Table 2.

Genetic nurture: Parenting mediates associations between mothers’ education polygenic scores and their children’s educational attainment independently of children’s polygenic scores.

	Parenting
	Cognitive stimulation	Warm, sensitive parenting	Low household chaos	Safe, tidy home	All mediators together
	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)
Total effect	.11 (.04, .18)	.11 (.04, .18)	.11 (.04, .18)	.11 (.04, .18)	.11 (.04, .18)
Direct effect	.03 (−.03, .09)	.09 (.02, .15)	.07 (.00, .13)	.09 (.02, .15)	.03 (−.03, .09)
Indirect effect	.08 (.04, .12)	.02 (−.01, .05)	.04 (.01, .08)	.02 (.00, .05)	.08 (.04, .12)
% Mediation	71%	20%	39%	22%	70%

Note: The table shows results of analyses testing whether the different aspects of parenting we measured (cognitive stimulation; warm, sensitive parenting; low household chaos; safe, tidy home) mediate associations between mothers’ education polygenic scores and their children’s educational attainment independently of children’s polygenic scores. Each column reports results from a model testing a different mediator; the last column reports results from a model testing all mediators jointly. Within each column, the ‘total effect’ is an estimate of the association before adding the parenting mediator(s); this does not differ across models. The “direct effect” is an estimate of how much of the association remains after adding the parenting mediator(s) (corresponding to path e in Figure 1). The “indirect effect” is an estimate of the amount of mediation through the parenting mediator(s); expressed as a percentage in the row “% Mediation” (corresponding to paths a*d in Figure 1). All estimates are standardized. Bolded estimates indicate statistically significant (p<.05) effects. 95% confidence intervals (CI) were obtained from 1000 bootstrap replications.