Clyde Hamilton Coombs spent most of his career at the University of Michigan, where he was a founding figure in mathematical psychology. His work was distinguished by deep concern for the logical structure of psychological data. His 1964 book A Theory of Data remains one of the most original contributions to the philosophy of psychological measurement.
Unfolding Theory
iff |x_i - y_j| < |x_i - y_k|
x_i = ideal point of person i
y_j, y_k = scale positions of stimuli j, k
Coombs' unfolding theory models preference as proximity to an ideal point. Each person has an ideal point representing their most preferred stimulus level, and preference decreases with distance from that ideal. The resulting preference order is a "folded" version of the underlying stimulus scale, and the measurement task is to "unfold" these orderings to recover both the stimulus scale and individual ideal points.
Coombs classified psychological data by the nature of observations (dominance vs. proximity) and entities compared (stimuli, individuals, or both). This taxonomy clarified that different data types require different scaling models, anticipating modern concerns about measurement validity and model selection.
Contributions to Decision Theory
Coombs contributed a portfolio theory of risky decision making, proposing that people have ideal levels of risk rather than simply maximizing expected utility. He co-authored with Dawes and Tversky the influential textbook Mathematical Psychology: An Elementary Introduction and trained numerous students who became leaders in the field.
Legacy and Impact
Unfolding theory has found applications in political science (ideal-point models of legislative voting), marketing (preference mapping), and psychometrics (unfolding IRT models). Coombs' emphasis on understanding data structure before computing remains a model of scientific rigor.