Artificial intelligence as a tool for item reduction in an organizational resilience questionnaire.

Abstract

Objectives. Considering that there is no standardized questionnaire for safety climate and resilience assessment, authors usually review a large number of questionnaires from the available literature, which results in a high number of questions distributed to respondents. As the questionnaire length increases, resistance from the respondents increases. Artificial intelligence (AI) tools until now have not been used for item reduction, besides the need for selecting and retaining only the most relevant and informative questions in the questionnaire with adequate accuracy. Methods. AI tools such as multiple linear regression analysis (MLRA) and the multilayer perceptron artificial neural network (MLP ANN) are used in the development of a model able to cluster respondents' ratings and to predict values of organizational resilience based on the respondents' ratings of the specific questions. Results. AI could be used as a valuable tool for item reduction, since the prediction accuracy for MLRA tools is 70.4-71.5% and for the MLP ANN it is 76.4%. Conclusions. This research proves that machine learning algorithms can be used to build predictive models that determine which survey questions are the most predictive for organizational resilience index calculation using safety climate factors.