Table 2.

Unstandardized estimates (SE in brackets) for the autoregressive and cross-lagged effects in studies 2a and 2b

Model	Study 2a		Study 2b
	β (SE)	P	β (SE)	P
Model 1
Level 1
contacttime1 → contacttime2	0.583 (0.031)	<0.001	0.624 (0.039)	<0.001
prejudicetime1 → prejudicetime2	0.611 (0.024)	<0.001	0.677 (0.027)	<0.001
contacttime1 → prejudicetime2	−0.055 (0.029)	0.059	−0.002 (0.004)	0.547
prejudicetime1 → contacttime2	−0.052 (0.026)	0.044	−0.560 (0.187)	0.003
Level 2
contacttime1 → contacttime2	0.910 (0.101)	<0.001	0.984 (0.169)	<0.001
prejudicetime1 → prejudicetime2	0.439 (0.173)	0.011	0.755 (0.209)	<0.001
contacttime1 → prejudicetime2	−0.298 (0.078)	<0.001	−0.039 (0.015)	0.009
prejudicetime1 → contacttime2	−0.101 (0.206)	0.624	2.848 (2.107)	0.177
Contextual effect	−0.243 (0.085)	0.004	−0.037 (0.016)	0.024
Effect size of contextual effect	0.29		0.17
Model 2
Level 2
contacttime1 → contacttime2			0.989 (0.160)	<0.001
prejudicetime1 → prejudicetime2			0.467 (0.221)	0.034
contacttime1 → prejudicetime2			−0.043 (0.015)	0.006
prejudicetime1 → contacttime2			4.028 (2.492)	0.106
normstime1 → normstime2			0.391 (0.146)	0.007
contacttime1 → normstime 2			0.084 (0.030)	0.005
normstime1 → prejudicetime2			−0.146 (0.057)	0.011
Contextual effect controlling for norms			0.036 (0.017)	0.033
Indirect effect of context			0.012 (0.007)	0.086

Note. Model 1 considers only the relationships between contact and prejudice; model 2 adds norms as a mediator. For model 2, level 1 results are shown in Table S1.