EXPECTATIONS AND SOCIAL INTERACTIONS OF CHILDREN WITH AND WITHOUT MENTAL RETARDATION

Mentally retarded and nonretarded perceiver children (n = 40) conversed by telephone with a child who was described as a special or regular education student. Perceivers reported that special and regular education telephone partners behaved differently during the conversation even though observers who were unaware of how telephone partners had been described did not detect behavioral differences between them. These same observers did detect differences in stereotype related social behaviors of mentally retarded and nonretarded perceivers, but only when perceivers thought they were speaking to a regular education student. Observer ratings also suggested that nonretarded perceivers "talked down" to special education telephone partners. These results suggest that stereotypes about children with and without learning problems may become self-fulfilling prophecies by altering how children treat one another and by affecting how they interpret each other's behaviors.

One important development in understanding children's educational experiences is the discovery of the Pygmalion effect (Rosenthal & Jacobson, 1968). The finding that teachers' expectations could affect children's academic performance stimulated vast amounts of research aimed at establishing (or challenging) the replicability and generalizability of this phenomenon and identifying the variables that mediate it (Braun, 1976; Brophy & Good, 1974; Dusek & Joseph, 1985; Harris & Rosenthal, 1985; Jussim, 1986; Miller & Turnbull, 1986).

Other researchers realized that self-fulfilling prophecies might be important in situations far removed from teacher-student classroom interactions. These researchers began to show how expectations about self and others could dramatically affect the course and outcome of adult social interactions in nonacademic settings. Expectancy confirmation has since become a core concept in several major theoretical analyses of social interaction (Darley & Fazio, 1980; Deaux & Major, 1987; Miller & Turnbull, 1986; Swann, 1984).

A major premise of these analyses is that expectations can become self-fulfilling prophecies because they affect how people interpret, remember, and behave toward the targets of these expectations who, by responding in kind, often confirm the expectations. This has been demonstrated in several well-controlled laboratory experiments in which, for example, people who were expected by another to be hostile (Snyder & Swann, 1978a), friendly (Cristensen & Rosenthal, 1982; Curtis & Miller, 1986; Farina, Allen, & Saul, 1968), introverted, or extroverted (Fazio, Effrein, & Falender, 1981; Snyder & Swann, 1978b) responded to the treatment they received in ways that confirmed these expectations.

Expectations often have been manipulated by identifying a person as a member of a stereotyped group. For example, Snyder, Tanke, and Berscheid (1977) led college men to believe that they would have a telephone conversation with a physically attractive or unattractive woman. Because of widespread prejudice against unattractive women (Adams, 1977; Berscheid & Walster, 1974; Patzer, 1985), Snyder et al. predicted and found that the men who thought they spoke to attractive woman were more friendly and sociable (according to naive observers who listened to recordings of the conversation) than were men who thought they spoke to unattractive women. Moreover, even though the target women actually did not differ in attractiveness, observers' ratings indicated that the women's behavior during the conversation differed in accord with the men's expectations. Similar results have been obtained in studies of the self-fulfilling nature of college students' expectations about blacks, women, and homosexuals (e.g., Kite & Deaux, 1986; Skrypnek & Snyder, 1982; Word, Zanna, & Cooper, 1974).

These demonstrations of the power stereotype-based expectations have in the social interactions of adults with their peers suggest that expectancy confirmation might also be important in children's social interactions with their peers. Recognition of this possibility is particularly important in understanding social relationships between mentally retarded and nonretarded children. In recent years the philosophy that mentally retarded people function best when integrated as completely as possible with nonretarded people has resulted in legislative mandates to integrate, or mainstream, mentally retarded children with nonretarded children in regular education classes (Fielder & Simpson, 1987). Although the major goal of mainstreaming is to improve the academic opportunities of retarded students, its proponents also hoped that increased contact between mentally retarded and nonretarded students would increase the acceptance of retarded students by their nonretarded peers (Birch, 1974; Exposito & Reed, 1986; Fisher & Rizzo, 1974; Sheare, 1974).

Research indicates, however, that integration with nonretarded students often fails to improve the social status of mentally retarded students (see Gresham, 1982; Ottensbacher & Cooper, 1984, for reviews) and may even result in greater rejection of them by nonretarded students (Corman & Gottlieb, 1978; J. Gottlieb, Semmel, & Veldman, 1978; Gresham, 1982). Although not usually phrased in terms of a self-fulfilling prophecy, it is commonly assumed that some of the problems faced by mainstreamed mentally retarded children in their social relationships with nonretarded peers stem from negative stereotypes about retarded children. These stereotypes may lead other students to ignore or reject children with mental retardation.

Although researchers have long been aware that stereotypes about retarded children might affect how nonretarded children treat them, there is surprisingly little research on this topic. Previous research typically has focused on documenting the existence of stereotyped expectations about retarded children, examining how these beliefs influence judgments and attributions about retarded children, and determining what effects increased contact with retarded children has on nonretarded children's attitudes and stereotypes (Corman & Gottlieb, 1978; J. Gottlieb, 1975; Siperstein & Bak, 1986). A limitation of such research is that it examines what children say and not what they do. Recognizing this problem, some researchers have attempted to investigate social behavior by assessing nonretarded children's preference for interacting with retarded versus nonretarded children (Aloia, Beaver, & Pettus, 1978; Bak & Siperstein, 1987; J. Gottlieb & Davis, 1973). However, because the interaction about which preferences were expressed did not actually occur, these studies do not capture the spontaneity, richness, and reciprocal nature of nonretarded children's social behavior toward children with mental retardation.

