Irrational Attention to Correlation in Selected Data
December 9, 2020 12:30 - 13:30
It is very rare that the causal link between actions and outcomes can be observed directly. Often, it is much easier to observe correlation. To derive causality from correlation, economists use elaborate techniques. However, little is known about individuals’ capabilities to infer from correlational data in their everyday lives. I test if individuals erroneously infer causality from correlation in a relevant labor market application. Using insights from selection neglect theory, I study if individuals mistake raw wage gaps between part-time workers and full-time workers for the causal effect of part-time work on wages. I elicit the baseline predisposition for selection neglect among a representative sample of N=1,115 German employees, and design and implement an information experiment among a second representative sample of N=911 respondents to explore the elasticity of beliefs about causal effects to average correlations. I estimate an elasticity of 0.2 to 0.4 of beliefs in response to raw wage gaps.