Evaluation of an intervention using cross-race friend storybooks to reduce prejudice among majority race young children

Early Childhood Research Quarterly 40 (2017) 110–122

Contents lists available at ScienceDirect

Early Childhood Research Quarterly

Evaluation of an intervention using cross-race friend storybooks to reduce prejudice among majority race young children Philip Jai Johnson, Frances E. Aboud ∗ McGill University, Canada

a r t i c l e

i n f o

Article history: Received 1 April 2016 Received in revised form 30 January 2017 Accepted 28 February 2017 Keywords: Prejudice reduction Attitude change Ethnic racial bias Early childhood Media intervention Program evaluation

a b s t r a c t We tested a cross-race friendship picture book intervention under two conditions that aimed to facilitate the reduction of racial prejudice that children of two age groups ascribe to adults reading the book (communicators) and their own prejudice. White majority children, 113 kindergarteners (M = 5.97 years) and 103 second graders (M = 7.93 years) were randomly assigned to an ingroup (White) or outgroup (Black) communicator and skill training. They were pretested on racial attitudes they ascribed to their photographed communicator and then trained in reconciliation or classification. They heard four cross-race friend stories; then were retested on communicator attitudes, own attitudes and cognitive elaboration. Results indicated only second graders, and those able to reconcile differences in perspectives, predicted on the basis of their photo alone that outgroup communicators would hold positive attitudes toward Blacks. After communicators read the stories and explicitly stated their antibias attitude, reports changed among second graders who now rated both communicators as holding more positive Black attitudes. Second graders also held more positive Black attitudes themselves. Kindergarteners consistently assumed both ingroup and outgroup communicators were pro-White. Children elaborated more about the Black story characters when read to by an outgroup (Black) communicator. In conclusion, cross-race friend storybooks are a promising way to expose children to other racial groups and to antibias attitudes, under certain conditions. © 2017 Elsevier Inc. All rights reserved.

1. Introduction Mixed outcomes have been found with past home and school interventions to reduce racial/ethnic prejudice in children under 8 years of age, underscoring the need to examine conditions that enhance their effectiveness. Two common strategies, namely direct contact with children from other racial groups and media and/or instruction that tell and show children about other racial groups, were similarly effective with majority race (mainly White) children in that two-thirds of the systematically reviewed studies yielded positive attitude change (Aboud et al., 2012). Because majority children in North America and elsewhere have few opportunities for direct contact with minority children (Derman-Sparks & Ramsey, 2011), indirect contact via picture storybooks, called media-mediated contact, is coming under closer scrutiny. Interventions to prevent and reduce prejudice among children are central

∗ Corresponding author at: Department of Psychology, 1205 Dr. Penfield Avenue, Montreal, QC H3A 1B1, Canada. E-mail addresses: [email protected] (P.J. Johnson), [email protected] (F.E. Aboud). http://dx.doi.org/10.1016/j.ecresq.2017.02.003 0885-2006/© 2017 Elsevier Inc. All rights reserved.

to efforts to promote a more equal and inclusive society, reduce bullying and violence, and set norms of respect for diversity. Media-mediated contact was operationalized here by showing and reading to White children picture storybooks about the exciting adventures of a White child with a Black friend (hereafter called cross-race friend storybooks or simply storybooks). It combines some features of direct contact with an outgroup friend with features of media interventions. Direct contact occurs naturally in integrated schools when children make friends with peers, and it has also been studied experimentally by putting children from different racial groups together to work or play. Media interventions expose children to picture storybooks, television or video depicting children of different racial groups playing together (Cameron, Rutland, Brown, & Douch, 2006), though sometimes a respectful, antibias message is stated explicitly (Verkuyten & De Wolf, 2007). Several controlled evaluations of Sesame Street programs (e.g. Cole et al., 2003; Fluent Public Opinion, 2008) and others with short exposures to video and/or stories (Kowalski, 1998; Lichter, Johnson, & Ryan, 1973; Persson & Musher-Eisenman, 2003; Verkuyten & De Wolf, 2007; Wham, Barnhart, & Cook, 1996) found mixed attitudinal outcomes for children over 6 years, and little impact on younger ones.

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

Still, the storybook interventions look promising for young children, in particular if we can identify reasons for their mixed effectiveness and overcome the challenges with specific educationrelated inputs. One possible reason for these mixed results is that children are distorting the antibias message of the stories, specifically by ascribing biased attitudes to the adult reading the story. Only a few studies asked children to report on the communicating adults’ attitudes (Johnson & Aboud, 2013; Pahlke, Bigler, & Suizzo, 2012; Verkuyten & De Wolf, 2007; Vittrup & Holden, 2010) and there were obvious gaps in the accuracy of their ascriptions. These gaps raise concerns about whether children are receiving the intended antibias message. Three variables that may improve the accuracy of these ascriptions are the child’s age, reconciliation skills, and the race of the communicator (the adult reading the story). Thus, the purpose of the present study was to determine if age, reconciliation, and the race of the communicator could improve the accuracy with which children assess the unbiased attitudes of a communicator. We used the basic cross-race friend storybook intervention with White children from kindergarten and second grade, the age group where findings have yielded the most mixed results. Even in comparison with same-race storybooks, those exposed to cross-race stories showed few benefits (e.g. Katz & Zalk, 1978; Lichter et al., 1973). The second variable, reconciliation skills, was expected to

111

enhance the impact of the stories if children could acknowledge the respectful antibias attitude of the adult reading the stories. The third variable, namely whether the communicator’s race was ingroup White or outgroup Black, was expected to influence the attitudes children ascribed to the communicator. Here we use the term “ingroup” to refer to the participants’ White racial group and “outgroup” to refer specifically to brown-skinned people or Blacks. Bias refers to attitudes or communicated messages that White targets are good and/or Black targets are bad; whereas antibias refers to attitudes or communicated messages that Black targets are good (White targets might be described as good or neutral). By combining a popular school activity, namely teacher-led story reading, with variations on the reading adult’s race, and variations in children’s reconciliation skills, we build on a conceptual framework (Bigler & Liben, 2006) that proposes age-related cognitive filters as the source of distortion leading to stereotype maintenance (see Fig. 1). Bigler and Liben identify variables relevant to distorted processing of stereotype-inconsistent information, and we add parallel variables relevant to distortion of attitudinal messages. The specific relevance of our selected variables is explained by more micro-level theories of attitude change and of how children ascribe attitudes to others. One is communication theory which explains change following exposure to antibias messages

Prejudice acquired concerning salient racial groupsby age 4 years

Encounter member of racial outgroup

Member possesses stereotypic-inconsistent aributes Member is seen with an ingroup or outgroup friend

Cognive filter applied to member’s aributes

Forget or distort individual’s aributes to make consistent with stereotype & retain stereotype

Age: 4-8 years old Receive racial atude message

Message is explicitly anbiased Mulple Classificaon skills naturally occurring or trained

Cognive filter not applied to member’s aributes

Acknowledge individual’s aributes; alter one’s stereotype & its generalizability

Reconciliaon skills naturally occurring or trained

Communicator is ingroup or outgroup

Cognive filter not applied to adult’s atude

Acknowledge adult’s anbias atude; alter one’s atude

Cognive filter applied to adult’s atude

Forget or distort adult’s atude to make it biased; retain bias

Fig. 1. Conceptual framework of processes involved in the maintenance or modification of racial prejudice (extended from Bigler & Liben’s, 2006, encounters with outgroup members, to include also receipt of antibias messages from Aboud, 2005).

