Mission Statement Topic Models: Coherence, Diversity, and Utility

Mission statements from 527 U.S. Federal agency and subagency websites were analyzed with factor analysis-based topic modeling. Solutions were evaluated with standard topic modeling metrics of Coherence and Diversity. A new Utility metric based on canonical correlation is introduced. The three metrics identified a 20-topic solution most useful with a specific survey response dataset. This approach generalizes to literary and historical research on relationships between texts and relevant quantitative metadata.