A Cognitive Bias Under Composite Percentage Relationship And Verification On Potential Explanations To The Bias

Since the end of last century, researchers, represented by Daniel Kahneman, have conducted a series of complicated and rigorous study in decision-making field to reveal that people’s decision-making process, under certain contexts, is not absolute rational. For example, framing effect, as a typical decision making anomaly, indicates that decision making could be greatly influenced by representation of problem rather than nature of the problem only.This paper explored a cognitive bias under frame of composite percentage relationship, which indicated that when people were exposed to multiple items which are expressed in a form of dual dimension of percentages they will overweight the salient dimension, and as a result they prefer big percentage in small subset over small percentage in big subset, even when net overall quantity is equated.Study 1 aimed to confirm the existence of the cognitive bias under frame of composite percentage relationship. The study built multiple items with vary representations of composite percentage relationship. Participants were asked to rating the items basing on appealing to themselves. Through comparing ratings to pairs of items with consistent inner value, participants’preference had been examined. The result shows that among pairs of equivalent items, participants prefer big percentage in small subset over small percentage in big subset.Study 2 and 3 aimed at verifying two potential explanations to the cognitive bias. One explanation is that preference over big percentages contribute to the bias, and another one is partially neglect of subset information. In study 2, a mixed-effect model was built to learn the way how people weight percentage in composite percentage frame. The result shows there is no extra preference upon big percentages distinguished, thus the hypothesis of preference over big percentages contribute to the bias was not supported. Study 3 employed continuous valuation task and anchoring point, which would smooth activation of System 2, to observe whether or not participants still have cognitive bias under frame of composite percentage relationship. The result indicated that the number of people who showed the bias in study 3 is significantly less than the number in study 2. It supports the hypothesis that partially neglect of subset information has contributions to the cognitive bias under frame of composite percentage relationship.