112

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

and explicitly when delivered by an ingroup or outgroup communicator (Elaboration Likelihood Model of Petty & Cacioppo, 1986; Slater & Rouner, 2002). However, because communication has not always been effective in bringing about change, we looked to variables from several other theories to enhance its effect. Contact theory suggests that recipients may be more receptive to this message if they can identify with a White ingroup friend or communicator who likes the outgroup story character (Allport, 1954; Wright, Aron, McLaughlin-Volpe, & Ropp, 1997). Finally, because most media messages are communicated by someone (e.g., a teacher), we argue that social-cognitive developmental theory may explain why children under 8 years are less receptive than older children to media interventions, namely because they lack reconciliation skills and so egocentrically distort messages about attitudes different from their own (Aboud & Doyle, 1996; Bigler & Liben, 1993; Piaget & Weil, 1951). They may be less likely to distort if the adult reading the story is from the outgroup and children are able to acknowledge the adult’s preference for an outgroup member. Consequently, we recognize that several theories of change are needed to explain why children of certain ages are more receptive to adults from the ingroup or outgroup who communicate antibias messages. We start by discussing theory and research related to the main intervention, namely communicating antibias messages with cross-race friend storybooks, before addressing the two conditions examined here, specifically reconciliation and communicator race. The important variable of age is expected to interact with the variables of reconciliation and attention to communicator’s race. 1.1. Cross-race friend storybook interventions Storybooks featuring friendships between ingroup and outgroup children may provide the kind of exposure, similar to having direct friendship contact, considered optimal for prejudice reduction. Stories and pictures can convey equal status, respect and affection, mutual goals, and parental support. However, in recent studies, only when the story characters looked similarly white-skinned (e.g., Cameron et al., 2006), but not when one was brown-skinned and one white-skinned (Johnson & Aboud, 2013), did White children 4–8-years of age express positive attitudes toward the outgroup after the exposure. The latter study found that children distorted the attitudes of the reading adult claiming that the adult held anti-Black attitudes. The distortion occurred especially when she was from their White ingroup and did not strongly express her own antibias attitude (Johnson & Aboud, 2013). This is in fact the common context for story reading in that parents and teachers rarely comment on race and assume that mentioning race might make their children biased (Pahlke et al., 2012; Vittrup & Holden, 2010). In brief, a number of factors converge to constrain children’s receptivity to an antibias message such as messages conveyed in cross-race friend storybooks. Many of these constraints are associated with age, particularly in the early childhood years prior to the age of 8, and include the levels of self-reported bias, ascription of bias to others, egocentrism, and superficial processing of communications. Specifically, children may already be quite biased by the age of 4 or 5 years, holding positive ingroup and/or negative outgroup attitudes. A systematic review of studies examining age differences in White children’s bias found that it declined after age 8; only after age 8 was bias reduced as a result of direct contact with members of visible racial minorities, hereafter called “outgroup” (Raabe & Beelmann, 2011). Second, because of an “egocentric” cognitive constraint (Selman, 1971), children younger than 8 years assume that others such as parents and teachers are likewise biased (Aboud & Doyle, 1996; Augoustinos & Rosewarne, 2001; Vittrup & Holden, 2010). Without an explicit antibias statement by the adult, children simply hear the adult read the story and do not think his or

her choice of book reflects personal attitudes or values (Johnson & Aboud, 2013). If the children think the adult is biased, then they will be receiving an unintended biased message rather than the adult’s intended antibias one. Thus, our main outcome was children’s reports of antibias attitudes of the adult reading to them. Our overall goal was to identify conditions that might enhance children’s depiction of the antibias attitudes of the adult reading to them, especially the adult’s positive attitudes toward Blacks. All children were to have exposure to an adult showing and reading them the cross-race friend stories, but under varying conditions. One relevant condition is how well children can take the differing perspective of another person and judge it to be a valid one (here called perspective-taking and reconciliation). For children to depict accurately the communicator’s attitude, they would need to control their own perspective and acknowledge the communicator’s differing attitude – conditions often not met with children under 8 years (e.g., Mares & Acosta, 2010; Pahlke et al., 2012). A second relevant condition is the race of the communicator. Children have been found to distort the expressed antibias attitudes of an ingroup communicator, but they might ascribe positive Black attitudes to an outgroup Black communicator. Building on past research, we first examined how ready young children were to receive an antibias message from adult communicators as a result of their perspective-taking and reconciliation skills. Theory and research on social development were used to derive hypotheses related to the question: How would children differing in perspective-taking and reconciliation skills predict attitudes of an ingroup and an outgroup communicator based on appearance alone before hearing the message? Second, we examined whether children were receptive to an antibias message from the story and the communicator. Theory and research on the Elaboration Likelihood Model were used to derive hypotheses related to the question: How would children change their reports of the communicator’s attitudes after hearing her message, as a function of the communicator’s race. Third, we asked whether training in reconciliation would help children more accurately represent the communicator’s antibias message. 1.2. Perspective taking and reconciliation skills for representing antibias messages Theory and research on social development were used to derive hypotheses related to the question: How would children differing in perspective-taking and reconciliation skills predict attitudes of an ingroup and an outgroup communicator based on appearance alone before even hearing the story? The ability to take the perspective of another and reconcile differing attitudes might make children more receptive to an antibias attitude despite their own bias; with these skills they would be prepared to ascribe pro-Black attitudes to a Black adult (Piaget & Weil, 1951). Researchers have generally assumed that children are sensitive to the race of an adult, that they understand an outgroup Black adult prefers his or her own group, and this explains why children report less biased attitudes to an outgroup experimenter (e.g., Katz & Zalk, 1978). However, these assumptions were not borne out in the Johnson and Aboud (2013) study where White children 4–8 years ascribed to outgroup Black communicators pro-White attitudes, and reported pro-White attitudes themselves. If unable to take an outgroup perspective, children might assume, on the basis of a photo alone, that both ingroup and outgroup communicators are more positive toward Whites than towards Blacks. Age matters here: Children are found to take the perspective of another person in predicting their racial attitudes at 5 or 6 years of age, but their judgment that the other person’s different attitude is valid comes several years later (Aboud, 2003). Consequently, because of these cognitive constraints, kindergarteners

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

may ascribe to both ingroup and outgroup communicators biased pro-White attitudes, whereas second graders may recognize on the basis of appearance alone that outgroup Black communicators would be pro-Black or unbiased. Taking the perspective of people from different racial groups was therefore measured, along with the ability to reconcile a different attitude and accept its validity (i.e., it is not wrong to like Blacks). The hypothesis was that based on the communicator’s appearance alone, second graders but not kindergarteners would predict more positive attitudes toward Black target children to the outgroup communicator than to the ingroup communicator. Positive Black attitudes ascribed to the outgroup communicator were expected to correlate with perspective-taking and reconciliation skills. 1.3. Communicator’s race and age differences in how messages are processed The communicator’s race may also determine how well his/her message is processed and therefore how influential it is in producing attitude change. If the communicator’s message is only superficially processed or attributed to external rather than internal motivations (e.g., the adult says that Blacks are nice but she is only saying what the books says, not what she thinks), then the communication will understandably not be convincing. Current theory and research on attitude change are therefore relevant to the second research question, which is: Would children change their reports of the communicator’s attitudes after hearing the message, as a function of the communicator’s race? The Elaboration Likelihood Model (ELM) explicitly addresses the communicator’s race as an important feature of the communicator, and indirectly implicates the child’s age. So we first present the contribution of ELM to explaining why ingroup communicators and older listeners produce more accurate perceptions of the message. From this we derive hypotheses for the context of this study. ELM explains the persuasiveness of messages in terms of how deeply they are processed and cognitively elaborated, operationalized as the number of target-related thoughts generated (Petty & Cacioppo, 1986). A message that is processed superficially, in terms of the communicator, the tone or the length of the message, is said to be processed through the peripheral route and not well elaborated. A message that is examined for its content and strength of argument is said to be processed through the central route and very well elaborated. Both theory and empirical findings support the claim that outgroup communicators are less influential than ingroup communicators (Mackie & Wright, 2001) because their messages are less likely to be processed by either the peripheral or the central route (Petty & Cacioppo, 1986; White & Harkins, 1994). Yet, there is some suggestion that an outgroup communicator’s message may be scrutinized more deeply if the arguments are strong and the message is about the outgroup itself (Horcajo, Petty, & Brinol, 2010; Study 2; Martin, Hewstone, & Martin, 2007). So there is slightly stronger evidence for ingroup influence but it is insufficient to make a firm hypothesis. To date, little research has examined these claims using communicators who differ in race and whose messages are about racial attitudes. Furthermore, most of the research testing the ELM is conducted with young adults who likely have access to both peripheral and central processing. Still, ELM theory and research would suggest that young children, being cognitively less able than adults, prefer the peripheral route (Te’eni-Harari, Lampert, & Lehman-Wilzig, 2007) and attend only to the ingroup communicator because they share superficial attributes such as race with the communicator. So the ELM prediction is that children will more accurately represent messages of the ingroup communicator than the outgroup communicator. Older children, such as second graders, if capable of deeper processing of the message itself might

