This table contains the highest-weighted words and LDA topics for predicting that tweets are from (+ row) and not from (- row) a given state. Some LDA topics have been manually assigned names (italicized) based on their apparent content for convenience, with prominent words listed in parentheses.