A flare up occurred in December about the integrity of Thomson’s IBIS database of analyst recommendations.

Some researchers got a feed covering analyst recommendations from 1993 to 2002 in 2002 and then got the same feed again in 2004 (i.e., covering the same 1993-2002 period). They discovered that almost 10% of the records had been changed in non random ways (i.e., analyst’s names were taken off of many bad calls).

You can find the paper at

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=889322

And various discussions here

Thomson’s response to date, was that the researchers received an incomplete feed and that the data was never changed. Unfortunately, since this is a private, expensive database, the number of people who have access who could see whether they too received “incomplete” feeds is small. So I don’t know whether this will ever get resolved. But it does put in quesetion for me, the validity of any research that is based on the IBIS analyst data.