@sivers one strategy I found to work well is to ask lots of people, then compile what was similar and different, the outliers and norms