Do educational reforms increase or decrease health inequalities: A matter of methods?

Evaluating whether social policies reduce health inequalities is complicated by the fact that these upstream determinants may also change socioeconomic distribution. Failure to account for compositional changes severely bias effect estimation procedure. In this article, we illustrate how a inequality impact assessment of policy (also) distribution produce biased results. First, show why analyses do not fail estimate correct counterfactual outcome. This problem most notably occurs when using repeated cross-sectional data, often only available option evaluate large-scale policies. Second, conducted microsimulation study magnitude under various conditions. The results showed actual on underestimated and even are in opposite direction causal policy. Future studies should explore new strategies, such as simulation methods, assess cause composition population, enable researchers accurately their inequalities. • important. Effect evaluations will be without correction changes. Results from microsimulations substantial.