113

accurately receive the message of the outgroup communicator. To explore the value of ELM in this context, we included a measure of elaboration about the story characters who were the focus of attitudinal messages. Assessing children’s elaboration would help decide whether the theory is applicable to children and whether there is an age difference. An alternative prediction is that young children do not even process the content of the message at all and so show no change in ascribed attitudes. If children use the peripheral processing route and attend to superficial features of the message such as the group affiliation of the communicator, they may hardly process the message content. In this case, children may report the communicator’s attitude based on appearance only, and maintain that view regardless of the message heard during the story reading. If processing is this superficial, then children will not change their reports after hearing the story. We hypothesized that second graders would ascribe to both communicators more positive attitudes toward Black targets after compared to before hearing the stories. Kindergarteners would show little or no change, meaning that they would ascribe to communicators non-positive attitudes toward Black targets even after hearing the stories. 1.4. Reconciling different attitudes and ELM The third research question was: Would children be more receptive to the adult communicator’s explicit antibias message conveyed while reading the storybooks and thereby ascribe to the communicator more positive Black attitudes if they were trained to take a different perspective than if they were trained in another social-cognitive skill? Based on social-cognitive developmental theory (Selman, 1971), we assumed that most of our participants would be egocentric and so would distort the communicated antibias message to make it fit with their own bias. This might negate or neutralize the storybook message. So training them to reconcile different attitudes before hearing the antibias stories might give them the skill to accept and not distort the antibias messages. For example, past research suggests that young children find antibias messages to be unexpected and non-normative when spoken by an ingroup White adult (Augoustinos & Rosewarne, 2001); their cognitive filter then distorts the message (Johnson & Aboud, 2013; Verkuyten & De Wolf, 2007). Consequently, we attempted to train children to reconcile attitudes different from their own before they heard the antibias story and message. This went beyond measuring reconciliation ability as it naturally occurred when it might influence their prediction of the communicator’s attitudes before hearing her message. Manipulating reconciliation experimentally would be a stronger way of showing its influence. Specifically, we studied the effects of reconciliation ability by training children who could not reconcile but whose age places them at the cusp of acquiring this ability. Kindergarten and second grade children who do not yet naturally possess this skill might be trainable. Previous studies have demonstrated that conservation, morally mature judgments, emotionally mature judgments, class inclusion, and other skills can be quickly adopted by children who initially do not possess these skills, if a rationale for the more sophisticated judgment is presented (e.g., Siegler & Svetina, 2006). For this reason, before exposure to the antibias message, children in the current study were given a convincing reason as to why another individual’s differing preference could be valid (e.g., why someone else might like to read a dictionary even though that is your least favorite pastime, why a Black child might prefer a Black friend even though you prefer a White friend). To equate exposure to racial groups, the comparison group received training in multiple classification, which might help to break down stereotypes (Bigler & Liben, 1992; see also Fig. 1), but was shown to have no effect on receptivity to cross-

114

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

race friend stories (Cameron, Rutland, & Brown, 2007). Still, both forms of training provide some exposure to racial groups, with reconciliation being directed specifically at valuing and not dismissing or distorting different attitudinal perspectives. The hypothesis was that non-reconcilers (egocentric children) who are trained to reconcile different perspectives before hearing the message would ascribe to communicators more positive attitudes to Black targets compared to those trained on multiple classification. This difference might be more pronounced for ingroup than outgroup communicators, because acknowledging their antibias attitude would otherwise be difficult. 1.5. The current study The goals of the current study were to examine the conditions under which a storybook intervention would enhance children’s assessment of unbiased attitudes held by ingroup compared to outgroup adult readers of storybooks, and whether this would be a function of the children’s grade level (kindergarteners or second graders), and their naturally-occurring or trained reconciliation skills. The three variables were studied in terms of their impact on children’s reports of communicators’ attitudes, particularly their attitudes to Black targets. It has been noted in past research that children distort the respectful attitudes of teachers and parents and thereby resist adopting these more desirable attitudes. Specific hypotheses are as follows: Hypothesis 1. Based on the communicator’s appearance alone (i.e., pretest ratings made while looking at the communicator’s photo and prior to training and story reading), second graders but not kindergarteners were expected to predict that the outgroup communicator would hold more positive attitudes toward Black target children than the ingroup communicator. This was based on second graders’ greater naturally occurring reconciliation skills. All children were expected to ascribe to the ingroup communicator positive attitudes toward White target children. Hypothesis 2. Second graders were expected to ascribe to both communicators more positive attitudes toward Black targets after compared to before hearing the stories. Kindergarteners were expected to show no change from before to after hearing the stories and so ascribe to communicators non-positive attitudes to Black targets. This is based on the ELM view that kindergarteners, unlike second graders, may not process the attitudinal content of the message and so maintain their “egocentric” view that all communicators are biased. Hypothesis 3. Trained reconcilers were hypothesized to ascribe more positive Black attitudes than those trained on multiple classification. This might occur more to ingroup communicators whose antibias attitude might otherwise be unexpected and distorted. Hypothesis 4. Children’s own attitudes were hypothesized to follow the attitudes they ascribe to communicators. 2. Methods 2.1. Participants Two hundred and sixteen White children (105 girls and 111 boys) from Kindergarten (n = 113, Mage = 5.97 years; SD = 0.61) and Grade 2 (n = 103, Mage = 7.93 years, SD = 0.62) received informed parental consent and were tested. This constituted approximately 75% of the eligible students. They were recruited from four small elementary schools selected because they were in predominantly White middle-class suburban neighborhoods, where the Black population was under 5%, and their teachers were White. The participating children across schools had similarly little exposure to

Black peers or Black adults. This sample size allowed for effect sizes of d = 0.50, with alpha <0.05 to be powered at 0.80 when evaluating grade differences, and powered at 0.70 when evaluating the interactions hypothesized to take place, for example Grade X Communicator X Pre-Post Reading or Training X Communicator X Pre-Post Reading. 2.2. Design The study used a 2 (Grade: kindergarten vs. grade 2) X 2 (Communicator Race: ingroup vs. outgroup) X 2 (Training Condition: reconciliation vs. multiple classification) X 2 (Time: pre- vs. posttest) design, with Time being a repeated measure (see Fig. 2 for the full design with subsample sizes). Stratified by grade, children were randomly assigned to the communicator race and training conditions prior to starting the procedure. To examine the effects of the intervention conditions on how children perceived the communicator’s attitudes to Black targets and White targets, they were tested at two time points – before and after the story reading. Children’s own attitudes and cognitive elaboration were assessed at post-test only. 2.3. Procedure The testing and intervention occurred over the course of four sessions, each four to five days apart. In the first session (pretest), children ascribed racial attitudes to their ingroup or outgroup communicator while looking at a colored photograph of her (essentially making a prediction based on the communicator’s appearance alone), and then were trained on reconciliation or classification by the same tester. In the second and third sessions (intervention), children met with their assigned communicator who read the cross-race friend stories and stated her positive evaluations. In the fourth session (post-test), the tester re-assessed children’s reports of their communicator’s attitudes, their own attitudes, and their cognitive elaboration of the message. Testers used in Sessions 1 and 4 were White female university students. Communicators were two ingroup White females and two outgroup Black females recently graduated from university, of similar age and attractiveness. 2.4. Session 1. Pre-test measures and training 2.4.1. Multiresponse racial attitude measure (MRA; Doyle & Aboud, 1995) The MRA is a 20-item measure that is used to obtain separate indices of participants’ attitudes towards Black and White target children. In the current study, it was also used to assess their reports of the communicator’s racial attitudes at pre- and post-test, while viewing a colored photograph of her. It consists of 8 positive (e.g., smart, friendly), 8 negative (e.g., selfish, naughty), and 4 neutral evaluations (e.g., likes to run, likes music) with behavioral examples of each. As described elsewhere (Tredoux, Noor, & de Paulo, 2009), children place cards depicting the evaluation into one, both, or neither of two boxes, tagged with a same-sex racial face. Attitude scores for analysis are derived for both White and Black stimuli by subtracting the number of negative evaluations from the number of positive ones to produce a score falling in the range of −8 to +8 (alphas 0.62 positive White to 0.91 negative Black) where higher scores indicate more positive evaluations. Lower alphas are often found for positive White evaluations because White children are less uniform with these, especially as they get older. 2.4.2. Reconciliation This measure assesses children’s ability to accept differing self and other judgments as equally valid (Aboud, 1981). It consists of a 60 cm liking board on which participants place 6 standing color

