I wouldn't say so. They did find high statistical significance (p=0.0004) in the first test, and then it's very legitimate to explore subsets to find out what drives this. (But in general there appears to be plenty of p-hacking here)Isn’t that classic p-hacking? Our first hypothesis didn’t significantly provide results since some cells didn’t fragment so we narrowed our search window and now the women’s group is extremely significant with all mitochondrial fragmentation.