This paper examines critically four habits of social scientists: a reliance on constructs for processes that do not specify the class of agent or the context of action, skepticism toward empirical truths that are inconsistent with political or social goals, a reluctance to use profiles or characteristics that have different weights to classify persons, and an aversion to treating subjects with extreme values as representing special categories. These points are made in the context of the author's research on temperaments in children.