Table 5.

Mean squared error of predicted and observed popularity for the three social models we compared for each of the conditions (C1–C5) in Experiment 2

	Mean Squared Error
Model	C1	C2	C3	C4	C5
Naive copying	0.0433	0.0424	0.0371	0.1078	0.1030
Social sampling (matching)	0.0542	0.1703	0.0825	0.1069	0.2196
Social sampling (Bayesian)	0.0366	0.0333	0.1188	0.0969	0.0960

The best performing model in each column is shown in bold.