you can do whatever you like. Whether your results will persuade people is the question ie whether it's seen as data dredging, or labelled as 'exploratory', which is not to discount the results entirely. Just be honest in describing all of the analyses performed and not just those that turned out to be statistically significant. You need to define the purpose of your analysis upfront, and how the subgroup are defined. Etc. Incidentally, it would make more sense to post this question eg here: https://stats.stackexchange.com/