Hello,
I personally stick with standard deviation unless determining reproducibility. The misuse of SEM has been mentioned repeatedly in several scientific articles. However, there's an open access journal (with impact factor) requiring statistics as in the example bellow. Computing one value of SEM for all the experimental groups, like all the means come from the same population. In that case I would expect that the higher SEM, the lower p value (the bigger difference between the experimental groups). However, it does not seem to work so (see the example bellow with very similar mean values in each table line).
Could anyone explain me please how does it work and what sense there is in using the SEM this way?
Thanks in advance.