News...
                        sponsored by

 


DIGITS

A battered polling industry takes stock

Send a link to a friend 

[September 12, 2013]  (AP) -- A new survey highlights the two-front problem facing the polling industry: People don't trust polls, and many say they are unlikely to want to participate in them.

Let's take trust first. The Kantar survey, conducted to assess how people view polls, shows that 75 percent of Americans think most polls they hear about are biased toward a particular point of view.

But Americans differentiate polls based on their source. Most say they trust polls from nonpartisan foundations and academic centers, slightly fewer trust polls from polling companies or news media organizations, and even fewer have faith in those from political parties or candidates.

These two findings combined -- broad questions about the veracity of polling coupled with greater trust in sources that seem to prioritize scientific methods -- suggest that Americans take a savvy approach to polling.

And in two key ways, they're right. Not every poll is worthy of trust, and there probably are more "biased" polls out there than unbiased ones.

Yet bias is not the same as duplicity. Most polls released publicly have not been designed to dupe the public. Bias in surveys can stem from factors beyond a pollster's control, and can emerge even when tried-and-true methods are used.

Two of the industry's most respected names -- Gallup and the Pew Research Center -- have been conducting experiments to refine their likely voter models after the 2012 election seemed to foil processes that had worked well in the past. That's not to say polls from Gallup or Pew ought to be dismissed as untrustworthy, but that even the best researchers can sometimes fall victim to unintended biases, making it very difficult for poll consumers to know whom to trust.

The other problem illuminated by Kantar's poll -- that Americans are often unwilling to participate in polls -- is thornier for the polling industry.

For decades, polling has relied on widespread telephone ownership, with phone numbers tied to a geographic location, to generate samples that are about as close to scientific randomness as possible.

Those samples worked remarkably well in the past because of their compatibility with the culture of the time. Almost everyone had a phone at home and answered when it rang. Until caller ID and answering machines were widespread, few screened their calls. It was a near-perfect environment for a methodology that requires the researcher to select the respondent rather than taking volunteers.

[to top of second column]

Communication today emphasizes choice. It has never been easier to reach people, yet people have more power to choose when, where and how they are reached. The technologies replacing landline telephones make it easier for people to avoid being asked to take a poll and harder for pollsters to select a random sample.

Even if pollsters can crack the code for sampling online or via cellphones, finding people willing to participate could remain the industry's biggest challenge. The Kantar poll shows people have no greater inclination to take polls online than on a landline phone, nor are respondents particularly excited about taking a survey via cellphone, mobile app or social media. Yet usage of those new communication tools grows daily.

As Gallup's Frank Newport said in an address to survey researchers in 2011, "The public, apparently more than ever, wants to give its opinion, just not necessarily in the way we want them to."

The polling industry's sharpest minds see their next tests ahead in the 2014 and 2016 elections, and are working to merge the new culture of sharing with the world of scientific surveys.

___

Kantar's Path to Public Opinion:
http://us.kantar.com/public-affairs/politics/
kantars-path-to-public-opinion/

[Associated Press; By JENNIFER AGIESTA]

Follow Jennifer Agiesta on Twitter at http://twitter.com/JennAgiesta.

Copyright 2013 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.

< Top Stories index

Back to top