Full Matching Approach to Instrumental Variables Estimation with Application to the Effect of Malaria on Stunting

摘要

Most previous studies of the causal relationship between malaria and stuntinghave been studies where potential confounders are controlled viaregression-based methods, but these studies may have been biased by unobservedconfounders. Instrumental variables (IV) regression offers a way to control forunmeasured confounders where, in our case, the sickle cell trait can be used asan instrument. However, for the instrument to be valid, it may still beimportant to account for measured confounders. The most commonly usedinstrumental variable regression method, two-stage least squares, relies onparametric assumptions on the effects of measured confounders to account forthem. Additionally, two-stage least squares lacks transparency with respect tocovariate balance and weighing of subjects and does not blind the researcher tothe outcome data. To address these drawbacks, we propose an alternative methodfor IV estimation based on full matching. We evaluate our new procedure onsimulated data and real data concerning the causal effect of malaria onstunting among children. We estimate that the risk of stunting among childrenwith the sickle cell trait decrease by 0.22 times the average number of malariaepisodes prevented by the sickle cell trait, a substantial effect of malaria onstunting (p-value: 0.011, 95% CI: 0.044, 1).