Table 3.

Fixed effects estimates (top) and variance estimates (bottom) for the multilevel linear regression of belief in the theory on the number of significant results, the composition of significant results, preregistration status, and the possible presence of p-hacking – for all participants in Study 1 (Models 1a and 1b), only participants in the preregistration condition in Study 2 (Models 2a and 2b), and all participants in Study 2 (Models 2c and 2d)

Parameters	Model 1A	Model 1B: 1A + Conceptual	Model 2A	Model 2B: 2A + Conceptual	Model 2C: 2A + Preregistration	Model 2D: 2A + p-hacking
Regression coefficients (fixed effects)
Intercept	-1.63 (0.074) *	-1.57 (0.076) *	-1.67 (0.07) *	-1.45 (0.075) *	-1.62 (0.06) *	-1.49 (0.06) *
Level 1
k	0.74 (0.028) *	0.74 (0.028) *	0.74 (0.26) *	0.69 (0.03) *	0.71 (0.02) *	0.72 (0.02) *
Conceptual	-	-0.13 (0.041) *	-	-0.23 (0.04) *	-	-
Preregistration	-	-	-	-	0.02 (0.06)	-
p-hacking	-	-	-	-	-	-0.31 (0.06) *
Variance components (random effects)
Residual	0.44 (0.66)	0.43 (0.65)	0.84 (0.92)	0.48 (0.70)	0.66 (0.81)	0.58 (0.76)
Intercept	2.15 (1.47)	2.15 (1.47)	2.03 (1.43)	2.42 (1.55)	2.07 (1.44)	2.06 (1.43)
Slope	0.28 (0.53)	0.28 (0.53)	0.30 (0.55)	0.34 (0.58)	0.23 (0.47)	0.24 (0.49)
r(intercept, slope)	-0.78	-0.78	-0.74	-0.73	-0.75	-0.77

Standard errors are in parentheses. k refers to the number of significant results within the scenario. Conceptual is a binary variable that takes on the value of 1 if the conceptual replication was significant and 0 otherwise. Preregistration is a binary variable that takes on the value of 1 if the participant was allocated to the preregistration condition and 0 if the participant was allocated to the regular condition. p-hacking is a binary variable that takes on the value of 1 if the participant indicated to have taken into consideration p-hacking in their responses and 0 if they did not indicate this

^* p $<$ .001