Even less attention has been given to the possibility that the self-fulfilling nature of stereotypes also could produce observable effects on the behavior of mentally retarded children. This is because most research on the social behavior of children with mental retardation focuses on identifying deficits in their social skills or other behavioral problems that alienate their nonretarded peers a Gottlieb et al., 1978; Gresham, 1982, 1984). Although this research often is flawed by the failure to realize that reports about the behavior of mentally retarded students- particularly reports made by untrained observers such as teachers and nonretarded classmates--may be biased by stereotypes, most available evidence suggests that mentally retarded children have less well-developed social skills than their nonretarded peers (Gampel, Gottlieb, & Harrison, 1974; J. Gottlieb et al., 1978; Gresham, 1982, 1984; Kitano, Stiehl, & Cole, 1978; Mundy, Sigman, Kasari, & Yirmiya, 1988; Peters, Pumphrey, & Flax, 1974; Taylor, Asher, & Williams, 1987). The existence of this difference often is interpreted as evidence that the social relationships of mentally retarded children can best be improved by social skills training (Davies & Rogers, 1985; Gresham, 1982, 1984; Strain & Odom, 1986).

The social skills of retarded children may partially explain their poor social relationships, but ignoring the effects of stereotypes on both their own and nonretarded children's behavior runs the risk of blaming the victims of stereotyping for their rejection by others. Even armed with the best of social skills, mentally retarded children might be ill equipped to deal with the social situations that con. front them. Because of prevailing stereotypes, nonretarded children may behave in a relatively unfriendly, or perhaps patronizing, fashion toward retarded children (Gibbons, 1981). Mentally retarded children may respond to such treatment by confirming the expectations others have about them. Moreover, because mentally retarded children are aware of what others expect of them and also know how they have been treated in the past, their own expectations may discourage them from behaving in a socially appropriate manner, even if they have the skills to do so. Thus, the anticipation of rejection might inadvertently produce it.

Purpose

We adapted the methodology used by Snyder et al. (1977) to see how the expectations of both mentally retarded and nonretarded perceiver children would affect their behavior during a brief telephone conversation with a target child who was described as being in special education or regular education classes. Perceivers' expectations about their telephone partners were assessed before the conversation began and after it was over. Audiotape recordings of the conversation were later evaluated by teacher, peer, and college student observers to determine how students' expectations affected their own and their telephone partners' behavior during the conversation.

Predictions

The first prediction was that mentally retarded and nonretarded children would behave differently toward telephone partners described as having learning problems than toward partners not so described. In particular, previous research has shown that nonretarded children may be overly lenient or patronizing toward retarded children (Crosby, Bromley, & Saxe, 1980; Jones et al., 1984; Katz, 1981), and that people often use "baby talk" when addressing people with mental retardation (DePaulo & Coleman, 1986). Such behaviors roughly correspond to children's stereotypes about what nonretarded children are like (Miller et al., 1989). Consequently, we predicted that the behavior of nonretarded perceivers would be more consistent with stereotyped expectations about nonretarded children when they thought they were speaking to a student in special education than when they thought they were speaking to a child in regular education. Similarly, research has shown that evaluation by a nonretarded audience impairs mentally retarded children's performance of an academic task (B.W. Gottlieb, 1982). We hypothesized that a nonretarded audience might also have a detrimental effect on the social behavior of children with mental retardation because of their apprehension about how they will be evaluated and treated. Consequently, we predicted that the behavior of mentally retarded perceivers would be more like the stereotype of a child with learning problems when they spoke to a regular education student than when they spoke to a special education student.

Our second prediction was that the behavior of the targets, who (unbeknownst to themselves) were described as special education students, would be more like the stereotype of a special education student than would the behavior of the telephone partners described as regular education students. This finding would indicate that behavioral confirmation of perceivers' stereotyped expectations had occurred.

Finally, perceivers completed a pre- and post-conversation measure of their perceptions about what their partners were like. If behavioral confirmation of perceivers' expectations did occur, this should affect perceivers' post conversation ratings of targets described as being in special or regular education classes. Even if behavioral confirmation did not occur, a large volume of research shows that perceivers remember and interpret the behavior of the targets in light of their stereotyped expectations about them, a process known as cognitive confirmation of stereotypes (Darley & Fazio, 1980; Miller & Turnbull, 1986). Accordingly, we predicted that the descriptions we provided about the targets might affect perceivers' post conversation ratings of them even if there were no real behavioral differences between targets described as being in special or regular education class.

METHOD

Subject Selection

Perceiver students were recruited from three schools located in a small city (population 45,000), four schools from suburban areas surrounding this city, and four schools from rural communities. Perceivers included 15 males and 5 females who had been diagnosed as mentally retarded and 20 nonretarded students who matched them in sex and the school they attended. All perceivers were 9-14 years old, with a mean age of 12.5 for mentally retarded perceivers (SD= 1.7) and a mean of 11.1 (SD=1.3) for nonretarded perceivers. There were no minority students in the sample, which reflects the total population of students in the schools we sampled.

