The BBC's Robert McKenzie with the Swingometer on Election Night in 1964. Assumptions about national and regional swings in polling play a key role in explaining why the 538 probability favours Trump more than other poll-based prediction models (Vox). One odd thing about the 538 model is that it's unusually sensitive to news that registers in post-news polling compared to other models. Thus, the attention that it gets as an apparent alternative to punditry and ground-reporting based assessments of the campaign is somewhat misplaced.
No comments:
Post a Comment