The other variables have little impact on the firm or the person effects in the presence of age and education controls.
The inclusion of age and education controls in our specification follows from our theoretical model to capture the average level of productivity across region and occupation cells.
The only statistically significant between-group difference at post-intervention was in pain severity, with the Headstrong group reporting significantly lower pain severity than education controls (M = 5.06 vs. 6.25; p = 0.03; ES = 0.7).
Specification (2) examines the impact of paternal income but excludes the parental education controls and suggests an income elasticity with respect to post compulsory schooling participation of 11.8% boys and for 8.8% for girls.
Treatment effects for women fall by approximately one fifth when language and age are added in specification 2, then climb marginally when education controls are included (from specification 2 to 3), and climb again when a broad range of controls are included (from specification 3 to 4).
There are four education controls, each measuring the share of the native population native in each region, occupation cell that have completed a university or college degree, a secondary school education, those who have not completed secondary school, and those who are students.
