Age differences in the effects of social influence on children’s eyewitness performance and their metacognitive monitoring

Journal of Experimental Child Psychology 94 (2006) 229–248 www.elsevier.com/locate/jecp

Age diVerences in the eVects of social inXuence on children’s eyewitness performance and their metacognitive monitoring Stefanie Schwarz a,¤, Claudia M. Roebers b a

Department of Psychology IV, University of Würzburg, 97070 Würzburg, Germany b University of Berne, CH-3012 Berne, Switzerland Received 27 July 2005; revised 24 January 2006 Available online 15 March 2006

Abstract The current work investigated the eVects of social inXuence on children’s recall accuracy and metacognitive monitoring. Two studies were conducted in which 8- and 10-year-olds were confronted with postevent information in an interview situation. An interviewer (Study 1) or a confederate (Study 2) provided postevent information with two levels of assertiveness, inducing (a) a variation of conformity pressure and (b) a variation of information credibility. Afterwards, children’s conWdence judgments were assessed. The results revealed signiWcant age diVerences in children’s ability to adequately cope with variations of social inXuence. Although conformity pressure was especially important for the 8-year-olds, eVects of informative social inXuence were independent of age. However, 10-year-olds were also able to act appropriately on low credibility, thereby demonstrating a more sophisticated consideration of social inXuence sources. Moreover, varying assertiveness also aVected the quality of children’s conWdence judgments by improving their metacognitive diVerentiation skills. © 2006 Elsevier Inc. All rights reserved. Keywords: Social inXuence; Memory; Conformity; ConWdence judgments; Eyewitness memory; Suggestibility

*

Corresponding author. Fax: +49 931312763. E-mail address: [email protected] (S. Schwarz).

0022-0965/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jecp.2006.01.003

230

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

Introduction Many cognitive factors mediate diVerences in children’s eyewitness reports and their suggestibility, including prior knowledge (Ornstein, Shapiro, Clubb, Follmer, & BakerWard, 1997) and metacognitive monitoring processes (Roebers, 2002). There is also growing consensus about the importance of social factors in memory accuracy. For example, reports by adult eyewitnesses can be contaminated when participants observe coparticipants being questioned (Wright, Self, & Justice, 2000), observe a confederate being questioned (Gabbert, Memon, & Allan, 2004), or read the written statement of someone they believe to be another participant (HoVman, Granhag, Kwong See, & Loftus, 2001; Walther et al., 2002). Developmental research on children’s susceptibility to natural sources of social inXuence on their memory reports is rare. Some research shows that children are sensitive to the social demands of an interview situation in that their performance diVers as a function of, for example, personal attributes of the interviewer (Goodman, Sharma, Thomas, & Considine, 1995; Jackson & Crockenberg, 1998) or the status and/or familiarity of the interviewer (Bjorklund et al., 2000). However, such studies typically do not consider age diVerences in the magnitude of social inXuence. As a result, it remains an open question whether, and to what extent, children of varying ages diVer concerning their vulnerability to the social inXuence of the interview situation. Studies examining conformity development using simple perception tasks (i.e., judgments of line lengths [Asch, 1956]) oVer evidence that, although conformity to peers shows an inverted U-shaped development with a peak during early adolescence (Brown, Clasen, & Eicher, 1986), conformity toward adults and authority Wgures decreases continuously from the early elementary school years until puberty (Berndt, 1979). In a recent study (Schwarz, Roebers, & Schneider, 2004), age diVerences in children’s susceptibility to social inXuences were investigated in the context of memory reports. In that study, 8- and 10-year-olds Wrst underwent a leading interview in which an adult confederate was present during the interview for half of the children, and the confederate answered half of the questions correctly and half incorrectly. For the other half of the children, no confederate was present. The 8-year-olds were signiWcantly aVected by leading questions and the confederate’s answers, whereas the 10-year-olds were aVected only by the confederate’s answers. Furthermore, when children were interviewed a second time using the same questions but without a confederate present, both 8- and 10-year-olds who had been interviewed with the confederate during the Wrst round achieved lower recall accuracy scores. Thus, although 10-year-olds were well able to resist the inXuence of misleading questions when the confederate was present, the confederate’s incorrect answers nevertheless impaired children’s subsequent memory reports (Schwarz et al., 2004). However, in a subsequent study, there were also indications of age diVerences with regard to the social inXuence stemming from a confederate. Roebers, Schwarz, and Neumann (2005) found that 10-year-olds in a strong social inXuence condition (misleading questions and a confederate yielding to the wrong suggestions) did not diVer signiWcantly from those in a control condition with no confederate. In contrast, 8-year-olds in the confederate condition were signiWcantly aVected. In other words, Roebers and colleagues (2005) found an age-dependent eVect of social inXuence, whereas Schwarz and colleagues (2004) did not. A possible reason for the divergent Wndings may lie in the way in which

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

231

social inXuence was implemented. In Schwarz and colleagues’ (2004) study, social inXuence was varied as a within-participants factor; thus, higher cognitive demands and more Wnely tuned adjustments were necessary to answer questions correctly. In Roebers and colleagues’ (2005) study, the correctness of the confederate’s answers to the critical questions was manipulated between participants. In actual eyewitness investigations, more subtle social factors than repeated blatantly misleading questions typically are at work; thus, on-line adjustments in response to verbal and nonverbal social cues are necessary. In the social psychology literature, the credibility of the source of information aVects adults’ answering behavior. For example, in a study by Smith and Ellsworth (1987), adult participants who were questioned about the content of a video were misled by an interviewer who had seen the video but not by an interviewer who had not seen the video. Concerning children’s ability to take credibility information into account, the existing studies have produced inconsistent Wndings. Preschool children cannot explicitly reXect on or judge the reliability of a source of information, yet their answers show that they more often believe a person who is knowledgeable because he or she actually saw something than a person who had no access to information (Robinson, 2000) and that they also comply more often to an interviewer who was present at the target event as opposed to an uninformed interviewer (Waterman, Blades, & Spencer, 2004). Furthermore, preschool children’s recall was aVected when a credible adult presented misinformation but not when the adult was discredited as “a silly man” (Lampinen & Smith, 1995). Findings of developmental studies with regard to the eVects of a warning about “tricky” questions range from a signiWcant decrease of yielding answers (Warren, Hulse-Trotter, & Tubbs, 1991) to no beneWt for witnesses’ performance and suggestibility across a wide age range (Beuscher & Roebers, 2005). Thus, credibility information as a social cue may be powerful for older children but not necessarily for younger children. Even though young children often react sensitively to social demands, their knowledge about social interactions and conversational rules is still developing and these emerging abilities may account for the inconsistent results (Saywitz, 2002). Moreover, within one conversation, social inXuence can vary considerably depending on the issue that is being discussed. For example, a conversational partner can be very certain about some details but less certain of others and may communicate this lower conWdence to the other person directly and/or indirectly. In the two studies presented in the current article, we investigated children’s ability to respond discriminately to a subtle variation of socially transmitted information. We sought an answer to the following question: Are children able to consider the assertiveness with which another person presents postevent information? As used here, assertiveness encompasses two related constructs: (a) variation of conformity pressure and (b) variation of information credibility. This diVerentiation relates to Deutsch and Gerard’s (1955) two perspectives on how the eVects of social inXuence can be viewed: as normative social inXuence (adopting presented information against one’s own better knowledge) and as informative social inXuence (adopting presented information to Wll one’s own memory gaps). In the two studies reported here, postevent information was given with diVering degrees of assertiveness, resulting in diVerent degrees of credibility (in the sense of informative social inXuence) and diVerent degrees of conformity pressure (in the sense of normative social inXuence). Social inXuence was realized as a within-participants factor that changed across questions. This was done to achieve similarity with a naturally occurring conversa-

