single post

Politics



It’s every bit as interesting to consider what people say when they’re asked who they definitely wouldn’t vote for, rather than simply who they would vote for

How to decode mixed messages from the latest US presidential election polls

As is so often the case, the latest US presidential election polls seem to be sending different messages about who’s up and who’s down. Here are a few words of analysis that might help you make sense of it all.

1) A CNN/ORC poll out today shows Donald Trump polling at 41% nationally while Texas senator Ted Cruz trails at a mere 19%.

Change: This is the highest level of support Trump has received since CNN/ORC began polling this Republican nomination race.

Method: A random national sample of 1,002 adults contacted by telephone between 21 and 24 January.

Margin of error: +/- 3 percentage points.

2) A national poll from the Washington Post/ABC News gives Trump 37% of support while Cruz comes in second with 21%. Change: Trump’s support is steady compared to a month ago.

Method: 356 Republican-leaning registered voters were contacted from 21-24 January.

Margin of error: +/- 5.5 percentage points.

3) Today’s Quinnipiac University poll of Iowa voters has Trump polling at 31% and Cruz at 29%.

Change: Virtually no change since their last Quinnipiac poll earlier this month.

Method: 651 likely Iowa Republican caucus participants were contacted between 18 and 24 January by real people via landlines and cellphones.

Margin of error: +/- 3.8 percentage points.

Analysis

Advertisement

That CNN/ORC poll might sound like the best of the bunch (decent number of respondents, low margin of error), but when you take a closer look, it gets tricky.

That’s because the topline numbers about Republican candidates are only based on 405 registered voters who are Republicans or Republican-leaning. Suddenly, the margin of error jumps to five percentage points.

That doesn’t change the fact that Trump appears to be significantly ahead of Cruz nationally (although the margin could be much narrower at 36-24), but it does show the fragility of polling numbers that are based on such a small number of US voters.

Here’s another thing that’s worth keeping in mind: It’s easy to focus on answers to the rather simplistic (and very hypothetical) question: “If the primary or caucus in your state were today, for whom would you vote?” But it’s every bit as interesting to consider what people say when they’re asked who they definitely wouldn’t vote for and who they might vote for.

single post In a write-up of their poll, the Washington Post added up respondents’ first and second choices and found that the race tightened a little, with Trump at 49% and Cruz at 39%.

One last point to keep in mind: research has repeatedly shown that one of the best ways to predict whether someone will vote is to look at whether they have voted in the past. As Ryan Struyk at ABC points out, those Quinnipiac polling numbers look very different when you only take previous voting attendance into account.

Still, whichever way you slice these numbers, there’s one clear takeaway: Trump is doing great.

>>>>>