,
5 tweets,
2 min read
Read on Twitter
I couldn't agree less. Exit polls are used around the world as validation tools - in spite of the very issues you point out. Too much eagerness not a problem (can't answer unless asked); nonparticipants (refusals) are recorded & factored in; lying is against interest and rare. 1/
That said, exit polls are certainly not perfect - there's quite a bit of art (the stratification algorithm, e.g.) to go with the science. But when we observe certain pervasive patterns (e.g., accurate for noncompetitive/nonsuspect races, red shift for competitive/suspect)...2/
...it speaks volumes and in fact is HIGHLY probative. Such second-order comparatives (see, e.g., 2016 National vs. Swing States attached) transcend the issues with a single exit poll. We would almost certainly NOT keep seeing such meta-patterns in the absence of targeted fraud.3/
@AnalyticsEqv 4/4 
@threadreaderapp pls unroll thx
