Nourafshan: The politics of polling


SocialTwist Tell-a-Friend

Mitt Romney’s highly anticipated and surprisingly deft pivot to the center during the presidential campaign signals his campaign’s understanding that the views espoused by the candidate during the Republican primary do not align with the values of the majority of the American electorate.

This conclusion is easily drawn looking at polling data on American attitudes in a number of policy areas: Romney’s attempt to retreat from his signature “more tax cuts for the wealthy” economic plan, for example, likely arises from data that show more than two-thirds of Americans support raising taxes on the top income bracket. With such reliance on polling data to inform the candidates’ views, it is quite surprising how far some on the right have gone to discredit polling data in general as biased and untrustworthy.

To begin, poll data are not opinion-based; they are aggregated statistical information. To suggest that some of the most reputable polling firms in the country, including the right-leaning Rasmussen, are getting things wrong as a matter of opinion is an irresponsible characterization of the purpose and nature of poll data.

Every social-science student understands the challenges of representative sampling and identifying and eliminating biases in the information-collecting process, and these factors are undoubtedly controlled for by firms with long histories of accurately forecasting election outcomes and reporting on demographic trends. The advent of organizations such as Unskewedpolls.com to correct a perceived bias in polling by reweighting every poll to achieve a desired outcome is the only example of manifestly unreliable polling data.

Possibly the most vulgar and unsophisticated critiques in the manufactured controversy over poll data are the attacks focused on Nate Silver of the widely read FiveThirtyEight blog. Silver’s transparent methodology consists of aggregating the most highly regarded public opinion surveys available and factoring in additional economic data to forecast election outcomes.

In the last election, Silver predicted 49 out of 50 states correctly, and the one state he got wrong, Indiana, he erroneously called in favor of John McCain. Despite his record of accuracy and a transparent modeling methodology, Silver has been the target of criticism from many right-leaning organizations, as an article by Henry Blodget of Business Insider explains: “Every time Silver opens his mouth, the ~45 percent of the country that is rooting for Romney accuses him of being an idiot, being ‘in the tank’ for Obama, or both. This criticism is asinine.”

The phenomenon of attempting to discredit sources simply because they do not conform to a given ideology suggests that the problem of confirmation bias is only intensifying for the American polity.

While it is true that news organizations may have expressed certain political biases in the past by emphasizing certain poll results and not others, this type of cherry-picking from legitimate data is quite different from inventing a polling methodology with the sole purpose of fulfilling an ideological agenda.

Irrespective of which polling firm ends up most accurately predicting the election outcome, the effect of challenging reliable polls will be to breed unfounded skepticism among some Americans, which will be difficult to eradicate in the future. At a time of particularly low levels of institutional faith, it seems dangerous to further undermine the credibility of important sources of information by introducing an unnecessary political dimension to otherwise objective data.

In today's issue:

Privacy Policy (8/15/07) | Terms of Use (4/28/08) | Content Submission Agreement (8/23/07) | Copyright Compliance Policy (8/25/07) | RSS Terms of Use

Copyright © The Daily Iowan, All Rights Reserved.