The Fighting 29th

All about New York's 29th Congressional District

The National Republican Congressional Committee has released the results of a January poll that show Randy Kuhl 20 points ahead of Eric Massa. As with all polls from candidates or their representatives, this one should be taken with a grain of salt.

The WHAM13 blog post on the poll has the detail released by the pollster. Here's the tell:

Interview selection was random within predetermined election units. These units were structured to correlate with actual voter turnout.

One of the most important characteristics of a poll is its likely voter screen, the questions that pollsters ask to judge whether the person responding will actually vote. These weasel-words dodge the question of how voters were selected. With a sample size of 300, a slight tweak of the voter screen will lead to a huge effect in the poll's outcome. That's where this ham was watered.

What's more interesting is why the NRCC would release a poll so old that the Presidential choice section doesn't even include the presumptive Democratic nominee. The answer is that campaigns release these polls when they want to goose fundraising by convincing donors that their candidate has a chance. Since Kuhl is trailing Massa in funds raised, this poll is no surprise. What's really interesting is the weakness of this effort. The NRCC is so broke that they can't even afford to pay for a fresh poll, so they release this chestnut.

Update: As Zabriskie points out in the comments, the poll was conducted by McLaughlin and Associates on January 27-28. On January 30, the Kuhl campaign paid $16,215 to McLaughlin and Associates for a "Survey". So the NRCC press release is leveraging an old Kuhl poll. And the following assertion reported in the 13WHAM post is laughable:

“As far as the NRCC, Randy is a completely separate entity from the NRCC and really has no idea what they will or won’t do,” said Meghan Tisinger of Kuhl’s office.

A new National Public Radio/Kaiser Foundation/Harvard poll on S-CHIP has been released, and pollster.com has a summary.   The results are somewhat different from the USAToday/Gallup cited by Randy Kuhl.   When asked about the possibility of S-CHIP leading to people dropping private insurance, 65% said still favored S-CHIP expansion.  64% said they supported the veto override, and 52% said they weren't concerned about S-CHIP expansion leading to government-run health care.   35% said they would be less likely to support their Member of Congress if he or she voted to sustain the veto.

The difference in the two surveys is in the phrasing of the questions.  This shows that one poll by itself proves very little, a lesson that all of us should remember when it comes to polls in the 29th next year.  The full detail on the NPR poll is here [pdf], and the USAToday/Gallup detail is here. 

In other news, the Hornell Evening Tribune carries a money roundup today, calling the money race "neck and neck". 

Finally, my inability to punch in a 5-digit code correctly prevented me from covering the Massa press conference yesterday.

Randy Kuhl's blog, and the Ontario GOP blog, take encouragement from a USA Today/Gallup poll on S-CHIP.  Both note that a majority of those polled (55%) said they were "somewhat concerned" (33%) or "very concerned" (22%) that S-CHIP would be an incentive for the middle class to drop private health insurance.   And another majority (52%) preferred Bush's plan to limit S-CHIP to a family of 4 making less than $41K, versus the Democrats' plan to make the upper limit $61K.

Both Kuhl the the Ontario GOP omit another interesting statistic from that poll.  When asked who they trust to get S-CHIP right, 52% say that they trust the Democrats.  Only 32% trust the President.   Another interesting fact:  17% of the respondents were following S-CHIP "very closely", and 34% "somewhat closely".  Almost half of the public (49%) are following the issue "not too closely" or "not at all".

So, when a generally uninformed public is presented specific facts on S-CHIP, they tend to agree with Republicans.  When those facts are absent, their general distrust of the President governs their opinions.  The news on S-CHIP could be worse, but this certainly doesn't qualify as good news for Republicans.

Also, it's interesting to see Republicans in the same position where Democrats usually find themselves.  The Republicans' position is based on a careful reading of the details.  Democrats are aiming straight for the gut.  When half of the public doesn't really care, the gut carries the day.

Update:  Via pollster.com, here's the Gallup writeup on the poll.

There's been a lot of poll analysis in the last couple of weeks. Today, Pollster.com posted a discussion of where their method of averaging available polls failed. The 29th was one of those races. No particular conclusions were drawn about the 29th, but it is clear that this district was very lightly polled. When there's hardly any data to start with, any average is going to be more suspect.

Turning to the actual polls, Rochesterturning has been posting graphical interpretations of election results. Today's post shows that a significantly smaller percentage of voters voted for Kuhl in 2006 than in 2004 in almost every county. This is especially interesting because the Conservative party fielded a candidate in 2004, as did the Independence party. Both of those candidates should have taken some votes from Kuhl in '04 that he regained in '06. In this analysis, and the other county-by-county post, keep in mind that the graphs are unweighted. So, for example, when you see that Kuhl lost roughly 15% of his vote in Yates county, remember that Yates had just under 7,000 votes cast in 2006. The 29th had about 200,000 voters in '06.

Rochesterturning also looked at correlations between census data and voting patterns. They found a positive relationship between income, education and a tendency to vote for Eric Massa. There are two ways to interpret this data. A partisan spin for Democrats is that smart people voted for Massa. Republicans could argue that the Democrats are turning into the party of the elites, rather than the party of the "average working man". I don't buy that. I agree with RT that those numbers show that more educated people are more likely to inform themselves about the candidates in an election, and therefore are more likely to vote for a less-well-known challenger like Massa. Also, the income split mirrors the North/South split in the 29th, with Monroe being the most wealthy county. By registration, that county is also the most Democratic of the bunch.

