I’ve found many errors in statistical methodology in published journals. The greatest error is not in the statistics, but in biased sample selection.
This study was studying the bias itself. It may not be precise, but then, psychology is essentially unquantifiable. If the "show of hands" cited is any example, there is a lot of wiggle room in their findings.
you also have “operative definitions” which taint the sample.
ie when the libs wanted to make all video games cause violence. They defined a violent act as “pressing a button” regardless of the effect caused. (reset, jump, pause etc)
So now the psychologists just define violent as having “conservative thoughts” and presto no sane person is conservative.
I still like the old study that found 45% of doing ANYTHING would have the same result as psychological treatment. (or that no training was required to interprit ink blot studies)
And those are called "Presidential tracking polls".
Regards,
GtG