Table 4.

Results of mixed-effects logistic regression models with SIC proportion scores by phase and stage predicting program start-up status

Outcome	Model	Predictor		Fixed effects					Variance componentsb,c
				β	SE	p	95% CI		Var	SE
Start-Upd
	Phase 1
		Intercept		-1.73	0.74	.019	[-3.17, -0.29]		8.86	3.83
		Proportiona		13.78	1.10	< .001	[11.63, 15.94]
	Stage 1
		Intercept		-0.97	1.09	.373	[-3.10, 1.16]		8.70	6.08
		Proportiona		8.33	3.39	.014	[1.68, 14.98]		63.19	59.55
	Stage 2
		Intercept		-1.51	0.38	< .001	[-2.25, -0.77]		1.92	0.87
		Proportiona		7.87	0.47	< .001	[6.94, 8.79]
	Stage 3
		Intercept		-1.23	0.83	.139	[-2.87, 0.40]		11.98	5.23
		Proportiona		13.33	1.11	< .001	[11.16, 15.50]
Competencee
	Phases 1 & 2
		Intercept		-4.53	0.45	< .001	[-5.41, -3.65]		1.1138	0.7959
		Phase 1a		3.07	0.83	< .001	[1.46, 4.69]
		Phases 1 × 2a		8.22	1.42	< .001	[5.45, 11.00]

^aProportion scores range from 0.00 to 1.00 and were grand mean centered prior to entry, thus the intercept reflects the log-odds of program start-up for an average proportion in the respective phase or stage

^bSite-level variance component estimates are not available for the Bernoulli outcome distribution and, as such, the reported estimates are limited to program-level variance

^cVariance components for the proportion predictor specified based on the likelihood ratio test

^dThe sample for this regression include N = 1287 sites with a known end-status (discontinued or achieved start-up) for program start-up

^eThe sample for this regression include N = 1105 sites with a known end-status (discontinued or achieved competence) for competency