ABSTRACT
To recapitulate, this book began by pointing out that people sometimes have problems when solving statistical tasks. Then, four explanations for why this is so were introduced-the pragmatic-implications, heuristics-andbiases, abstractrules, and adaptive-algorithms approaches. These theoretical approaches allow us to predict whether and how people’s problems with statistical tasks can be remedied by suitable training programs. Using the results of training programs to evaluate theories is a novel methodological procedure that enabled a direct comparison of the four approaches, an endeavor that had not yet been attempted, probably because these approaches focus on different aspects of statistical reasoning. The pragmatic-implications approach examines differences between natural language and probability language, the heuristics-and-biases approach focuses on the demonstration of judgmental errors, the abstract-rules approach concentrates on training, and the adaptive-algorithms approach explores conditions under which the mind can be expected to solve statistical tasks without problems.
