3 ms·
As long as the statistics are good, any sample size suffices. Even n=1 can give a statistically relevant result (for the hypothesis that a given event never ha
by Someone 5y ago
As long as the statistics are good, any sample size suffices.
Even n=1 can give a statistically relevant result (for the hypothesis that a given event never happens)
I haven’t read the paper, but n=51 can be more than high enough to get significant results for more realistic null hypotheses, if the effect is large enough.
Having said that, effects rarely are large. If you have a weaker effect, you need more subjects (picking a somewhat random paper from a google search: https://www.journalofcognition.org/articles/10.5334/joc.72/ https://www.journalofcognition.org/articles/10.5334/joc.72/)
I’m surprised by the fraction of male subjects they managed to recruit. A well-known joke is that the typical experiment in psychology is either done with mice or with teenage girls as participants.