, with higher scores indicating more positive evaluations. Trait Stereotyping: Individual Differences Additional analyses examined whether children s self-esteem and classification skill were related to trait stereotyping scores. Specifically, hierarchical multiple regression analyses were used to examine the unique contribution of self-esteem and classification skill to trait stereotyping, apart from their links to age and gender. Separate models were tested for positive and negative trait ratings of the in-group and out-group. Positive traits. Results indicated that after each of the variables (i.e., age in months, gender, composite classification skill, and self-esteem) had been entered, pretest self-esteem was a significant predictor of positive trait rating for the in-group,.25, t(82) 2.37, p.05. Children with higher self-esteem evaluated their own group more positively than did those with lower self-esteem. Age was also a significant predictor of positive trait ratings of the in-group,.26, t(82) 2.26, p.05. Younger children evaluated their own group more positively than did older chil-

1158 Child Development dren. None of the four variables were significant predictors of positive trait ratings of the out-group. Negative traits. After each of the variables (i.e., age, gender, classification skill, and self-esteem) had been entered, pretest self-esteem was a significant predictor of negative trait ratings for the in-group,.33, t(82) 3.03, p.05. Children with higher selfesteem evaluated their own group less negatively than did those with lower self-esteem. Of the four variables (i.e., age, gender, classification skill, and self-esteem) entered into the model of negative trait ratings of the out-group, only gender was a significant predictor,.25, t(82) 2.35, p.05. Girls evaluated the out-group more negatively than did boys. Self-esteem threat. To test the hypothesis that high self-esteem is linked to bias because high self-esteem children are threatened by group manipulations and, therefore, motivated to develop biased views of the outgroup, we examined whether self-esteem interacted with group status to produce increased intergroup bias. In other words, analyses were performed to examine whether high and low self-esteem children responded differentially to the group status manipulations. Contrary to expectations, a 2 (status: high versus low) 2 (self-esteem: higher versus lower) 2 (target: in-group versus out-group) ANOVA showed no significant main effects or interactions for positive or negative trait ratings. Results were identical whether the analysis was performed using children in both experimental groups or only children in the posters plus functional use group. Evaluation of In-Group versus Out-Group: Experimental and Control Conditions Chi-square analyses were used to compare the frequency with which high- and low-status children in the two experimental conditions (posters only and posters plus functional use) reported that: (1) they wanted to keep their shirt colors, and (2) a new schoolmate would like their shirt color. Results indicated no significant difference for either variable. Overall, 76% of the children wanted to keep their shirt color and 32% reported that a new schoolmate would select their shirt color. The frequency with which children in the control (no posters) group chose to keep their shirt color did not differ from chance (66%), whereas the frequency with which children in the high-status posters plus functional use group chose to keep their shirt color was significantly above chance (86%). The frequency with which children reported that a new schoolmate would like their shirt color was not different from chance for either condition. Next, the number of in-group biased, out-group biased, and neutral ( tie ) answers given by children in response to the questions about group attributes (questions 3 through 7) was tallied. A 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger versus older) ANOVA indicated no significant effects for any of the three types of responses. The parallel analyses of the control condition (i.e., no posters versus highstatus posters plus functional use) produced no significant results. Peer Preferences: Experimental and Control Conditions Because gender has been found to affect peer preferences, gender of peer was included as a factor in this analysis. Specifically, children s peer ratings were analyzed by a 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger versus older) 2 (group membership: in-group versus out-group) 2 (gender of peer: same-gender versus other-gender) ANOVA, with the last two factors as repeated measures. The four-way interaction of condition, status, group membership, and age was significant, F(1, 41) 6.02, p.05, as were several of the two- and three-way interactions. These interactions were, however, largely uninterpretable and unrelated to the theoretical interests of the present study. Overall, there was no evidence of in-group bias in peer ratings. Instead, the clearest effect concerned gender of peer. The main effect for gender of peer was significant, F(1, 41) 55.19, p.001, with same-gender peers rated higher than other-gender peers (M 3.79, SD.86, and M 2.43, SD 1.18, for same-gender and other-gender peers, respectively). The parallel analyses of the control condition (i.e., no posters versus high-status posters plus functional use) produced a similar pattern of results. Posttest Self-Esteem To examine the effect of the experimental conditions on self-esteem, a 2 (condition: posters only versus posters plus functional use) 2 (status: high versus low) 2 (age: younger versus older) ANCOVA was conducted, using pretest self-esteem as the covariate. The only significant effect was for the covariate, F(1, 39) 17.88, p.01. Overall, self-esteem remained high at posttest (M 32.29, SD 5.03). Multiple hierarchical regression analyses were used to examine whether children s trait ratings were predictive of changes in children s self-esteem from pre- to post-

