This study examines whether league context is associated with systematic differences in how fans explain and evaluate outcomes in event-time Finals discourse, comparing NBA and WNBA conversations on X. Specifically, we assess whether the leagues differ in the locus of attribution emphasized (internal vs. external) and in the composition of evaluative sentiment, distinguishing supportive versus non-supportive positivity and hostile versus non-hostile negativity.
We collected public X posts from the 2025 NBA and WNBA Finals using a standardized game-centered observation window and restricted analyses to original posts. Posts were annotated for (a) attribution locus and (b) sentiment. Labels were generated at scale using a supervised, fine-tuned multi-task large language model, with GPT-4.1 mini as the base model. League differences in category prevalences were summarized as percentage-point contrasts with bootstrap confidence intervals.
NBA Finals discourse emphasized internal attributions more frequently (49 vs. 42%), whereas WNBA Finals discourse emphasized external attributions more frequently (48 vs. 41%), indicating a league-linked shift in accountability orientation. Sentiment patterns also diverged: NBA discourse was more negative overall (50 vs. 41%), driven by both hostile and non-hostile negative expressions, while WNBA discourse was more positive overall (59 vs. 50%), with the positivity difference concentrated in supportive, affiliative expressions (29 vs. 19%) rather than non-supportive positivity.
The study advances a comparative account of league-differentiated fandom environments by treating league context as an interpretive framework for online discourse. It refines sport social media sentiment measurement beyond valence and links attribution and sentiment as interlocking components of fan discourse.
