It wasn’t ‘leaked,’ but presented at a conference, according to both the post article and the snippet that the blog post included.

I read another link that said they had not yet submitted this study to peer review, but like their others ones, will do so. I can’t find that link now.

I looked at the survey regarding race and guns or other dangerous weapons. The experiment shows you pictures, and asks you to select one choice or the other, quickly, in response to the photos. One item I found was that, the way the sequence was setup, when the task was to select non-threatening and blacks, the ordering was changed from the previous two of the four test sequences. What this leads to is that, as might be expected, the first time the ordering was changed, there is a tendency to automatically push the same key you used for the previous two tests, even though that was “wrong.” It was easy to automatically press the wrong key, which in this case leads to the impression that one associates dangerous items with blacks more than with whites.

This was only one test, but in order to take it you had to identify a political affiliation, which may have impacted the order and precedence of the data.

Right now, I’m dubious at best. Plus, psychological and sociological testing relies on so many assumptions and is based on such fuzzy, difficult to quantify metrics, it’s almost impossible to take most of it seriously. Within the realm of the “hard” sciences, the sociologists and psychologists are not that well respected.

Be aware that skewing data doesn’t have to be deliberate, the same subconscious bias they are looking for can, and often does, easily influence the researcher.

They could just be making everything up. But you think they skewed the data?

“The study does not give even remotely enough information about their methodology and data sets to allow anyone to tell whether or not there was pollution of the dataset, or misanalysis of the result, which in itself says something. ”

You can read about their methodology at http://implicit.harvard.edu They also there link to their published papers. This study is currently unpublished.

Let me postulate an example. I want to “prove” that conservatives are biased against minorities. I skew the dataset so that conservatives are given pictures of aggressive looking minorities, while liberals are given pictures that are calm and smiling. That’s one way to skew data.

The study does not give even remotely enough information about their methodology and data sets to allow anyone to tell whether or not there was pollution of the dataset, or misanalysis of the result, which in itself says something. Quite often, lack of information means that that info will tell you something the study proponents don’t want you to know.

What facts did they ignore? And what is their ‘theory’? As far as I saw, there was no theory, just correlations.

The study in question has not been peer reviewed, nor do I think it would pass an honest peer review. In case you don’t know, peer review is not an ideological litmus test, at least it’s not supposed to be. It’s a review of the basic scientific methodology and an examination of the mathematics/statistical analysis used, to make sure there are no errors, and that the experiment is repeatable.

Scientists lie and distort all the time, sad to say. I was taught that this is tantamount to original sin when I was in school, but sadly more and more scientists are doing it to get funding, make a point, support an idea they believe in.

were you talking about the political views of the washington post or the political views of the scientists?

I’m talking about the scientists. Why do their views matter? they just ran correlations on data.

Case in point, local paper has article of local reacton to the SOTU on top of front page. Only picture is that of the local Democrat Central Committee chairman. No GOP officials pictured. Only democrats are written about on front page. After the jump, then 1 very junior level Republican is written about and his views questioned by the reporter where as all the democrats’ views were reported unchallenged and unedited.

Which of their articles have their political views in them? Every one? Can you list a few and show me how?

Did they do that? They just ran some correlations on data.

Why do their political views matter?