The rich data approach removes the endogenous part of the predictor with control variables, similar to a partial correlation. However, this addresses solely measurable omitted variables.
Fixed effects approaches remove the endogenous part of the predictor with dummy controls based on the longitudinal and cross-sectional data structure. However, this addresses solely omitted variables that are constant for each fixed effect.
