Benefits of extensive recruitment effort persist during follow-ups and are consistent across age group and survey method. The TRAILS study

Table 3 Multivariate multinomial logistic regression analysis predicting T1-hard-to-recruit-retainers, T1-easy-to-recruit-retainers, T1-easy-to-recruit-dropouts and T1-hard-to-recruit-dropouts. Retention and dropout observed at T4

	T1-easy-to-recruit-retainer			T1-hard-to-recruit-retainer (ref.)	T1-easy-to-recruit-dropout			T1-hard-to-recruit-dropout
	( n = 1,549)			( n = 55)	( n = 222)			( n = 33)
	n(%)	OR	95% CI	n(%)	n(%)	OR	95% CI	n(%)	OR	95% CI
T1 Sociodemo-graphics
Non-western immigrant	102 (6.6)	0.40*	0.19-0.86	10 (18.2)	40 (18.0)	1.03	0.47-2.29	14 (42.4)	2.72	0.98-7.58
(univariate effects)		0.32*	0.16-0.65			0.99	0.46-2.13		3.32*	1.25-8.77
Lower education mother (≤lower tracks of sec. educ.)	491 (31.7)	0.44*	0.24-0.79	33 (60.0)	132 (59.5)	1.06	0.56-2.00	22 (66.7)	1.25	0.48-3.24
(univariate effects)		0.31*	0.18-0.54			0.98	0.54-1.79		1.33	0.54-3.29
Low family income (<€1,135)	183 (11.8)	0.48*	0.25-0.91	17 (30.9)	60 (27.0)	0.83	0.42-1.65	15 (45.5)	1.35	0.51-3.57
(univariate effects)		0.30*	0.17-0.54			0.83	0.44-1.58		1.86	0.76-4.55
Low IQ (wisc < 85)	213 (13.8)	0.55	0.30-1.03	18 (32.7)	65 (29.3)	0.88	0.45-1.73	13 (39.4)	1.08	0.41-2.81
(univariate effects)		0.33*	0.18-0.59			0.85	0.45-1.60		1.34	0.55-3.28
	mean (SD)	OR	95% CI	mean (SD)	mean (SD)	OR	95% CI	mean (SD)	OR	95% CI
T1 Psychopathology
CBCL Internalizing problems	0.25 (0.19)	1.33	0.96-1.84	0.24 (0.17)	0.25 (0.20)	1.07	0.76-1.52	0.30 (0.24)	1.34	0.84-2.16
(univariate effects)		1.00	0.76-1.31			1.03	0.76-1.38		1.26	0.85-1.86
CBCL Externalizing problems	0.23 (0.19)	0.70*	0.53-0.91	0.31 (0.27)	0.30 (0.26)	0.95	0.71-1.27	0.28 (0.23)	0.74	0.47-1.17
(univariate effects)		0.71*	0.56-0.89			0.97	0.76-1.23		0.89	0.61-1.30

*p < .05; Odds ratios (OR) and 95% confidence intervals (CI) from multivariate multinomial logistic regression analysis; CBCL = Child Behavior Checklist (parent report); CBCL scores standardized; univariate effects of single multinomial regression analyses italicized. Nagelkerke R² = 0.145.

ISSN: 1471-2288