Are you a statistician? I'm not.
If you have a normal distribution in a population, that may seem to be a tiny sample, but if there is a biphasic distribution according to gender, the results seem intriguing at the least.
Revenge 'more satisfying for men'
When the "fair" players were shocked, both female and male volunteers showed increased activity in the pain-related centres of the pain - the fronto-insular and anterior cingulate cortices.
When the "unfair" actor received a shock, the women taking part in the experiment showed a similar empathy with them.
In contrast, the male volunteers showed no increased activity in the empathy-related pain areas.
They did, however, show a surge of activity in the reward centre of the brain - the nucleus accumbens.
P.S. When I worked in Quality Control for Wella AG, a medium sized, hair care products company two decades ago, filling weights were usually checked with two dozen samples for each lot whose usual total daily production numbered about 3,000 - 5,000 units.
The reference to the actual article is Nature (DOI: 101038/nature04271). I'm going to the library to see if I can locate the abstract and an associated P value.
Actually I have a masters degree in math/statistics. 32 subjects/ presumably 16 male, 16 females in a paired t-test is a VERY small sample. Like I said, earlier, details of the study should have been revealed before making general statements like this author did. Even though the results were to be published in a peer reviewed journal such as Nature, how do we know this author is reporting the results correctly?
fyi, sample sizes for maintaining Quality Control or SPC (Statistical Process Control) can be different than a required sample size for hypothesis testing especially when inferences such "men are hungrier for revenge" are involved. In reality, in QC, cost is often a factor in determining sample size. In hypothesis testing, statisitical significance is often the determinant.
Well thirty years ago I taught stat and studied it fairly extensively. No study would be considered valid with such a small sample. There is simply too much chance that you would not have a random selection with such a small sample.
The test for filing levels is not the same kind of thing.
I am not saying the conclusion is wrong just that it is not based upon a sufficiently large sample.