Kai Larsen: Understanding Behavioral Constructs: Toward Theoretical and Disciplinary Integration

Dr. Kai R. Larsen, Leeds School of Business

Word co-occurrences and similarity in survey prompts (questions) carry lexical and semantic information. This presentation demonstrates the potency of using such embedded information by demonstrating that behavioral theories can be reconstructed significantly through Natural Language Processing (NLP) approaches. It is suggested that, possibly, part of the reason for the high statistical validity and correlational relationships between many constructs in behavioral theories may be attributed to the lexical and semantic closeness of measurement items. After demonstrating this phenomenon in several theories, two competing areas of research are outlined: 1) The use of NLP to predict the correlation matrices for behavioral theories, and potentially adjust the relationship between constructs and behaviors to better predict future behavior; 2) the large-scale analysis of thousands of survey prompts to create maps and better understand their relationships. The presentation will show how NLP can be used to solve Thorndike’s (1904) Jingle Fallacy and Kelley’s (1927) Jangle Fallacy, as well as to create the nomological networks suggested by Cronbach and Meehl (1955).

Dr. Kai R. Larsen is an associate professor of information systems within the Division of Organizational Leadership and Information Analytics at the Leeds School of Business with a courtesy appointment as an Associate Professor in Information Science in the College of Media, Communication and Information. He is conducting research to create a transdisciplinary "backbone" for behavioral research. His research has implications for our understanding of all human behaviors, including technology utilization, investor decisions, and cancer prevention behaviors. He was the 2015 recipient of the Association for Information Systems’ Technology Challenge Award.