causal

Measurement Error Biases Correlations Toward Zero

Because both GDP and happiness are noisily measured, and measurement error in two variables biases their estimated correlation toward zero, finding a correlation as high as 0.8 means the true relationship between income and happiness is even stronger than measured; thus Russ's point about data sloppiness actually strengthens rather than weakens the case against Easterlin.

causalpending

Speaker

Justin Wolfers

Evidence Quote

“the more that biases the estimated correlation towards zero”

Source

Betsey Stevenson and Justin Wolfers on Happiness, Growth, and the Reinhart-Rogoff Controversy— EconTalk
Created: 6/13/2026, 12:04:40 AM

Your notes

Loading notes...