Table 2.

Logistic regressions for demographics characteristics predicting presence vs. absence of goal formulation

Predictor	β	SE	P	OR (95% CI)
Intercept	–6.49	1.37	<0.01
Aged 6–12 vs. 0–5	–0.50	0.25	0.05	0.61 (0.37, 0.99)
Aged 13–18 vs. 0–5	–0.23	0.28	0.41	0.79 (0.46, 1.38)
Autism present vs. absent	–0.24	0.38	0.52	0.78 (0.37, 1.66)
Hyperactivity present vs. absent	–0.44	0.44	0.31	0.64 (0.27, 1.53)
Conduct problems present vs. absent	–0.52	0.33	0.12	0.59 (0.31, 1.14)
Self-harm present vs. absent	–1.11	0.85	0.20	0.33 (0.06, 1.74)
Learning disability present vs. absent	2.10	0.28	<0.01	8.13 (4.72, 14.14)
Hyperactivity × conduct difficulties present vs. absent	2.60	0.75	<0.01	13.44 (3.10, 58.56)
Emotional problems × self-harm present vs. absent	1.58	0.92	0.09	4.84 (0.80, 29.46)
Autism × learning disability present vs. absent	1.00	0.60	0.10	2.72 (0.84, 8.81)

N = 3757. A stepwise model selection was used, meaning predictors were only retained in the model if they improved the model fit

SE standard error, OR (95% CI) odds ratio (with 95% confidence interval)

Service-level random effects variance (standard deviation): 24.8 (4.98)