We consider a method to analyze how individuals in a certain group make a decision or judgement based on the observations on phenomena and what characteristics individuals have regarding their decision. Suppose we have two different types of datasets. One dataset is of the observations on real phenomena for two classes of subjects. The other is of the controlled values that imitated the phenomena, and decisions/judgements of the classes by individuals in a certain group. Using these datasets, we would like to analyze what variables are used for judgement by individuals and estimate the distribution of individuals' recognition of misclassification costs. As a concrete example, we analyze the data on some species of fireflies (Takatsu, Minami, Tainaka and Yoshimura, 2012) to figure out what information male fireflies use to find out female fireflies, and how they recognize misclassification costs.