With the development of new digital human-machine interfaces, many discussions in the nuclear industry have focused on the human factors issues that arise from the interfaces. To quantitatively characterize the effects of the interfaces on human reliability, we collected empirical data from a full-scope simulator of the APR1400 nuclear power plant using the Human Reliability Extraction (HuREX) framework. From the numerous variables in the collected data describing the contexts of the performance influencing factors (PIFs), including crew experience, task complexity, and procedure quality, the significant variables were identified by three techniques comprising both qualitative and quantitative analyses. Based on the selected variables, the nominal error probabilities and PIF multipliers were then estimated by logistic regression analysis. This paper interprets the meanings of the estimates and discusses the advantages of the employed variable selection techniques.