The Harvard article probably uses the National Crime Victimization Survey NCVS. Cook and Ludwig, who have published studies showing 108 thousand DGU from the NCVS and 4.7 million DGU from the NSPOF (National Survey on Private Ownership and Use of Firearms), pointed out that everyone in the NSPOF survey sample were asked if they had used a gun defensively, but very few of the NCVS survey sample are ever asked a DGU question.
Personally I think Lott has been swiftboated because a lot of people did not like his conclusions. Or that his facts contradict a precious political Tru[sup]th[/sup].
Awhile back at Amazon I commented on a review of Lott's book on court selections where the critic mostly talked about Lott's gun research using hackneyed talking points. (I use a pseudonym at Amazon because I think the focus should be on the issue not on the person.) Rather than answer my arguments, the critic snarked "You're right, Mary, er, John, ..." and spendt most of the discussion badgering me for my true ID, rather than debating the issues, and went on his personal blog to accuse me of being John Lott. I have also been accused of being other people by another person who preferred to attack me personally rather than answer my criticism rationally. Which is why a lot of people, not just John Lott, have used pseudonyms to argue on the internet (especially at Amazon).
Lott's two self-defense surveys 1996 and 2002 are actually middle-of-the-road in their conclusions. As Kleck & Gertz pointed out about their 1994 NSDS National Self-Defense Survey that got 2.4 million Defensive Gun Uses (DGU) per year, at that time there were a dozen national and three state surveys that got 764 thousand to 3.6 million DGU per year. Lott's surveys confirmed about 2 million DGU per year with the vast majority non-shooting.
Peter Hart research associates did a 1981 survey on guns that got DGU figures very similar to Lott's. By the time Gary Kleck asked Peter Hart about the survey for his 1988 Social Problems article, Hart no longer had the survey data. That was one of the reasons Kleck & Gertz later did their NSDS study. Hart's reputation has not been trashed for losing track of a survey data set after the results were published.
Lott's credibility is attacked because he lost the 1996 DGU survey data. A half-a-dozen academics did confirm that Lott had a massive data loss involving work coauthored with them in the summer of 1997 and he had to recover their shared data from their backups. Apparently Lott considered an external harddrive to be a permanant backup and since that 1997 mishap has adopted a multiple backup policy. Lost data was the one survey for the 2 million DGU figure and an draft article where the co-author did not have backups either. Not lost was the larger data set of the Right-To-Carry concealed weapons econometric regression (co-author David Mustard had his own copies which he supplied to Lott), the core of "More Guns, Less Crime" and the data set that Lott has shared with anyone who wants to download it (and it is massive).