232

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

tion in which each question or message can be accompanied by social cues that provide information about the partner’s evaluation of and conWdence in the provided information. This method provides a within-participants variation of social inXuence, allowing us to look at children’s ability to make Wne-tuned adjustments in their answering behavior during an ongoing interrogation. A cognitive factor that has been shown to inXuence recall accuracy and, at the same time, to be vulnerable to social inXuences is an individual’s ability to monitor ongoing cognitive processes metacognitively. Typically, metamemorial monitoring abilities can explain signiWcant amounts of variance in children’s recall accuracy (Roberts, 2000), but even in adults the relation between conWdence judgments and recall accuracy can vary substantially depending on task demands, recall topics, and social factors (for reviews, see Penrod & Cutler, 1995; Perfect, 2002). SpeciWcally, in an interrogation with a mock cowitness, adults’ conWdence judgments decreased or increased as a function of the correspondence between the participant’s answers and the cowitness’s answers (Luus & Wells, 1994). In addition, Moises, de Carvalho, and Masamichi (2001) showed that participants’ metacognitive monitoring judgments varied systematically in accordance with the level of a coparticipant’s conWdence. As for children’s ability to give adequate conWdence judgments, the developmental literature on children’s metacognitive monitoring abilities points, on the one hand, to pronounced improvements during the elementary school years with regard to the overall level of conWdence and also with respect to the ability to adequately give higher conWdence judgments after correct responses than after incorrect responses (Roebers, 2002). On the other hand, there are certain conditions under which developmental progression takes place earlier. For example, by 8 years of age, children can diVerentiate between correct and incorrect answers to unbiased questions, but their monitoring is less adequate for misleading questions (Roebers, 2002; Roebers & Howie, 2003). In addition, a mix of question formats, as opposed to a “bombardment” with misleading questions, improves children’s metacognitive diVerentiation (Roebers & Howie, 2003). Taken together, these studies clearly suggest that children’s metacognitive monitoring abilities undergo developmental progression but are also aVected by variations of social inXuence. However, it has not yet been explored systematically how and to what extent variations of social inXuence aVect children’s conWdence judgments. Because social inXuence was manipulated in the current studies, an inclusion of conWdence judgments seemed warranted and promising. In addition, there is a strong similarity between assertiveness and conWdence in the current studies: Assertiveness was varied by adding information as to how certain the interviewer or confederate was that her answer or suggestion was correct. This strongly resembles conWdence judgments. By assessing participants’ conWdence judgments following the exposure to social inXuence, we were able to inspect whether the variation of assertiveness aVects children’s metacognitive diVerentiation. To answer the question of whether children can consider an even Wner variation of socially transmitted information by varying assertiveness and the question of whether and how assertiveness inXuences children’s conWdence judgments concerning their answers, we conducted two separate studies using two diVerent experimental situations. In Study 1, we used misleading questions as the source of postevent information. Assertiveness was manipulated by additional social cues of the interviewer regarding her own certainty accompanying each suggestion. In Study 2, a female confederate of the interviewer, giving her answers before the children gave their answers, was used as the source of postevent information. For each of her answers, the confederate provided additional qualiWers indi-

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

233

cating how sure she felt that her answer was correct. In both studies, recall measures as well as conWdence judgments were considered. To continue our work on age diVerences in response to social inXuences, we again tested 8- and 10-year-olds. The existing literature so far is dominated by age comparisons of preschool and school-age children in a way suggesting that there are no more age diVerences in children’s event reports over the primary school years. However, many studies, including our previous and current studies, document important developmental changes during the early school years in children’s ability to react appropriately on various forms and sources of social inXuence, all of which are relevant to children’s recall in an interview (for reviews, see Bruck & Ceci, 1999; Saywitz, 2002). Our experimental design does not allow an absolute distinction between normative and informative social inXuence. However, it can be assumed that for the confederate’s positively leading answers, mainly informative inXuence should be at work and conWrm children’s recall or improve their recall performance by Wlling memory gaps. Predominantly normative social inXuence, in contrast, can be expected for the confederate’s misleading answers, contradicting children’s own recall. In any case, the two kinds of answers given by the confederate are very likely to result in diVerent degrees of social inXuence, and we aimed to study developmental diVerences concerning the eVects of these diVerent impacts. We expected that a combination of high assertiveness and misleading information would aVect younger children more strongly than it would aVect older children due to the former’s greater vulnerability to conformity pressure. Children of both age groups, in contrast, should beneWt from a combination of high assertiveness and positively leading information, with high assertiveness implying high credibility resulting in the perception of reliable informative social inXuence. In contrast, low assertiveness should lead to the perception of lower credibility, thereby reducing the exerted conformity pressure. Thus, we expected smaller eVects of informative and normative social inXuence. Furthermore, we anticipated that the variation of assertiveness itself should aVect children’s metacognitive monitoring. Because Study 1 included a misleading question format and Study 2 included an unbiased question format, we expected—based on the existing empirical evidence—signiWcant metacognitive diVerentiation between correct and incorrect answers in Study 2 for both age groups but not necessarily in Study 1. The additional information about the confederate’s varying certainty resembles, as a cognitive factor, conWdence judgments and could unspeciWcally inXuence children’s own metacognitive monitoring. Moreover, varying certainty is also a social factor, in the sense of social pressure and credibility, that can aVect metacognitive performance.

Study 1: Misleading questions as source of postevent information Method Design overview The design of the Wrst study was a 2 (Age: 8 or 10 years) £ 2 (Assertiveness of Interviewer: high or low) factorial design with age as a between-participants factor and assertiveness as a within-participants factor. Dependent measures were accuracy of event recall and conWdence judgments.

