Michael LaCour, the UCLA graduate student at the center of an epically bizarre science scandal, apparently made up data in a different study about political bias in the media.
This study made the controversial claim that people don't have a particularly biased media diet.
Gregory Martin, a political scientist at Emory University, published a report Thursday outlining several concerns about LaCour's study, which was unpublished but frequently cited at scientific conferences.
According to Martin's analyses, LaCour fudged his data to make it look much cleaner. LaCour also claimed that seven news shows were included in an archive of show transcripts that, in fact, were not.
"He was very bold," Martin told BuzzFeed News. "These shows just don't exist in the database, and you don't need a fancy statistical analysis to see that."
Martin also studies political bias in the media, and has used one of the methods that LaCour described in his study. That method estimates the ideology of news shows by comparing certain phrases they use to phrases used by members of Congress. It's a fairly crude method. "It generally puts Fox News to the right of MSNBC," Martin said, but isn't more accurate than that.
In LaCour's paper, however, this same method led to very precise ideological estimates. So nearly a year ago, Martin's team reached out to LaCour to try to find out why the method was working so much better for him.
LaCour gave Martin access to his code, but Martin couldn't get the code to work. So Martin wrote his own version of the model, based on the description LaCour had given in his paper. When he did this, the ideological analysis gave him the same crude results that he had always gotten, and nothing like the clean data LaCour had reported in his paper.
"My analysis just shows that the method he purports to have used to do this identification is not anywhere near as precise as he claims it is," Martin said.
At the time, Martin gave LaCour the benefit of the doubt, figuring there must have been some discrepancy in the underlying datasets that they used to write their respective codes. But after hearing about the fraud allegations of his same-sex marriage paper last week, Martin thought it was useful to publish these irregularities.
"This is a pretty influential paper that a lot of people have seen," Martin said. But given these troubling findings, he added, "this is probably also fake."
On Thursday afternoon Science officially retracted LaCour's 2014 study about swaying voters on same-sex marriage, following two weeks of fraud allegations by LaCour's co-author and others.
This flurry of allegations also calls into question LaCour's unpublished study of abortion views, which claimed that voters might be persuaded on reproductive rights issues after having a conversation with a woman who revealed she had had an abortion. BuzzFeed News has reached out to several social scientists and statistical experts who say there is not enough information in that study to determine whether the data have irregularities.
At this point, other political scientists are not surprised to hear of problems in LaCour's work.
"I would say, without having a degree in psychology or the medical field, that this is borderline pathological," Mark Major, associate director of the McCourtney Institute for Democracy of Pennsylvania State University, told BuzzFeed News. "And the hubris behind this, too, of being able to fabricate — and not just outright falsify data, but also make it so public. That's what I find so strange."
BuzzFeed News has reached out to LaCour for comment. He has previously said that he will provide a "definitive response" on or before May 29.