This repository contains the manuscript and code for paper "Transformation-Induced Bias: Unbiased Coefficients Do Not Imply Unbiased Quantities of Interest."
You can find the latest version here.
To replicate the results, simply run the file
do-all.R. If you prefer to replicate only some of the results, this file points to the relevant files in the directory
Here's the abstract:
Political scientists commonly focus on quantities of interest computed from model coefficients rather than on the coefficients themselves. However, the quantities of interest, such as predicted probabilities, first differences, and marginal effects, do necessarily not inherit the small sample properties of the coefficient estimates. Indeed, unbiased coefficients estimates are neither necessary nor sufficient for unbiased estimates of the quantities of interest. I characterize this transformation-induced bias, calculate an approximation, illustrate its importance with two simulation studies, and discuss its relevance to methodological research.