Class analysis is not useful because it hides nuance which is usually crucially important to the issue that one is attempting to understand. Look at things in terms of populations / demographics is obviously a very useful technique, but class analysis as usually performed isn't nearly granular enough to get at the truth.
For example, saying "men, as a class, are more violent than women" is true. It's also close to useless at giving any insight, and leads to inaccurate generalisation and misdiagnosis of the actual issues.
Aggression, as a psychological character trait, is not too different on average between men and women. Men are slightly more aggressive on average. But the two populations overlap hugely - so much so that basing any policy or analysis on "men vs women" is meaningless and misleading.
At the tail ends of the distribution, however, there is enormous difference. The most violent people are almost always men. The distribution of aggression among the male population is hugely variable.
This is a subtle, but crucially important distinction, because the question becomes not "why are men more violent than women" but "why are a small proportion of men way more violent than everyone else".
Class analysis is fine when it tries to incorporate the required level of detail and granularity in its workings. But most people who wield it don't, because they're not interested in truth, only in narrative.