234

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

Sample A total of 60 children (30 girls and 30 boys) participated. There were 30 8-year-olds in Grade 2 (16 girls and 14 boys, mean age D 8 years 4 months, SD D 5 months) and 30 10year-olds in Grade 4 (14 girls and 16 boys, mean age D 10 years 4 months, SD D 4 months). Children were recruited from an elementary school in a suburban area in northern Bavaria, Germany, and came from predominantly lower to upper middle-class family backgrounds. Written parental consent was obtained for all children. Procedure In small groups of 10 to 15, participants watched a short video (7 min) that had been used previously (Roebers & Fernandez, 2002; Roebers & Howie, 2003). Before the video started, children were told that the experimenter would be interested in their opinions of the Wlm later (this was done to prevent deliberate rehearsal) and that they should pay attention. The Wlm shows a group of children visiting a family in the countryside. They learn that an old castle once stood there and had a secret passageway. Hoping to Wnd a treasure, they start to search for the passage’s entrance. In the end, they climb down into underground tunnels and Wnd a dungeon and the treasure. When the video ended, the participants were asked brieXy for their opinions and thanked for their cooperation. One week later, an unfamiliar adult questioned the children individually about the contents of the video in a quiet room in the school. The interviewer Wrst built rapport and then explained that she wanted to ask some questions about the video. She remarked that she had also seen the Wlm and that she could remember some things quite clearly but could not remember others so well. Furthermore, she stressed that it was very important to give only correct answers and encouraged the child to respond with “I do not know” whenever she or he was uncertain. The interview began when the child indicated that she or he understood these accuracy instructions. After 22 questions about the Wlm (described subsequently), the interviewer introduced a 5-point scale for conWdence ratings. A 10 £ 40-cm dark blue piece of cardboard that showed Wve diVerent yellow “smiley faces” was placed in front of the child. The smiley faces’ mouths progressed from a very sad face on the far left to a very big smile on the far right. The interviewer explained to the child the meanings of the diVerent faces, from left to right, as being very unsure (coded as 1), not so sure (2), neither unsure nor sure (3), pretty sure (4), and very sure (5). To minimize socially desirable conWdence judgments, the interviewer told the child that she wanted to Wnd out which questions were hard and which were easy for her or him so that she could change them accordingly for kindergarten children. Each child was asked a minimum of three training questions that were independent of the recall questions and that required diVerent conWdence ratings. Training questions included, for example, “How much is 2 plus 2? And how sure are you that your answer is correct?” for the very sure conWdence rating, “How many hairs do you have on your head? And how sure are you that your answer is correct?” for the very unsure conWdence rating, and “What are you going to do this afternoon? And how sure are you that your answer is correct?” for the neither unsure nor sure category. If a child did not point to the appropriate smiley face, she or he was corrected, given a rationale for the appropriate rating, and asked another training question. Children generally learned the use of the conWdence scale with ease. When a child had given three appropriate conWdence ratings, the interviewer started

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

235

with the conWdence ratings concerning the recall questions. She repeated each question, repeated the child’s answer, and then asked the child to point to the smiley face best representing how sure she or he was that the given answer was correct. The child’s conWdence judgments were recorded on a protocol sheet. In the end, after 22 conWdence judgments, the child was praised for her or his good performance, given a small gift, and returned to the classroom. Materials The recall questionnaire consisted of 22 questions (16 target questions and 6 Wller questions) that covered the entire content of the video. All questions were in child-appropriate wording and had been used in previous studies (Roebers & Fernandez, 2002; Roebers & Howie, 2003). The questions focused on various aspects of the video (e.g., external characteristics, setting of the event, main actions and dialogues). The 16 target questions were asked in a strongly misleading format. The interviewer varied social pressure across these 16 misleading questions by expressing her own conWdence in each of the suggestions, asking 8 of the misleading questions with high assertiveness (e.g., “This I can remember quite clearly, the boys came by car to the farmhouse, right?”) and asking the other 8 with low assertiveness (e.g., “I’m not at all sure about this, but the children used candles to see in the dark tunnel, didn’t they?”). Questions were arbitrarily assigned to two sets of 8 questions, and assignment of each set to high and low assertiveness was varied across children such that questions that were asked with high assertiveness in Set 1 and were asked with low assertiveness in Set 2 and vice versa. Questions with high and low assertiveness were presented to each child in a mixed order. The six Wller questions were asked in an unbiased, open-ended format without further comment from the interviewer (e.g., “How did the dog get down into the tunnel?”). These questions were interspersed among the target questions to reduce the uniformity of question format and to avoid bombarding children with misleading questions (Roebers & Howie, 2003). All answers were recorded on the protocol sheet and later were coded as correct, incorrect, or “I do not know.” Results All main eVects and interactions are reported at the p < .05 level if not otherwise speciWed. Preliminary analyses revealed no diVerences in any of the dependent measures between boys and girls. Therefore, data were collapsed across gender. Recall For each participant, a score of recall accuracy was computed: correct answers/ [correct + incorrect answers] (see, e.g., Koriat, Goldsmith, & Pansky, 2000). However, the identical pattern of results was found for corresponding analyses with children’s ability to correctly resist the false suggestions (percentage correct answers). Table 1 presents the mean recall accuracy scores for the misleading questions with high versus low assertiveness of interviewer and for the two age groups separately. An analysis of variance (ANOVA) with level of assertiveness as a repeated measure and age as a between-participants factor was conducted. Results revealed a main eVect of assertiveness, F (1, 60) D 4.99, 2 D .08, and

236

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

Table 1 Mean recall accuracy scores in Study 1 as a function of age and level of assertiveness

High assertiveness Low assertiveness

8-year-olds

10-year-olds

.61 (.29) .73 (.22)

.74 (.19) .74 (.23)

Note. Standard deviations are in parentheses.

a signiWcant interaction between assertiveness and age, F (1, 60) D 5.40, 2 D .09. The 8-yearolds were less accurate when the interviewer’s assertiveness was high than when it was low, t (29) D 3.02, p < .01, but there was no eVect of assertiveness for the 10-year-olds, t(29) < 1. ConWdence judgments Table 2 shows the mean conWdence ratings for correct and incorrect answers separately as a function of age and level of assertiveness. For each child, the conWdence judgments after correct and incorrect answers to the misleading questions were compared to examine the ability to monitor the child’s recall processes. Indicators of adequate metamemorial competencies are higher conWdence ratings after correct responses than after incorrect responses (Roebers, 2002). Furthermore, we were interested in eVects of level of assertiveness on conWdence judgments. We conducted an ANOVA with correctness of answer and level of assertiveness as within-participants factors and with age as a between-participants factor. Results revealed a main eVect of correctness of answer, F (1, 41) D 147.86, 2 D .78, and an interaction between correctness of answer and age, F (1, 41) D 16.62, 2 D .29. Both 8- and 10-year-olds were more conWdent following correct answers than following incorrect answers, ts (26) 7 3.84. However, the diVerence between conWdence ratings on correct and incorrect answers was larger for 10-year-olds than for 8-year-olds, t (52) D 3.16, suggesting better metamemorial competence in the older children. There was no signiWcant main eVect or interaction concerning the level of assertiveness. Discussion Study 1 aimed to explore whether 8- and 10-year-olds are aVected by a subtle variation of socially transmitted information, that is, by variations of the assertiveness with which the

Table 2 Mean conWdence judgments (1 D very unsure, 5 D very sure) in Study 1 as a function of age, level of assertiveness, and correctness of answer 8-year-olds

10-year-olds

High assertiveness Correct Incorrect

4.3 (0.7) 3.2 (1.0)

4.5 (0.3) 2.7 (0.9)

Low assertiveness Correct Incorrect

4.0 (1.0) 3.2 (1.1)

4.5 (0.4) 2.6 (0.7)

Note. Standard deviations are in parentheses.

