Parsing the poll: NH Senate race (Updated: PPP numbers suspect)

*
In his column Wednesday, The Atlantic's liberal-leaning politics editor Marc Ambinder dumps a cold bucket of reality on the left for getting all wee-wee'd up over the latest poll from New Hampshire. "Progressives" are using the survey by Democrat pollsters PPP to argue that GOP Senate candidate Kelly Ayotte is somehow doomed... doomed! because Sarah Palin endorsed the Republican last week.

According to PPP, Ayotte had a seven-point lead over Democrat Paul Hodes in the survey the company conducted in April. PPP polled again over the three-day span of July 23 to July 25, and the Republican is now up by three points. Considering that PPP's methodology has a margin of error of plus or minus 3.6 percent, the actual difference between the April and July polls could be as insignificant as six-tenths of a point. As Ambinder reasons:

I don't see the connection, or much actual movement. Logically, one cannot connect any appreciable drop in Ayotte's numbers to one thing that happened so recently; this is the post hoc ergo propter hoc fallacy. PPP covers themselves by modifiers, as in, "Palin's endorsement may well be playing a role in this." Maybe. The poll suggests that most independents -- those famous New Hampshire independents, that 50 percent of New Hampshire voters -- are less willing to vote for a candidate who gets the support of Sarah Palin. That's an interesting data point, but it screams for more information: of Republican-leaning independents, how salient is this belief? And since PPP last polled in April, why isolate Palin's endorsement as the thing that has damaged Ayotte's standing among moderates? Nothing happened for three months, and suddenly Palin makes Ayotte a Mama Grizzly, and suddenly Ayotte's ratings drop? Eh. More information is needed.

Democrats are in a tizzy about this, as anything involving Sarah Palin's interaction with a candidate is like catnip for them. Truth is, New Hampshire might be one state where Palin's endorsement hurts a Republican if you go by the conventional picture that people have in their minds of New Hampshire voters, but beyond that, there's not much to say or conclude. Truth is, Palin has endorsed some winning candidates and some losing candidates, but she's done so carefully, strategically, and more cannily that she's been given credit for. Truth is, as PPP points out, the Palin endorsement might have helped Ayotte among Republicans. It's an open primary and there's no competition on the Democratic side, so plenty of true independents can vote in the GOP contest if they want. So Palin may have an effect. She may not. We don't know, and won't know, for a while.

Ambinder's voice of reason is sure to be tuned out by hysterical Democrats who see their 2010 prospects sinking faster than the Titantic. They are desperate to find any life saver to cling to, given that the latest weekly generic congressional ballot polling by Rasmussen Research shows Republicans up by ten points over the Dems among likely voters.

In Tuesday's GOP gubernatorial primary in Oklahoma, Palin-backed Congresswoman Mary Fallin was victorious in a 25-point blowout of her nearest rival to win her party's nomination for governor. That makes Gov. Palin a record of 10-4 in her 2010 endorsements for candidates in primary, runoff and special elections. President Obama would give his golf clubs for a record like that. In fact, by Ian Lazaran's analysis:

Palin's endorsement in a state that is allegedly a bad fit for her brings just as many New Hampshire voters to a candidate endorsed by her as the number of voters that Obama brings to a candidate that he endorsed in his home state of Illinois.

- JP

First Update: The same PPP, the pollsters who planted the meme that Sarah Palin's endorsement may be playing a role in what turned out to be very little movement in Ayotte's numbers against Democrat Paul Hodes is now saying that her lead in the GOP primary is "pretty thorough."

On the PPP blog, Tom Jensen reports on some of the poll internals:

She's been able to unite the various factions of the party around her. She's up 48-12 with people who think the party's too liberal, 38-14 with people who think it's too conservative, and 53-15 with ones who think it's the fine way it is. She's the overwhelming favorite of Tea Partiers but she's also up 46-17 with folks who don't consider themselves Tea Party members and 41-10 with those who aren't sure. And she pairs a 43-16 advantage with conservatives with a 48-14 [favorability] with moderates.

Sarah Palin's endorsement has received a lot of attention of late but how important that really is to Ayotte's advantage is unclear. 38% of primary voters say they're more likely to vote for a Palin endorsed candidate to 28% who say that would turn them away from a candidate and 34% who say it doesn't make a difference either way.

Public opinion poll numbers need to be viewed with a critical eye. Most important is the pollster's methodology. What was the size of the sample? What was the margin of error? What are the demographics of the sample, i.e., how is it composed by party affiliation, age groups, and other factors? Did the polling firm weight the same by party affiliation? If so, it is using current statistics or older data? The number of voters who are registered Republicans has been rising since 2009, for example, but some polls are using years-old data which favors Democrats). Was the sample drawn from a sub-population of likely voters or just adults from the general population? Poll results are meaningless unless these questions have been answered. Don't trust any poll which tries to hide this vital information.

h/t this update: roy y

Second Update: Jim Geraghty asked those questions and a few more, and the answers explain a lot about PPP's results. It appears as though PPP should be an acronym for "Patently Phony Polling."

h/t this update: Brian F

Third Update: From a Wednesday editorial in the Concord Monitor:

Public Policy Polling released the results of a New Hampshire survey it conducted Friday through Sunday. The headline on the press release read, "Republican Chances are Good in New Hampshire Senate Race." The release emphasized the low favorability rating of U.S. Rep. Paul Hodes, the only candidate who filed for the Democratic primary, and said he is "behind the most popular Republican candidates Kelly Ayotte and Bill Binnie."

But PPP, as the polling outfit is known, also has a blog. There, under the headline, "NH looking more competitive," the first paragraph read: "Kelly Ayotte's seen her appeal to moderate voters crumble in the wake of her endorsement by Sarah Palin and her lead over Paul Hodes has shrunk to its lowest level of any public polling in 2010 - she has a 45-42 advantage over him, down from 47-40 in an April PPP poll."

Look, the last thing we want to do is take these numbers seriously because they don't even purport to measure the primary campaign currently being waged. But even if you do give credence to this hypothetical match-up, it's preposterous to draw bold conclusions from numbers whose variation from April to now could just reflect statistical noise.

The kicker is, the same PPP blog post ends with this: "Numbers we'll release tomorrow on the Republican primary show that the Palin endorsement has certainly helped Ayotte on that front."

No kidding. Maybe Washington Post blogger Greg Sargent didn't read that far. He blogged yesterday under the headline "More evidence Palin should stay inside her bubble." Whereupon a web editor created a link to Sargent at the top of the Post's home page. And so for a while yesterday, one of the most prominent headlines on the website of one of the nation's largest newspapers read: "Poll: Palin endorsement backfires."

Sure it has.

- JP