Ian Ayres & John Donohue

5/8/03:

Lack of data sharing etiquette

------ Forwarded Message
From: "Ian Ayres"
Reply-To:
Date: Tue, 26 Feb 2002 14:39:08 -0500
To: "John R. Lott, Jr."
Subject: request

I finally got round to reading the epilogue to your second edition and will try to make sure that we respond to the work that you have done there so that we don't make the error of responding to just the last edition.

Because of the massive amount of additional work that you have done in creating the second edition, there are now several types of editional data that you have that would help us in assessing your work.

I think listed in order of importance, Donohue and I would be inteested in obtaining:

the county spillover data
the 10 state permit data
tje city level data
the fee and training data
the 2SLS instrument data

You have been incredibly generous with us and others already (and Levitt and I have not responded in kind) so I will understand if you are to busy to provide any of the foregoing pieces of info. Even putting aside issues of confidentiality, it takes a lot of time to dig up and present data in a usable way. So don't worry if you can't do this. And either way, Donohue and I will give you all of the do files and datasets for the regressions that we report in our paper.

It is amazing the number of hours that Donohue and I have put into this and it can only be a fraction of what you have invested.

Professor Ian Ayres
William K. Townsend Professor of Law
Yale Law School
PO Box 208215
New Haven, CT 06520
203.432.7101 (voice)
203.432.4769 (fax)
ian.ayres@yale.edu
www.law.yale.edu (publications)

Note: 1) I contacted the company (Lojack) that had provided the data to Ayres and Levitt, but the company informed me that they had been asked not to give out the data to other researchers. Levitt told me that this statement was wrong and that the company probably told me this to keep me from bothering the company for the data. 2) I believe that Ayres was the person who contacted me about providing more data because Donohue's record of sharing data was much worse than Ayres. Shortly before this Donohue and Levitt had refused to respond to e-mails where I had asked them for explanations of how they had constructed some crucial parts of their abortion study data. Of course, it took 19 months for Donohue and Levitt to provide any of the data for that paper and even then it was only a partial set of the data. Besides John Whitley not being able to replicate some of their data in the Donohue and Levitt abortion study (possibly there was a mistake, but neither Donohue nor Levitt would answer the questions that were sent them and help us figure out what was happening here), Donohue and Levitt also had the problem that they assumed that no legal abortions took place prior to Roe (with the exception of five states). The problem is that abortions were allowed when the life or health of the mother was considered to be endangered and that could be interpreted quite broadly.