The Taylor Swift effect

Trump is on trial! If each of you reading this can kick in $10 or $25, it'll help keep Palmer Report firing on all cylinders at this crucial time in our nation's history: Donate now
-----
Palmer Report readers: sign up for our free mailing list here


Yesterday I pointed to a new Reuters poll which showed that Democratic candidate Phil Bredesen is just three points behind Republican candidate Marsha Blackburn in the U.S. Senate race in Tennessee. This stands in sharp contrast to the New York Times poll which came out a few days ago and showed that Bredesen was down by fourteen points. Some of you have responded by asking whether this is the Taylor Swift effect.

First, it’s important to keep in mind that these polls are apples vs oranges. Reuters and the NY Times are using two entirely different methodologies. So the logical conclusion isn’t that eleven percent of Tennessee voters changed their minds within a few days, but that one (or perhaps both) of these polling outlets is using a flawed methodology.

As I explained earlier, these are two out of five polls in this race, and the polling average – the only number that matters – has Bredesen down by five points. There are two ways to measure the Taylor Swift effect going forward. One would be to watch how the polling average shifts as new polling data arrives. That may be tricky, because there hasn’t been a ton of polling in this race.

The other measure would be to compare apples to apples. The current Reuters poll shows Phil Bredesen down by three points. Unfortunately, this is the first time Reuters has polled this race, so there’s no way to know if the Reuters model had Bredesen higher, lower, or the same prior to Taylor Swift’s endorsement. That said, if Reuters conducts another poll in this race next week, that’ll be a good solid comparison to see if Bredesen is on an upward trend.

The bottom line, for now, is that we don’t have a solid way of measuring the Taylor Swift effect in Tennessee. You can look at the huge spike in Tennessee voter registration after she made her endorsement, and assume that most of them registered with the intention of voting for Bredesen, and then try to calculate how many of them will actually show up and vote, but that involves a lot of assumptions. As always, election night is the ultimate arbiter of these things. Update: shortly after this article was published, a new Vanderbilt University poll was released which has Bredesen up by one point.

Trump is on trial! If each of you reading this can kick in $10 or $25, it'll help keep Palmer Report firing on all cylinders at this crucial time in our nation's history: Donate now
-----
Palmer Report readers: sign up for our free mailing list here