Margolis: “Robo” poll gives Welch, Leahy wide edge, while Shumlin gets narrow win

Congressman Peter Welch. VTD/Taylor Dobbs

Maybe – just maybe – Vermont really is different from America.

In that other place (America), the latest polls show that disapproval of Congress is all but unanimous.

In this place (Vermont), voters love their incumbent members of Congress. They approve of Sen. Patrick Leahy and Rep. Peter Welch (both Democrats) by two-to-one majorities and of Independent Sen. Bernie Sanders by even more.

In America, incumbent members of both houses of Congress seem eligible for listing under the Endangered Species Act. In Vermont, Sanders and Welch are poised to coast to re-election. America’s mood is to throw the rascals out. Vermont’s is to throw its rascals right back into office.

All this assumes that the latest Vermont poll is accurate, which explains the “just maybe” caveat above. The poll was taken by Public Policy Polling (PPP) based in Raleigh, N.C., between July 28 and July 31, and it was released last week.

Like all Public Policy Polling surveys, this was an automated, or “robo,” poll. The Vermonter answering his or her phone call from the polling firm did not talk to a human being but was instructed by a recorded voice to push buttons on his or her telephone keypad to indicate approval or disapproval of candidates or issues.

And just what is wrong with this technique?

Maybe nothing. But the major established pollsters such as Gallup and the television networks reject it. Several respected news organizations, including NBC News and The New York Times, will not use “robo-poll” results in their news reports. Andrew Kohut, director of the Pew Research Center for People and the Press, is on record as being “not a fan” of “robo-polls.”

One possible flaw of automated polls is that more people immediately hang up on them (“non-response bias,” in the jargon). Some potential respondents refuse to cooperate with conventional polls that use interviewers, too. But not as many; basic politeness creates a disincentive for slamming down the phone while talking to an actual person.

In addition, “almost all (robo-polls are) done by landline only, which produces a potential bias because of the exclusion of the approximately 30 percent of adults who can be reached only by cell phone,” said Scott Keeter of the American Association of Public Opinion Research (by email).

On the other hand, Public Policy Polling does not have a bad record of forecasting election results. Self-appointed poll ranker Nate Silver of FiveThirtyEight.com lists PPP as among the more accurate polling firms (though some experts doubt his assessments, too.) Keeter’s email said the association “does not stipulate that any particular type of survey methodology is required.” And the large sample in the Vermont (poll 1,233 voters) should offset the immediate hang-up problem at least to some extent.

But the sample might have tilted Republican. That would not reflect deliberate bias. Public Policy Polling is run by Democrats. In this case, though, 37 percent of the respondents pressed “1” on their telephones identifying themselves as Democrats, 24 percent pressed “2” indicating they were Republicans.

That’s a mere 11-point margin for the Democrats. A Gallup Poll last year showed a 26-point margin of self-identified Democrats over Republicans in Vermont.

Still, there may be nothing wrong with the Public Policy Polling poll. Perhaps there is something wrong with reporting on the poll without informing readers/listeners/viewers that many public opinion survey experts doubt the soundness of its method.

In Vermont, many news organizations reported the poll’s results. None, according to an Internet search, mentioned those doubts.

Even a doubter, though, would have to be impressed by the approval ratings of Sanders, Leahy and Welch. Only if the poll were completely incompetent (and even robo-poll critics don’t claim that) could it be exaggerating the popularity of these officials very much. In fact, if the sample is more Republican than the actual electorate, the two Democrats and the sort-of Democrat Sanders might be even stronger than the survey numbers indicate.

Where the numbers are tighter, the meaning becomes less clear, and the numbers are tighter when it comes to that other key incumbent up for re-election next year, Democratic Gov. Peter Shumlin.

Not that the poll results were bad for Shumlin. They showed him leading his 2010 (and possible 2012) opponent, former Lt. Gov. Brian Dubie, by 48-to-40 percent, not overwhelming, but a lot bigger than Shumlin’s actual victory margin last year.

The two men had almost identical approval-disapproval ratings (45-36 for Shumlin, 48-33 for Dubie; in polling, that’s effectively a tie). Only in incumbent-loving Vermont would Shumlin’s approval-disapproval ratings be considered a problem. As Public Policy Polling noted in announcing its results, Shumlin’s “margin puts him at about the median for the 42 sitting governors on which PPP has polled.”

Furthermore, Dubie may not run, and the poll gives Shumlin huge margins against any other potential Republican challenger.

The poll also indicated that Vermonters are closely divided on whether the Vermont Yankee nuclear power plant should continue operating, with 45 percent in favor and 41 percent agreeing with Shumlin that the plant should shut down when its original license expires next year. Again, this is survey research, based on probabilities, not precise absolutes. That result is really an even split.

On another contentious issue, 40 percent of respondents said they supported and 35 percent opposed “Vermont’s single payer health care law.”

Vermont has no such law. The bill passed by the Legislature this year, strongly backed by the governor, creates a process aimed at rationalizing and simplifying the health care system. A single-payer plan is one of several possible outcomes.

These results spawned some speculation that the split verdicts on these issues helped explain Shumlin’s apparently unimpressive favorability rating. The speculation should be greeted with some skepticism. First of all, Shumlin’s numbers are not at all bad, especially for a governor who is: (a) trying to do a lot (doing arouses opposition) and (b) is not warm and cuddly, making him more admired than loved even among his supporters.

Second, the connection between an official’s position on specific issues and his overall approval rating is murky. Most Americans disagreed with Ronald Reagan on a host of issues but elected him twice anyway. Polling measures opinions, but not the passions by which they are held. By all indications, most Vermonters don’t give a hoot whether Vermont Yankee stays or goes. They care that when they flick the switch, the light goes on, and that the cost of keeping it on doesn’t rise abruptly. Considering that New England has abundant electric generating capacity at reasonable prices for at least the next few years, Shumlin is not likely to pay a political price next year for opposing the plant.

The health care initiative does pose a potential political risk for him, but probably not for next year. If the board created by the new law decides that it can’t agree on how to overhaul the system, the governor will be vulnerable for having wasted time and money. But that decision won’t come until after Election Day.

If the board does propose a new health care plan, and the plan leads to higher prices and/or a bollixed health care system, Shumlin will of course pay a political price. On the other hand, if a new system controls prices and leads to better care, he will become more popular than ever.

And not just in Vermont.