Date of Award

2014

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Applied Cognitive Science and Human Factors (PhD)

College, School or Department Name

Department of Cognitive and Learning Sciences

Advisor

Edward T. Cokely

Co-Advisor

Rocio Garcia-Retamero

Abstract

Numeracy—i.e., one’s practical understanding of mathematics in context—is one of the strongest predictors of people’s general decision making skill, independent of other cognitive abilities (e.g., intelligence, working memory, attentional control). Despite notable scientific progress on the nature of numeracy and decision making, the cognitive and decision sciences have yet to investigate individual differences in numeracy components (e.g., algebra versus probability). In this dissertation, I report on my efforts to develop new measurement technology and quantitative models of cognitive and decision skills. Analyses include the first known investigations of the relations between the major adult component numeracy skills and general decision making skills. Specifically, Study 1 built on an established theoretical framework from adult education and used a two-parameter logistic IRT model to create the Berlin Numeracy Components Test (BNT-C) for college educated samples. Behavioral and analytic results indicated that the test efficiently measured full-scale adult numeracy and component numeracy skills (i.e. operations, probability, geometry, and algebra), with superior psychometric performance (e.g., difficulty, discriminability, and sensitivity). In Study 2, I investigated the links between the BNT-C, other numeracy tests, general cognitive abilities (e.g., intelligence, cognitive impulsivity), and general decision making skills (e.g., Adult Decision Making Competencies, Risk Literacy). Predictive modeling of behavioral data revealed that with few exceptions the BNT-C explained all types of decision skill better than any other individual ability assessment (e.g., intelligence v. impulsivity v. other numeracy tests). The BNT-C additionally outperformed the optimal linear combination of all combined ability tests when predicting overall general decision making skill, a finding that is consistent with a causal cognitive account of the relations between numeracy and general decision making skill. In accord with leading theory, component analyses indicated that operations and probability skills were robustly and uniquely tied to risk literacy and to nearly all general decision making sub-skill competencies independent of the influence of other cognitive abilities (e.g., intelligence). Analyses also provided the first evidence that algebra and geometry sub-skills are uniquely linked to some essential general decision skills in educated adults (e.g., confidence calibration, sunk costs, ratio bias). Discussion focuses on theoretical implications and factor analytic modeling of the relations between numeracy, its component skills, and superior decision making. Adaptive test construction and potential applications in training and personalized decision support are also briefly discussed.

Share

COinS