Table 3.

Prediction of the mean performance in two psychosocial OSCE stations (OSCE) by first impression (FIR) and MMI performance (MMI)

	R2	P(F)	Beta	Sig. (beta)	First-order corr	Semi-partial corr	R2 change	Sig. (R2 change)
Model 1 First Impression only	.057	.012*
FIR			.238	.012*	.238
Model 2 MMI performance only	.065	.007**
MMI			.255	.007** *	.255
Model 3 FIR and MMI	.084	.009**
FIR			.153	.142	.238	.136	.019	.142
MMI			.185	.076	.255	.165	.027	.076

The predictors FIR and MMI are mean ratings across two raters. The coefficients in this table are representative of 8 multiple regression runs obtained with different permutations of designations as rater 1 or rater 2. The differences between these runs are small and do not change any conclusion from the data. If instead of means across two raters scores from a single rater are used as predictors explained variance drops by a small amount but the pattern of coefficients stays the same

The semipartial correlation is the correlation of residuals with the OSCE-criterion where the effect of MMI is taken out of FIR and vice versa