4 ms·
Perhaps I am a bit dense but I still don't see why small sample sizes aren't problematic. If the data source we tap is valid (i.e.: it does give us the informa
by cJ0th 9y ago
Perhaps I am a bit dense but I still don't see why small sample sizes aren't problematic.
If the data source we tap is valid (i.e.: it does give us the information that we really need) then we need a big sample size to gain precision. If, otoh, the data is not valid in the first place then neither a small nor a big sample size are helpful.
- jonex 9y agoSample size is relevant to which opinions to form. But I think the GP wanted to point out that just having a small sample size does not invalidate the results of a study given equal confidence scores. It's a common misconception that the conclusions of a large study always are more trustworthy than from a small study. A larger study will however be able to detect smaller effects and will, as you mention, provide more precision.
- Obi_Juan_Kenobi 9y agoThe issue is effect size. If the effect is large or easily detected, then you don't need many individuals to determine significant differences in the broader population. What you seem to be referring to is whether or not a sample is representative of a population, which is indeed a critical factor in statistics, but not really related to the issue here.