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

237

interviewer confronted children with misleading information. With regard to recall accuracy, we found no eVect of the interviewer’s level of assertiveness for 10-year-olds. Regardless of the interviewer’s apparent conWdence about the correctness of her recollections, 10-year-olds strongly resisted false suggestions. Moreover, their level of accuracy (overall M D .74) was comparable to that in previous studies using the same video and questions without variations of assertiveness (e.g., Roebers & Schneider, 2002: M D .73; Schwarz et al., 2004: M D .79). In contrast, 8-year-olds’ recall was signiWcantly lower when the interviewer’s assertiveness was high. This result is in line with our expectations that younger children would be more vulnerable to strong social pressure in the interview situation than would older children (Bruck & Ceci, 1999). However, compared with the results of our previous studies in which the interviewer asked the same misleading questions without any comments regarding her assertiveness (e.g., Roebers & Schneider, 2000: M D .56; Schwarz et al., 2004: M D .48), high assertiveness did not have an additional signiWcant negative impact on children’s level of accuracy (M D .61). The unexpected result was that younger children’s recall was surprisingly accurate when the interviewer’s assertiveness was low. Here the typical age diVerences in recall in response to misleading questions vanished completely, with 8year-olds achieving recall accuracy equivalent to that of 10-year-olds. Thus, the interviewer’s expression of low assertiveness might have served as an icebreaker that helped younger children to withstand the social pressure exerted by misleading questions. A comparable “inoculation eVect” against 8-year-olds’ yielding to suggestions was also found in Schwarz and colleagues’ (2004) study whenever the confederate gave a correct answer to a misleading question before the child gave her or his own report. This clearly supports the interpretation that younger children’s low recall accuracy when confronted with biased questions is mainly a conformity phenomenon (Gee, Gregory, & Pipe, 1999). Considering children’s conWdence judgments, 10-year-olds were more accurate than 8year-olds. This is in line with previous Wndings concerning the development of metacognitive monitoring abilities (Roebers, 2002). There was no signiWcant main eVect and no interaction concerning the level of assertiveness for children’s metacognitive monitoring. However, the variation of assertiveness itself aVected metamemorial competencies. Children at both ages diVerentiated between their correct and incorrect answers. In contrast, prior studies have shown repeatedly that for misleading questions, 10-year-olds do not diVerentiate between correct and incorrect answers and 8-year-olds sometimes even show a tendency to be overconWdent; that is, the latter give somewhat higher conWdence judgments to incorrect answers than to correct answers (Roebers, 2002). For example, Roebers and Howie (2003, Study 1) found mean diVerences in conWdence judgments of ¡.3 and .1 for 8- and 10year-olds, respectively. Thus, although the current Study 1 lacked a strict control group, there is evidence that varying assertiveness can lead to patterns of metacognitive diVerentiation that have not been observed in studies where assertiveness was not manipulated. There are several possible reasons that may explain such a Wnding. From a cognitive point of view, the interviewer’s comments about her own conWdence may have promoted children’s metacognitive involvement, resulting in better metamemorial performance. In a way, the interviewer gave a social model to the children by verbalizing the manner in which she was thinking about her memories. From a social point of view, the interviewer’s confessions of her own uncertainty with some questions might have reduced the overall level of perceived social pressure. Furthermore, the interviewer’s statements qualifying each answer probably also inXuenced the general interview atmosphere by making it much more personal and friendly for the children. The assumption that children’s deWcient metacogni-

238

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

tive diVerentiation is a result not of their immature metacognitive skills but rather of the task demands is also supported by studies assessing conWdence judgments to unbiased questions: For an unbiased question format, even 8-year-olds typically achieve an appropriate diVerentiation between their correct and incorrect answers, whereas they do not do so for misleading questions (Roebers, 2002; Roebers & Howie, 2003). At Wrst blush, the lack of a control condition without any variation of assertiveness seems to be a weakness of Study 1. However, because our main foci were within-participants comparisons between the two levels of assertiveness, we did not see the necessity to include a control group. In addition, given the overwhelming and consistent empirical evidence on age diVerences when responding to misleading questions, the inclusion of a control group without variations in assertiveness did not warrant any signiWcant insights. In Study 2, we chose a diVerent experimental paradigm to assess the eVects of varying assertiveness on children’s recall accuracy and conWdence judgments. The variation of assertiveness was accomplished using a female confederate, who was interviewed together with the child and answered each question before the child answered. Furthermore, we used an unbiased, open-ended question format. Consequently, the confederate was the only source of both informative and normative social inXuence.

Study 2: Confederate’s answers as source of postevent information Method Design overview The design of the second study was a 2 (Age: 8 or 10 years) £ 2 (Condition: confederate or control) factorial design with two additional within-participants factors for the confederate condition: correctness of the confederate’s answer (correct or incorrect) and assertiveness of the confederate’s answer (high or low). Dependent measures were accuracy of event recall and conWdence judgments. Sample A total of 79 children (38 girls and 41 boys) took part in the experiment. As in Study 1, we recruited second and fourth graders from an elementary school in a suburban area in northern Bavaria. There were 40 8-year-olds (20 girls and 20 boys, mean age D 8 years 4 months, SD D 5 months), with 20 of these children being randomly assigned to the control condition (interview without confederate) and the other 20 being assigned to the confederate condition. The group of 10-year-olds consisted of 39 children (18 girls and 21 boys, mean age D 10 years 4 months, SD D 5 months), with 19 of these children being randomly assigned to the control condition and the other 20 being assigned to the confederate condition. Written parental consent was obtained for all children. Procedure The procedure was identical to that in Study 1 except that a confederate was introduced to the children in the experimental group. All children watched the same video with the same

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

239

instructions as in Study 1. One week later, an unfamiliar adult questioned the children individually about the contents of the video in a quiet room in the school. In the experimental condition, the interviewer’s confederate was introduced as a university student who had also watched the video and who needed to answer the same questions as the child. To save time, the interviewer would question both together. The interviewer established that the adult always should answer each question Wrst so that she would not get mixed up writing down the adult’s and child’s answers. When the child indicated that she or he understood the procedure, accuracy instructions were given and questioning started. When the questions were Wnished, the interviewer announced that she had another task for the adult and child that they had to do alone. She decided to start with the child and asked the confederate to wait outside. Thus, for assessing conWdence ratings, children in both conditions were alone with the interviewer. The procedures from Study 1 for obtaining conWdent judgments were used. After giving the conWdence ratings, children were thanked for their help and given a small gift. Materials The recall questionnaire in Study 2 consisted of 24 questions, all of which were asked in an unbiased, open-ended format. Of these 24 questions, 22 related to the same contents of the video as the 22 questions used in Study 1. The 2 additional questions had also been used in previous studies (Roebers & Fernandez, 2002; Roebers & Howie, 2003) and were included to achieve four equally sized question subgroups in the confederate condition, representing four diVerent levels of social inXuence situations: (a) 6 questions that the confederate answered correctly with high assertiveness, (b) 6 questions that the confederate answered incorrectly with high assertiveness, (c) 6 questions that the confederate answered correctly with low assertiveness, and (d) 6 questions that the confederate answered incorrectly with low assertiveness. In a previous study (Schwarz et al., 2004), these four groups of questions were comparably diYcult. In the case of high assertiveness, the confederate stressed high conWdence in her own answer (e.g., “I can remember this quite clearly. It was the grandfather who came to the kitchen!”), whereas in the case of low assertiveness, she expressed uncertainty in her own answer (e.g., “I am not at all sure about this. But perhaps they had suitcases in their hands.”). Questions with high versus low assertiveness and correct versus incorrect answers of the confederate were presented in mixed order so that every child was confronted with all four variations of social pressure. Children in the control condition answered the 24 unbiased questions in the same order as did the experimental group. All answers were written down on the protocol sheet and later were coded as correct, incorrect or “I do not know.” Results For each dependent variable, we Wrst report analysis for the experimental group, followed by planned comparisons with the control group. Recall Table 3 presents the mean recall accuracy scores as a function of age group, condition, and levels of assertiveness and correctness of the confederate’s answer (in the case of the confederate condition). Regarding the confederate condition, we conducted an ANOVA

