Table 3.
The logistic regressions modelling whether social learners chose %. Predictors included (a) the centred number of demonstrators who chose % in their final period, (b) each combination of the similarity and reliability information, minus the omitted category of reliably incorrect–similar signals and (c) the interactions between each of these dummies and the centred proportion of demonstrators who chose %. We centred the proportion so that any block where 3/6 demonstrators chose % became the omitted category of the regression and thus were reflected in the intercept. The robust standard errors given in parentheses were clustered on the social learner to reflect the multiple observations gathered per learner. See Appendix 7 for the regressions with control predictors. As the only significant control predictor was an increased likelihood to choose % as the blocks progressed during the game against nature only, the models reported below just focus on the social information of interest. We also give the 95% confidence interval lower and upper bounds for each estimate.
| Parameter | Estimate (game against nature) | Estimate (coordination game) |
|---|---|---|
| Intercept | 0.337 ** (0.12) 95% CI [0.10, 0.57] |
0.308 * (0.120) 95% CI [0.07, 0.54] |
| Centred proportion of demonstrators choosing % | −1.493 * (0.581) 95% CI [−2.63, −0.36] |
−1.400 ** (0.492) 95% CI [−2.36, −0.44] |
|
Reliably incorrect–different dummy [signal indicates different and is correct with 0.1 probability] |
−0.177 (0.145) 95% CI [−0.46, −0.11] |
−0.112 (0.175) 95% CI [−0.45, 0.23] |
|
Uninformative–same dummy [signal indicates same and is correct with 0.5 probability] |
−0.251 (0.143) 95% CI [−0.53, −0.03] |
−0.007 (0.154) 95% CI [−0.31, 0.29] |
|
Uninformative–different dummy [signal indicates different and is correct with 0.5 probability] |
−0.274. (0.158) 95% CI [−0.58, −0.03] |
−0.065 (0.154) 95% CI [−0.37, 0.24] |
|
Reliably correct–same dummy [signal indicates same and is correct with 0.9 probability] |
0.006 (0.152) 95% CI [−0.29, 0.30] |
−0.089 (0.197) 95% CI [−0.47, −0.30] |
|
Reliably correct–different dummy [signal indicates different and is correct with 0.9 probability] |
−0.034 (0.137) 95% CI [−0.30, 0.14] |
−0.025 (0.168) 95% CI [−0.35, 0.30] |
| Centred proportion of demonstrators choosing % × reliably incorrect–different dummy | 2.371 ** (0.750) 95% CI [0.90, 3.84] |
2.911 *** (0.751) 95% CI [1.44, 4.38] |
| Centred proportion of demonstrators choosing % × uninformative–same dummy | 4.161 *** (0.802) 95% CI [2.59, 5.73] |
4.242 *** (0.752) 95% CI [2.77, 5.71] |
| Centred proportion of demonstrators choosing % × uninformative–different dummy | 1.785 * (0.695) 95% CI [0.43, 3.14] |
1.630 * (0.677) 95% CI [0.31, 2.95] |
| Centred proportion of demonstrators choosing % × reliably correct–same dummy | 5.714 *** (0.992) 95% CI [3.77, 7.65] |
6.700 *** (0.989) 95% CI [4.77, 8.63] |
| Centred proportion of demonstrators choosing % × reliably correct–different dummy | −0.438 (0.640) 95% CI [−1.69, −0.81] |
−0.576 (0.642) 95% CI [−1.83, 0.68] |
Asterisks denote the level of significance of our p-values, with the following key: *** p < 0.001; ** p < 0.01; * p < 0.05; trend, p = 0.05–0.10 significance.