Bigler, Brown, and Markell 1159 test. In a model that included the four types of trait ratings (i.e., positive trait ratings of the in-group and out-group, plus negative trait ratings of the in-group and out-group), none of the four variables were significant predictors of the self-esteem change scores. The experience of developing biased intergroup attitudes was not, therefore, associated with changes in self-esteem. Finally, ANOVAs using both posttest self-esteem and self-esteem change scores were performed to examine whether children s level of classification skill might mediate the effects of the group status on selfesteem. Results indicated that those children with more advanced classification skills were not more likely than those children with less advanced classification skills to show changes in self-esteem as a result of the status manipulation. (The possibility that children developed more advanced classification skill as a result of the summer school curriculum cannot be ruled out. If this were true, children s pretest classification scores would not be expected to mediate the relation between group status and self-esteem.) DISCUSSION The primary purpose of the present study was to examine whether implicit links between social groups and attributes that convey status affect the formation of intergroup attitudes and behavior in children. The findings indicate that children s intergroup attitudes are affected by the presence of implicit links between social groups and status-conveying attributes in some situations. Specifically, the high- versus low-status manipulation affected children s intergroup attitudes when social groups were used in a functional manner by authority figures in the environment. Children in experimentally manipulated high- and low-status groups did not develop either in-group or out-group biases when teachers ignored the presence of these groups in the classroom. In contrast, high-status (but not low-status) children developed in-group biases when teachers made use of the new social categories in the classroom. The finding that the functional use of a perceptually salient social group is necessary for the formation of in-group bias is consistent with previous research. In a similar field experiment, Bigler et al. (1997) reported that children developed biased intergroup attitudes only when novel social groups were both perceptually salient and used by classroom teachers to label children and organize the environment. This finding was replicated in the present study. Children whose teachers made functional use of the novel color groups (in the absence of information about status) developed stereotypic beliefs about the groups that favored their own color group. Research within the field of educational psychology also indicates that teachers verbal statements about social groups can have important effects on children s intergroup attitudes and behavior (Cohen & Lotan, 1995). It is interesting that the development of biased intergroup attitudes in the present study was dependent on the functional use of social groups, even though the environment provided models suggesting that the novel social groups actually did differ in their attributes. The poster manipulation in this study might, for example, be considered a kernel of truth manipulation, in the sense that there was evidence of a veridical covariation between status and the color groups. The failure of children to develop intergroup biases in this condition is important because it suggests that children will not necessarily form stereotypes for which there is some basis in the environment. Overall, these data suggest that authority figures should not use group membership for labeling and organizing the social environment when they wish to impede the development of social stereotypes, and that this is true whether social groups do or do not actually differ. How does functional use of a social category affect children s intergroup attitudes? One possibility is that it does so indirectly by affecting the salience of the social groups (e.g., see Brewer, 1979). This idea seems unlikely to account for the effects reported within this research paradigm, however, given that the presence of the two shirt colors in the classroom was extremely salient. Instead, Bigler et al. (1997) argue that the functional use of a social group by adults leads children to develop hypotheses concerning the differences between social groups. The expectation that the two social groups must differ in some way may, in turn, affect children s interpretation of information concerning covariation (see Alloy & Tabachnik, 1984). It is interesting to note that little is known about how children interpret information about social groups and their attributes. Future research might benefit from asking children why certain attributes covary with social groups. As in earlier studies, children s intergroup bias stemmed from the differential evaluation of the positive (rather than negative) traits (Bigler et al., 1997; Brewer, 1999). That is, intergroup bias took the form of believing that one s in-group was nearly perfect and the out-group was merely good. It is important to note, however, that the status manipulation used here concerned primarily positive attributes. For ethical reasons, attempts were not made to convey a strong

