Main Results

Before the intervention


Large gender gap in math before the intervention in favour of boys (0.21 standard deviation)

  • larger than the national average at the same age

  • in line with INVALSI results for the same classes (0.24 s.d)

Checks

Balance

  • Children characteristics: pre-test score, special education needs and disability, gender

  • Family backgorund: parental education, migratory background

  • Teacher's c characteristics: education, experience, type of contract

  • Class characteristics: average pre-test score and variability in the classclass size

Differences between the treatment and the control groups are small and mostly not statistically significant

Compliance

  • All classes in the treatment group were exposed to the intervention

  • All classes in the control group were not exposed to the intervention

  • Large majority of the children actually attended the majority of the lab sessions (98.6% attended at least 3 out of 5 sessions)

Attrition

  • About 5 percent of children not present at the post-test

    • No difference treated/controls and boys/girls

  • About 15 percent of children not present at pre- or post-test

    • Pre-test during flu peak

    • Significant difference treated/controls

    • Balance after attrition is confirmed

    • No impact on results

Results

Core results

  • Large and significant improvement of girls' math skills (+0.14 s.d.)

    • Large and policy-relevant

  • Zero effect for boys

  • Reduction of the GGM by more than 40 percent

By prior math skills

  • No effect for boys, irrespective of their starting level

  • Large effect for girls, increasing with prior achievement