Retrospective versus Prospective Assessments of Childhood Parenting: Evidence from a 26-year Longitudinal Study

• In Event: The Convergence of Retrospective and Prospective Measures: An Examination of Normative and Atypical Caregiving Experiences

Abstract

Introduction. Across the full range of psychological science scholars have shown interest in the degree to which childhood caregiving experiences shape mental and physical health in adulthood. However, across the field, there is a strong reliance on the use of retrospective, self-report, measures that often assume that they are a proxy for prospective evidence of caregiving. Recent evidence from the maltreatment literature has suggested that there is low agreement between retrospective self-report and prospectively acquired data on maltreatment (κ = .19; Baldwin et al., 2019). However, research on the quality of parenting within the normative range and the extent to which retrospective, self-reported assessments by adults converge with prospective assessments of those experiences is in short supply in the psychological literature. In order to examine the validity of retrospective reports against prospective data, data were drawn from the NICHD Study of Early Child Care and Youth Development, which contains extensive prospective caregiving data from birth to 18 years as well as a retrospective assessment of parental caregiving at age 26 years.

Aims. The analysis was organized around three scientific aims: (1) the extent to which the retrospective reports of maternal and paternal emotional availability converged with prospectively documented evidence of mother-child and father-child relationship quality, respectively, (2) the extent to which the retrospective measures demonstrated discriminant validity in relation to prospective data provided by the other parent, and (3) the extent to which a small set of potential sources of bias (selected a priori) in retrospective reporting of caregiving collected at the outcome assessment accounted for the retrospective reports relative to the prospective data on parent-child relationship quality (i.e., current closeness to mother/father, family cohesiveness, and depressive symptoms).

Study Population. Participants were drawn from the SECCYD, an on-going longitudinal study of participants and their families followed from birth to age 26 years. The subsample for the present study included 805 participants (53% female, 81% White/non-Hispanic) who completed the Emotional Availability Scale (EAS; Sorhagen et al., 2019) at the age 26-year follow up assessment for either their primary maternal or paternal caregiver in childhood.

Methods. Prospective measures included: observed parent-child sensitivity tasks assessed 9 times from 6-months to age 15 years; parent-reported parent-child relationship quality assessed 8 times from 54-months to age 18 years; child-reported parent-child relationship quality assessed 3 times from 6th grade to 18 years. Retrospective measures included the EAS administered to target participants at age 26 years. Measures of potential bias were also administered at age 26 and included current closeness to mother/father, family cohesiveness, and depressive symptoms.

Results. Retrospective reports of childhood emotional availability demonstrated weak convergence with composites reflecting prospectively acquired observations (R2s=.01-.05) and parent-reports (R2s=.02-.05) of parenting quality. Retrospective parental availability was more strongly associated with prospective assessments of child-reported parenting quality (R2s=.24-.25). However, potential sources of bias (i.e., age 26 family closeness/cohesiveness and depressive symptoms) accounted for more variance in retrospective reports (39-40%) than did prospective measures (26%), suggesting caution when using retrospective reports of childhood caregiving quality as a proxy for prospective data.

Authors

• Marissa Nivison, University of Minnesota - Twin Cities
  Presenting Author

• Deborah Lowe Vandell, University of California - Irvine
  Non-Presenting Author

• Cathryn Booth-Laforce, University Of Washington
  Non-Presenting Author

• Glenn I. Roisman, University of Minnesota - Twin Cities
  Non-Presenting Author