Table 4.

Study 3: IAT D-scores regressed onto a participant’s prediction for their own score (left), the same participant’s prediction for the average participant and their own score simultaneously (2^nd column), IAT scores regressed onto explicit thermometer ratings (3^rd column), and both self-predictions and explicit ratings simultaneously (right).

Parameters (DV: IAT D-scores)	Prediction model estimates	Self-vs-aver. model estimates	Imp.-exp. model estimates	Sim. reg. model estimates
Fixed effects
IAT score predictions self	.59***	.34***		.58***
self predictions × order	−.02	−.04		.07
IAT score predictions other		.34***
other predictions × order		−.00
Explicit therm. ratings			.27***	.03
Exp. Therm. rating × order			.07	.07
Random effect variances
IAT score predictions self	.028	.009		.019
IAT score predictions other		.033
Explicit therm. ratings			.111**	.077**
Residuals	.505***	.452***	.655***	.451***
Goodness of fit
−2 log likelihood	1313.49	1268.38	1501.44	1306.61

^**p<.01

^***p<.001

All level-1 variables are standardized for each individual participant before they are entered in the analysis. “Order” represents a level-2 (between-subjects) condition assignment, one half predicted their own scores first (assigned code −1), another half predicted the score of the average participant first (coded 1).