A new Majority Watch poll in the 29th, taken from October 24 to October 26 and released today, shows little movement from the last poll taken ten days earlier (52%/40%).  The crosstabs [pdf] are almost identical to the last one, and the same grim conclusions can be drawn for the Kuhl campaign. 

Looking through the races at pollster.com's House page, it looks like NY-20 is still one of the area races where Majority Watch and other pollsters disagree.  In that more heavily polled race, Majority Watch leans farther toward the Democrat, Kirsten Gillibrand, than even her own polls.  Yet there are a number of other races were Majority Watch is consistent with other non-partisan polls.  So this one could be dead on, or out to lunch.  We'll find out a week from tomorrow.

Kuhl's spokesman, Bob VanWicklin, spins wildly on the Majority Watch poll:

It's a left-leaning group. I wouldn't call them independent. They're a group that wants to see a Democratic majority in Congress, from what I understand.

Majority Watch is a project of Constituent Dynamics and RT Strategies.  Constituent Dynamics is a recently established polling company trying to perfect IVR polls, or robo-polling.  RT is a joint project of one Democrat and one Republican.  They're the same firm that does polling for independent, well-respected political analyst Charlie Cook.  Many have called their polls wrong, nobody's called them biased.

And we don't know anything about the methodology of the poll, other than the little information they gave, so we're sort of skeptical about the questions they may have asked, or how they asked them.

A full discussion of methodology, as well as every question asked in the poll [pdf], is easily accessible on the Majority Watch site.  As for how the questions are asked, it's a robo-poll, so every question is asked the same way by a pre-recorded voice.

It doesn't at all compute with any of our internal polling [...] So we'll just sort of discount it.

Finally, after two heaping helpings of bullshit, a little bit of truthful spin.  I expect spin from a flack, but Mr. VanWicklin's bull-to-spin ratio is chronically high. 

As noted in this morning's earlier post, NY-20 was the major outlier in the races where other polls could be compared to Majority Watch's work.  Today's Siena Poll in that race adds more reason for skepticism.  It shows Gillibrand trailing Sweeney by 14 points.  The Majority Watch poll, taken around the same time, shows her ahead by 13, a 27-point spread.

Is Majority Watch full of crazy talk?  That seems to be the theme of some postings here and on other blogs.  While keeping in mind that old chestnut "the only poll that matters is on election day", let's look how Majority Watch polls stack up in 8 9 other contested districts with multiple recent polls.

It looks like there are three four districts where Majority Watch is in the ballpark:

• IL-06:  MW calls it Duckworth (D) by 5.  Other recent polls are tied or Duckworth by 5.
• MN-06:  This is the Wetterling race, the mother of an abducted child who gave the Democrats' radio address on the first weekend of the Foley scandal.  Polls taken early in the month show Bachmann (R) by 3.  The MW poll taken on 10/12 shows Wetterling by 5.  A newspaper poll taken 10/16 shows Wetterling by 8.
• NM-01:  In this heavily polled race, both MW polls are squarely in the middle of the other polls.
• WA-08: MW is within a couple of points of a recent SurveyUSA poll, and both show Reichert (R) with a slim lead.

Here are races that have only partisan comparison polling, but MW seems reasonable:

• CT-05:  MW has Johnson (R) by 6.  The other polls are all partisan, with the Republican poll showing the Republican by 10, and the Democratic poll showing the Democrat by 1 or by 5.
• FL-13:  The Democratic poll has Jennings (D) up by 8 last month and 12 this month.  MW has him up by 3 last week.

Here are some districts where MW may be an outlier:

• CO-07:  MW has it as a tie.  Other recent surveys have Perlmutter (D) up by 6 or 11.
• NY-20:  Gillibrand's (D) own poll has her up by 1.  MW's, taken two days later, has her up by 13.
• NY-26:  MW's Davis (D) by 16 is the widest margin so far, even though Zogby's poll taken at about the same time has Davis by 15.

Other than NY-20, I don't see a poll where Majority Watch looks way out of sync with the rest of the professional pollsters.  But, as with the first poll in this post, I recommend that all readers who are interested in polls read the excellent analysis at pollster.com, which makes it clear that polling in House races is far less accurate than Presidential race polling.

(Update:  Forgot WA-08)

The crosstabs [pdf] for the Majority Watch poll show how important the Monroe/Ontario suburbs (the "Volvo-Donut") are to the Massa effort. Majority Watch divides the district into the North (Monroe and Ontario) and the South (the rest). In the North, Massa leads Kuhl by 26 points. In the South, Massa's ahead by one percent.

According to this poll, everything's worse for for Kuhl in the North. The "voter motivation index", which measures how likely it is that voters are going to vote, is the highest in there (7.76/9.0). Voters in the North are the most certain in their choice (83%), and the least pleased (67%) with the job that Bush is doing.

Overall, the bad news keeps on coming for Kuhl. 82% of voters in the entire district are "certain" in their choice. Only 9% of the voters are undecided. The voter motivation index for Massa voters is higher than that of Kuhl voters. Democrats are more motivated to vote than Republicans.

With undecided, certainty and motivation numbers like these, it's hard to see light at the end of the tunnel for the Kuhl campaign. If this survey is right, voters in the 29th have made up their minds, they've chosen Massa, and they're going to vote that choice.

Majority Watch, an independent, non-partisan polling project, has Eric Massa leading Randy Kuhl by 12 points in a poll taken Sunday and Monday in the 29th. The poll uses IVR (touch tone response) technology and has a margin of error of 3%.

This is the first year of the Majority Watch polling project, the methodology hasn't been proven, and House races are hard to poll.  Nevertheless, this is huge news for the Massa campaign. 

Full crosstabs [pdf] are available and worth a closer look.