Mentally retarded perceivers. Special education coordinators in each participating school were asked to identify students who were moderately mentally retarded (IQ range 55-75) and who were mainstreamed in at least one regular education class. The 26 students they identified all scored at least 1.5 SD below the mean on IQ and/or achievement tests, which is the state's criteria for classification as mentally retarded. To obtain a uniform estimate of the perceiver children's cognitive abilities, we obtained parental consent and the schools' permission to administer the Slosson Intelligence Test (Slosson, 1984) to each student approximately 2 weeks prior to his or her participation in this experiment. Parents of 2 children did not consent to their children's participation, and 2 other children were eliminated because their Slosson scores did not fall within the 55-75 range. The average Slosson score of the 22 students who participated was 67.6 (SD= 4.7).

Nonretarded perceivers. The special education coordinators also identified for each mentally retarded participant one regular education classroom in which the retarded perceiver was mainstreamed. Forms for obtaining parental consent were then sent to parents of all children in that class. We administered the Slosson to all students for whom consent was obtained and who were the same sex and approximately the same age as the perceiver child. The student whose Slosson score was closest to 100 was selected to be the nonretarded match for the mentally retarded perceiver. The scores of the nonretarded perceivers selected by these procedures ranged from 90 to 115, with an average of 103.7 (SD=7.2).

Test of recall. In addition to testing the perceivers' intelligence, we also administered a test of their ability to remember a verbal description of another child. The experimenter showed each student three pairs of portrait style photographs of same sex children who were approximately the same age as the perceivers and who were rated in a pretest as average in physical attractiveness. For each pair, the experimenter pointed to one randomly selected picture and said, "This child needs special help because he or she has trouble learning." The experimenter then pointed to the other picture and said, "This child does not have trouble learning." The experimenter then asked the student to indicate which child does and does not have trouble learning. The students had to correctly identify the child with and without learning problems for all three pairs of pictures to be included in this study. All nonretarded students we tested met this criterion, as did 20 of the 22 mentally retarded students we tested. The order and position of the picture said to represent the child who has trouble learning were varied across the three pairs of pictures.

Telephone Conversation

Telephone partners were children who were recruited by sending consent forms to parents of all fourth, fifth and sixth graders attending a local school that none of the perceivers attended. Signed consent forms were returned for 52 of 134 children. For each perceiver, the same-sex child who was the most similar in age was selected for participation as a telephone partner. None of the telephone partners actually had learning problems, so their behavior provides a test of the hypothesis that children treated in accord with stereotypes about children with learning problems will behave in ways that confirm those stereotypes even if the children do not, in fact, have learning problems. Although it also would have been informative to have mentally retarded children speak to perceivers who had been told they did or did not have learning problems, this would be tantamount to labeling mentally retarded children as retarded to other children. We did not think the benefits of doing this outweighed the risks.

Pre-conversation. An experimenter took each perceiver student to a quiet room in the school attended by that student. The experimenter explained that he or she was interested in finding out how children use the telephone to get to know one another, and that the student would be talking on the telephone to a child from another school.

The experimenter then showed the perceiver child a portrait-style photograph of a same-sex child approximately the same age as the perceiver. Mentally retarded perceivers and their nonretarded matches saw the same photograph, and the photograph was always different from the ones used in the pretest of their ability to remember descriptions of children. Perceivers were told that this was a picture of the child they would be talking to over the telephone, and that this child likes to play different games and is looking forward to summer vacation. A randomly selected sample of the retarded perceivers (and their nonretarded matches) were also told that the child they would be speaking to "is in a special class for children who have trouble learning" and the remaining perceiver students were told that the child "is in the same grade you are." The experimenter asked perceiver students to pretend that they would be meeting their telephone partner sometime soon and that they had 5 minutes to get to know that person over the telephone.

The experimenter then asked the perceiver students to make predictions about the social characteristics and behaviors of their telephone partners. They were asked whether (yes or no) the telephone partner would make fun of them, help them out, be someone they want to be like, get along with their friends, think he or she is better than they, get special favors, and be someone they would feel sorry for. These seven questions were selected from a previous study in which the perceiver students had participated (Miller et al., 1989). Results of that study showed that both mentally retarded and nonretarded perceivers expected a special education student to be more likely than a regular education student to get special favors and be someone they would feel sorry for, whereas they expected a regular education student to be more likely than a special education student to display the remaining behaviors.

Each question was assigned a score of 1, which indicated that telephone partners were expected to be like a (stereotyped) regular education student, or a score of 2, which indicated that perceivers expected their telephone partners to be like a (stereotyped) special education student. The seven questions were then summed to produce a score that could range from 7 (least like stereotypes about special education students) to 14 (most like stereotypes about special education students). This score is referred to as an index of stigmatizing social behaviors because a high score indicates that a child is perceived as being similar to stereotypes associated with members of a stigmatized group (i.e., children with learning problems).

Perceivers also were asked to predict whether (yes or no) their telephone partner would be slow to learn new things. This question about the telephone partner's academic ability was embedded among the questions they answered about their partner's social behaviors.

While the experimenter prepared the perceiver student for the telephone conversation, a second experimenter met the perceiver's assigned telephone partner at the school attended by the partner. This experimenter gave the telephone partner the same explanation given to the perceiver students, except that the telephone partners were always told that the other child was in the same grade they were. Students were asked to use fictitious names for themselves and for their schools, but were otherwise free to talk about anything they wished.

Post-conversation. Immediately after the conversation, the experimenters asked perceivers and telephone partners to report what the child they spoke to was like on the same questions used to assess perceivers' pre-conversation expectations. The experimenters then explained the purpose of the study to the children.

Observer Ratings

The experimenters audiotaped each child's contribution to the telephone conversation. These recordings were rerecorded in a random order. Three different types of observers then listened to and evaluated the recordings of the students' behavior.

