3 ms·
If I understood it right they checked that against two groups of 20 people (40 in total only). The study itself seems super interesting and important but such r
by caiobegotti 6y ago
If I understood it right they checked that against two groups of 20 people (40 in total only). The study itself seems super interesting and important but such ridiculously small sample size to me, a layman following COVID-19-related research. Is 40 acceptable in this case? Since they are on Cell.com I suppose it is. Bad sample size in COVID-19 research has been discussed recently already and it is a known problem. Another thing is that the sample size was a hidden lead in page 3 and a figure shown only in page 11 of the paper. I wish scientific papers had something of a "nutrition facts" label with basic info like that in their first page.
- ramraj07 6y agoThis is not a clinical trial though, merely a mechanistic study. The types of assays they have done typically take years to get right with just a handful of samples so it's actually quite impressive. Importantly this is a T cell study not exactly an antibody assay study (t cell responses are a prerequisite for antibody responses).
- Someone 6y agoYes, the paper could’ve been written better. The normal presentation order is abstract-introduction-methods-results-discussion. Here, they mix the methods and the results/hide the methods inside the results. I don’t understand the paper well enough to even try to check the statistics (if there is any; I didn’t see clear claims, so this may just be a ‘here is what we observed’ paper), but you can’t disqualify a paper on sample size alone. 20 can be large enough if the effect is large enough. For example, if you do heads/tails 20 times with a fair coin, the probability to get more than 15 heads is less than 1%, so if you get, say, 17, the hypothesis that the coin is fair can be rejected.