Table 6. Cross-lagged and autoregressive associations between parents’ positive and critical comments and children’s outcomes.

Mothers	N	β	p	Fathers	N	β	p
T1 M_CC = > T2 EXT	141	.01	.94	T1 F_CC = > T2 EXT	157	.02	.83
T1 EXT = > T2 M_CC		.32	.002	T1 EXT = > T2 F_CC		.46	< .001
M_CC autoregressive		.18	.11	F_CC autoregressive		-.10	.30
EXT autoregressive		.66	< .001	EXT autoregressive		.67	< .001
T1 M_CC = > T2 INT	141	.08	.27	T1 F_CC = > T2 INT	157	-.08	.37
T1 INT = > T2 M_CC		.15	.15	T1 INT = > T2 F_CC		.23	.02
M_CC autoregressive		.33	.001	F_CC autoregressive		.05	.64
INT autoregressive		.64	< .001	INT autoregressive		.63	< .001
T1 M_PC = > T2 EXT	141	.03	.66	T1 F_PC = > T2 EXT	157	-.17	.002
T1 EXT = > T2 M_PC		-.13	.18	T1 EXT = > T2 F_PC		-.18	.07
M_PC autoregressive		.51	< .001	F_PC autoregressive		.32	< .001
EXT autoregressive		.67	< .001	EXT autoregressive		.65	< .001
T1 M_PC = > T2 INT	141	-.04	.44	T1 F_PC = > T2 INT	157	-.20	.001
T1 INT = > T2 M_PC		-.40	< .001	T1 INT = > T2 F_PC		-.11	.21
M_PC autoregressive		.42	< .001	F_PC autoregressive		.35	< .001
INT autoregressive		.65	< .001	INT autoregressive		.58	< .001

In all of these models, we also included child gender, and depressive symptoms, educational attainment and age of the relevant parent as covariates (for presentational clarity, we do not include these coefficients here). The pattern of significant findings was identical when a categorical research diagnosis of depression was used instead of the continuous measure of depression symptoms (PHQ-9 score).

INT = internalizing (emotional problems), EXT = Externalizing (behavioural problems); M = Mothers, F = Fathers; CC = Critical comments, PC = Positive comments; T1 = Time 1, T2 = Time 2; autoregressive = T1 variable predicting T2 variable.