Table 2.

Potential outcome means, discrete change of mental health over survey waves, and estimated causal effect among students attending 43 secondary schools with linked survey data across three waves of the COMPASS study

			Depression		Anxiety		Flourishing—reversed
			Model 2	Model 3	Model 2	Model 3	Model 2	Model 3
Potential outcome mean	2018	Mean [95% CI]	7.64 [7.36, 7.92]	7.65 [7.37, 7.94]	5.28 [5.02, 5.55]	5.29 [5.03, 5.55]	7.01 [6.61, 7.40]	6.99 [6.61, 7.37]
	2019		8.61 [8.40, 8.83]	8.62 [8.42, 8.82]	6.05 [5.82, 6.28]	6.06 [5.84, 6.28]	7.61 [7.30, 7.93]	7.59 [7.30, 7.89]
	2020		9.09 [8.66, 9.53]	9.09 [8.66, 9.52]	6.36 [6.04, 6.68]	6.32 [5.97, 6.66]	7.70 [7.27, 8.13]	7.61 [7.14, 8.08]
Average discrete change	2019–18 (Pre)	Dif. Mean [95% CI]	.97 [.81, 1.14]	.97 [.79, 1.14]	.77 [.62, .91]	.77 [.62, .92]	.61 [.42, .79]	.61 [.42, .80]
	2020–19 (Per)		.48 [.10, .87]	.47 [.10, .83]	.30 [.07, .53]	.26 [.02, .50]	.09 [−.24, .41]	.01 [−.35, .38]
Estimated causal effect	Per - Pre	Dif.-In-Dif. [95% CI]	−.49 [−.93, −.05]	−.50 [−.94, −.05]	−.47 [−.74, −.19]	−.51 [−.83, −.20]	−.52 [−.92, −.13]	−.59 [−1.05, −.14]

Model 1 (Not shown): Base model. Fixed effect model controlling for time-invariant unobservables; the lagged variable is the outcome; covariates used: sex interaction with age at entry.

Model 2: Model 1 but constraining sex/age effects on outcomes to be fixed across time and sample selection correction with predictors of selection: age, sex, and weekly spending money.

Model 3: Model 2 with predictors of selection: age, sex, weekly spending money, province, and school connectedness (predictor IMR).

Depression: CESD-10, range 0–30; Anxiety: GAD-7, range 0–21; Flourishing—reversed: Diener's 8-item Flourishing Scale, range 0–32 with lower numbers depicting higher flourishing.

CI = confidence interval; COMPASS = Cannabis use, Obesity, Mental health, Physical activity, Alcohol use, Smoking, and Sedentary behavior.