Candidate Name: Paul Amari
Program: Organizational Science​​​​​​​
Committee Chair: Dr. George Banks
Committee MemberLeader behavior has long been recognized as a central component of the leadership process, yet much of the leadership literature continues to emphasize traits, perceptions, and broad leadership styles rather than the specific behaviors leaders enact. Furthermore, leader behavioral taxonomies are often examined in isolation, limiting our understanding of how leader behaviors co-occur and vary across organizational contexts. To address these gaps, this dissertation introduces an algorithmic and configurational approach to studying leader verbal behavior. Four fine-tuned machine learning classification models were developed to identify 30 leader verbal behaviors spanning charismatic, ethical, transformational, and destructive leadership taxonomies. The models were then applied to two large corpora of naturally occurring organizational communication. Dataset 1 consisted of 26,405 organizational emails containing 94,625 sentences produced by 182 unique senders representing multiple leadership levels. Dataset 2 consisted of 69 virtual team meeting transcripts representing 257 participants, including 69 assigned leaders and 188 followers. Results demonstrated meaningful variation in leader verbal behavior, behavioral configurations, and communication participation across contexts and leadership levels. Although several frequently occurring behavioral configurations generalized across contexts, the structure and strength of co-occurrence patterns differed between organizational emails and virtual team meetings, suggesting that contextual factors play a central role in shaping behavioral expression. Collectively, these findings support a configurational approach to leadership that places leader behaviors and communication context more centrally within future theory building and refinement efforts while simultaneously demonstrating the utility of artificial intelligence and large-scale text analysis for advancing behavioral research. Implications for leadership theory, evidence-based practice (e.g., leadership coaching and development), as well as future avenues of research in leadership science are discussed.
Keywords: leader behavior; leader communication; behavioral configurations; leadership theory; leadership processes; machine learning; artificial intelligence; verbal behavior​​​​​​​s: Dr. Scott Tonidandel; Dr. Eric Heggestad; Dr. Wenwen Dou
Abstract: