EXPLANATORY AND PREDICTIVE ASSESSMENT OF IOT SECURITY BEHAVIOR USING PMT: A HYBRID SEM–AI PERSPECTIVE FROM KARACHI

Abstract

This paper examines the predictors of the IoT security behavior by combining a Protection Motivation Theory (PMT) and a hybrid analytical system that incorporates Structural Equation Modeling (SEM) and Artificial Intelligence (AI). Based on findings accumulated by the users of IoT in Karachi (N 132), the research investigates the role of perceived severity, vulnerability, response efficacy, self-efficacy on intentions and actual security behavior. The findings of SEM results confirm that the threat and coping appraisal are very important predictors of security intention and response efficacy and self-efficacy have been identified as the most significant ones. To improve predictive accuracy, Artificial Neural Networks (ANN) and other AI algorithms were utilized and it was found that better nonlinear predictive power existed with Artificial Neural Networks (ANN) and artificial intelligence algorithms than with SEM alone. The hybrid SEM-AI method has a high explanatory and predictive capability and offers a more solid picture of the behavior of IoT security in a new digital environment. The results emphasize the need to address users' confidence and competence, enabling them to implement protective measures, and focus more on threats posed by IoT. The paper is a contribution to theoretical driven cybersecurity research and it represents a methodological improvement in that it combines behavioral modeling and intelligent prediction systems.