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

115

Fig. 2. Flow Diagram.

photographs of same-sex stimulus children (i.e., a pair from each of White, Black, and Chinese groups) in order of their preference. The tester then proceeds to show the preferences of a recently interviewed, photographed, Black peer by arranging a duplicate set of photos on a second board in a mirror-image of the participant’s. Participants are then asked the reconciliation question, “Are both of you right, or is someone wrong?” If children correctly judge both arrangements to be right, the tester asks them for a justification. A correct judgment and an appropriate justification (e.g., she is different from me so her preferences are different; people can like whomever they want) each receive one point for a maximum score of 2. The tester then repeats the task with the photo of a White role peer. Thus the total score on the reconciliation measure ranges from

0 to 4 (alpha 0.85). Children scoring 3 or 4 are classified as natural reconcilers. 2.4.3. Reconciliation training Children were trained over four trials using the same liking board, where they were presented with opposing preferences and a reconciliation rationale for why both preferences were valid. The person who administered the MRA and Reconciliation measure conducted this training, using a script. In the first of four trials, children were asked to arrange pictures of various activities (e.g., swimming, painting, bowling) in terms of their preferences, with the most liked activity closest to them. The tester then arranged her preferences of the same activities on a second liking board, in a

116

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

mirror image of the child’s arrangement, ostensibly reflecting her own preference. They were then asked the reconciliation question, “Are both of us right, or is one of us wrong?” Regardless of their response, children were provided with the reconciliation rationale that both their preferences were correct because people can have different likes and dislikes based on their own feelings and ideas. In the second trial, children’s preferences for pictures of ingroup and outgroup teachers were contrasted with the mirror-arrangement preference of a gender-matched ingroup peer (who was shown to equally prefer White and Black stimulus teachers). Children were not asked the reconciliation question, and were immediately presented with the reconciliation rationale. In the third and fourth trials, children’s arrangement of ingroup and outgroup stimulus peers was contrasted with the preferences of an outgroup child followed by an outgroup adult (who equally preferred a White and Black stimulus peer). As a manipulation check on successful training, they were asked the reconciliation question on the fourth training item and then given the rationale. 2.4.4. Multiple classification training The person who administered the MRA and Reconciliation measure conducted this training, using a script. This is a 4-trial training process requiring simultaneous attention to 2 or 3 attributes of objects and people. The first two trials dealt with pictures of objects differing in shape and color, and the remaining two items with photographs of people differing in race, age, and gender. The tester placed a picture in three of the quadrants of a 2 × 2 matrix such that the attributes were varied systematically, e.g., by shape along the columns and color across the rows for the objects, by race along the columns and age along the rows (Aboud, 2003). The child was then required to choose from four response options an object/person to go in the empty quadrant to complete the puzzle. Regardless of children’s response, they were provided with the rationale for the correct 2 × 2 arrangement at the end of each trial. Answers to the last two trials (with people, where race was one of the attributes) were scored. The training in this study differs from Bigler and Liben (1992) and Cameron et al. (2007) where children sorted cards into two separate piles sequentially and were not given a rationale for the correct arrangement. 2.5. Sessions 2 and 3. Story intervention with communicator manipulation The intervention phase consisted of two 15-min sessions where the assigned communicator read four stories, each approximately 800 words, featuring cross-race friendships between White and Black children, two male and two female pairs. In each session, children heard two stories that were presented in counterbalanced order. The stories were based on the exciting adventures of two best friends, and were adapted from store-bought picture books; they were appropriate for both age groups. A professional artist modified the illustrations of Black children in two of the books to clarify race-appropriate features. Stories were also customized to ensure that the friendships were unconditionally positive, that the cross-race friends appeared together on every page, and that at least once their parents were shown expressing explicit approval of the friendship (according to Allport’s criteria for positive intergroup contact, 1954). Communicator race was manipulated by having a White (ingroup) or Black (outgroup) young woman introduce herself as a teacher from another school, and read the stories in a quiet room to two or three children at a time. Communicators were trained to read in a similarly engaging way, and to show warmth and acceptance toward the children. They spoke to children and answered their questions, but refrained from engaging in extensive discussions beyond the script. Children sat beside her holding the book

and turning the pages in order to look closely at the pictures. The communicator introduced the stories by mentioning that they were about two friends who did exciting things together, even though they looked different. After she finished reading a story, she provided a scripted epilogue (Slater & Rouner, 2002) that reminded children of the positive nature of the cross-race friendship (e.g., “Alex and Zack are such good friends, they really had fun together, there was a lot of excitement on the raft”), the White character’s fondness for his or her Black friend (“Alex really likes being with Zack, he thinks Zack is pretty smart getting them to shore”), and the communicator’s respect for the Black character’s interpersonal attributes (“I really like Zack and Alex; I think Zack is a really good friend to have”). Training took place over several hours as communicators practiced reading the stories aloud at the right pace, as well as saying the extra scripted phrases. They practiced on their own and then were tested in a role play until communication was appropriate in terms of intonation, pace, responsiveness to questions, and ability to speak the scripted phrases in a conversational tone. 2.6. Session 4. Post-test measures In the fourth and final session, the original tester readministered the MRA to individual children, first asking them to report on the communicator’s attitude and then to report their own attitudes towards the Black and White target children. Children’s own attitudes were assessed at post-test and only after they reported on their communicator’s attitudes. This was done in order to minimize the salience of their own attitudes. We had two reasons for wanting to minimize the salience of children’s attitudes. If children expressed their own attitudes at pretest they might want to appear consistent throughout the testing – a source of resistance to change. Also, their own attitudes might be less likely to change than reports of communicators’ attitudes, given that the communicators were explicitly stating their antibias evaluations. If their own attitudes were salient, they could influence perceptions of communicators’ attitudes – a potential source of resistance to changing their reports of communicators’ attitudes. The cognitive elaboration measure was administered as described below. 2.6.1. Cognitive elaboration measure Comparable to a thought-listing task for adults (Clark & Wegener, 2009), this measure elicited children’s thoughts and feelings about the characters. The tester shows children pictures of the storybook characters, and as a warm up, the tester asks if they can recall the characters and their names. The elaboration questions ask whether or not they like the White and the Black characters, and then they are asked to give examples of attributes they liked or did not like (i.e., positive and negative thoughts about the characters). All answers were recorded verbatim and rated. Personal preferences were given a point for affirmative responses. Elaborations of attributes and activities were coded as positive, neutral, or negative, with one point for each response. Answers of 20 children were coded by two independent raters. Correlation coefficients were calculated for the number of positive, neutral, and negative responses, which showed high agreement between the two raters: r(20) = 0.85 to 0.98; all p < 0.01. Thereafter, one researcher coded all responses of the study sample. Across stories, alphas ranged from 0.72 to 0.79. 3. Results 3.1. Plan of analysis Pretest data analyses were conducted on natural reconciliation scores to examine grade differences. We also analyzed children’s

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

pretest predictions, based only on the photograph of the communicator, of both ingroup and outgroup communicators’ attitudes to Black and White targets. The analyses to examine the effects of grade (2), communicator’s race (2), and training condition (2) were conducted on the following dependent variables: (i) children’s reports of Black and White attitudes ascribed to ingroup and outgroup communicators’ at pre- and post-test; (ii) children’s own Black and White attitudes assessed at post-test; and (iii) children’s cognitive elaboration of the message also assessed at post-test. Cohen’s effect sizes d were calculated with their 95% confidence intervals (CI). There were no missing data. Because our hypotheses focus on attitudes toward Blacks and how children rated communicators’ attitudes toward Blacks, we conducted follow-up post hoc analyses, using simple main effects analyses, of significant effects on the two target races separately without necessarily requiring target race to interact significantly with other variables. This strategy is based on the understanding that both sets of attitudes might become more positive but that it would be more important for attitudes toward Blacks to change as a result of the intervention. Separate grade analyses were likewise performed because of a priori hypotheses about second graders’ change.

