zlacker

The issue is that other sciences are not held back by these problems. We still have bridges, semiconductors, space flight and new cancer treatments. This is partly because it is much easier to measure and verify experimental phenomena (you call this a bias, but it is merely a fact?), but also because being correct matters. However many crappy experiments are done or results fabricated, at the end of the day you still have to put your rocket on the launch pad and fire the engines. The reproducibility issues in these sciences is a matter of efficiency.

Psychology is different in the sense that it doesn't seem to matter to the field very much whether they are correct. I say this because they keep doing experiments which lack a robust design, and then proceed to use complex statistical models to infer many unsupported claims which never seem to replicate. Then they meta-analyse these results to conclude that if there is an effect it is small or only applies to certain people. The 'power pose' study I linked is a case in point. All this is done in a very rigorous way by rigorous people using advanced equipment and statistics, who spend much time spinning careful narratives about their work in long Discussion sections. But you don't get a medal for trying in science.

It is hardly a compliment to psychology that they gave birth to the meta-analysis [1]. Note that getting some result in an experiment, then having the result overturned in a meta-analysis is the functional equivalent of not having done the experiment at all. In fact, it is inferior to doing nothing, because you have just wasted everyone's time.

[1] https://www.sciencemag.org/news/2018/09/meta-analyses-were-s...