Scott Alexander examines how different statistical presentations of the same data in social science studies can lead to vastly different interpretations, potentially misleading readers.
Longer summary
Scott Alexander discusses the importance of understanding and interpreting statistical measures in social science studies, particularly focusing on correlation, percent variance explained, and visual representations of data. He examines two studies: one on IQ and state wealth, and another on wealth inheritance. Alexander highlights how different presentations of the same data can lead to vastly different interpretations, potentially misleading readers. He emphasizes the need for a good grasp of statistical concepts and realistic expectations when evaluating social science research, noting that correlations above 0.4 are rare in this field.
Shorter summary