Peccadillos Of The New Political Polling

Polling has always been an interesting mix of marketing and science. Traditional polling is one of those subjects that only the few seniors or grad students who took courses like “Survey Sampling 501” can truly understand. For example, how could a poll of only 540 people realistically determine who was going to vote for Mitt Romney in Ohio?

Without disputing, or even challenging, the science of polling, we may see distinct evidence of the blending of marketing and science in the upcoming elections.

Maybe it is society’s obsession with immediate gratification, but it subjects like data, polling and keeping score are sexy these days. Earlier in the month David Gregory was ousted as the host of “Meet the Press.” His reported replacement, Chuck Todd, is an NBC analyst whose specialty has been interpreting and reporting on public opinion and the polls that accompany elections. Todd has been successful at giving viewers the crisp analysis and the “score” the network is looking for.

Until recently political polling has been based on random digital dialing that carefully fills a statistical sample that claims to be representative of the electorate. Venerable organizations like Pew Research, Quinnipiac, Rasmussen, and Gallup go to great lengths to manage their respondents and generate an accurate representation of the base voters. So what’s changed?

Over time the random, statistically accurate American has stopped answering the phone. Pew Research reported that only 9% of people contacted for political polls now respond, down from 28% five years ago. Mobile phones give the consumer more choice on who they talk to and when. The number of households with landlines is down dramatically to fewer than 60% in some states. Landline polls also create a problem as research shows more Republicans than Democrats have them. Caller-ID means fewer pick-ups, and those that do answer are disproportionately elderly, established voters. Controlling the sample has consistently degenerated.

The New York Times and CBS News recently announced that they will now use online polling, based on survey panels from YouGov that include preselected groups of more than 100,000 members. YouGov builds demographically diverse panels that map to the American population and can be reached online. The folks at The Times and CBS believe that this approach will mimic the traditional telephone sampling method but, through scale, address some of the partisan survey results that are so prevalent, especially at the state level. The key to this move, however, is that the poll will provide a snapshot of the electorate at the moment when the survey data is collected and thus be more responsive to the trends of the election.

This blending of marketing and science suggests three opportunities for the local and state level campaigns in midterms and beyond:

1. Timing: The use of the online polling now has a level of authenticity and credibility that it may not have had in the past. Polls can now be done more frequently and at a lower cost. In the past, campaigns would commission a $100K+ poll at the start of the campaign and ask 500 voters 50 to 100 questions on every issue imaginable. The results (too) often served as the blueprint for a candidate’s position on issues or with constituent groups. While for the most part scientifically accurate these mega-polls can be a trap of rigidity that keeps the candidate on topic, but not topical, until it is too late to adjust to changing voter opinions. Online polling allows the candidates to be more responsive.

2. Reach: One of the advantages of online polling is reach at scale. Regardless of whether the campaign uses email or online advertising to drive responders it is important that all segments of the electorate are polled. Email lists are prized possessions in a campaign, but they rarely provide scale. Buying key words helps build a list of folks that are already aligned because they opted-in. Buying cookie-based on-boarded lists have a relatively low average match rates of 30%, which limits reach especially at a local level or on mobile devices. There are, however, solutions for geo-demographic targeting or neighborhood level targeting through programmatic display that fit well with the goal of near 100% reach of the voter base.

3. Flash Polling: A new concept in online polling, this keeps a candidate in touch with the pulse of the community. It is also a cost-effective tool for engagement. The results of mega-polls may suggest that certain constituencies; ethnic groups, social categories or neighborhoods are a lost cause. As a result, these groups often get ignored until the final television buy based on the logic that name recognition and TV personality alone will secure enough of these voters.

Flash polling is “digital kitchen table politics at scale.” It uses online display advertising to target and invite voters at a neighborhood level to share their opinion on questions that are relevant to them. Asking one or two questions in an online form on a weekly or monthly basis allows the candidate to tabulate the results, inform their position and share them with the voter. Engagement begins. Packaged correctly as an engagement tool - that may or may not include social media and is language appropriate and issue sensitive - online ads will draw in voters at an unprecedented scale and at a very low cost. Flash polling can help create dramatically improved levels of public awareness of a candidate across every community and social group.

Voters still care about who is winning and why; but we are likely to see these new polling concepts in action over the coming weeks. Online polling gives campaigns access to a new tool set. Instead of shying away or limiting engagement with the non-aligned voter they can now reach out to all of them. Instead of using advertising to exclusively run attack ads they can also be used to engage voters with positive messages through polling. With these new tools campaigns can find out what voters really think in real time.

MediaPost article: http://www.mediapost.com/publications/article/233751/peccadillos-of-the-new-political-polling.html#comments