Observers. Observers were currently employed elementary and secondary teachers and school administrators enrolled in a summer postgraduate seminar (n = 30), nonretarded middle school students (n = 22), and college students enrolled in introductory psychology (n = 80) (see Note 1). The middle school students were recruited from a school that none of the perceivers or telephone partners attended. Consent forms were sent to parents of all students in sixth, seventh, and eighth grades. The first 26 students for whom consent was obtained were scheduled, and 22 of these (9 females, 13 males) kept their appointments.

The purpose of these ratings was to see what impressions the participants' behavior made on people similar to those with whom they might interact. We deliberately did not train these observers because we wanted to see how naive observers might evaluate the children's behavior. We included different types of observers because age and expertise could affect perceptions of children's behavior.

Procedures. Procedures for obtaining ratings from teacher, peer, and college student observers were similar. All three types of observers were told only that they would be listening to a series of tape recordings of children's telephone conversations and asked to rate their impressions of each child. They were given no information about how the conversations were conducted or about the identity and characteristics of the children involved. Each type of observer listened to the conversations in small groups (7-10 observers). Perceivers' and targets' contributions to the conversation had been recorded separately, and each group of observers listened to a randomly selected subset of these recordings. After the group listened to a child's contribution to the conversation, each observer rated the child on a rating form. Thus, each perceiver's and each telephone partner's recording was independently evaluated by separate groups of teachers/administrators, peers, and college students.

Stigmatizing social behaviors. Observers rated each child they heard on the seven stigmatizing social behaviors described above. They indicated the likelihood that each behavior would characterize the child's interaction with other children on a 5-point scale (1 =very unlikely; 5=very likely). Ratings on each scale were averaged across observers within each type of observer group. Before these averages were computed, all items were scored so that higher scores indicated more similarity between the child's behavior and stereotypes about children with learning problems. For example, high scores indicated that other children would feel super" for to the child and would not want to be like the child. These average ratings were then summed within each observer group. This produced three indices of stigmatizing social behavior for each child which could range from 7 (least like stereotypes about special education students) to 35 (most like stereotypes about special education students). Each index represented the average impression his or her behavior created on a different group of observers.

Academic ability. Observers also were asked to estimate each child's grade level and to rate on 5-point scales the likelihood that the child would be slow to learn new things, would perform well in school, and would have above average intelligence. The average rating made by each type of observer was computed for each item. Then, the ratings that had been made on 5-point scales (i.e., ratings of the children's ability to learn, school performance, and intelligence) were summed to form an index of academic ability which could range from 3 (low ability) to 15 (high ability). Thus, each child received scores representing average teacher/ administrator, peer, and college student estimates of his or her grade level and academic ability.

RESULTS

Children's Ratings of Each Other

Perceivers' ratings of their telephone partners on the index of stigmatizing social behaviors were analyzed to see (a) whether perceivers had stereotyped expectations about their partners prior to the conversation and (b) whether perceivers maintained their stereotype based perceptions after conversing with the telephone partners. The analysis used to address these questions was a 2 (Perceiver Child Classification) x 2 (Telephone Partner Description) x 2 (Pre- vs. Post-telephone conversation) analysis of variance in which the last factor was within subjects.

There was a significant main effect for telephone partner description, F(1,36) = 3.99, p =.05. This effect indicated that telephone partners described as special education students were rated higher on the index of stigmatizing behaviors (M=11.1) than were telephone partners who were not so described (M=10.4). Table 1 presents perceivers' mean pre- and post conversation ratings of telephone partners in each experimental condition. Although examination of these means suggests that differences between ratings of special education and regular education telephone partners were more pronounced before the conversation began than after it was over, the interaction between timing of the ratings and telephone partner description was not significant, F(1,36)=.96, ns. There was a significant main effect for the timing of the ratings which indicated that, regardless of condition, perceivers thought their telephone partners had displayed less stigmatizing behavior (post-conversation M= 10.4) than they had expected them to (pre-conversation M = 11.0), F(1,36) = 11.37, p < .01. Mentally retarded and nonretarded perceivers did not differ significantly in their ratings of the telephone partners, F(1,36) =.48, ns. Thus, both retarded and nonretarded perceivers cognitively confirmed their stereotyped expectations about their telephone partners.

The factors described above also were used in an analysis of variance of perceiver's ratings of whether their telephone partners would be slow to learn new things. Results revealed a significant interaction between perceiver child classification and telephone partner description, F(1,36) = 6.21, p < .05. Simple effects tests showed that nonretarded perceivers rated special education telephone partners as slower to learn new things than regular education partners (Ms = 1.1 and 1.7, respectively; 1=yes and 2=no), F(1,36)=25.4, p < .001. In contrast, mentally retarded perceivers' ratings of special and regular education telephone partners (Ms = 1.1 and 1.3, respectively) did not differ significantly, F(1,36) = 2.34, ns. For both groups of perceivers, ratings made before and after the telephone conversation were virtually identical.

Finally, Table 1 also shows mean scores for telephone partners' ratings of perceivers on the index of stigmatizing social behaviors. These ratings were analyzed to see if telephone partners detected behavioral differences between mentally retarded and nonretarded perceivers and between perceivers who thought they were special education or regular education students. An analysis of variance in which perceiver classification and telephone partner description were the only factors revealed no significant effects. A similar analysis of telephone partners' ratings of the perceivers' ability to learn new things also revealed no significant effects.

Observers' Ratings of Stigmatizing Social Behaviors

Observers' ratings of the children were used to see if people not directly involved in the telephone conversation could detect any behavioral differences resulting from perceivers' expectations about their telephone partners or from differences in the social skills of mentally retarded and nonretarded perceivers. Ratings on the index of stigmatizing social behaviors were analyzed by a 2 (Perceiver Child Classification) x 2 (Telephone Partner Description) x 2 (Child's Role: perceiver vs. telephone partner) x 3 (Type of Judge) unweighted means analysis of variance. The child's role in the conversation was treated as a within-subjects' factor because the behaviors of perceivers and telephone partners were interdependent, making the dyed the unit of analysis. Similarly, type of judge was treated as a within subjects factor because to do otherwise would greatly inflate the degrees of freedom and power of the analysis, and because the children, not the observers, were the unit of analysis (see Note 2).

This analysis revealed a significant interaction between perceiver classification, telephone partner description, and child's role in the telephone conversation, F(1,36) = 5.57, p < .05. Table 2 presents mean ratings of perceivers and telephone partners and summarizes the results of simple effects comparisons between means.

As might be expected from previous research on social skills, mentally retarded perceivers engaged in more stigmatizing behaviors than nonretarded perceivers. However, this was true only when perceivers spoke to a regular education student, F(1,36) = 11.07, p < .01, not when they spoke to a special education student, F(1,36)=.52, ns.

Although telephone partner description significantly affected the behavior of mentally retarded and nonretarded perceivers, the differential treatment telephone partners received did not affect their behavior. That is, simple effects tests revealed no significant differences in the behavior of telephone partners who supposedly were in special education or regular education classes. Finally, the joint effects of telephone partner description and perceiver child classification were similar across all three groups of observers, F(2,34)=.46, ns.

Observers' Ratings of Academic Abilities

Observers' ratings on the index of academic afility and estimated grade level were submitted to analysis of variance in which the factors were the same as those used to analyze the index of stigmatizing behaviors. We had predicted that nonretarded perceivers would talk down to a special education student, and results for estimated grade level suggested that they did. The interaction between perceiver child classification, telephone partner description, and child's role in the conversation was significant for estimates of the child's grade level, F(1,36) = 4.26, p < .05, but was not significant for the index of academic ability, F(1,36) = 2.41, ns.

Simple effects tests on estimated grade level (see Table 2) revealed that nonretarded perceivers who spoke to a special education telephone partner were rated nearly two grade levels lower than nonretarded perceivers who spoke to a regular education partner, F(1,36)= 11.70, p<.01. Moreover, nonretarded and retarded perceivers were rated as being in approximately the same grade when perceivers thought their partners were in regular education classes, F(1,36) = 1.83, ns, but nonretarded perceivers received lower grade level ratings than their mentally retarded counterparts when perceivers thought they were speaking to a special education student, F(1,36)=4.53, p<.05.

This difference in the way perceivers treated special education and regular education students did not produce any significant effects on telephone partners" estimated grade level.

Although analysis of the index of academic ability revealed no effects relevant to self-fulfilling prophecies, there was a significant interaction between perceiver child classification and the child's role in the conversation, F(1,36) = 4.64, p <.05. Simple effects tests indicated that mentally retarded perceivers received lower academic ability ratings than did nonretarded perceivers (Ms = 8.8 and 9.9, respectively, F(1,36)= 14.74, p<.001, but that academic ability ratings of the telephone partners who spoke to mentally retarded and nonretarded perceivers did not differ (Ms = 10.3 and 10.6, respectively), F(1,36) =.68, ns. Because mentally retarded and nonretarded perceivers actually did differ in academic abilities, whereas their telephone partners were randomly assigned to condition, this finding is not surprising.

DISCUSSION

The mentally retarded and nonretarded perceivers who participated in this study engaged in a relatively naturalistic conversation with a child whom they believed was a special education or regular education student. Impressions of the conversation were assessed from multiple vantage points: perceivers' reports about their partners, telephone partners" reports about perceivers, and naive observers' ratings of both perceivers and partners.

Cognitive Confirmation of Perceivers' Expectations

From the perceivers' point of view, the conversation was relatively straightforward. Prior to the telephone conversation, those children who were told they would be speaking to special education students expected their partners to behave in a relatively stigmatizing fashion. After the conversation was over, they reported that their partners had behaved in a relatively stigmatizing fashion. This occurred despite the fact that observer ratings of telephone partners revealed no behavioral differences between partners who supposedly did and did not have learning problems. Thus, both mentally retarded and nonretarded perceivers cognitively confirmed their stereotyped expectations about special education and regular education students.

Because previous research has indicated that labeling a child as mentally retarded often has relatively weak effects on perceptions when information about the child's behavior also is available, it is surprising that perceivers in the present experiment who had the opportunity to converse with the target children persisted in their stereotyped beliefs (Bak & Siperstein, 1987; Gibbons & Gibbons, 1980; Siperstein & Bak, 1985; van Bourgondien, 1987). One important difference between the present study and previous research is that perceivers actively participated in a real interaction with the target of their stereotypes, whereas participants in prior research typically were exposed to filmed or written portrayals of another child's behavior. Being involved in the give-and-take of a conversation with a real child probably is a more vivid experience than is reading about or watching another (often fictitious) child's behavior. Although this could make the targets" behavior in the present study even more salient than it usually is, the relative richness, complexity, and spontaneity of an actual interaction could also leave more room for biased interpretation and remembering of behaviors.

