New Poll: Schansberg Spoiler Speculation

Taking center stage.
SurveyUSA is out with another poll done on behalf of WHAS11 on the 9th District race. It continues to show a right race largely within the margin of error, with challenger Baron Hill leading by four points over incumbent Mike Sodrel.

Yet as Hill's lead has widened, his own standing within the polling declined from the last SurveyUSA poll, taken twelve days ago. Hill lost one point, Sodrel three. Schansberg gained three, up from two. SurveyUSA concludes from this that Schansberg is stealing votes from Sodrel. I am not sure that this is the entirely the case. Some of the internals on the poll are different from the last one, and that likely explains some of the movement. More on that later.

Schansberg has had a good week. As much of a good week as could be had by a Libertarian third-party candidate in one of the most hotly-contested Congressional races in the country, at least. The News & Tribune did a flattering column about him over the weekend. Notable are his website and campaign blog.

I suspect that the five percent number for the Libertarian is a statistical blip. It hasn't been reflected in any other poll (all of which have shown a narrowing and now close race between Hill and Sodrel consistent with the rest of this poll). If presented with a well-articulated message about the national importance of the race, I find it unlikely that many conservatives will bleed from Sodrel to Schansberg on election day.

The sorts of conservative voters that could vote Libertarian as a protest tend to be more educated, and thus more understanding that such votes could have consequences--like Democrats running Congress--that they most certainly do not want. Sodrel is hardly Lincoln Chafee; it is difficult for him to get much of a revolt on his right given that doing so could result in something even worse from the ideological perspective of such voters.

Many Republicans, and some Democrats, will vote Libertarian as a protest vote against Senator Dick Lugar. But that is a place were such a vote sends a clear message without shooting yourself in the foot. Conservatives voting for Schansberg to punish Sodrel are cutting off their nose to spite their face. It would be no different with liberals voting for Baron Hill if there was a Green candidate in the race.

Now on to the poll, and a look at how and where the numbers differ from last time.

Gender
Women remain largely unchanged. There is some erosion from Sodrel to Schansberg among men, but there is also erosion from Sodrel to Hill among men. It is among men that this poll is differing from the prior one. Women appear to have made up their minds. It will be interesting to see if the NRCC's recent set of anti-Hill sex and violent video game ads will have an impact on mothers in the 9th District.

Age
A statistically-significant smaller number of people in the 18 to 34 age group were polled in this poll than last time. Sodrel carried this group big last time. This time, he loses it to Hill, and Schansberg makes big gains. That movement to Schansberg, however, is of six people when you look at the numerical response counts instead of the percentage counts. This is probably a blip.

Sodrel holds the 35 to 49 demo, but it appears that he has lost voters to Schansberg while gaining them from Hill. Sodrel holds steady in the 50 to 64 demo, with Hill losing to Schansberg. The senior demo holds steady for everyone. Movement to Schansberg from Hill is illogical, if the conservative desertion thesis of Tom Wolf (in the News & Tribune article) is true.

Party
Sodrel lost five points among self-described Republicans. But those voters didn't go to Schansberg, who pretty much held steady. They became undecided; could be lingering anger over Foley or general base irritation. Minor variation among Democrats (Sodrel up to 13% from 11%, Hill down from 83% to 82%, Schansberg up 1%) are probably a statistical blip. I retain my question from the prior poll about party affiliation in the 9th District. Sodrel has bled independents to Schansberg. The former dropped 8%, and the latter gained 11%. This fits with my own thesis that conservative Republicans are unlikely to desert Sodrel given the importance of the race, though independents could.

Ideology
The conservative and moderate identification numbers have flipped since the last poll, though this is probably anomalous. Sodrel lost 3% of Conservatives (three people in the actual respondent count); 2% went to Schansberg and 1% became undecided. Sodrel lost 7% in moderates. Hill gained 3% and Schansberg gained the rest. Baron Hill lost 3% of liberals to Schansberg, and Sodrel gained 2%. This last result does not follow from the Wolf thesis, and I think indicates something slightly wonky at the margins of the polling.

Education
Sodrel lost 3% to Schansberg among those with no college education. He lost 8% among those with some college; 6% went to Schansberg and 2% became undecided. Sodrel lost 2% among those with college education. Baron Hill lost 6%. Six percent of those went to Schansberg, 2% became undecided. Six percent of grad school voters moved from Sodrel to Schansberg.

Income
This one is surprising. Among those making less than forty thousand a year, Sodrel has lost 2%. Baron Hill, however, lost 7%. They pretty much all went to Schansberg. The Libertarian's rhetoric about the working poor must really be helping. Either that or the poll is genuinely screwy. Three percent in the 40K to 80K demo moved from Sodrel to Schansberg. Sodrel lost six percent among those making more than 80K; they split between Schansberg and the undecideds.

Bush Job Approval
Schansberg, again, makes gains. Four percent of those approving of Bush moved from Sodrel to Schansberg. Two percent of those disapproving of Bush moved from Hill to Schansberg. Quite curious.

Generation
As noted before, fewer Gen Y voters were polled in this poll than in the prior one. Indeed, that entire demographic flipped entirely from Sodrel to Hill, with nominal gains for Schansberg. That makes no sense. Are the NRCC ads attacking Hill over violent video games upsetting those that actually play violent video games, or is the smaller sampling of this demographic distorting the results? I suspect the latter.

Sodrel has erosion in Gen X to both Hill and Schansberg. The so-called Jones generation has flipped; Hill lost 15%, Sodrel gained 6%, and Schansberg picked up the rest. Sodrel lost 4% to 5% of the Baby Boomers to Schansberg. The mature demo remains unchanged.

Concluding Thoughts
It is worth bearing in mind that the internals of the poll can be swayed by a handful of respondents. The margins of error within the internals of the polling are actually greater than the rest of the poll itself, because the samples of each demographic are correspondingly smaller. For example, the Gen Y sample in the current poll was only 44 individuals. That's a very large margin of error. This explains a lot of the wonky results, like the flip-flop of twenty-somethings.

There is movement overall to Schansberg from Sodrel within the poll. The question is whether this is an actual phenomenon, or whether it is a result of wonkiness in the polling. The latter is a serious possibility, when you consider some of the illogical areas where Schansberg was making gains, such as in middle-aged voters from Hill to Schansberg.

I split the difference. I think that Schansberg has made some gains, and those gains came mostly at the expense of Sodrel and are a virtue of his recent (unsustainable because of limited funding) barrage of radio ads. I think that other elements of the Schansberg gain in this poll are in the margin of error. The next SurveyUSA poll will tell whether this is an enduring trend or a blip, or a mixture of the two.