Table 2. Top models (ΔAIC_C<4) predicting SAIL mental health metrics.

	Predicting SAIL Mental Health Domain
Model	df	AICC	ΔAICC	Weight	Rank	R 2
1p) T* + SW + SB + P*	12	1890.17	.00	.60	1	.90
2p) T* + SW + SB + P* + SB:P	13	1892.85	2.68	.16	2	.90
3p) T* + SW + SB + P* + T:SW	13	1893.29	3.12	.13	3	.90
	Predicting SAIL Population Coverage
Model	df	AICC	ΔAICC	Weight	Rank	R 2
4p) T* + SW* + SB + PW + PB	13	463.62	.00	.67	1	.96
5p) T* + SW* + SB + PW + PB + T:PW	14	467.48	3.86	.10	2	.96
6p) T* + SW* + SB + PW + PB + T:SW	14	467.49	3.87	.10	3	.96
	Predicting SAIL Continuity of Care
Model	df	AICC	ΔAICC	Weight	Rank	R 2
7p) T* + S + P* + S:P	12	3529.58	.00	.41	1	.74
8p) T* + S + P*	11	3529.71	0.13	.39	2	.73
	Predicting SAIL Experience of Care
Model	df	AICC	ΔAICC	Weight	Rank	R 2
9p) T* + SW* + SB + P	12	2473.44	.00	.42	1	.85
10p) T* + SW* + SB + P + T:SW	13	2474.00	0.55	.32	2	.85
11p) T* + SW* + SB + P + T:P	13	2476.26	2.82	.10	3	.85

Notes:

*Denotes that a random slope was included for the predictor.

All models included a random intercept. T = Time. S = Staffing (overall). SB = Staffing (between facilities). SW = Staffing (within facilities). P = Productivity (overall). PB = Productivity (between facilities). PW = Productivity (within facilities). R² = pseudo-R² as proposed by Snijders and Bosker [17].