240

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

Table 3 Mean recall accuracy scores in Study 2 as a function of age and experimental condition (correctness of confederate’s answer and level of assertiveness)

Confederate condition Confederate incorrect/high assertiveness Confederate incorrect/low assertiveness Confederate correct/low assertiveness Confederate correct/high assertiveness Control condition

8-year-olds

10-year-olds

.41 (.27) .48 (.21) .75 (.23) .91 (.16) .60 (.13)

.70 (.24) .54 (.22) .65 (.19) .87 (.14) .65 (.10)

Note. Standard deviations are in parentheses.

with level of assertiveness and correctness of the confederate’s answer as within-participants factors and age as a between-participants factor. The analysis revealed a main eVect of correctness of the confederate’s answer, F (1, 38) D 45.74, 2 D .55, a main eVect of assertiveness, F (1, 38) D 12.74, 2 D .25, a two-way interaction between correctness and assertiveness, F (1, 38) D 6.30, 2 D .14, a two-way interaction between age and assertiveness, F (1, 38) D 4.87, 2 D .11, and a two-way interaction between age and correctness of the confederate’s answer, F (1, 38) D 10.87, 2 D .22. The main eVects showed that recall was more accurate when the confederate had answered correctly (M D .79) than when she had not answered correctly (M D .53) and that recall was more accurate when the confederate indicated her answer with high assertiveness (M D .72) than when she indicated her answer with low assertiveness (M D .61). The interaction between assertiveness and correctness of the confederate’s answer revealed that the eVect of the level of assertiveness was signiWcant only for the confederate’s correct answers (high assertiveness mean D .89, low assertiveness mean D .70), t (39) D 5.47, and not for her incorrect answers (high assertiveness mean D .56, low assertiveness mean D .51), t (39) < 1. Furthermore, the analysis revealed age-speciWc eVects for the level of assertiveness as well as for the correctness of the confederate’s answers. Although 10-year-olds were less accurate after low assertiveness (M D .60) than after high assertiveness (M D .78), t (19) D 4.28, 8-year-olds were comparably accurate (high assertiveness mean D .66, low assertiveness mean D .62), t (19) < 1. The correctness of the confederate’s answers, in contrast, aVected both 8- and 10-year-olds’ accuracy, ts(19) 7 2.58, but the eVect was larger for 8-year-olds (Ms D .83 and .44 for 8-year-olds after the confederate’s correct and incorrect answers, respectively, whereas corresponding Ms D .72 and .66 for 10-year-olds). Then we conducted an ANOVA with age as a between-participants factor for the control condition only. The results revealed no signiWcant diVerence between 8- and 10-yearolds’ recall. Subsequent t tests for independent samples were conducted to compare the eVects of assertiveness and correctness of the confederate’s answer on recall accuracy in the control condition. The analysis revealed diVerent patterns of results for the two age groups. There was a signiWcant eVect of condition in the 8-year-olds as a function of the confederate’s correct and incorrect answers, with 8-year-olds in the confederate condition being more accurate than those in the control group when the confederate gave correct answers, high assertiveness t (38) D 4.82, low assertiveness t (38) D 3.41. When the confederate gave incorrect answers, 8-year-olds’ recall was signiWcantly less accurate than that of children in the control group, high assertiveness t (37) D 2.56, low assertiveness t (38) D 2.41.

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

241

In contrast, for 10-year-olds, the only experimental condition in which recall diVered from the control condition was when the confederate answered correctly with high assertiveness, t (32) D 4.26. In all other experimental conditions, 10-year-olds’ recall did not diVer from that of control children, ts < 1. ConWdence judgments Table 4 shows the mean conWdence ratings as a function of age and condition as well as for the confederate condition only as a function of assertiveness and correctness of the confederate’s answers. To explore the inXuence of the confederate’s assertiveness and the correctness of her answers on children’s ability to diVerentiate between their own correct and incorrect answers, we conducted an ANOVA with correctness of the confederate’s answers, level of assertiveness, and correctness of the children’s answers as within-participants factors and with age as a between-participants factor for the confederate condition only. Results revealed a main eVect of age, F (1, 11) D 10.29, 2 D .48, with higher conWdence judgments for the 8-year-olds (M D 4.1) than for the 10-year-olds (M D 3.8), a main eVect of correctness of the children’s answers, F (1, 11) D 34.69, 2 D .76, and a two-way interaction between correctness of the children’s answers and age, F (1, 11) D 12.58, 2 D .53. Both 8- and 10-year-olds were more conWdent after correct answers than after incorrect answers, ts (19) 7 7.1. Yet the diVerence between conWdence ratings on correct answers and those on incorrect answers was signiWcantly smaller for 8-year-olds than for 10-year-olds, t (31) D 2.99, indicating better metacognitive diVerentiation in the older children. To examine the possibility that the confederate had a general unspeciWc eVect on children’s conWdence judgments, we also compared the levels of conWdence in the experimental and control conditions. For children in the experimental condition, mean levels of conWdence that collapsed across the various manipulations involving the confederate were computed. An ANOVA with correctness of the children’s answers as a within-participants factor and with age and condition as between-participants factors revealed a main eVect of age, F (1, 75) D 7.17, 2 D .09, a main eVect of correctness of the children’s answers, F (1, 75) D 154.03, 2 D .67, and a two-way interaction between age and correctness of children’s answers, F (1, 75) D 9.05, 2 D .11. In addition, there was a two-way interaction between correctness of children’s answers and condition that approached signiWcance, F (1, 75) D 3.66, p D .06, 2 D .05, and a three-way interaction among correctness of Table 4 Mean conWdence judgments (1 D very unsure, 5 D very sure) in Study 2 as a function of age, condition, and correctness of answer 8-year-olds

Confederate condition Confederate incorrect/high assertiveness Confederate incorrect/low assertiveness Confederate correct/low assertiveness Confederate correct/high assertiveness Control condition Note. Standard deviations are in parentheses.

