The Stanly News and Press (Albemarle, NC)

September 24, 2012

For today’s polls, accuracy is credibility

By Scott Mooneyham for the SNAP

Monday, September 24, 2012 — RALEIGH — In my other job as editor of the state government newsletter, “The Insider,” the topic of how to deal with polls never goes away.

A couple of decades ago, news organizations had pretty cut-and-dried rules about polls.

Then, polls came in two main varieties: those commissioned by news outfits, and conducted by companies like Gallup and Mason-Dixon, that would be reported on and disseminated by the media; and those commissioned by the candidates’ campaigns that wouldn’t be disseminated by the media for fear of being manipulated by the campaigns.

Campaign consultants would sometimes discuss their private polling numbers with reporters, but reporters typically approached those discussions with a good dose of skepticism.

We live in a different world today.

As automated polls be-came more refined and gained more credibility, more companies sprang up conducting polls.

And the motives of those involved became a bit more complex.

Today, some of the same polling firms that work for candidate campaigns conduct other surveys that they distribute freely. Some political consultants have started their own polling operations, and distribute some findings to the news media. Some advocacy or-ganizations commission polls while also trying to influence elections.

At “The Insider,” we err on the side of publishing most North Carolina poll results that come across the desk.  

With a specialized readership focused on state government issues, we want subscribers to have information that is out there that might affect them. It’s also a sophisticated readership that understands bias in polls.

If I were a newspaper editor, I might have a different take on the publication of poll results because of the potential for bias, bad data and manipulation.

That issue of bias recently came up after the conservative Civitas Institute issued poll findings showing Republican Mitt Romney leading Democratic Pres-ident Barack Obama by 10 percentage points in North Carolina.

It’s possible, right? It might be except that the poll showed Romney enjoying the support of 30 percent of African Americans and 60 percent of voters between the ages of 18 and 34.

In any other polls, in this or any other state, Obama enjoys much stronger support among those demographics.

The poll is no different than one that might show Obama leading among white retirees. It’s a bad poll.

Instead of publishing it, the good folks over at Civitas should have tossed it in the trash and demanded their money back from the polling firm.

The other major distributor of polls in North Carolina is a Democratic-leaning polling firm called Public Policy Polling.

That affiliation has Republicans suspicious of its findings, and there was at least one instance in 2010 when its demographic weighting looked strange.

Of course, the bottom line regarding these polling outfits’ credibility is their accuracy.

If Civitas publishes too many more polls like its most recent offering, I guess I’ll have another conversation about polls.    

And if, unlike in 2008, PPP’s surveys in 2012 prove highly inaccurate, I’ll have yet another such conversation.