Nate Cohn

@Nate_Cohn ↗

Deleted No

Hibernated Yes

Last Checked Sept. 30, 2020

Created

Tue Dec 24 00:43:26 +0000 2019

Likes

Retweets

Source

Twitter Web App

View Raw Data

JSON Data

View on Twitter

Likely Available

Nate Cohn @Nate_Cohn

The pattern tracks fairly closely with the overall decline in error/bias at the end, including the possibly-Kavanaugh-related rough stretch in early October https://t.co/Yegi99wdGH — PolitiTweet.org

With the results final, average error in Upshot/Siena polls over last 3 weeks was 3.32 pts, R+0.96 bias (44 polls);… https://t.co/yz9C4xgPF6

Posted Dec. 24, 2019 Hibernated

Preceded By

Nate Cohn @Nate_Cohn

Self-reported turnout steadily became more accurate in the final stretch in Times/Siena polls, mainly because you could start to believe self-reported unlikely voters (validated turnout rate among those who said they were likely held pretty constant) https://t.co/ern8QdimDC — PolitiTweet.org

Posted Dec. 24, 2019 Hibernated

Followed By

Nate Cohn @Nate_Cohn

Anyway, interesting choices here. Would you rather a poll used the final LV probabilities, in the spirit of 'if the election were held today'? Or would you rather something that evolves as the election got closer, and gradually gave more weight to self-report over a model — PolitiTweet.org

Posted Dec. 24, 2019 Hibernated