The finding that perceivers maintained their stereotyped beliefs about their partners after the conversation was over does not mean that perceivers were oblivious to their telephone partners' behavior. There was a highly significant difference between perceivers' pre- and post-conversation ratings that indicated that, regardless of condition, perceivers rated their telephone partners' social behavior during the conversation as less stigmatizing than they had expected it to be. This finding suggests that the relative difference between perceivers' postconversation ratings of special education and regular education telephone partners is not simply the result of their being unwilling to change what they had said prior to the conversation. Perceivers did change their ratings of their partners in general, but this did not eliminate the effects of telephone partner description on post conversation ratings. Nevertheless, it would be desirable in future research to include conditions in which stereotype measures are administered only after the interaction is over. This would reduce some of the demand characteristics that may have contributed to perceivers' ratings in the present experiment.

Effects of Expectations on Perceivers' Behavior

According to the observers, mentally retarded perceivers' displayed more stigmatizing social behaviors than did nonretarded perceivers only when perceivers thought they were speaking to a regular education student. Because observers did not know which perceivers were retarded and, in fact, were never told that any of the children they heard were retarded, their judgments about the social behavior for of the perceiver students could not have been biased by their own preconceived ideas about mentally retarded and nonretarded children. Thus, the finding that observers detected differences in the social behavior of retarded and nonretarded perceivers who spoke to regular education telephone partners indicates that the two groups of perceivers actually behaved differently.

Different levels of social skill development could account for this behavioral difference, but the finding that observers detected no differences between the social behavior of mentally retarded and nonretarded perceivers who thought they were speaking to a child with learning problems provides an alternative explanation. Perhaps some of the social skill "deficits" of retarded children are a consequence of stereotyped expectations.

Ratings of perceivers' grade level were consistent with the hypothesis that nonretarded perceivers would "talk down" to a child with learning problems. When nonretarded perceivers thought they were speaking to a special education rather than a regular education student, their estimated grade level dropped from approximately the seventh to the fifth grade. In fact, when speaking to a special education student, nonretarded perceivers apparently went to such great lengths to accommodate their partners' lower cognitive abilities that they appeared to be in a lower grade than did retarded perceivers in the same condition.

These findings are consistent with previous research showing that people use baby talk when speaking to retarded people (DePaulo & Coleman, 1986). They also are consistent with results of some recent studies which suggest that nonretarded children may overcompensate for the cognitive impairments of retarded children by patronizing them (Gibbons, Sawin, & Gibbons, 1979).

Although telephone partner description affected grade level ratings of nonretarded perceivers, it did not affect ratings on the index of academic ability. This suggests that, in adjusting their behavior to accommodate the needs of a child who supposedly had learning problems, nonretarded perceivers acted relatively younger and/or less academically advanced, but not necessarily less academically able.

In contrast to nonretarded perceivers, there was no evidence of expectation effects in estimates of mentally retarded perceivers' grade level. One reason for this may be that, although mentally retarded perceivers expected special education and regular education telephone partners to differ in social behavior, they expected both types of partners to be equally slow to learn new things. Lacking differential expectations about their telephone partners' learning abilities, mentally retarded perceivers had no reason to talk down to telephone partners in special education classes. Alternatively, mentally retarded perceivers may have little control over those aspects of behavior that communicate academic progress. There were real differences between the academic abilities of retarded and nonretarded perceivers, and these differences were detectable by the observers. This suggests that retarded perceivers may have been unable to modify ability-related behaviors even if they had been motivated to do so by their expectations.

Effects on Telephone Partners' Behavior

Because perceivers' beliefs about their telephone partners' learning abilities affected their own behavior, the telephone partners' behavior should have been affected as well, but it was not. It may be that the telephone partners simply were unaware of the treatment they received. The finding that there were no differences in ratings made by telephone partners of perceivers who thought they were speaking to special education or regular education students is consistent with this notion. Another possibility is that telephone partners were aware of how they were treated but, due to self-presentation concerns (Baumeister, 1982), were unwilling to accurately report their impressions of the perceivers or to respond in kind to the treatment they received.

In addition, because stereotypes about mentally retarded and nonretarded people are not simple, uniformly positive or negative expectations (Miller et al., 1989; Willey & McCandless, 1973; Williams, 1986), perceivers probably were communicating a relatively complex set of expectations. Results of a number of studies indicate that the targets of such expectations often fail to confirm them in their behavior (see Deaux & Major, 1987, for a review). Finally, the social interaction we observed was a brief telephone conversation. Longer or more frequent interactions might be necessary to establish the conditions that would create a self-fulfilling prophecy.

Comments on Methodology

This study was not designed to determine which of the perceivers' expectations were the most important determinants of their behavior, and there may have been some important expectations that were not included on the measures we used. Also, because variations in verbal labels, physical appearance, and behavior can affect children's evaluations of supposedly retarded and nonretarded children (see Siperstein & Bak, 1986, for a review), it is possible that using a different way to communicate information about a child's learning abilities might produce somewhat different results.

