A Comparative Analysis of Human and AI Approaches to Thematic Coding in Youth Development Evaluation

As artificial intelligence (AI) tools gain prominence in evaluation practice, questions emerge about their effectiveness in qualitative analysis. This paper presents a comparative study of human-coded and AI-assisted thematic analysis using longitudinal data from 2,627 Louisiana 4-H senior participants. The dataset spans 5.5 years and focuses on youth development outcomes, particularly career readiness skills such as leadership, communication, time management, and perseverance. Through side-by-side analysis, this study examines the extent to which AI-generated codes align with or diverge from those produced through traditional human coding. The presentation explores both the methodological strengths and limitations of each approach, including issues of interpretive nuance, contextual sensitivity, and analytical efficiency. Specific examples of coded outputs will be discussed, offering insight into the implications of integrating AI into qualitative evaluation. This session contributes to methodological discourse while supporting evaluators seeking to balance innovation with rigor and inclusivity in data interpretation.

Christina Zito-Hebert – Louisiana 4-H Youth Development