10-year-olds

Correct

Incorrect

Correct

Incorrect

4.3 (0.8) 4.0 (0.7) 4.6 (0.6) 4.6 (0.4) 4.5 (0.4)

4.0 (0.7) 3.6 (0.8) 3.8 (0.9) 4.3 (0.4) 4.0 (0.7)

4.2 (0.7) 4.0 (0.8) 4.4 (0.7) 4.4 (0.6) 4.4 (0.3)

3.6 (1.2) 3.2 (1.1) 3.0 (0.8) 3.0 (1.5) 3.7 (0.6)

242

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

children’s answers, age, and condition that also approached signiWcance, F (1, 75) D 3.15, p D .08, 2 D .04. Again, 8- and 10-year-olds were more conWdent after correct answers than after incorrect answers, ts (39) 7 8.5, and metacognitive diVerentiation increased with age, as indicated by a signiWcantly smaller diVerence between conWdence ratings for correct and incorrect answers in the 8-year-olds than in the 10-year-olds, t (64) D 2.95. Further analyses revealed that the age eVect for conWdence judgments after incorrect answers was signiWcant only in the experimental condition, with 10-year-olds giving lower conWdence judgments than 8-year-olds, t (38) D 3.55. There was no similar age eVect for conWdence judgments after incorrect answers in the control condition, and there was no age eVect for conWdence judgments after correct answers in either condition, ts (37) < 1. Discussion The aim of Study 2 was to explore age diVerences in the eVects of varied assertiveness on children’s recall. As expected, the results revealed a greater sensitivity of the younger children to the content of postevent information. When the confederate responded incorrectly, 8-year-olds in the confederate condition recalled less than did those in the control group. For the confederate’s correct answers, the eVects were reversed; that is, 8-year-olds in the confederate condition recalled more than did those in the control condition. The variation of the confederate’s assertiveness, however, had no global eVect on the 8year-olds’ memory performance: Overall, they achieved a comparable mean accuracy for high and low assertiveness. Yet the signiWcant interaction between level of assertiveness and correctness of answers revealed for the confederate’s correct answers a signiWcant inXuence of assertiveness, with high assertiveness leading to better recall accuracy than low assertiveness. In contrast, the confederate’s incorrect answers, independent of the level of assertiveness, led to decreases in recall performance that fell below chance level. The Wndings in this age group are consistent with those in previous studies demonstrating that young children can be strongly inXuenced by the content of false postevent information, especially when it is communicated by an unknown adult (Roebers, Howie, & Beuscher, in press; Schwarz et al., 2004). The 10-year-olds were signiWcantly less aVected by the content of the confederate’s answers. Compared with those in the control condition, their performance improved only for questions that the confederate had answered correctly and stated with high assertiveness. Regarding the overall eVects of assertiveness, however, the 10-year-olds generally achieved better recall accuracy when the confederate stated her answer with high assertiveness. This Wnding was expected for the confederate’s correct answers and shows an adequate consideration of the credibility of the confederate’s answers with an avoidance of uncertain information in that age group. Yet for the confederate’s incorrect answers, this “positive” eVect of high assertiveness on 10-year-olds’ recall accuracy was unexpected. We could speculate that the feeling of being strongly pressured to give an incorrect answer resulted in a kind of reactance, opposing answering behavior, in that age group. This interpretation would be in line with interpretations in previous studies with adult participants showing that blatantly false suggestions can have a positive eVect on the accuracy of witnesses’ reports (Loftus, 1979). However, reactance more typically is observed in adults and not necessarily in children (Roebers, Rieber, & Schneider, 1995). Another possible explanation for the higher memory

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

243

accuracy for misleading information presented with high assertiveness could be a challenging eVect of the confederate’s expression of uncertainty. The confederate’s low assertiveness may have decreased children’s conWdence in their own recall and encouraged them to reXect on the possible answers and perhaps to conform more easily. In sum, the 10-yearolds showed a more adult-like pattern regarding the eVects of social inXuence; they were able to disregard socially transmitted postevent information when it did not match their own recollections, but they were able to use informative social inXuence to the beneWt of their recall when their own recollections failed. Informative and normative social inXuences may mediate the eVects of the confederate’s incorrect answers. Considering the inXuence of the confederate’s correct answers (a source of correct information), both informative social inXuences and recognition eVects may play a role (i.e., the child recognizes the information presented by the confederate as stemming from the video). Recognition eVects can be expected whenever the confederate answers correctly, independent of the level of assertiveness. However, the very speciWc beneWt to older children’s recall only when a correct answer was given with high assertiveness indicates that recognition eVects cannot be considered as the main mediator, at least for the older children. Rather, an adequate consideration of informative social inXuence may be responsible for the observed recall improvement; children adopted externally provided information only when the confederate stressed remembering it very well, whereas they were able to ignore the confederate’s answers as soon as she stressed her own uncertainty. Because this ability to diVerentiate was not as apparent in the 8year-olds, our results point to important developments, namely that the ability to make Wne-tuned adjustments in one’s answering behavior on the basis of social inXuence factors, and at the same time being able to adequately consider informational inXuences, seems to develop at a very fast pace during the late elementary school years and has numerous implications for children’s everyday social lives. With respect to the conWdence judgments, older children again proved to have more sophisticated metacognitive skills than younger children. The 10-year-olds diVerentiated appropriately between their correct and incorrect answers in the confederate and control conditions. The 8-year-olds, in contrast, were not able to discriminate between correct and incorrect answers when the confederate’s assertiveness was high; independently of the correspondence between these children’s own answers and the confederate’s answers, the confederate’s high assertiveness produced a stronger feeling of certainty in the 8-year-olds when they gave their subsequent conWdence judgments. Only for the 10-year-olds was there better metacognitive diVerentiation in the confederate condition than in the control condition. The statements and the variation of the confederate’s assertiveness possibly mediated this beneWcial eVect by activating older children’s own metacognitive monitoring competencies. Overall, the results regarding recall accuracy as well as metacognitive monitoring suggest again that 8-year-olds are more globally inXuenced by social factors, whereas 10year-olds can adequately diVerentiate and decide whether sources of social inXuence aVect them or not.

