The underlying idea is simple but powerful. If we are trying to explain some phenomenon, X, then we need to identify variations in the likelihood of X or the rate of X, and look for potential causes that (1) vary across the relevant circumstances in a way that could explain X and (2) that we can connect to the outcomes for X in some way. For example, with the gender distinctive clothing question, some ways to better specify the question and look at it through comparisons are:
Natural supplements won't hurt you. They can only help. They won't make chemotherapy or radiation therapy less effective. Instead they amplify the power of whatever cancer treatments you may be doing by attacking cancer in other ways, or supporting the health of the body and immune system. For example, research studies have shown that when you oxygenate cancer cells, radiation therapy is more effective at killing those cells. And beta glucan, when used in conjunction with some types of chemotherapy, produced marked improvement over chemotherapy alone.