A fraud score calculating program primarily for use in determining the possibility of credit card fraud can calculate a score reflecting the reliability of a model created based on Bayesian theory. Various factors are extracted from authorization data, and cases are selected for various combinations of the factors. A fraud score is calculated for each case, and a result of score calculation is specified using at least one of the calculated scores. The result of score calculation may be an average of a plurality of the calculated scores. When each score reflects reliability, the result of score calculation may be the largest of the calculated scores.