General discussion The current studies are among very few studies investigating the eVects of social inXuence on children’s recall accuracy and on metacognitive monitoring. We explored age-

244

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

related diVerences in children’s competencies to simultaneously consider diVerent informational cues and to respond sensitively to subtle variations of social inXuence. Slight variations of social inXuence were implemented as variations of assertiveness, either of the interviewer or of the confederate, inducing variations in credibility and conformity pressure by which the postevent information was provided. Overall, the current studies document signiWcant age diVerences in children’s responses to diVerent sources of social inXuence. Compared with 8-year-olds, 10-year-olds were signiWcantly less aVected by the interviewer confronting them with misleading questions (Study 1) as well as by the content of the answers of a mock coparticipant (Study 2). These Wndings are generally consistent with the social psychology literature concerning the development of conformity toward authorities (Berenda, 1950) as well as eyewitness studies of children’s suggestibility (Bruck & Ceci, 1999). With respect to children’s ability to make Wne-tuned adjustments to their answering behavior when confronted with diVerent levels of assertiveness of a social inXuence source (high vs. low assertiveness), the results showed that children at both ages respond in a sensible manner to this additional information. In our studies, 8-year-olds proved to be able to process their own memory recall, the presented postevent information, and the variation of assertiveness at the same time. Although this age group somehow solved this demanding task, there was nevertheless an important age-related improvement regarding the eVects of the two forms of normative and informative social inXuence. The 8-year-olds were mostly aVected by misleading postevent information when provided with high assertiveness, pointing to the importance of normative social inXuence (Deutsch & Gerard, 1955). The 10-year-olds, in contrast, were not aVected by normative social inXuence; when misleading postevent information was provided with high assertiveness, it did not impair their recall. In Study 2, this even produced an answering behavior that resembled a kind of reactance eVect, with better memory performance after high assertiveness than after low assertiveness. However, because there was no comparable result in Study 1, this interpretation of a reactance remains speculative. In sum, regarding children’s ability to resist conformity pressure, a signiWcant development seems to take place during the elementary school years. The Wnding that normative social inXuence plays a signiWcant role for younger children is also in line with Wndings in other research, for example, the Wnding that 4-year-olds are more susceptible to misleading questions when interviewed by a stranger as opposed to their mother (Goodman et al., 1995; Jackson & Crockenberg, 1998). In contrast, considering informative social inXuence, children of both age groups signiWcantly beneWted from correct postevent information in terms of their recall accuracy. However, there again were hints of age-related diVerences in acting on these inXuences appropriately; older children avoided adopting information that was associated with low conWdence, whereas younger children had problems in taking the provided information and the level of assertiveness into account simultaneously and thus often also gave in to uncertain suggestions. The Wnding that informative social inXuence has a strong and rather age-independent eVect, whereas normative social inXuence is especially important for younger children, is supported by previous research results as well. Roebers and colleagues (in press) also suggested a greater impact of informative social inXuence than of normative social inXuence. In their Wrst study, children in one group were asked to whisper their answers to misleading questions into a discreet teddy bear’s ear while the other group was questioned in a normal face-to-face interview situation. The

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

245

results showed that the private answering option did not decrease children’s suggestibility. Thus, the acceptance of postevent information stemming from suggestive questions appears to derive mainly from children’s susceptibility to informational social inXuence sources because children use the provided information to Wll their own memory gaps and not so much due to their reluctance to contradict the adult interviewer. Overall, our results, as well as those of the other research, suggest that older children are more sophisticated in their ability to deal with social inXuence factors, especially regarding exerted conformity pressure. For younger children, in contrast, normative social inXuence plays an important additional role, and they often fail to act appropriately on unreliable informative social inXuence. Yet it is also interesting that the Wndings regarding age-dependent eVects of assertiveness between the two studies diVered from each other. For 10-year-olds, level of assertiveness aVected recall in Study 2 but not in Study 1. This corresponds to the results of Schwarz and colleagues (2004) study in which inXuence exerted by suggestive questions (as in Study 1) had a small eVect on 10-year-olds’ answering behavior, whereas inXuence exerted by the answers of a confederate (as in Study 2) strongly inXuenced their recall performance. For 8-year-olds, in contrast, both sources of social inXuence appeared to signiWcantly aVect their recall performance. Thus, the source of social inXuence seems to play an important role when considering age-dependent eVects on children’s recall. The results of children’s conWdence judgments replicated earlier Wndings of developmental changes in metacognitive monitoring skills in children between 8 and 10 years of age (Roebers, 2002; Roebers & Howie, 2003). Interestingly, the variation of assertiveness also aVected the quality of children’s conWdence judgments. In spite of a misleading interview, 8- and 10-year-olds showed good metacognitive diVerentiation between their correct and incorrect answers (Study 1), and 10-year-olds in the experimental condition (varying assertiveness) outperformed those in the control condition (no varying assertiveness) (Study 2). What are the possible mechanisms by which social inXuence in the interview situation aVects conWdence judgments? For example, is the previous Wnding of missing metacognitive diVerentiation when answering misleading questions the result of cognitive processes undermining the metacognitive judgments? Or, is it a more socially inXuenced phenomenon in that children justify their yielding to suggestions by expressing high conWdence? The results of Roebers’ and colleagues (in press) second study seem to speak for this Wrst reason. In that study, misleading questions had the same deleterious eVect on metacognitive monitoring when children were allowed to give their conWdence judgments privately as when they gave them while facing the interviewer. Yet because children’s yielding to suggestions did not decrease in private reporting, their experimental manipulation was not eVective in reducing the perceived social pressure. In the current Study 1, in contrast, we found lowered suggestibility as well as improved metacognitive diVerentiation when the level of assertiveness was low and thus social pressure was reduced. Our results seem to argue for a more socially driven phenomenon. A reduction of the usually exerted normative pressure through misleading questions by presenting some of the questions with low assertiveness (Study 1) seemed to promote children’s metacognitive performance, leading to unexpected good metacognitive diVerentiation, even for the 8-year-olds. In Study 2, in contrast, the coparticipant’s statements that she felt very certain with an actually incorrect answer increased children’s conWdence in their own incorrect answers to a level that was even higher than the

246

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

levels reported in previous studies. In this case, the informative social inXuence of another person’s certainty resulted in a reduction of children’s metacognitive performance. In summary, the results of the current studies show that by 8 years of age, children can simultaneously consider diVerent informational cues and respond adequately to subtle variations of social inXuence. Within these abilities, however, our results point to marked developmental improvements concerning children’s ability to react appropriately to social inXuences. Although children of all age groups can be strongly aVected by informative social inXuence, older children appear to be able to disregard postevent information whenever it is not trustworthy. Moreover, children’s ability to resist conformity pressure increases signiWcantly with age, leading to a more adequate consideration of the social demand characteristics of the interview situation. Overall, however, informative social inXuence seems to play a more important role than does normative inXuence. More generally, our Wndings indicate that social factors should not be underestimated in their detrimental eVects on children’s recall performance. Social inXuence, as realized in the current studies, aVected not only children’s memory reports but also their metacognitive monitoring performance.

Acknowledgments We thank Wolfgang Schneider, who made this research possible, and Barbara Bjorklund for her valuable comments and linguistic help on an earlier version of this article. We are also grateful to research assistants Agnes Renner, Natalia Maltseva, and Nicole Museiko for their help in conducting the interviews. Thanks are especially due to all of the principals, teachers, parents, and children for their cooperation and participation.

