Judgment in forecasting is claimed to emanate from an underlying reasoning process of constructing arguments. To understand the reasoning upon which judgments are based, a theoretically-grounded typology of arguments is presented that organizes argument types according to the knowledge that is used to warrant movement from evidence to claim. The chapter then focuses on arguments that are based on the direct application of category knowledge. Of particular interest are the distributional characteristics of people's category knowledge, which have impacts on the way that reasoning is performed. To explore these impacts, the chapter develops a theoretical understanding of the roles of typicality and variability in probability assessment. Perceptions of variability--within and between categories--have been generally neglected by researchers. Such perceptions are expected to have impacts on the selection and use of evidence for probability forecasting and decision making, and these expected impacts are expressed in a series of propositions.