15581-m-nanninga

CHAPTER 5 92 with nine items on a five-point Likert scale (maximum of three missing items, Cronbach’s α=.92). The subscale was categorized as low (score 9-36), medium (37-42) and high (43- 45) level of support based on tertiles of data distribution. Parenting skills were measured with the three subscales of the nine-item version of the Alabama Parenting Questionnaire (APQ) [40]. The subscales poor supervision , inconsistent disciplining and positive parenting were answered with a five-point Likert scale (maximum of one missing item, Cronbach’s α respectively: 0.28, 0.70, 0.85). The item ‘ Your child fails to leave a note or to let you know where he/she is going’ caused the subscale poor supervision to have a low reliability (0.28) and, therefore, it was not included, leading to a Cronbach’s α of 0.57. Poor supervision and inconsistent disciplining were dichotomized as either good/consistent (highest 25%; respectively scores 1-2, 1- 2.67) or poor/inconsistent (respectively scores 2.5-5, 3-5). Positive parenting was dichotomized to ‘low’ (lowest 25%, score 1-3.67) and ‘high’ (score 4-5). Relevant background characteristics concerned children’s age, gender, ethnicity, parental educational level and family composition. Parental educational level was based on the highest educational level achieved by either one of the parents/caregivers [41]. Family composition was assessed by asking the parent with whom the child lived and was categorized as ‘biological two-parent family’ or ‘other’ (e.g. living with one parent, a foster family or living in a residential care facility). Children’s psychosocial problems were measured using the total difficulties score of the ‘Strengths and Difficulties Questionnaire’ (SDQ) based on the past 6 months (Cronbach’s α=.73) [42-45]. The score consists of 20 items describing positive and negative attributes of children to the following dimensions: emotional symptoms, conduct problems, hyperactivity/inattention and peer problems. The scale was dichotomized into the ‘normal’ range and the ‘borderline to abnormal’ range of the SDQ. Statistical analyses First, we described the characteristics of the children enroled and not enroled in psychosocial care. Second, we used logistic regression analyses to assess the association between family social support, parenting skills and children’s psychosocial care enrolment. We repeated these analyses with adjustment for background characteristics. Third, we assessed whether the associations differed for children with and without psychosocial problems or whether the associations were mediated by the child’s psychosocial problems. We tested this using moderator and mediator analyses, according to the procedure proposed by Baron and Kenny [46]. Relevant mediation was defined as at least 10% change in the regression coefficient after adjustment [47]. Finally, we included participants from the community sample whose children contacted a professional