References Asch, S.E., 1956. Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs, 70(whole No. 416). Berenda, R. W. (1950). The inXuence of the group on the judgments of children. New York: Kings Crown. Berndt, T. J. (1979). Developmental changes in conformity to peers and parents. Developmental Psychology, 15, 608–616. Beuscher, E., & Roebers, C. M. (2005). Does a warning help children to more accurately remember an event, to resist misleading questions, and to identify unanswerable questions? Experimental Psychology, 52, 232–241. Bjorklund, D. F., Cassel, W. S., Bjorklund, B. R., Brown, R. D., Park, C. L., Ernst, K., & Owen, F. A. (2000). Social demand characteristics in children’s and adults’ eyewitness memory and suggestibility: The eVect of diVerent interviewers on free recall and recognition. Applied Cognitive Psychology, 14, 421–433. Brown, B. B., Clasen, D. R., & Eicher, S. A. (1986). Perceptions of peer pressure, peer conformity dispositions, and self-reported behavior among adolescents. Developmental Psychology, 22, 521–530. Bruck, M., & Ceci, S. J. (1999). The suggestibility of children’s memory. Annual Reviews of Psychology, 50, 419–439. Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social inXuences upon individual judgment. Journal of Abnormal Psychology, 51, 629–636. Gabbert, F., Memon, A., & Allan, K. (2004). Say it to my face: Examining the eVects of socially encountered misinformation. Legal & Criminological Psychology, 9, 215–227. Gee, S., Gregory, M., & Pipe, M-E. (1999). “What colour is your pet dinosaur?” The impact of pre-interview training and question type on children’s answers. Legal & Criminological Psychology, 4, 111–128. Goodman, G. S., Sharma, A., Thomas, S. F., & Considine, M. G. (1995). Mother knows best: EVects of relation status and interviewer bias on children’s memory. Journal of Experimental Child Psychology, 60, 195–228.

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

247

HoVman, H. G., Granhag, P. A., Kwong See, S. T., & Loftus, E. F. (2001). Social inXuence on reality-monitoring decisions. Memory & Cognition, 29, 394–404. Jackson, S., & Crockenberg, S. (1998). A comparison of suggestibility in 4-year-old girls in response to parental or stranger misinformation. Journal of Applied Psychology, 19, 527–542. Koriat, A., Goldsmith, M., & Pansky, A. (2000). Toward a psychology of memory accuracy. Annual Review of Psychology, 51, 481–537. Lampinen, J. M., & Smith, V. L. (1995). The incredible (and sometimes incredulous) child witness: Child eyewitnesses’ sensitivity to source credibility cues. Journal of Applied Psychology, 80, 621–627. Loftus, E. F. (1979). Retrieving information from memory. In E. F. Loftus (Ed.), Eyewitness testimony (pp. 88– 109). Cambridge, MA: Harvard University Press. Luus, C. A. E., & Wells, G. L. (1994). The malleability of eyewitness conWdence: Co-witness and perseverance eVects. Journal of Applied Psychology, 79, 714–723. Moises, K., de Carvalho, F., & Masamichi, Y. (2001). The eVects of social cues on conWdence judgments mediated by knowledge and regulation of cognition. Journal of Experimental Education, 69, 325–343. Ornstein, P. A., Shapiro, L. R., Clubb, P. A., Follmer, A., & Baker-Ward, L. (1997). The inXuence of prior knowledge on children’s memory for salient medical experiences. In N. Stein, P. A. Ornstein, B. Tversky, & C. J. Brainerd (Eds.), Memory for everyday and emotional events (pp. 83–111). Mahwah, NJ: Lawrence Erlbaum. Penrod, S., & Cutler, B. (1995). Witness conWdence and witness accuracy: Assessing their forensic relation. Psychology, Public Policy and Law, 1, 817–845. Perfect, T. J. (2002). When does eyewitness conWdence predict performance. In T. J. Perfect & B. L. Schwartz (Eds.), Applied metacognition (pp. 95–120). Cambridge, UK: Cambridge University Press. Roberts, K. P. (2000). An overview of theory and research on children’s source monitoring. In M. Blades & K. P. Roberts (Eds.), Children’s source monitoring (pp. 11–57). Mahwah, NJ: Lawrence Erlbaum. Robinson, E. J. (2000). Belief and disbelief: Children’s assessments of the reliability of sources of knowledge about the world. In M. Blades & K. P. Roberts (Eds.), Children’s source monitoring (pp. 59–83). Mahwah, NJ: Lawrence Erlbaum. Roebers, C. M. (2002). ConWdence judgments in children’s and adults’ event recall and suggestibility. Developmental Psychology, 38, 1052–1067. Roebers, C. M., & Fernandez, O. (2002). The eVects of accuracy motivation on children’s and adults’ event recall, suggestibility, and their answers to unanswerable questions. Journal of Cognition and Development, 3, 415–443. Roebers, C. M., & Howie, P. (2003). ConWdence judgments in event recall: Developmental progression in the impact of question format. Journal of Experimental Child Psychology, 85, 352–371. Roebers, C.M., Howie, P., Beuscher, E. (in press). Can private reports enhance children’s event recall, lower their suggestibility, and foster their metacognitive monitoring compared to face-to-face interviews? Computers and Human Behavior. Roebers, C. M., Rieber, F., & Schneider, W. (1995). Zeugenaussagen und Suggestibilität als Funktion der Erinnerungsgenauigkeit: Eine entwicklungspsychologische Studie [Eyewitness testimony and suggestibility as a function of recall accuracy: A developmental study]. Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie, 27, 210–225. Roebers, C. M., & Schneider, W. (2000). The impact of misleading questions on eyewitness memory in children and adults. Applied Cognitive Psychology, 14, 509–526. Roebers, C. M., & Schneider, W. (2002). Stability and consistency of children’s event recall. Cognitive Development, 17, 1085–1103. Roebers, C. M., Schwarz, S., & Neumann, R. (2005). Social inXuence and children’s event recall and suggestibility. European Journal of Developmental Psychology, 2, 47–69. Saywitz, K. J. (2002). Developmental underpinnings of children’s testimony. In H. L. Westcott, G. D. Davies, & R. H. C. Bull (Eds.), Children’s testimony: A handbook of psychological research and forensic practice (pp. 3–20). Chichester, UK: Wiley. Schwarz, S., Roebers, C. M., & Schneider, W. (2004). Entwicklungsveränderungen in Konformität und in kognitiven Folgen sozialer BeeinXussung [Development of conformity and cognitive eVects of social inXuence]. Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie, 36, 173–181. Smith, V. L., & Ellsworth, P. C. (1987). The social psychology of eyewitness accuracy: Misleading questions and communicator expertise. Journal of Applied Psychology, 72, 294–300. Walther, E., Bless, H., Strack, F., Rackstraw, P., Wagner, D., & Werth, L. (2002). Conformity eVects in memory as a function of group size, dissenters, and uncertainty. Applied Cognitive Psychology, 16, 793–810.

248

S. Schwarz, C.M. Roebers / Journal of Experimental Child Psychology 94 (2006) 229–248

Warren, A., Hulse-Trotter, K., & Tubbs, E. C. (1991). Inducing resistance to suggestibility in children. Law and Human Behavior, 15, 273–285. Waterman, A. H., Blades, M., & Spencer, C. (2004). Indicating when you do not know the answer: The eVect of question format and interviewer knowledge on children’s don’t know responses. British Journal of Developmental Psychology, 22, 335–348. Wright, D. B., Self, G., & Justice, C. (2000). Memory conformity: Exploring misinformation eVects when presented by another person. British Journal of Psychology, 91, 189–202.