117

connection between natural reconciliation skills and recognizing that outgroup communicators are positive toward Black targets partly addresses Hypothesis 1. To examine children’s predictions of communicators’ attitudes towards the Black and White target children at pretest, i.e., when the communicator’s photograph was presented but no story reading or training had occurred, a 2 (Grade: kindergarten vs. grade 2) × 2 (Communicator Race: ingroup vs. outgroup) × 2 (Training condition: reconciliation vs. multiple classification) × 2 (Target Race: Black vs. White) analysis of variance was conducted, with repeated measures on the target variable. There was a main effect of Target Race, F (1, 208) = 65.08, p < 0.0001, Á2 = 0.24, which was qualified by two interaction effects. The first was a Grade x Target Race effect, F (1, 208) = 11.90, p = 0.0007, Á2 = 0.05, and the second was a 3-way Grade × Communicator Race x Target Race effect, F (1, 208) = 5.38, p = 0.02, Á2 = 0.03 (see Table 1). No other effects were statistically significant. Analyses were conducted separately by target race (see Table 1). Concerning attitudes toward the Black target, a significant Grade × Communicator effect, F (1, 208) = 6.70, p = 0.01, yielded a simple Communicator effect for kindergarteners, F = 4.33, p = 0.04, indicating they predicted more positive Black attitudes to the ingroup compared to the outgroup communicator (Cohen’s d = 0.42, CI 0.03, 0.81), but no Communicator effect for second graders (p = 0.11). There was also a simple main Grade effect for the outgroup communicator, F = 15.48, p = 0.0001, indicating more positive Black attitudes predicted by grade 2 compared to kindergarten children (Cohen’s d = 0.82, CI 0.43, 1.21). Predictions made by children regarding their communicators’ attitudes toward the White target showed no significant differences for grade, communicator, training or any interaction (all p’s > 0.10). Both ingroup and outgroup communicators were expected to hold highly positive White attitudes. Thus, on the basis of the communicators’ appearance alone, children assumed that both ingroup and outgroup communicators would hold more positive attitudes toward Whites than Blacks. Only second graders, and this appeared linked to reconciliation ability (recall the correlation of r = 0.32, p = 0.007), predicted that outgroup communicators would have positive Black attitudes; kindergarteners expected ingroup communicators to have more positive Black attitudes than outgroup communicators but these were generally neutral (M = 1.29 and −0.46).

3.2. Preliminary analyses Analyses were conducted on ascribed attitudes with school and participant gender as between-subject variables. No effect of school was found. However, girls made more positive ratings of White targets than boys. As no effects of gender were found with Black attitudes, the outcome of interest in this study, data were pooled for girls and boys during subsequent analyses. 3.3. Hypothesis 1. Predicting communicators’ attitudes at pretest from photographs Pretest scores on the reconciliation test before training correlated as expected with age (M = 1.65, SD = 1.58; r = 0.38, p < 0.0001). Among kindergarteners 13.27% obtained the passing score of 3 or 4 out of 4, whereas 50.49% of second graders passed. Reconciliation scores correlated positively with children’s predictions about the outgroup Black communicator’s attitude toward Blacks (r = 0.32, n = 115, p = 0.0007) but not with predictions about the ingroup communicator’s attitude toward Blacks (r = 0.13, n = 101, p = 0.18). This

Table 1 Mean (SD) Reports by Children of Communicators’ Black and White Attitudes as a Function of Grade, Communicator Race, and Time (pre vs. post) (n = 216). Communicator Race Ingroup

Outgroup

Grade

Pre

Post

Pre

Post

Sample sizes Kindergarten n = 113 Grade 2 n = 103 Total N = 216

41 60 101

41 60 101

72 43 115

72 43 115

I. Ascribed Black Attitudes Kindergarten

1.29a (4.50)

−0.46ab (3.96)

−0.43 (5.24)

c

1.22 (5.16) c

Grade 2

1.43 (4.28)

3.08 (4.66)

2.84bd (4.41)

4.07d (4.25)

Total

1.37 (4.35)

2.33 (4.93)

0.77 (4.42)

1.25 (5.34)

II. Ascribed White Attitudes Kindergarten

4.61 (3.37)

4.85 (3.96)

4.43 (3.00)

4.42 (3.68)

Grade 2

3.95 (3.66)

4.08 (3.95)

3.58 (3.84)

4.16 (4.36)

Total

4.22 (3.54)

4.40 (3.96)

4.11 (3.35)

4.33 (3.93)

Note. Cell sizes were the same for the repeated factors of Time (Pre, Post) and Target (attitudes toward Blacks, attitudes toward Whites). The theoretic range for attitudes is −8 to +8. Means with the same superscript were significantly different at p < 0.05; effect sizes varied from d = 0.32 to 0.82 (exact values in text).

118

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

3.4. Hypothesis 2. Overall influence of the intervention on ascribed attitudes.

3.5. Hypothesis 3. Effect of training condition

To test the hypothesis that children would ascribe more positive Black attitudes after hearing the stories, the data were submitted to a 2 (Grade) × 2 (Communicator Race) × 2 (Training) × 2 (Target Race: Back vs. White) × 2 (Time: pre- vs. post-test) ANOVA, with repeated measures on the last two variables. To examine changes in ascribed attitudes, main effects of Time and its interactions were identified (see Table 1 and Fig. 3 ). There was a main effect of Time, F (1, 208) = 8.43, p = 0.004, Á2 = 0.04, indicating that children ascribed to communicators increasing positivity in attitudes from pre-test to post-test. This was superseded by an interaction effect of Time x Grade, F (1, 208) = 6.77, p = 0.01, Á2 = 0.03. There were no other significant effects of Time, notably no significant Time x Communicator interactions indicating that pre-post changes, or lack of changes, applied to both ingroup and outgroup communicators. Follow-up tests showed that there was no change in kindergarteners’ ascribed attitudes from preto post-test (p = 0.84) for Black targets (M = 0.41 and 0.39, respectively) and for White targets (M = 4.52 and 4.63, respectively). However, follow-up analyses showed that second graders significantly ascribed more positive Black attitudes to both communicators’ from pre-test to post-test (M = 2.02, SD = 4.36; M = 3.50, SD = 4.50, F = 13.00, p < 0.0001; effect size d = 0.32, CI 0.05, 0.60). No Time effects were found for second graders’ ascription of White attitudes to communicators (p = 0.21). Thus, the intervention was successful only with second graders, who improved in the Black attitudes they ascribed to both communicators after exposure to the antibias message. Specifically, they ascribed greater positivity to the ingroup communicator only after explicit exposure to her message, and greater positivity to the outgroup communicator in addition to what they had predicted based on her appearance alone. Kindergarteners’ remained resistant to the antibias message, showing no change in the biased attitudes toward Blacks they initially ascribed to communicators, based on appearance alone.

First we present the analysis including all children. Second, we present effects of training for those who were initially nonreconcilers to see whether successful training increased their receptivity to the antibias stories. This subsample analysis was necessary because some children, mainly second graders were already able to reconcile and did not require training, and because some children even after training were unable to reconcile. Those successfully trained in reconciliation, compared to unsuccessfully trained, were expected to be better at ascribing positive Black attitudes to communicators. The first analysis used the full sample with a 2 (Grade) × 2 (Communicator) × 2 (Training) × 2 (Target Race) × 2 (Time) analysis. Training had no significant main effect or interaction with time or target when analyzing the full sample (all p > 0.20). This is not surprising, as those with reconciliation skills at the start would not be expected to change with training. Training appeared to have an effect only when analyses were conducted on the subsample of children initially classified as non-reconcilers (n = 149 out of 216, or 69%), of whom 68 had been randomly assigned to receive reconciliation training and 81 to receive multiple classification training. Among those trained in reconciliation, 46 out of 68 (67.65%) were successful in passing the final reconciliation item; whereas 40 out of 81 (49.38%) of those trained in multiple classification were successful in passing the final classification item. A 2 (Communicator Race) × 2 (Training: reconciliation vs. multiple classification) × 2 (Outcome: successful vs. unsuccessful) ANCOVA was conducted on communicators’ ascribed Black attitudes at post-test, covarying pre-test (see Table 2). Data from the two grades were combined (n = 149) because the second grade sample was small (n = 51) and like kindergarteners they were initially unable to reconcile. There was a significant Training × Outcome interaction effect, F (1, 138) = 5.10, p = 0.02, Á2 = 0.03. Post-hoc analyses showed that children successfully trained in multiple classification ascribed to communicators significantly more negative Black attitudes compared to children unsuccessfully trained

6

Atude score ascribed to communicator

Pre, based on seeing her photo 5

Post, aer she read the story and stated her atude

4.07

4

3.08 2.84 3

2

1.29 1.22

1.43

1

0

-0.46 -0.43 -1 Kinder

Grade 2

Atude of Ingroup Adult

Kinder

Grade 2

Atude of Outgroup Adult

Fig. 3. Means of Black attitudes ascribed by children in Kindergarten and Grade 2 to their ingroup White or outgroup Black communicator, before and after receiving the antibias message.

