If science depends on observation of empirical data, how can social scientists study non observable phenomena, such as though patterns or imagery?

Have questionnaires for subjects to explore these phenomena or have behavioral criteria that allow one to infer these phenomena.