Group- and Individual-Level Information Affects Children’s Playmate Choice

• In Event: The Statistics of Group Cognition: New Evidence for Children’s Statistical Learning of Social Groups

Abstract

Having high quality and stable friendships is greatly beneficial for children (Berndt, Hawkins, & Jiao, 1999; Ladd, Kochenderfer, & Coleman, 1996). However, less is known about how children form friendships. Here, we focus on the first step of this process – how children choose their playmates. Past research has shown that children prefer individuals based on their social group memberships (race and gender: Shutts, Pemberton Roben, & Spelke, 2013; linguistic groups: Kinzler, Dupoux, & Spelke, 2007). Furthermore, according to the adaptive sampling strategy (Denrell & Le Mens, 2012), children would approach individuals whom they had positive experiences with, and avoid individuals whom they had negative experiences with. In the present studies, we examined how 5- and 6-year-old children use statistical information – both information about a group and the past behaviors of an individual – in their playmate choices.

Children imagined that they had just transferred to a new classroom. In Experiment 1 (n = 24), they learned that a majority of children in the new classroom were nice (16 nice and 4 mean children; Majority Nice classroom). In Experiment 2 (n = 24), a majority of children in the classroom were mean (16 mean and 4 nice children; Majority Mean classroom). Next, a target child from the classroom showed either 5 nice behaviors and 1 mean behavior (target nice trials) or 5 mean behaviors and 1 nice behavior toward the participant (target mean trials). Then, participants chose a playmate from two options: the target child or a new child randomly selected from the classroom. Experiment 3 (n = 48) was a replication of Experiments 1 and 2, except that children were randomly assigned to the Majority Nice (MN) classroom condition or the Majority Mean (MM) classroom condition, and the target child showed behaviors toward another child from the classroom, instead of toward the participant.

The percentages of children who chose the target child as a playmate are shown in Figure 1 (Expts 1 & 2) and Figure 2 (Expt 3). Whether the target child showed behaviors toward participants themselves (Experiment 1 and 2) or toward another child from the classroom (Experiment 3) did not affect children’s playmate choices. Therefore, we combined the data from all experiments in our analysis. Comparison of generalized linear mixed models (GLMMs) revealed an effect of trial type and an effect of classroom type on playmate choice. Children were more likely to choose the target child as a playmate in target nice trials than in target mean trials (p < .001). That is, regardless of the classroom composition, a child who has shown nice behaviors is preferred over a child who has shown mean behaviors. Moreover, the base-rate information about the classroom affected children’s playmate choices. A new, random child from the MN classroom is more likely to be nice than a new, random child from the MM classroom. Indeed, children in the MN classroom were more likely to choose the new, random child as a playmate than children in the MM classroom (p = .02).

Authors

• Rongzhi Liu, University of California - Berkeley
  Presenting Author

• Gil Diesendruck, Bar Ilan University
  Non-Presenting Author

• Fei Xu, University of California - Berkeley
  Non-Presenting Author