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

119

Table 2 Mean (SD) Black and White Attitudes Ascribed to Communicator and Reported by Children for Themselves as a Function of Training (Reconciliation, Multiple Classification) and Outcome of Training (Succeed, Not Succeed) (n = 149).

Attitude

Reconciliation Succeed (n = 46)

Not Succeed (n = 22)

Classification Succeed (n = 40)

Not Succeed (n = 41)

I. Black Attitudes Ascribed to Communicator Children’s Own Attitude

1.74 (5.30) 0.65 (5.28)

−0.86 (4.84) −0.95 (4.84)

−1.05a −0.45 (4.56)

2.32a (5.47) 2.00 (5.12)

II. White Attitudes Ascribed to Communicator Children’s Own Attitude

4.33 (3.92) 5.02 (3.08)

3.86 (3.78) 4.05 (4.01)

3.95 (3.69) 4.02 (3.43)

3.90 (4.88) 4.34 (3.92)

Note. Sample sizes apply to all data presented in that column. The theoretic range for attitudes is −8 to +8. Means with superscript a were p = 0.01, d = −0.68.

(M = −1.05 vs. M = 2.32, respectively, F = 6.40, p = 0.01; effect size d = −0.68, CI −1.13, −0.23). In contrast, post-hoc analyses showed that children successfully trained in reconciliation ascribed to communicators nonsignificantly more positive attitudes toward Blacks compared to those unsuccessfully trained (M = 1.74 vs. M = −0.86, respectively, F = 0.75, p = 0.37; effect size d = 0.50 or moderate but imprecise with CI −0.01, +1.02). No main or interaction effects were found when White attitudes were analyzed (all F < 1.0). Thus, successful training in multiple classification had the unintended effect of making children think that communicators were more negative toward Black children. Because of the subsample size, the interaction effects noted were powered at 54%. 3.6. Hypothesis 4. Effects on children’s own attitudes Children’s own post-test attitudes were analyzed similarly using the full sample of n = 216 and the subsample of children who were initially unable to reconcile (n = 149). Because we had only postintervention attitude ratings for children themselves, we covaried the pre-intervention rating made for the communicator (correlations between self and pre-intervention communicator ratings varied from r = 0.60 to 0.65 except when rating an ingroup communicator’s attitude toward Whites, r = 0.35). The 2 (Grade) × 2 (Communicator Race) × 2 (Training) × 2 (Target Race) ANCOVA (see Table 3) yielded an expected grade effect, F(1, 208) = 9.73, p= 0.002; an effect of Target, F(1, 208) = 12.58, p = 0.0005, and Grade × Target interaction, F(1, 208) = 5.81, p = 0.02. The interaction showed that second graders were more positive to Blacks than were kindergarteners (F = 11.52, p = 0.0008), but children from the two grades were not different in their attitudes toward Whites (F = 0.26, p = 0.60). There were no effects of training or communicator on children’s own attitudes. Likewise, the subsample of non-reconcilers who were subsequently trained also showed no effect of training (see Table 2).

Scrutiny of Table 2 indicates that children’s own attitudes were similar to those ascribed to communicators, but somewhat more positive: own attitudes toward Blacks correlated r = 0.65 and 0.68 with ingroup communicator’s and outgroup communicators’ ascribed attitudes, respectively; while own attitudes toward Whites correlated r = 0.53 and 0.72, with communicators’ attitudes, respectively. 3.6. Children’s cognitive elaborations Thought favorability ratios for Black and White characters were computed by subtracting the total number of negative thoughts about the story characters from the total number of positive thoughts, and dividing by the total number of thoughts (Clark & Wegener, 2009). The data were submitted to a 2 (Grade) x 2 (Communicator Race) x 2 (Training Condition) × 2 (Target Race: Black vs. White) ANOVA, with repeated measures on the last variable. A main effect of Grade was found, F (1, 208) = 4.38, p = 0.038, Á2 = 0.02, indicating that second graders (M = 0.71, SD = 0.47, n = 103) thought more favorably about the characters compared to kindergarteners (M = 0.60, SD = 0.65, n = 113, d = 0.19, or small and imprecise according to the CI −0.07, 0.46). There was a main effect of Communicator Race, F (1, 208) = 7.71, p = 0.006, Á2 = 0.04, and of Target Race, F (1, 208) = 19.87, p < 0.001, Á2 = 0.09, but this was superseded by a Communicator Race × Target Race interaction, F (1, 208) = 4.75, p = 0.03, Á2 = 0.02. Follow-up analyses revealed that children elaborated more on the Black characters when read to by the outgroup compared to the ingroup communicator (M = 0.68, SD = 0.53, n = 109 vs. M = 0.47, SD = 0.64, n = 107, respectively, p = 0.01, d = 0.36, CI 0.09, 0.63). Children also elaborated more on the White characters when read to by the outgroup versus ingroup communicator but the difference was nonsignificant (M = 0.77, SD = 0.40 vs. M = 0.69, SD = 0.48, d = 0.18, CI −0.09, 0.45). As such, the outgroup communicator was found to provoke more positive elaborations than the

Table 3 Mean (SD) Children’s Own Black and White Attitudes as a Function of Grade, Communicator Race (Ingroup, Outgroup), and Training (Reconciliation, Classification) (n = 216). Ingroup Reconciliation

Classification

Outgroup Reconciliation

Classification

−1.12 (5.53) n = 17 2.34 (4.90) n = 29

0.25 (5.01) n = 24 2.55 (4.62) n = 31

−0.38 (5.16) n = 39 3.87 (4.07) n = 23

−0.03 (4.89) n = 33 2.10 (5.29) n = 20

−0.24a (5.06) n = 113 2.71a (4.70) n = 103

1.06 (5.36) n = 46

1.54 (4.88) n = 55

1.19 (5.18) n = 62

0.81 (5.10) n = 53

1.16b (4.32) n = 216

II. Children’s White Attitudes Kindergarten 5.29 (3.20)

4.29 (3.77)

4.85 (3.32)

3.76 (3.65)

4.48 (3.50)

Grade 2

5.07 (2.78)

4.70 (3.59)

3.61 (4.57)

4.25 (3.45)

4.47 (3.50)

Total

5.15 (2.91)

4.53 (3.64)

4.38 (3.85)

3.94 (3.55)

4.48b (3.50)

Grade I. Children’s Black Attitudes Kindergarten Grade 2 Total

Totals

Note. Cell sizes were the same for Target (Black attitudes, White attitudes). The theoretic range for attitudes is −8 to +8. Means with superscript a were different at p = 0.0008. Means with superscript b were different at p = 0.0005.

