Early on in my graduate career, a professor hired me to do some data cleaning on a set of historical data she and a coauthor had collected. Eventually, my data analysis and stata skills became more useful than my data cleaning skills and at some point, she asked me to perform some sort of regression or matching analysis. I did it sort of slap-dash and sent it out, returning later to find at least one big mistake. Though I presented the corrected version to her in a meeting later, she had already seen the incorrect version and begun to make changes to the paper to fall in line with it. What followed was a 30 minute lecture on how I needed to be careful, how she couldn’t write me a good recommendation if I wasn’t careful, how specific employers wouldn’t want me if I wasn’t careful.
It is a conversation I’ve relived several times throughout my still very new career as a PhD economist, admonishing myself to be careful and diligent in all my work, but more than ever in the past week or two as the Reinhart-Rogoff Excel error uncovered by a UMass-Amherst grad student has come to light. It hasn’t gone well for them and according to some, may even be changing the debate on austerity in politics.
While I understand the excitement of finding something big, it seems that the bigger a deal this paper was to be, the more careful they would have been. I once asked Robert Barro whether he thought people went easy on him because he held so much sway. Not at all, he told me, if anything, they’re harder on me.
And we should be, hard on each other that is. We should demand transparency and replication, and not just by chance in some random graduate classroom. If evidenced by nothing else than the number of “you’re an economist, aren’t you used to being wrong?” jokes I heard this weekend, we need to be more careful.