In this study, the possibility of decision error is investigated to identify and improve the reliability of participants in the process of conducting the sound quality analysis for laser printers. So far, there is not a way to identify and express the possibility of individual participant quantitatively. Thus, the decision error model is proposed which is based on the expectation value between the perceived sounds. Through the experimental verification on the laser printers, it was found that the possibility of decision error is affected according to the normalized difference. The possibility of decision error has inversely proportional to the normalized difference between the perceived sounds. When the normalized difference becomes small value, the uncertainly between decisions is inversely increase, and then it is difficult to obtain the proper result in the process of the jury evaluation for laser printers. For this reason, in this study, the proposed decision error model is added in the previous step of the correlation verification. Comparing to the conventional process only using the correlation based method, after the reliability of each participant is verified, the correlation with the mean response of participants is verified. It was found that the participants who were recognized as having unusual preferences are actually identified as having the reliability problem. Based on the results of this study, the proposed decision error model will be helpful to identify and improve the reliability of participants in the following study for the sound quality analysis.