We also chose to test our hypotheses by measuring the impressions the children's behavior made on observers rather than by measuring the occurrence of specific behaviors. A disadvantage of this method is that it reveals the existence of behavioral differences without pinpointing precisely what it was the children did that created the impression they made. The major advantage of this method is that it is sensitive to relatively subtle differences in behavior that otherwise might be uncodable. For example, the difficulty of including all relevant and important behaviors in a coding system may explain why systematic observations of the behavior of mentally retarded and nonretarded children often fail to document differences between them, despite the widespread assumption, supported by reports from teachers and classmates, that such differences exist (Gampel et al., 1974; J. Gottlieb et al., 1978; Peters et al., 1974; Taylor et al., 1987). Although some researchers believe that the occurrence of rare, but critical, behaviors might account for this discrepancy (e.g.,J. Gottlieb et al., 1978), we suspect that many behavioral differences that affect children's acceptance of one another are too subtle to assess by counting the frequency and duration of specific behaviors.

Finally, we sampled only children with moderate mental retardation who were mainstreamed in at least one regular education classroom. Characteristics of the participating schools and the communities they serve are likely to have influenced the results we obtained. Consequently, stereotypes might have different effects for other populations of children with mental retardation.

CONCLUSIONS

This study demonstrates that expectations have observable effects on mentally retarded and nonretarded children's social interaction with another child. It indicates that nonretarded children may adopt simple speech when addressing a child with learning problems, and suggests that some caution should be used when attribusing differences in social behavior exclusively to differences in the social skills of retarded and nonretarded children. Because recent research has shown that the actual behavior of children labeled as retarded or nonretarded is a more potent determinant of how other children react to them than is the stigmatizing label (Gibbons & Gibbons, 1980; Siperstein & Bak, 1986; van Bourgondien, 1987), it is tempting to conclude that stereotypes may be less important than was previously assumed. However, the findings of the present study indicate that stereotypes may lead children to alter their behavior toward others. By anticipating what a retarded or nonretarded child is like, both types of children may precipitate reactions from others that otherwise would not occur.

Authors' Notes

This research was supported by Grant No. G008400642 from the U.S. Department of Education.

Notes

1. College student observers made separate ratings (in a random order) of the first and second half of each conversation. A preliminary analysis of variance in which the first and second halves of the conversations were compared produced no theoretically relevant effects. Consequently, we used each college student's average ratings of the first and second halves of the conversations for all subsequent analyses to make it possible to compare their ratings with those made by teacher and peer observers.

2. We are grateful to Robert Rosenthal for suggesting this analysis.

References

Adams, G.R. (1977). Physical attractiveness research: Toward a developmental social psychology of beauty. Human Development, 20, 217-240.

Aloia, G.F., Beaver, R.J., & Pettus, W.F. (1978). Increasing initial interactions among integrated EMR students and their nonretarded peers in a game playing situation. American Journal of Mental Deficiency, 82, 573-579.

Bak, J.J. & Siperstein, G.N. (1987). Effects of mentally retarded children's behavioral competence on nonretarded peers' behaviors and attitudes: Toward establishing ecological validity in attitude research. American Journal of Mental Deficiency, 92, 31 -39.

Baumeister, R. (1982). A self-presentational view of social phenomena. Psychological Bulletin, 91, 3-26.

Berscheid, E., & Walster, E. (1974). Physical attractiveness. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 7, pp. 158-216). New York: Academic Press.

Birch, J.W. (1974). Mainstreaming: Educable mentally retarded children in regular classes. Reston, VA: Council for Exceptional Children.

Braun, C. (1976). Teacher expectations: Sociopsychological dynamics. Review of Educational Research, 46, 185-213.

Brophy, J., & Good, T. (1974). Teacher-student relationships: Causes and consequences. New York: Holt, Rinehart & Winston.

Christensen, D., & Rosenthal, R. (1982). Gender and nonverbal decoding skill as determinants of interpersonal expectancy effects. Journal of Personality and Social Psychology, 42, 75-87.

Corman, L., & Gottlieb, J. (1978). Mainstreaming mentally retarded children: A review of research. In N.R. Ellis (Ed.), International review of research in mental retardation (Vol. 9, pp. 251-275). New York: Academic Press.

Crosby, F., Bromley, S., & Saxe, L. (1980). Recent unobtrusive studies of black and white discrimination and prejudice. Psychological Bulletin, 87, 546-563.

Curtis, R.C., & Miller, K. (1986). Believing another likes or dislikes you: Behaviors making the beliefs come true. Journal of Personality and Social Psychology, 51, 284-290.

Darley, J.M., & Fazio, R.H. (1980). Expectancy confirmation processes arising in the social interaction sequence. American Psychologist, 35, 867-881.

Davies, R.R., & Rogers, E.S. (1985). Social skills training with persons who are men tally retarded. Mental Retardation, 25, 186-196.

Deaux, K., & Major, B. (1987). Putting gender into context: An interactive model of gender related behavior. Psychological Review, 94, 369-389.

DePaulo, B.M., & Coleman, L.M. (1986). Talking to children, foreigners, and retarded adults. Journal of Personality and Social Psychology, 51, 945-959.

Dusek, J., & Joseph, G. (1985). The bases of teacher expectancies. In J. Dusek (Ed.), Teacher expectancies (pp. 229-250). Hillsdale, NJ: Erlbaum.

Exposito, B., & Reed, T.M. (1986). The effects of contact with handicapped persons on young children's attitudes. Exceptional Children, 53, 224-229.

Farina, A., Allen, J.G., & Saul, B. (1968). The role of the stigmatized person in affecting social relationships. Journal of Personality, 36, 169-182.

Fazio, R.H., Effrein, E.A., & Falender, V.J. (1981). Self perception following social interaction. Journal of Personality and Social Psychology, 41, 232-242.

