Betsey Stevenson on Economists and Data

Economics is fundamentally an empirical subject. Prof. Betsey Stevenson recently wrote on economists and data. 

Betsey Stevenson

First, let’s review how economists do research. We use the scientific method:

1. Notice something you’re interested in studying.
2. Do some research to learn more about the subject.
3. Develop a mathematical model of your subject.
4. Test the model against real world data
   1. If the model passes your test, write about it and wait for comments.  Others will test it using different data.  Some may improve the model.  This is the basis of the progress of knowledge in economics.
   2. If the model fails your test, write about that.  The scientific method works by rejecting hypotheses. There is a well-known bias in scientific fields towards publishing papers that support the hypothesis.  In addition to needing more replication studies, we also need desperately to learn from our (and others’) failed models.

For better or worse, step 4B has been, to some extent, replaced by “p-hacking,” adjusting the model to better fit the data.  This is, of course, a violation of basic statistical methodology.  You’re supposed to take one shot at the data.  If the model fails, either do something else or find a new (independent) dataset to use for testing.

But we rely on data.  Lots and lots of data.  Forthwith, Prof. Betsey Stevenson has a few things to say about data.