A colleague writes:
Have you seen it? this article Was this written by Nate Cohn of the New York Times?
A few things looked odd in there. First he writes:
The tendency of recall votes to overstate the winners of past elections means that weighting recall votes has a predictable effect. In other words, support for the party that lost in the last election increases.
Is this always true? I think there is a small algebraic example where this is not the case. Plus his table here seems to contradict that?
Out of curiosity, I sent the following message to Nate Cohn:
A colleague pointed me to your article and asked a question. See below. Was there a mistake in the article?
I also recommend the following articles on the general topic of the benefits of party identification mediation:
From 2001: http://stat.columbia.edu/~gelman/research/published/aprvlRv1.pdf
From 2016: http://stat.columbia.edu/~gelman/research/published/swingers.pdf
From 2016: https://www.nytimes.com/interactive/2016/09/20/upshot/the-error-the-polling-world-rarely-talks-about.html
From 2016: https://slate.com/news-and-politics/2016/08/dont-be-fooled-by-clinton-trump-polling-bounces.html
No reply! There were so many links in my message that I had the terrible feeling that they were caught in his spam filter. So perhaps posting a blog post is the best way to communicate this.
Anyway, I haven’t looked into this particular issue of coordinating past votes. There may be some subtleties here that I’m not aware of. In general, I think it’s a good idea to adjust the way we measure partisanship to some degree (reminiscent of Lohr and Brick’s reanalysis of the famous Literary Digest poll from 1936), including party identification, party registration, and recall of past votes. It is important to have relevant information on these variables at the state level. But yes. These measurements themselves are subject to error, so the best adjustment is not simply “weighting.”
P.S. My colleague adds the following comment:
