Abstract
We examined the extent to which mother and father attributions for child behavior problems predict child behavior problems over time, accounting for the other parent's attributions, initial child problems and the child's attention-deficit/hyperactivity disorder (ADHD) status. Parents of 7- to 12-year-old boys with (n = 26) and without (n = 38) ADHD participated. Parents completed the Strengths and Difficulties Questionnaire (SDQ) as a measure of their son's behavior problems as well as the Written Analogue Questionnaire, reporting their attributions for child behavior problems. Parents completed the SDQ a second time 7 months later. Both mother and father attributions were associated with child behavior problems at Time 1 and again 7 months later. However, when ADHD status and the other parent's attributions for child behavior were controlled, only father attributions predicted child behavior problems, and continued to be uniquely predictive of child behavior problems at Time 2 even with initial child behavior problems controlled. Father attributions provide unique information above and beyond mother attributions when considering current and future child behavior problems.
Notes
1In an initial sample of 65 families, one outlier was observed. This mother responded inconsistently both within and across questionnaires and rated herself as relatively low on acculturation. As a result, we omitted this family.
2There was also a question assessing parent blame, but this was omitted from analysis as it does not focus on attributions for the child.
3At Time 2, four fathers did not complete the SDQ, and their scores were imputed as equal to their wives. We reran analyses with these four families omitted, and this did not change the pattern of results, so the imputed results are reported.
4Data were collected as part of a larger project that required mothers and children to attend a lab visit.
5In addition, analyses using hierarchical linear modeling (HLM) with parent nested within family were conducted, and the results were nearly identical to those presented here. This was due to an ICC of 1.42e-11, indicating that variance due to family membership was negligible. Due to the small sample size, hierarchical linear regressions are presented rather than the HLM results.
6An interaction term between mother and father WAQ scores was included in the analyses for each regression but was not significant in any instance and therefore not discussed further.