View a PDF of the paper titled On Binscatter, by Matias D. Cattaneo and three different authors

View PDF

Summary:Binscatter is a well-liked methodology for visualizing bivariate relationships and conducting casual specification testing. We research the properties of this methodology formally and develop enhanced visualization and econometric binscatter instruments. These embody estimating conditional means with optimum binning and quantifying uncertainty. We additionally spotlight a methodological drawback associated to covariate adjustment that may yield incorrect conclusions. We revisit two purposes utilizing our methodology and discover considerably completely different outcomes relative to these obtained utilizing prior casual binscatter strategies. Common goal software program in Python, R, and Stata is offered. Our technical work is of unbiased curiosity for the nonparametric partition-based estimation literature.