engleski

Measuring rationality: The relationship between the modified CRT and reasoning task performance

The common version of the Cognitive Reflection Test (CRT) contains three problems which fit into the classical dual-process approach to thinking which posits two types of processing. Type 1 which is based on heuristics, and Type 2 which is based on analytical thinking. The CRT is designed to distinguish between participants according to the ability to overcome heuristic, and activate analytical thinking. Items from the CRT have become widely used and increasingly familiar among prospective participants which increases the demand for alternative forms based on similar principles. Thus, we modified the method (multiple choice with a clear heuristic choice available) while also changing the contents. The aim was to relate performance on this alternative CRT to other reasoning tasks in order to determine whether it captures similar processes. Participants (N = 590) had to solve a test with four novel CRT tasks, alongside three reasoning tasks: the Linda problem, the Base rate neglect task and the Covariation detection task. Five groups of participants were formed based on the number of correct CRT responses (0-4). One-way ANOVAs were conducted to determine whether reasoning task performance changed as a function of CRT performance. Results show significantly better reasoning performance as CRT performance increased. In the Linda task this manifested as less representativeness bias (F(4, 582) = 9.54, p< .01, ηp2= .06). In the Base rate neglect task we observed less stereotype-based thinking (F(4, 582) = 7.35, p< .01, ηp2= .05) and in the Covariation detection task it was reflected in less bias towards absolute values compared to ratios (F(4, 582) = 8.29, p< .01, ηp2= .05). An additional analysis grouped the participants according to whether they were mostly responding according to heuristics (3 or more heuristic-based responses) or analytics (3 or more analytic-based responses) in the CRT, creating more extreme groups for comparisons. The, so called, rational group performed better on all three reasoning tasks when compared to the heuristic group (all t(374) > 4.94, p< .01, d> 0.46). This modification of the CTR makes it resemble modern approaches in reasoning research and does well to differentiate heuristic-from analytic-based reasoners. However, even though there is a robust relation between the CRT and reasoning tasks future tests should incorporate more problems in order to cover the various processes involved in classical reasoning tasks.

dual-process theory, Cognitive Reflection Test, base rate neglect task, Linda problem, covariation detection task

nije evidentirano