Table 4.

Comparisons of regression coefficients.

	LME_INT	LME_INT+SLP	ComBat	ComBat-GAM

	*Effect of Age*
LME _INT	-	0.6e-04	−0.6e-04	10.8e-04
LME _INT+SLP	0.703	-	−1.2e-04	10.2e-04
ComBat	0.677	0.125	-	11.4e-04
ComBat-GAM	<0.001	<0.001	<0.001	-
	*Effect of Diagnosis*
LME _INT	-	0.4e-03	1.2e-03	6.6e-03
LME _INT+SLP	0.277	-	0.7e-03	6.2e-03
ComBat	<0.001	0.011	-	5.4e-03
ComBat-GAM	<0.001	<0.001	<0.001	-
	*Age by Diagnosis Interaction*
LME _INT	-	3.8e-05	−0.3e-05	−44.4e-05
LME _INT+SLP	0.262	-	−4.1e-05	−48.2e-05
ComBat	0.999	0.197	-	−44.1e-05
ComBat-GAM	<0.001	<0.001	<0.001	-
	*Effect of Sex*
LME _INT	-	−1.8e-03	0.6e-03	1.6e-03
LME _INT+SLP	<0.001	-	2.4e-03	3.4e-03
ComBat	0.012	<0.001	-	1.1e-03
ComBat-GAM	<0.001	<0.001	<0.001	-
	*Sex by Diagnosis Interaction*
LME _INT	-	6.0e-04	−0.4e-04	12.0e-04
LME _INT+SLP	0.056	-	−6.3e-04	6.1e-04
ComBat	0.999	0.037	-	12.4e-04
ComBat-GAM	<0.001	0.049	<0.001	-

Note: The upper triangle of the matrix are the differences between regression coefficients from methods in rows and columns. Higher values mean that the method in columns lead to more negative (i.e., weaker positive coefficients, or stronger negative coefficients) estimates than the method in rows. The lower triangle represents the corresponding p-values (Bonferroni corrected).