Survey Availability Bias
By Arnold Kling
What do the following beliefs have in common?
A. Belief that Kerry won in 2004, and that a vote-counting conspiracy took place.
B. Belief that men have more sex partners than women.
C. Belief that epidemiology is equivalent to a clinical trial.
D. Belief that happiness does not rise with income beyond a certain point.
In each case, people are believing survey results rather than other indicators of phenomena.
In the case of the election, people believe the exit polls instead of the election results, even though that means you have to believe that there was a widespread conspiracy to miscount votes that took place across numerous different precincts, election supervisors, and voting mechanisms.
People believe surveys which say that men have many more heterosexual sex partners, even though this is not mathematically possible.
Newspapers eagerly report epidemiological study results, even though time and time again such results fail to hold up in clinical trials.
And some economists take seriously the notion that people are not happier at higher income levels, even as they point out that people have a choice of whether to earn higher income or take more leisure.
I call this “survey availability bias,” because the availability of a survey result makes us want to comment, analyze, and interpret the survey, even though the most plausible interpretation is that the survey itself is flawed.
In my view, survey availability bias creates large negative externalities. Surveys add noise rather than signal to our society. If there were less survey availability bias, people would not be so prone to report, pass along, analyze, and comment on surveys. Survey-takers would be less well rewarded, and I believe we would all be better off.
We need a large Pigovian tax on surveys.