Addressing possible confounds

In order to examine whether participants may have had a distinct preference (i.e., bias) for one of the adults in the videos, or for one of the adult–child pairs, statistical analyses were conducted comparing the scores for each dependent variable between the two adult–child pairs, regardless of the coordination or non-coordination condition. Wilcoxon ranks sum tests for independent samples were conducted for the empathy, closeness, similarity, and quality of interaction measures, and a Chi-Square test for independent samples was performed for the social connection section.

To examine whether the age or gender of participants affected the results, the participants were divided according to age (above vs. below 10.5 years) or gender (boys vs. girls). Here too, Wilcoxon ranks sum tests for independent samples were conducted for the empathy, closeness, similarity, and quality of interaction measures, and a Chi-Square test for independent samples was performed for the social connection section.

In addition, to make sure that children are not responding based on demand characteristics (i.e., how they think they are expected to respond, rather than how they truly feel), we ran a separate control experiment, with a small sample of children (N = 15; average age 10.7 years, SD = 0.9; 4 boys and 11 girls) who did not participate in the main study. After watching the two coordinated and non-coordinated video excerpts, they were asked the following question regarding the two adults: “Who do you think is smarter?” The coordination condition is not expected to influence the children’s choice of who is smarter. However, if a bias is found in the children’s responses, then this could indicate a role for demand characteristics.