120

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

ingroup communicator, and particularly of the Black storybook characters. Children elaborated least about the Black characters when stories were read by the ingroup communicator. 4. Discussion The results of the current study indicate support for three of the four hypotheses, particularly in identifying reconciliation skills and outgroup communicators as important factors. Importantly, the storybook intervention was successful in improving children’s reports of communicators’ attitudes toward Blacks at post-test (Hypothesis 2). This was due to second graders ascribing more positive attitudes to communicators after hearing them read the stories compared to before. Second graders and natural reconcilers showed a prior readiness, elicited by the communicator’s photo, to acknowledge that others held positive Black attitudes, particularly outgroup communicators (Hypothesis 1). Second graders extended this inference to ingroup communicators after hearing their antibias message, providing some support for the ELM-derived claim that older children would process the content of the antibias message while younger ones would not. Findings did not confirm the ELM hypothesis that the ingroup communication would be better received. Training in reconciliation only partly facilitated receptivity. Cognitive elaborations of Black story characters were higher when read by the outgroup compared to the ingroup communicator. These findings have important implications for the contribution of both social development and ELM theories to explaining attitude change and the conditions under which storybook interventions are effective. 4.1. Cross-race friend storybook intervention effects The first two research questions addressed the crux of this evaluation, specifically would children change the Black attitudes they ascribed to communicators after seeing and hearing the story, and would the change be greater for second graders compared to kindergarteners and for outgroup compared to ingroup communicators? The finding that children’s reports of communicators’ Black attitudes improved at post-test is encouraging, in light of disappointing outcomes of past media-mediated interventions to reduce prejudice in young children (see Aboud et al., 2012 for a review). The increase in positive attitudes ascribed to communicators by second graders is consistent with other research showing that older children are more receptive to an antibias message (Cameron et al., 2007; Katz & Zalk, 1978; Verkuyten & De Wolf, 2007). The lack of change among kindergarteners parallels the outcomes of previous media-mediated interventions (Johnson & Aboud, 2013; Kowalski, 1998; Mares & Acosta, 2010; Persson & Musher-Eizenman, 2003). Thus, at least one explanation for the lack of effects of attitudechange communications with 4–6 year olds is the distorted attitude they attach to the communicating adult, namely that he or she does not like Blacks. Several conceptual frameworks were useful in interpreting the grade difference in outcome. Social development theory has a long history in identifying perspective taking and reconciliation as important skills developing between 4 and 8 years (Piaget & Weil, 1951; Selman, 1971). Here we found that second graders and those who could reconcile naturally predicted more positive attitudes for the outgroup communicators on the basis of appearance alone, before hearing their message. This may have mentally prepared them for the antibias message they heard from this outgroup communicator, thereby making them more receptive to the message from the story and from the communicator herself. Attitudes toward Black but not White targets were affected, as expected. Kindergarteners, only 13.27% of whom had reconciliation skills

at the start, assumed all communicators would be more positive to Whites regardless of their group status. Thus, developmentally acquired, in contrast to trained, reconciliation skills helped establish a readiness in some children to process the communicator’s antibias message. Training in reconciliation was largely ineffective. Those trained successfully how to reconcile different perspectives were only slightly more positive in their reports of communicators’ attitudes than those not successfully trained who were negative. Again, this applied to both communicators’ attitudes toward Blacks but not toward Whites. Children who are several months or years away from consolidating this skill may need more training given at the same time as they hear the antibias message so they are able to apply it. The ELM helped to explain how the messages were processed and led to a change in how children viewed the communicators’ attitudes. According to their reports, second graders processed the actual message of both communicators, whereas kindergarteners did not. The former showed a significant increase in positive Black attitudes ascribed to the outgroup communicator and also to the ingroup communicator who was initially thought to be strongly pro-White. Although past ELM research has found greater receptivity to ingroup over outgroup communicators, we found the opposite (Mackie & Wright, 2001; Martin et al., 2007). From the start, second graders appeared to take note of the outgroup communicator by adjusting their predictions; this was further demonstrated in children’s cognitive elaborations about Black story characters which were higher among those who encountered the outgroup communicator. The ELM therefore requires modification to the effect that outgroup communicators may be more effective when they are seen as having unique personal experience with and expertise about other outgroup members upon which they are basing their communicated attitudes. Kindergarteners’ lack of change is consistent with what has previously been reported (Johnson and Aboud, 2013; Persson & Musher-Eizenman, 2003; Verkuyten & De Wolf, 2007). They continued to report that both ingroup and outgroup communicators were positive to Whites and negative to Blacks. Lacking reconciliation skills, they did not predict different attitudes for the communicators on the basis of appearance and they did not change after hearing her antibias message. It appears that they not only lacked reconciliation skills, but they were unable to process the content of the story message or the communicators’ explicit personal antibias message. If anything, they predicted less bias from an ingroup compared to outgroup communicator. The consistent lack of positive outcomes with younger children might raise the issue of whether it is even useful to deliver a program to them, as they may not yet possess the cognitive structures to engage with an antibias perspective. Findings on incubation effects suggest that although the benefits of such interventions may not be immediately apparent (yet do not result in increased bias), at the very least, exposing children to antibias information may help prime them to be more sensitive to subsequent communications (Howe, McWilliam, & Crosse, 2005). For example, the teacher-child exchange reflected in our procedure may be repeated on numerous occasions with other adults; to the extent that these occasions are similar to the initial exchange they will trigger the communicator’s remarks about the story characters, her explicit antibias attitudes, and her solutions to the reconciliation challenge. Given that twothirds of children were able to reconcile by the end of training, these repeated experiences could strengthen their reconciliation skill and its application to antibias messages received from adults and peers.

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

4.2. Effect of reconciliation and classification training How reconciliation skills, both naturally occurring and trained, would facilitate the accurate ascription of positive Black attitudes to communicators was the subject of the first and third research questions. The ability to take the perspective of someone who expresses different attitudes and to see those attitudes as valid was considered central to whether children would accurately receive the antibias message. Naturally occurring reconciliation may have been partly responsible for second graders’ lower levels of bias in their prediction of the outgroup communicator’s attitudes based on her photo, and for their own attitudes (Hypothesis 4). Brief training in reconciliation was effective in raising reconciliation scores among 70% of children regardless of grade, and was only weakly applied by children to communicators’ Black attitudes but did not influence their own attitudes. Non-reconcilers who were successfully trained on multiple classification unexpectedly reported more negative communicator attitudes. Our training program differed from past research in that children were required to classify simultaneously individuals along two dimensions, i.e., age and race, or gender and race, rather than sequentially which was likely easier (Cameron et al., 2007; Bigler & Liben, 1993). Both training formats may draw attention to salient outgroup characteristics, which in the absence of discussion, may only serve to exacerbate prejudice rather than decrease it. 4.3. Limitations and contributions One limitation is the number of sessions; due to logistical constraints only two intervention sessions were iincluded. Ideally this storybook program would be embedded into the curriculum and continue for many sessions. However, when we asked teachers to read the stories to their class, despite coaching they were unable to explicitly mention ethnicity or race. Consequently children reported strong bias in their teachers. Another limitation is that the testers who conducted both the pre- and post-tests were not blind to the children’s conditions and so may have biased the administration of measures. Cell sizes for the analyses where training and outcome of training were used as factors were small and underpowered to find more than a 2-way interaction, as found. Finally, the reconciliation training applied to only four items and so may have been too brief and too far ahead of children’s exposure to antibias messages. The current research provides new insights into the mixed messages that children process when hearing an antibias cross-race friend story from an adult. One is the antibias message from the story and the communicator’s explicit attitudes, and the other is the unintended child-held presupposition that the communicator is biased. Educators and parents who use media to expose children to positive attitudes toward visible minorities need to understand that children come to the activity with some assumptions; they are not a blank slate. Children come with assumptions about others’ attitudes, and their assumptions that adults are biased, however unfounded, distort the media’s and the adults’ message of respect. This speaks to the need to consider children’s presumptions about adults’ attitudes and to address these explicitly. For example, kindergarteners may need to have both ingroup and outgroup adults reading to them – ingroup first because as we found, kindergarteners attributed more positive Black attitudes to their ingroup adult than to the outgroup adult. However, introducing them to outgroup adults who read these stories would provide an opportunity to prepare children ahead by having them predict the outgroup adult’s attitudes. Allowing children to listen to the assumptions of peers who have more mature reconciliation skills and to hear their justifications is often better than having children hear the justifications provided by adults. Likewise, second graders would benefit

121

most by having an outgroup adult reading cross-race friendship stories to them and engaging them in dialogue about his/her attitudes and children’s own attitudes. Elaborating on positive attributes of the story characters and their friendship helps to cement attitudes. Generalizing these positive Black attitudes to other people in the stories and in their lives would help to create the impression in young children that it is normative to respect and be friends with people who look different. Teachers may need training in how to engage children in these discussions and explicitly express their own attitudes of respect without sounding imposing or hypocritical. Television programs for children tend to avoid explicit reference to race and racial attitudes, but one could imagine an adult reading a cross-race friendship story on such a program, showing parents how to do it comfortably without fear of introducing prejudice. Adults who interact with children, specifically parents, teachers, religious educators and librarians, need guidance on what sort of commentary is required and how to engage children patiently in an open dialogue about friendship and diversity. They need guidance on how to address “racist-sounding” comments from children. Finally, story writers and publishers can help by providing more picture storybooks about the exciting adventures of children with a cross-race friend.

