Hacker News Reader

Measure five different things. Use underpowered design. Compare everything to everything using many fancy-sounding statisticsl tests. Fish for p<0.05 (and the occasional "marginally significant" p=0.07).

Remember, most psychology research cannot be reproduced: http://m.sciencemag.org/content/349/6251/aac4716

View on HN

We have simply got to remind our friends here on Hacker News that university press releases are almost never a good guide to what research shows, but rather are part of the science hype cycle that confuses the public about what research shows.[1]

[1] http://www.phdcomics.com/comics.php?n=1174

It's good to be critical, but you need to evaluate on a case-by-case basis or you risk never learning anything new.

I did. I looked at the paper and my description of it is accurate, including the reported p=0.07.

Because of the underpowered design, we learned nothing. Underpowered studies are worse than nothing: http://www.wiringthebrain.com/2015/12/on-literature-pollutio...

Reply on news.ycombinator.com