General Discussion
In reply to the discussion: Body Language Expert Dissects Iconic Pelosi/Trump Photo [View all]ECSkeptic
(62 posts)Statistical tests are designed to avoid bias; experimental designs are carefully structured and documented to allow other researchers to replicate the results and check them for bias.
When you're dealing with observer judgments of human behavior, you have to be particularly careful about observer bias. I once worked as a lab assistant in a set of studies that relied upon listeners' judgment of the mental state of speakers who were speaking under specific conditions. We had to control for listeners' ability to hear and understand spoken English, we had to carefully control the conditions under which we made recordings of our speakers, we controlled how the speech was presented to listeners, and we had to ensure that all speakers were saying the same things under the same conditions. And it was absolutely critical that the listeners didn't know which group of speakers was which (control vs. experimental). If the listeners knew what group was what, that would make the resulting data useless.
Similarly, the analyst here already knew much of the result of this meeting in the White House. It's nearly impossible to screen out the bias inherent in the analysis, because much of it is unconscious; the analyst simply isn't aware of it. That's how brains work.