Summary: Lott now claims that an incriminating file where he had been caught cooking his results was not meant to have been on his website and was only there because his webmaster screwed up. Unfortunately, his latest story is full of holes.
Way back in September last year I detailed how, after Ayres and Donohue showed that correcting Lott's coding errors made his results go away, Lott changed his model to bring his results back. Then when I asked him questions about the changed model, he tried to cover up the change by replacing the file at johnlott.org which contained the changed model with a new, different one. And in what appeared to be a botched attempt to rewrite history and backdate the replacement, the date on the new file was set in the future. Lott's response to all the questions about the changes to his model was to duck the most important one (why change the model?) and focus on the date on the file from the future. He dismissed the idea that he could have been trying to backdate the replacement as a "conspiracy theory" and argued that because he used a Mac he could not have messed up when changing the date on the file. This argument was met with wide derision and Lott soon removed it from his blog. Over a month later he replaced it with a new explanation: That incriminating file? It was posted accidently and wasn't meant to be there. The new file that was created in September? That was supposed to have been there from May. Yes, after calling the idea that he was trying to backdate the new file a "conspiracy theory" his latest explanation involves backdating the new file. Read on for all the details.
On Oct 19 Lott posted this explanation from Jeff Koch, the webmaster for johnlott.org:
I'm hesitant to enter this debate with your critics. But since I am the one who screwed things up, I guess the least I can do is to explain what happened. No doubt, the scales will fall from the eyes of your critics after they read this.
On August 20, at 5:23 p.m., I sent you an e-mail telling you that I had added, at your request, a link, "Statement regarding the data". This was located just above the link "Confirming Figures and Tables 1 * 5.doc". However, I was away from my "home" computer when I made this revision, and so rather than using my normal web-editing tool, I just used the little editor that is built into my browser, and made the change manually.
On August 31, at 5:30 a.m., you forwarded an e-mail to me, saying that some fellow named Mooney could not download the "Tables 1 *5" file. Apparently, in the process of editing the page to add the link, "Statement regarding the data", I broke the link to "Confirming Figures and Tables 1 * 5.doc"
I was unavailable to fix the broken link right away, so on August 31 you added a link pointing to your JohnRLott.com site and offered the "Tables 1-5.doc" file there.
It appears that I fixed the link to the "Tables 1-5.doc" file on September 1, 12:03 p.m. However, apparently when I did this, I clicked on an old copy of the file that you had corrected last spring. I still had it floating around on my computer, and didn't pay attention to what I was doing, and so clicked on the wrong file name.
Apparently, you noticed my error almost immediately; because you sent me an e-mil at 7:20 a.m. on September 2, with a new copy of the file, asking that I please correct it. It looks like I uploaded the corrected file on September 2 at 9:47 a.m.
So it appears that I was distributing the old, uncorrected file, for less than 24 hours. Does that sound right? When I started looking into this, I thought I'd served the wrong file for a couple weeks. But it looks like it was less than a day.
If someone has any questions, I can be reached at webmaster@JohnLott.org.
There are severe problems with Koch's story. The "corrected" file, which was supposed to have been present before Sep 1, was actually not created till Sep 2. He wasn't serving the "wrong" file for a day or even for two weeks. David Powell downloaded it in May. I downloaded it in June. Not only did Chris Mooney download it in August, but Lott emailed him a copy of the same file. From Mooney's article:
On the website, Lott claimed the "corrected" table used "clustering," when it did not. In a heated interview on August 19 (transcript), Lott said this labeling claim must be an error. But the very next day, he e-mailed a file containing precisely the same table, claiming that all the tables on his website were "clearly and properly labeled."
Both Chris Mooney and I emailed Koch to ask for an explanation. The only immediate response was that the invitation to send him questions was removed (you can see the current version on Lott's site here), but eventually, weeks later, Koch responded. He asked that his email not be posted, so I'll summarize it instead:
Koch conceded that his explanation was incorrect, but had no explanation for the fact that the "wrong" file had been served up since May. He conjectured that Lott had emailed the "wrong" file to Mooney because Lott had coincidentally also accidently clicked on the wrong file. He said that he thought that the "wrong" file might have been been created by a research assistant of Lott's called James. He stated that as well as the "wrong" file (dated May 9) and the current file (dated Sep 2) there was a third file dated May 1 and that the Sep 2 file was just the May 1 file converted from Word to Acrobat. This May 1 file was the one that was supposed to have been available on the web site the whole time.
I asked him to send me a copy of this May 1 file, since no-one had ever seen it. He did not respond. Nor did he respond to repeated requests for the file. And even though he conceded that his explanation was incorrect, no correction has been made.
Some readers might be wondering if there really is a Jeff Koch or whether he is yet another one of Lott's Internet personas. A whois search revealed that johnlott.org was registered to Jeff Koch and gave a Texas phone number. Calling the number got an answering machine with a message from Koch, so I think he is a real person.
Could the "wrong" file have been posted by mistake? Well, since it was the one at johnlott.org since May, the only way this could have happened is if Lott sent Koch the wrong file in the first place perhaps because he thought it contained the version of the table that is currently available on his web site. This could also explain why he sent that same file to Chris Mooney in August. Unfortunately, this explanation founders on Lott's own statement about what the table shows. On August 20, he wrote:
As to the claim that "correcting his errors did eliminate his finding." The data used in the Plassmann and Whitley paper can be downloaded at www.johnlott.org and one can readily see from the corrected tables and figures that while there are some changes in the results, this statement is false.
Lott can't be referring to the table that he currently has on his site, because that table uses his original, uncorrected data. The only table that he has produced using the corrected data is the one he is now trying to write out of history.
So what really happened? There isn't any need to invoke a conspiracy between Lott and Koch. Lott needed to pretend that the incriminating table never existed, so he just told Koch that Koch had put the wrong file up at johnlott.org. Koch took Lott's word that it was his fault (notice how in his explanation Koch assumes that it was his fault and he tries to deduce when he made the mistake), and took the blame for the "error", when really it was just another one of Lott's attempts to rewrite history.
Thanks for these detailed reports on John Lott. You do us readers a great service and we appreciate your efforts.