Justin Wolfers has skeptical comments on Gordon and Dahl’s new paper on economists’ ideology, featuring this cool graph:


Since I remained a little unclear about the origin of the graph, I emailed Wolfers.  He graciously responded, and gave me permission to share.  Enjoy.

The details were:

1. I looked at each name, and rated them from 0 to
10, using my completely subjective judgment.  Typically it was based on
op-eds each had written, which administrations they had worked for,
conversations I have had with them, general impressions based on their
research, and in one case, whether their facebook page involved a Tie
Dye shirt (seriously).  It’s purely subjective.  (When I knew very
little, I was systematically more likely to guess at a number towards
the center.)

2. I had someone else (whose name I can’t reveal) do the same.  Our guesses were correlated at about 0.7, which isn’t bad.

3. I averaged ’em, then converted to z-scores.
Unfortunately I promised my fellow rater that I would never reveal his scores, otherwise I would send them to you.
My guess is that there is tons of measurement error in my measure.
Also, the underlying construct “ideology” is not well defined
(although perhaps there’s an “I know it when I see it” sense when
talking about folks between Mankiw and Krugman).  But someone like you
would be very difficult to place in such a spectrum, even though there’s
lots of public info about your views.

I replied:

I’m confused.  Didn’t the NBER URL graph your scores against his?

Justin’s clarification:

The graph shows my ratings and those of the other rater, but not the identity of who we were rating.