Thursday, February 18, 2016


This seems like a big deal:
Bernie Sanders now tops Hillary Clinton in the race for the Democratic nomination.

The latest Fox News national poll finds 47 percent of Democratic primary voters now back the Vermont senator, up from 37 percent in January. Clinton gets 44 percent, down from 49 percent a month ago.
Cue the headlines: "Bernie Sanders Just Surged Past Hillary Clinton in a National Poll for the First Time."

But is there something suspicious about this?

Fox posted the complete results of the poll. That means you can read the questions in the order they were asked.

Question #1 was straightforward: "Do you approve or disapprove of the job Barack Obama is doing as president?" Then there's question #2:
2. If you could send just one of the following two messages to the federal government right now, would it be “lend me a hand” or would it be “leave me alone”?
And question #3:
3. For you and your family, does it feel like the economy is getting better or getting worse?
There's nothing wrong with asking these questions -- but it's very bad form to ask questions like this early in a survey about an election, before any questions about the candidates in that election.

The conscientious folks at Pew explain:
Once the survey questions are developed, particular attention should be paid to how they are ordered in the questionnaire. The placement of a question can have a greater impact on the result than the particular choice of words used in the question.

When determining the order of questions within the questionnaire, surveyors must be attentive to how questions early in a questionnaire may have unintended effects on how respondents answer subsequent questions. Researchers have demonstrated that the order in which questions are asked can influence how people respond; earlier questions -- in particular those directly preceding other questions -- can provide context for the questions that follow (these effects are called “order effects”).
That's why, if you're trying to get an accurate read on an election, you ask respondents which candidate they're voting for early, without other questions that might influence their thinking. It's a basic rule of polling -- so basic that Fox's violation of it must have been deliberate. Any pollster would know that questions like #2 and #3 would risk skewing the results.

Question #4 asks which primary the respondent intends to vote in, and then question #5 asks whether the respondent is sure about that choice. And then, for Democratic voters, comes question #6: Sanders or Clinton? And Sanders wins.

Now, why might questions #2 and #3 skew the results? Across all party affiliations, as the results show, voters tend to want the government to leave them alone -- but more progressive voters presumably want the government to offer more assistance. And question #3 taps into a Sanders theme, that the recovery isn't helping a lot of ordinary people. So the pollsters are trying to get voters to think in a Bernie-esque way.

I don't know if you believe poll-question order affects results this way. But it's a standard belief in the polling world. So, yes, I think Fox was trying to manipulate this poll to favor the Democrat Fox thinks is more likely to lose in November.


Republicans really do believe Sanders would be far easier to beat. They're so sure of this that in Nevada they're taking advantage of a quirk in election laws by openly urging Republican voters to caucus as Democrats.

Will anyone do this? Could it work? And are they right about the weakness of Sanders in a general election? On all three, hard to know. But pumping up Sanders seems to be a key Republican dirty trick. Make of that what you will.


Feud Turgidson said...

It's a trick, sure, and it's very much in the Donald Segretti sweet spot as such tricks go. But it's been around a long time, lots of Ds have played it on lots of Rs, just because there's a push doesn't mean it works, and finally - Steve, there's a REASON the Rs don't have any shot at winning in the presidential election in Nevada: they really don't have a handle on the peculiar labor politics there. The STATE politics is up for grabs and often Republican because the in-state Rs promise the hospitality industry workers they won't ever ever ever ever ever ever ever raise their taxes, which of course means Nevada social services programs are the absolute craps. But the same workers who vote R or don't vote at all locally or in state politics, have just loved voting for Obama for president.

What the R's DO have going for them here is the idea of pushing Sandwitches as the more likely to give them free stuff. But bear in mind that HRC won the popular vote in NV in 2008 but lost the caucuses to Obama due to better ground work in the caucuses. You think that's gonna happen again BECAUSE OF R PROPAGANDA? I don't think so.

Chai T. Ch'uan said...

Sure, I don't trust Fox News' either. But that's an awfully slim thread for the PUMAs to yell about without another poll to point at, isn't it? Is it really plausible that leading questions can turn a strong Hillary lead around, or just a virtual tie into a slight Sanders edge?
Am I the only one smelling a whiff of Clinton desperation in this sort of "unskewing the polls" spin?

Mark Jeffery said...

Um, this: