Article summary
Fixed effects models help reduce selection bias in studies by using only within-unit variation. However, their substantive importance is often overstated if researchers don't properly consider plausible counterfactuals for the independent variable being studied.