Skip to main content
. Author manuscript; available in PMC: 2022 Apr 1.
Published in final edited form as: J Exp Child Psychol. 2019 Jan 2;180:39–54. doi: 10.1016/j.jecp.2018.11.019

Fig. 3.

Fig. 3.

Modeling social influence. (A) BICs for model comparison. The relative difference in BIC values for each model compared with the model with the lowest overall BIC value is shown. The Bayes factor for comparing the best model (lowest BIC) and second-best model is 6494, which indicates that the best fitting model is very strongly favored over all other models tested. Note that this model is also the winning model if we performed these comparisons separately at the level of age groups. (B) Parameter estimates for the prior + bonus dual RL model. Median estimates presented separately for the three age groups and the two learning rates (αgain and αloss), exploration (θ), prior (ρ), and bonus (β) are shown. Error bars represent standard errors of the mean. Recovery analyses further confirmed these parameter estimates (see online supplementary material and Fig. S2). (C) Age differences in the parameter estimates of the prior + bonus dual RL model. Age represents the linear trend, Age2 the adolescent peak, and Age3 the adolescent emerging trend. *p < .05; **p < .01.