Revisiting Misadventures in Interpreting Democratic Primary Polling Data (LIZLADS ARE THE WORST)

Since we’re discussing the Democratic primaries, I’ll began with my typical disclosure that, were I voting today, I would vote Warren.

A month ago, Emerson Polling released a poll, which led to wailing and gnashing of teeth in certain quarters:

Anyway, the thing that has people freaking out is that 27/103 Sanders supporters in the primary say they would support Trump over Warren, were she to be the nominee (columns AU & AV, rows 90-92). On the other side of the ideological spectrum, 13/85 Biden supporters say they would choose Trump over Warren. If you’re wondering why sample size and weighting matter, fifty percent of Klobuchar supporters say they would choose Trump over Warren (ZOMG!): the result is actually 2/4.

The emphasized part was viewed as evidence that Sanders supporters are the WORST EVAR in the ENTIRE HISTORY OF WORSTYNESS. Well, the same polling outfit released a new poll with the same question, and guess what? Only 12/107 of Sanders supporters in a Trump versus Warren contest would vote for Trump (down from 27/103 a month ago). WORST EVAR?

Interestingly, that’s no different than the percentage of Warren supporters who would vote for Trump over Warren, 3/42 (Fisher’s exact test, p = 0.55). Yes, you read that correctly. LIZLADS ARE THE WORST EVAR.

Somehow, I don’t think that point will be trumpeted (SEE WHAT I DID THERE?) very much.

This entry was posted in Democrats, Polling, Statistics. Bookmark the permalink.