Meta-analysis finds little evidence for the Macbeth effect

digest.bps.org.uk

25 points·laurex··8 comments

And the Replication crisis continues... [1]

Seems like the big secret is small initial sample sizes. For example, the original three Macbeth studies: 60, 27, 32. Later tried with N=153, 148, 266, 210, etc. and surprise surprise...

Another example, the "Power posing will make you act bolder" study, originally with only N=42. Later tried with N=200 by researchers at the Univirsity of Zurich, and the effect disappeared.

And the list goes on [2]

[1] https://en.wikipedia.org/wiki/Replication_crisis [2] https://digest.bps.org.uk/2016/09/16/ten-famous-psychology-f...

Is there a place where major scientific findings are searchable to easily see the level of evidence supporting, whether they have been replicated, etc?

Examine.com does a good job for nutrition/supplements Example: https://examine.com/supplements/melatonin/

I wish there were something approachable like that for other fields.

For health, the Cochrane Library is a pretty good start: https://www.cochranelibrary.com/cdsr/about-cdsr

No, the best you can do is a find a recent review article.

Only rarely will you find direct replications in the case of medical and social research though, so as a heuristic you can just assume the evidence for everything is weak.

Also, this reminds me of that paper where they "discovered" that if you spray fart smell in a room that people are more disgusted by the thought of a guy rubbing "his bare genitals along the kitten’s body". https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2562923/

I wish that rather than being shown alphabetically, the image would show a funnel plot (https://en.wikipedia.org/wiki/Funnel_plot)

Deep down, I think I've always known it. There's no way the whole theater will come down just because you mention the play.

*her

Reply on news.ycombinator.com