Adult non-response bias from a child perspective. Using child reports to estimate father’s non-response

• Published in: Social science research Vol. 49; pp. 31 - 41
• Main Authors: Bastaits, Kim, Pasteels, Inge, Ponnet, Koen, Mortelmans, Dimitri
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.01.2015; Academic Press
• Subjects: Adolescent; Adult; Age Factors; Belgium; Bias; Child; Child reports; Children; Children & youth; Demographics; Divorce; Educational Status; Father reports; Father-Child Relations; Fathers; Female; Humans; Hypothesis; Indicators; Level of education; Multi-actor data; Non-response; Parenting; Parents & parenting; Research Subjects; Social research; Belgium; Multi-actor data; Father reports; Child reports; Non-response
• ISSN: 0049-089X; 1096-0317; 1096-0317
• DOI: 10.1016/j.ssresearch.2014.07.004

•We use child reports from dyadic survey data to estimate selectivity in paternal non-response.•We use SEM to look at the double bias of socio-demographic data on paternal survey participation & paternal involvement.•We find that more involved fathers are more participating in the survey.•Open father–child communication mediates between the father’s age/educational level & paternal survey participation. Most parenting research on paternal involvement uses data from a father perspective. Nevertheless, research on bias in father non-response is scarce. In this study, we examine the non-response bias of fathers, hypothesizing that fathers who engage in parenting studies might already be fathers who are more involved with their children than fathers who do not engage in these studies. Furthermore, we expect a double non-response bias by socio-demographic characteristics of the father, which impacts both paternal participation as well as paternal involvement. Using the multi-actor dataset from the “Divorce in Flanders”-project, which provides data from children whose fathers actually participated (N=461) as well as data from children whose fathers did not (N=137) with children reporting on paternal involvement, we are able to test our hypotheses. Results confirm our first hypothesis, indicating that non-participating fathers are significantly more uninvolved than participating fathers. Regarding our second hypothesis, an indirect effect of father’s educational level and age on non-response was revealed for one out of three indicators of paternal involvement.