Acknowledgements We appreciate the financial support given to the second author from the Social Science and Humanities Research Council of Canada [grant number 410-2011-0146]. We would like to thank the children, teachers, principals and research assistants who helped with this research. This study was a part of a doctoral thesis. It received ethics approval from McGill University.

References Aboud, F. E. (1981). Egocentrism, conformity, and agreeing to disagree. Developmental Psychology, 17, 791–799. http://dx.doi.org/10.1037/0012-1649. 17.6.791 Aboud, F. E. (2003). The formation of ingroup favoritism and outgroup prejudice in young children: Are they distinct attitudes? Developmental Psychology, 39, 48–60. http://dx.doi.org/10.1037/0012-1649.39.1.48 Aboud, F. E. (2005). The development of prejudice in childhood and adolescence. In J. F. Dovidio, P. Glick, & L. A. Rudman (Eds.), On the nature of prejudice: Fifty years after Allport (pp. 310-326). Malden: MA: Blackwell. Aboud, F. E., & Doyle, A. B. (1996). Parental and peer influences on children’s racial attitudes. International Journal of Intercultural Relations, 20, 371–383. http://dx. doi.org/10.1016/0147-1767(96)00024-7 Aboud, F. E., Tredoux, C., Tropp, L. R., Brown, C. S., Niens, U., & Noor, N. M. (2012). Interventions to reduce prejudice and enhance inclusion and respect for ethnic differences in early childhood: A systematic review. Developmental Review, 32, 307–336. http://dx.doi.org/10.1016/j.dr.2012.05.001 Allport, G.W. (1954/1979). The nature of prejudice. New York, NY: Perseus Books. Augoustinos, M., & Rosewarne, D. L. (2001). Stereotype knowledge and prejudice in children. British Journal of Developmental Psychology, 19, 143–156. http://dx. doi.org/10.1348/026151001165912 Bigler, R. S., & Liben, L. S. (1992). Cognitive mechanisms in children’s stereotyping: Theoretical and educational implications of a cognitive-based intervention. Child Development, 63, 1351–1363. Bigler, R. S., & Liben, L. S. (2006). A developmental intergroup theory of social stereotypes and prejudice. In R. V. Kail (Ed.), Advances in child development and behavior (vol. 34) (pp. 39–89). San Diego: Elsevier. Cameron, L., Rutland, A., Brown, R., & Douch, R. (2006). Changing children’s intergroup attitudes toward refugees: Testing different models of extended contact. Child Development, 77, 1208–1219. Cameron, L., Rutland, A., & Brown, R. (2007). Promoting children’s positive intergroup attitudes towards stigmatized groups: Extended contact and multiple classification skills training. International Journal of Behavioral Development, 31, 454–466. Clark, J. K., & Wegener, D. T. (2009). Source entitativity and the elaboration of persuasive messages: The roles of perceived efficacy and message discrepancy. Journal of Personality and Social Psychology, 97, 42–57. http://dx.doi.org/10. 1037/a0015450 Cole, C. F., Arafat, C., Tidhar, C., Tafesh, W. Z., Fox, N. A., Killen, M., & et al. (2003). The educational impact of Rechov Sumsum/Shara’a Simsim: A Sesame Street television series to promote respect and understanding among children living

122

P.J. Johnson, F.E. Aboud / Early Childhood Research Quarterly 40 (2017) 110–122

in Israel, West Bank, and Gaza. International Journal of Behavioral Development, 27, 409–432. Derman-Sparks, L., & Ramsey, P. G. (2011). What if all the kids are White? Anti-bias and multicultural education with young children and families. New York, NY: Teachers College. Doyle, A. B., & Aboud, F. E. (1995). A longitudinal study of white children’s racial prejudice as a social-cognitive development. Merrill-Palmer Quarterly, 41, 209–228. http://www.jstor.org/stable/23090532 Fluent Public Opinion (2008). Assessment of educational impact of Rruga Sesam and Ulica Sezam: Report of findings prepared for Sesame Workshop, New York. Horcajo, J., Petty, R. E., & Brinol, P. (2010). The effects of majority versus minority source status on persuasion: A self-validation analysis. Journal of Personality and Social Psychology, 99, 498–512. Howe, C., McWilliam, D., & Cross, G. (2005). Chance favours only the prepared mind: Incubation and the delayed effects of peer collaboration. British Journal of Psychology, 96, 67–93. Johnson, P. J., & Aboud, F. E. (2013). Modifying ethnic attitudes in young children: The impact of communicator race and message strength. International Journal of Behavioral Development, 37(3), 182–191. http://dx.doi.org/10.1177/ 0165025412466522 Katz, P. A., & Zalk, S. R. (1978). Modification of children’s racial attitudes. Developmental Psychology, 14, 447–461. Kowalski, K. (1998). The impact of vicarious exposure to diversity on preschooler’s emerging ethnic/racial attitudes. Early Child Development and Care, 146, 41–51. Lichter, J. H., Johnson, D. W., & Ryan, F. L. (1973). Use of pictures of multiethnic interaction to change attitudes of White elementary school students toward Blacks. Psychological Reports, 33, 367–372. Mackie, D. M., & Wright, C. L. (2001). Social influence in an intergroup context. In R. Brown, & S. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes. Oxford: Blackwell Publishers Ltd. Mares, M., & Acosta, E. E. (2010). Teaching inclusiveness via TV narratives in the US. Journal of Children and Media, 4, 231–247. Martin, R., Hewstone, M., & Martin, P. Y. (2007). Systematic and heuristic processing of majority- and minority-endorsed messages: The effects of varying outcome relevance and levels of orientation on attitude and message processing. Personality and Social Psychology Bulletin, 33, 43–56. Pahlke, E., Bigler, R. S., & Suizzo, M. (2012). Relations between colorblind socialization and children’s racial bias: Evidence from European American mothers and their preschool children. Child Development, 83, 1164–1179. Persson, A., & Musher-Eizenman, D. R. (2003). The impact of a prejudice-prevention television program on young children’s ideas about race. Early Childhood Research Quarterly, 18, 530–546.

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer. Piaget, J., & Weil, A. M. (1951). The development in children of the idea of the homeland and of relations to other countries. International Social Science Journal, 3, 561–578. Raabe, T., & Beelmann, A. (2011). Development of ethnic, racial, and national prejudice in childhood and adolescence: A multinational meta-analysis of age differences. Child Development, 82, 1715–1737. http://dx.doi.org/10.1111/j. 1467-8624.2011.01668.x Selman, R. L. (1971). Taking another’s perspective: Role-taking development in early childhood. Child Development, 42, 1721–1734. Siegler, R. S., & Svetina, M. (2006). What leads children to adopt new strategies? A microgenetic/cross-sectional study of class inclusion. Child Development, 77, 997–1015. http://dx.doi.org/10.1111/j.1467-8624.2006.00915.x Slater, M. D., & Rouner, D. (2002). Entertainment-education and elaboration likelihood: Understanding the processing of narrative persuasion. Communication Theory, 12, 173–191. Te’eni-Harari, T., Lampert, S. I., & Lehman-Wilzig, S. (2007). Information processing of advertising among young people: The elaboration likelihood model as applied to youth. Journal of Advertising Research, 47, 326–340. http://dx.doi.org/ 10.2501/S0021849907070341 Tredoux, C. G., Noor, N. M., & de Paulo, L. (2009). Quantitative measures of respect and social inclusion in children: Overview and recommendations. Effective Education, 1, 169–186. http://dx.doi.org/10.1080/19415530903522568 Verkuyten, M., & De Wolf, A. (2007). The development of in-group favoritism: Between social reality and group identity. Developmental Psychology, 43, 901–922. http://dx.doi.org/10.1037/0012-1649.43.4.901 Vittrup, B., & Holden, G. W. (2010). Exploring the impact of educational television and parent-child discussions on children’s racial attitudes. Analyses of Social Issues and Public Policy, 10, 192–214. Wham, M. A., Barnhart, J., & Cook, G. (1996). Enhancing multicultural awareness through storybook reading experience. Journal of Research and Development in Education, 30, 1–9. White, P., & Harkins, S. (1994). Race of source effects in the elaboration likelihood model. Journal of Personality and Social Psychology, 67, 790–807. Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The extended contact effect: Knowledge of cross-group friendships and prejudic. Journal of Personality and Social Psychology, 73, 73–90.