Recommendations for assessing differential susceptibility with secondary data

• In Event: 3-003 - Studying Biosocial Developmental Processes with Secondary and Administrative Data

Abstract

Since Boyce (1995) and Belsky (1997) proposed ideas of differential susceptibility, a wealth of evidence has been accumulating that susceptible children, compared to non-susceptible peers, have characteristics that make them vulnerable to developmentally poor environments, but also give them an advantage in developmentally rich contexts. This evidence has emerged across a range of outcomes (academic, behavioral and physical) and environments (parenting, child care and school) moderated by measures of susceptibility at several levels of biology (temperament, adrenocortical reactivity, brain circuitry and genetics).

Despite a theoretical grounding in biology and empirical evidence for differential susceptibility, literature examining the theory is inconsistent. For example, a recent review (Rabinowitz & Drabick 2017) of studies testing differential susceptibility to parenting found 33 (44%) interactions supported differential susceptibility, whilst 38 (51%) and 5 (6%) interactions, respectively, supported the diathesis stress or vantage sensitivity models.

Explaining these inconsistencies would advance developmental science by improving our understanding of when differential susceptibility can be observed. It would also guide research into how child characteristics moderate effects of the environment.

Secondary data is one avenue for additional research to clarify when differential susceptibility is observed. This data often contains a range of socio-ecological information on thousands of children followed over long periods of time.

In this presentation, results of three studies investigating differential susceptibility in Australia using secondary data are communicated. These studies focus on difficult temperament as a mechanism of susceptibility that moderates the effects of parenting and early child care on behavioral and academic outcomes at ages 4 to 10.

Preliminary results did not reveal evidence for differential susceptibility. For example, across 4544 children at age 10, difficult temperament significantly moderated the effects of parenting efficacy on externalizing behavior (p<0.00001; Figure 2), but the interaction did not cross-over within the range of observed parenting efficacy indicating differential susceptibility was absent (Roisman et al. 2012).

Several recommendations for studies wanting to investigate differential susceptibility with secondary data, however, became apparent and are provided, including; 1) focus on accurate measurements of the environment, outcome and mechanism of susceptibility; 2) investigate plausible mechanisms of susceptibility; 3) use outcomes and environments sufficiently positive and negative; and, 4) employ correct statistical methodology.

By exchanging the results and recommendations of this research with other conference members, secondary data can be better utilized in developmental science to investigate how children vary in their response to the environment.

Authors

• Peter Sheldon Rankin, University of Queensland, Institute for Social Science Research, ARC Life Course Centre
  Presenting Author

• Mark Western, University of Queensland

• Michele Haynes, Australian Catholic University

• Karen Thorpe, University of Queensland