Genetically diverse vs. genetically identical model pools – we massively overweight findings due to identical models.
Control group fuckery – Most control groups are segregated by “have trait” or “don’t” which requires pre-sorting and biasing. If trait exists, it should be visible in control group and it’s visibility in the control group helps validate the trait group. Far too much manipulation of control vs. test for statistical purposes and RCT only obviates a portion of this.
Lack of naturalistic research deprives us of interaction contexts.
Far too much research based around a desired goal.
Inadequate verification of prior results.
Myopic research goals.
Open Science in research assessment. An overview of quantitative and qualitative approaches