Fiedler, C.R., & Simpson, R.L. (1987). Modifying the attitudes of nonhandicapped high school students toward handicapped peers. Exceptional Children, 53, 342-349.

Fisher, C.T., & Rizzo, A.A. (1974). A paradigm for humanizing special education. Journal of Special Education, 8, 321-329.

Gampel, G.H., Gottlieb, J., & Harrison, R.H. (1974). A comparison of the classroom behaviors of special class EMR, integrated EMR, low IQ and nonretarded children. American Journal of Mental Deficiency, 79, 16-21.

Gibbons, F.X. (1981). The social psychology of mental retardation: What's in a label? In S.S. Brehm, S.M. Kassin, & F.X. Gibbons (Eds.), Developmental social psychology (pp. 249-270). New York: Oxford University Press.

Gibbons, F.X., & Gibbons, B.N. (1980). Effects of the institutional label on peer assessments of institutionalized EMR per sons. American Journal of Mental Deficiency, 84, 602-609.

Gibbons, F.X., Sawin, L., & Gibbons, B.N. (1979). Evaluations of mentally retarded persons: "Sympathy" or patronization? American Journal of Mental Deficiency, 84, 124-131.

Gottlieb, B.W. (1982). Social facilitation influences on oral reading performances of academically handicapped children. American Journal of Mental Deficiency, 87, 153-158.

Gottlieb, J. (1975). Public, peer, and professional attitudes toward mentally retarded persons. In M.J. Begab & S.A. Richardson (Eds.), The mentally retarded and society: A social science perspective (pp. 99-125). Baltimore, MD: University Park Press.

Gottlieb, J., & Davis, J.E. (1973). Social acceptance of EMR children during overt behavioral interactions. American Journal of Mental Deficiency, 78, 141-143.

Gottlieb, J., Semmel, M.I., & Veldman, D.J. (1978). Correlates of social status among mainstreamed mentally retarded children. Journal of Educational Psychology, 70, 396-405.

Gresham, F.M. (1982). Misguided main streaming: The case for social skills training with handicapped children. Exceptional Children, 48, 422-433. Gresham, F.M. (1984). Social skills and self efficacy for exceptional children. Exceptional Children, 51, 253-261.

Harris, MJ., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects: 31 meta-analyses Psychological Bulletin, 97, 363-386.

Jones, E.E., Farina A., Hastorf, A.H., Markus, H., Miller, D.T., Scott, R.A., & French, R.S. (1984). Social stigma: The psychology of marked relationships. New York: W.H. Freeman.

Jussim, L. (1986). Self-fulfilling prophecies: A theoretical and integrative review. Psychological Review, 93, 429-445.

Katz, I. (1981). Stigma: A social psychological analysis. Hillsdale, NJ: Erlbaum.

Kitano, M.K., Stiehl, J., & Cole.,J. (1978). Role taking: Implications of special education. The Journal of Special Education, 12, 59-74.

Kite, M.E., & Deaux, K. (1986). Attitudes toward homosexuality: Assessment and behavioral consequences. Basic and Applied Social Psychology, 7, 137-162.

Miller, C.T., Malcarne, V.L., Clarke, R.T., Lobato, D., Fitzgerald, M.D., & Brand, P. (1989). What mentally retarded and nonretarded children expect of one another. American Journal of Mental Retardation, 93, 396-405.

Miller, D.T., & Turnbull, W. (1986). Expectancies and interpersonal processes Annual Review of Psychology, 37, 233-256.

Mundy, P., Sigman, M., Kasari, C., & Yirmiya, N. (1988). Nonverbal communication skills in Down Syndrome children. Child Develop meet, 59, 235-249.

Ottensbacher, K., & Cooper, H.M. (1984). The effect of class placement on the social adjustment of mentally retarded children. Journal of Research and Development in Education, 17, 1-14.

Patzer, G.L (1985). The physical attractiveness phenomena. New York: Plenum Press.

Peters, E.N., Pumphrey, M.W. & Flax, N. (1974). Comparison of retarded and non retarded children on the dimensions of behavior in recreation groups. American Journal of Mental Deficiency, 79, 87-94.

Rosenthal, R., & Jacobson, L. (1968). PygmaIion in the classroom. New York: Holt, Rinehart & Winston.

Sheare, J.B. (1974) Social acceptance of EMR adolescents in integrated programs. American Journal of Mental Deficiency, 78, 678-682.

Siperstein, G.N., & Bak, J.J. (1985). Effects of social behavior on children's attitudes toward their mildly and moderately men tally retarded peers. American Journal of Mental Deficiency, 90, 319-327.

Siperstein, G.N., & Bak, JJ. (1986). Under standing factors that affect children's attitudes toward mentally retarded peers. In C.J. Meisel (Ed.), Mainstreaming handicapped children: Outcomes, controversies, and new directions (pp. 55-75). Hillsdale, NJ: Erlbaum. Skrypnek, B.J., & Snyder, M. (1982). On the self-perpetuating nature of stereotypes about women and men. Journal of Experimental Social Psychology, 18, 277-291.

Slosson, R.L. (1984). Slosson Intelligence Test for children and adults. East Aurora, NY: Slosson Educational Publications.

Snyder, M., & Swann, W.B., Jr. (1978a). Behavioral confirmation in social inter actions: From social perception to social reality. Journal of Experimental Social Psychology, 14, 148-162.

Snyder, M., & Swann, W.B., Jr. (1978b). Hypothesis-testing processes in social interaction. Journal of Personality and Social Psychology, 36, 1202-1212.