Poster #137 - Challenging the Challenge Hypothesis on Testosteron in Parenting:A Meta-Analysis

• In Event: 1-169 - Poster Session 04
  In Poster Session: PS 04 Section - Parenting & Parent-Child Relationships

Integrative Statement

In many species, higher Testosterone (T) levels are linked to increased mating efforts , whereas lower T levels are associated with increased parenting effort. Indeed, it has been suggested that also in humans mothers and fathers have lower T levels than non-mothers and non-fathers (Barret et al., 2013; Gray, Yang, & Pope, 2006; Kuzawa, Gettler, Huang, & McDade, 2010), and that in men caregiving is related to a decrease in T (Storey, Walsh, Quinton, & Wynne-Edwards, 2000). These findings are in line with the Challenge hypothesis that T levels increase in the context of competition and decrease in the context of caregiving (Archer, 2006). To investigate the association between parenting and T, we did a systematic literature search and meta-analysis.
The first screening of 1,275 studies was based on title and abstract. All dimensions of parenting were included: being a parent (comparing parents with non-parents), parenting behavior, parental perceptions, cognitions, and emotions. We included studies with testosterone administration, studies with baseline T levels in blood or saliva, or T reactivity. We excluded clinical samples. Two coders (WMM and MJBK) agreed 100% upon including 30 studies with 43 study outcomes (total N = 3,481) (see Figure 1). The coders independently coded all studies using a standardized coding system. Average agreement between coders was .94. The outcomes were categorized into (1) intervention studies (including 7 RCTs), (2) studies on the association between T and parenting quality, and (3) studies that compared T levels of parents and non-parents.
Using the program Comprehensive Meta-Analysis (Borenstein et al., 2013), results of individual studies were transformed into common metrics (r) and overall effect sizes per outcome category were computed. Study outcomes of overlapping samples were never combined in one meta-analysis. The combined effect size for 10 intervention studies (N=428, mostly testing the effect of behavioral interventions on T levels) was not significant. RCT vs. non-RCT did not moderate the effect size. For 19 study outcomes on parenting quality (N = 1,819), the combined effect size was not significant either, and applying trim-and-fill showed that 4 studies needed to be trimmed and filled, resulting in an adjusted correlation that was negligible. The 12 studies (N = 1,199) comparing T levels of parents with non-parents showed a significant effect size. Effect sizes were significant for both males and females, with parents showing lower T levels than non-parents. However, trim-and-fill indicated 4 studies to be trimmed and filled resulting in an adjusted non-significant effect. We are currently exploring the role of child age in this association.
Overall, our results appear to challenge the Challenge hypothesis: To date, there is insufficient evidence for a robust association between T and parenting. Measurement issues and heterogeneity may account for the absence of a clear link between the various dimensions of parenting and T. We suggest that more RCTs (currently only 16% of the studies) are needed to test the bidirectional association between parenting and T.

Authors

• Willemijn Meijer, Vrije Universiteit Amsterdam
  Non-Presenting Author

• Marian J Bakermans-Kranenburg, Vrije Universiteit Amsterdam
  Presenting Author