This paper aims to make an empirical and methodological contribution to the literature on family policies. Evaluating three reforms to the UK maternity leave legislation which successively extended pay and job protection benefits, the paper documents an impact of the extension of pay benefits on children’s test scores and trade-offs in the respective impacts on mothers and children. The paper also warns of attenuation bias affecting Regression Discontinuity Designs in the presence of measurement error in the running variable–a problem to which studies of family benefits are prone when the eligibility cutoff is imprecisely observed. We propose a practical and transparent solution leveraging validation data.