Poster #120 - Measurement, analyses and patterns of outcome: Systematic review of child by environment models of development

• In Event: 3-106 - Poster Session 11
  In Poster Session: PS 11 Section - Methods, History, Theory

Integrative Statement

Across the last decade efforts to understand tailored intervention strategies have generated significant interest in child X environment interactions on developmental outcomes. Analyses have tested the long-standing diathesis stress model (Zuckerman 1999), against emerging hypotheses of vantage sensitivity (Pluess & Belsky 2013) and differential susceptibility (Belsky & Pluess 2009) / biological sensitivity to context (Boyce & Ellis 2005). Empirical testing includes measures of genetic, endophenotypic and phenotypic characteristics of children as moderators, yet with varying levels of empirical support. This study focuses on parenting environments and behavioural outcomes. Several prior reviews on parenting and behavioural outcomes provide a strong base to understand the state of play (Kiff et al. 2011; Rabinowitz & Drabick 2017; Slagt et al. 2017). However, these reviews have not fully evaluated the association of measurement with support for different theoretical models nor systematically examined analytic approaches and decision pathways. This study aimed to review the literature to examine:

- Empirical support for different child X environment models
- The number of significant child X environment interactions
- Measurement techniques associated with theoretical models
- Analytical approaches used

The study systematically evaluated the literature and identified 30 articles and 538 interactions that evaluated if moderating effects were consistent with the diathesis stress, differential susceptibility or vantage sensitivity hypotheses. The methodological approach surrounding evidence for each of hypothesis was assessed, including measurement type (genotype, endophenotype and temperament), measurement quality and analytical approach.

Of the 538 reviewed interactions, 84 were statistically significant and indicated 89 incidences of support for theoretical models as some interactions supported multiple models. Of these 89, 40 (45%) supported diathesis stress, 33 (37%) supported differential susceptibility, 12 (13%) supported vantage sensitivity and 4 (4%) supported a rarely documented hypothesis of contrastive effects (Belsky et al. 2007). This demonstrates a leniency towards diathesis stress and differential susceptibility.

The large number of non-significant interactions is concerning. However, two studies were notably exploratory and accounted for 292 interactions. Excluding these studies improved estimation with 80 (33%) significant and 166 (67%) non-significant.

There were few reliable measurement quality trends. Survey versus observations of parenting and behaviour inconsistently influenced support for the hypotheses. Differential susceptibility was observed more for genetic moderators, whilst temperament moderators supported diathesis stress more frequently.

Several indices showed analytical quality of the interactions was low, and this quality did not explain support for the hypotheses. Studies performed well in terms of independence of measurement and performing regions of significance tests. However, studies rarely adjusted for multiple statistical comparisons, there was limited sensitivity analysis for variable create, raw data was inconsistently visualised to evaluate results and there were inconsistent effects within studies.

Overall, these results highlight substantial variation in support for evidence of the diathesis stress, vantage sensitivity and differential susceptibility hypotheses. As such, a systematic exploration of evidence for the hypotheses is required. Improved analytical rigour and a focus on only changing one or two measurements can help build an evidence base that achieves the methodological requirements for accurately assessing the hypotheses.

Authors

• Peter Rankin, University of Queensland
  Presenting Author

• Mark Western, ARC Life Course Centre, Institute for Social Science Research, University of Queensland
  Non-Presenting Author

• Michele Haynes, Institute for Learning Sciences and Teacher Education, Australian Catholic University
  Non-Presenting Author

• Karen Thorpe, University of Queensland
  Non-Presenting Author