1160 Child Development link between one color group and negative attributes. It is possible, however, that such a manipulation might have produced intergroup biases on negative trait scales. Although the functional use of the social groups affected children s intergroup attitudes, the type of effect differed for children in the high- versus low-status groups. High-status children rated their in-group more positively than their out-group. In contrast, low-status children rated the in-group and out-group equivalently. This finding is consistent with the results of other studies in which children and adults were given explicit feedback concerning group status (Von Knippenberg, 1984). For example, Brown and Abrams (1986) reported that 12-year-olds did not display in-group (or out-group) biases on an evaluation measure of school performance after having been told that another school showed quite a bit better competence in math and English than did their own school. The finding that membership in the low-status group eliminated intergroup bias is also consistent with the stereotyping patterns of boys versus girls, and African American versus European American children. The mechanism by which membership in low-status groups affects intergroup attitudes may be a simple one. The tendency for children to view their in-group in an extremely positive light may be reduced among those individuals who receive negative information about their own group, thus eliminating intergroup biases in their attitudes. Consistent with this notion, Doyle and Aboud (1995) report that European American children develop more positive nonbiased racial attitudes over time as a function of increasing the attribution of negative traits to their own racial group, as well as increasing the attribution of positive traits to other racial groups. If such a mechanism is at work, it suggests some interesting (and counterintuitive) ideas for intervention strategies. Young children might, for example, benefit from exposure to information that emphasizes the shortcomings of those highstatus groups to which they belong. Before any such interventions are considered, however, it is important to know how negative information about one s in-group affects children s overall self-esteem. Although membership in the low-status group reduced bias in intergroup attitudes, we found no evidence that membership in the low-status group affected children s self-esteem. Posttest self-esteem did not vary as a function of experimental condition or group status. Thus, consistent with Spencer (1985) and Cross (1991), there was little support for the idea that evaluative information about social groups affects children s self-esteem. Although the overall size of the intergroup bias reported here is relatively small, it is surprising that in-group biases developed at all. The children in this study were exposed to a great deal of individuating (and stereotype-disconfirming) information about their peers, and they had a great deal of contact with peers under conditions that are usually associated with low levels of prejudice (e.g., cooperative learning situations; see Hewstone, 1996). The tenacity and rigidity of stereotypic beliefs in the face of inconsistent and individuating information is, however, a common finding in the stereotyping literature. The finding that children showed biased trait ratings and, simultaneously, nonbiased peer preferences is also consistent with findings from the school integration literature. Most research indicates that cooperative learning programs successfully reduce racial bias in peer interactions and preferences, but do not diminish stereotyping of racial groups (see Hewstone, 1996). The ability of the status information to inhibit intergroup bias is equally surprising. The status messages in this study were limited to six posters in children s school classrooms. Status messages concerning actual social groups (e.g., racial and ethnic minorities) differ from those used here in that they are typically (1) present in multiple environments (rather than at school alone), (2) highly salient because they involve live models (rather than posters), and (3) present over a long period of time (rather than a few weeks). The effect of group status on children s attitudes toward actual social groups is likely, therefore, to be much stronger than that reported here. Although these data are valuable because they provide clues about the development of social stereotyping in actual social groups, it is important to use caution in interpreting these findings. As is true of most research conducted in naturalistic settings, there are limitations to this study. Most obviously, the results are based on a relatively small and homogeneous sample of summer school students, who may differ from regular classroom students in some systematic ways. It is also important to note that the novel groups used in this study differ from actual social groups in many significant ways. It will be necessary for future research to replicate these findings using more diverse samples and to extend this research by examining other characteristics (e.g., power and degree of social segregation) associated with actual social groups. ACKNOWLEDGMENTS This research was conducted while the first author was a Visiting Associate Professor of Psychology at the University of Wisconsin at Madison. The authors