In Polls Regarding Policy, Words Can Be Deciding Factor

In his blog post detailing the likely public response to President Obama's new jobs proposal, Nate Silver pointed out an important variable in the polling data we often take as fact: the wording of questions. He asks, "Are Americans going to be tolerant of proposals for new spending after having spent six months hearing about deficit reduction?" His answer? "Well, it depends on how you ask them."

According to Silver,"when the issue is framed as one of jobs against deficits, jobs win." He writes: "On average, those polls that ask Americans to prioritize 'creating jobs' or 'reducing unemployment' against 'cutting spending' or 'reducing the deficit' have had 57 percent of respondents coming out on the jobs side, against 36 percent who prioritize the deficit."

But wording can make all the difference. "The answer changes, however, when the conflict is instead framed as stimulus or recovery spending against deficits. In polls that employ the term 'spend' or 'spending' in describing the additional stimulus, its support drops to an average of 44 percent, with 50 percent saying that deficit reduction is the higher priority."

It appears President Obama has taken notice. Silver found "that in his speech on Thursday night, Mr. Obama used the term 'job' or 'jobs' 39 times, often preceded by 'create' — but never uttered the word 'stimulus.'" And when he used to word 'spend' or 'spending', he was typically referring to his efforts to reduce spending--not plans to spend more. Nor did "Obama specify the cost of his program."

Silver anticipates more "semantic scrum" to follow in the coming weeks of partisan policy debate. And he has a suggestion for pollsters and readers: "I would advise [pollsters] to use multiple question variants where possible, taking a larger sample and splitting it into halves or thirds, and I would advise readers to be suspicious of articles that cherry-pick one or two polls without discussing the broader context."


NH cf