Table 4.

Results of multilevel models predicting change in discrepancy score over the 32-week training system

	β	B (SE)	p-value	95% CI
Mean discrepancy score, all itemsa	− 0.36	− 0.00 (0.00)	.15	[− .01, .00]
Univariate modelsb
Time-invariant predictors (baseline)
Family therapy allegiance	− .01	− .00 (.05)	.94	[− .08, .07]
Experimental condition	− .01	.01 (.08)	.93	[− .12, .14]
Digital literacy	− .29	− .11 (.06)	.05	[− .20, − .02]
Time-varying predictors
Video vignette utility	− .05	− .06 (.03)	.04	[− .11, − .01]
Percent weeks completed	− .06	− .00 (.00)	.40	[− .00, .00]
Number of submitted checklists	− .02	− .01 (.01)	.10	[− .02, .00]
Final conditional model
Time-invariant predictors (baseline)
Digital literacy	− .36	− .13 (.06)	.02	[− .22, .04]
Time-varying predictors
Video vignette utility	− .02	− .04 (.03)	.28	[− .09, − .02]
Number of submitted checklists	− .04	− .02 (.18)	.03	[− .03, − .00]

Negative estimates indicate a decrease in discrepancy score over time

SE standard error

^aThe unconditional model tested the average discrepancy score slope

^bVariables were entered one at a time to screen for potential significant effects. Variables exceeding a threshold of p < .10 were included in the final model. Interactions between experimental condition and the three time-varying predictors were also screened for inclusion in the final conditional model and did not exceed the p < .10 threshold