Irregularities in LaCour (2014) [pdf]

stanford.edu

1 pointscott_s3 comments

This is perhaps a confusing title, because the "(2014)" is part of it, as it is a reference to a paper.

I'm posting this because many of us here will have heard the media reports about this study's retraction. (NYT: http://www.nytimes.com/2015/05/26/science/maligned-study-on-...; This American Life: http://www.thisamericanlife.org/blog/2015/05/canvassers-stud...). But this is the statistical detective work that other researchers did to convince themselves and one of the original study's coauthors that the data is fabricated. Their methods are clear, showing R snippets, graphs, code comments explaining where to find the data sets, and explanations of the significance of each step. I am impressed with this report.

By the way, the author who likely fabricated his results (which is a conclusion I feel confident in, based on this report), implied this report is not to be trusted because it is not peer-reviewed. (http://www.mikelacour.com/5yxcfv8vmowjjyimtqkfb65i7owz7y) I find this position funny, because while this report has not gone through the peer-review process, it is thorough, and I see no room for them to hide any tricks. If there are problems with the report, we will know soon. But I doubt there are.

Discussed a few days ago: https://news.ycombinator.com/item?id=9579881.

Ah, I missed that. I really wanted to point out the report itself, though, not just the conclusions. It's a great example of statistical detective work, independent of the larger issues going on (academic honest, peer-review, etc).

Reply on news.ycombinator.com