Table 2.

Logistic and linear regression models on the association between family childcare types and conduct problem behaviors in young children.

Family childcare types	Total, n (%)	Conduct problem behaviors
		Logistic regression (Taken as a categorical variable)			Linear regression (Taken as a continuous variable)
		Case, n (%)	Crude model†	Adjusted model	Score, Mean ± SE	Crude model†	Adjusted model
			Crude OR (95% CI)	Adjusted OR (95% CI)		Crude β (95% CI)	Adjusted β (95% CI)
Parents together	1228 (13.2)	21 (1.7)	Ref	Ref	0.470 ± 0.009	Ref	Ref
Mother alone	347 (3.7)	14 (4.0)	2.42 (1.22, 4.80)*	2.53 (1.27–5.04)**	0.523 ± 0.019	0.053 (0.013, 0.093)**	0.059 (0.019, 0.099)**
Mother with others (except father)	1530 (16.5)	56 (3.7)	2.18 (1.32, 3.63)**	2.55 (1.52–4.27)***	0.523 ± 0.009	0.053 (0.028, 0.078)***	0.063 (0.037, 0.088)***
Grandparents	320 (3.4)	10 (3.1)	1.85 (0.86, 3.98)	2.18 (1.01–4.71)*	0.502 ± 0.019	0.032 (−0.009, 0.074)	0.043 (0.002, 0.085)*
Changing caregivers	5864 (63.1)	208 (3.5)	2.11 (1.34, 3.33)**	2.33 (1.47–3.69)***	0.539 ± 0.004	0.069 (0.048, 0.089)***	0.073 (0.051, 0.094)***

^†Model was unadjusted for any confounding variables.

‡Model was adjusted for confounders including child age, gender, parental education level, parental age at pregnancy, marital status, and family income.

^*P < 0.05,

^**P < 0.01,

^***P < 0.001.