Thomas Lumley • 4/14/2024

Assumptions

This article examines the concept of 'assumptions' in statistics, highlighting two key problems: distinguishing necessary from sufficient conditions, and confusing assumptions with modeling choices. It uses examples from linear regression to show how different assumptions (e.g., normality, independence) enable different conclusions and properties of estimators, emphasizing that stronger assumptions allow for more inferences but are not always required for validity.

0 comments

#statistics #Linear Regression #Assumptions