Sunday, October 09, 2005

Kirk Poll Results, Revisted

   The poll conducted by Penn, Schoen, and Berland on behalf of the CD-10 Democrats and released on Friday has had a less than warm reception in the blogosphere(including in our post on the subject here), but does raise valuable insights, if only in providng affirmation that defeating Congressman Mark Kirk(R-Highland Park) would require Heruclean efforts.
   The fact that only 401(what, did they get a bonus if they had more than 400) voters were surveyed and that the margin of error is well over 4% alone should raise eyebrows. Also, the breakdown between Cook and Lake counties confirmed conventional wisdom that, in the 10th, Cook is considerably more conservative, almost curiously so. Kirk enjoyed higher favorability ratings in Lake County, but in the section labeled "Congressional Horserace" voters in Cook County were consistently more likely to self identify as voting for Kirk, with degrees of certainty ranging from "possibly voting for" to "definitely voting for." Their counterparts in Lake County were consistently more likely to identify as voting against Kirk, again with varying degrees of certainty. (The totals of "undecided" and "don't know"-I don't know how the two categories differ-were 32% for both counties.) In 2000, Kirk carried Cook County and Gash, Lake, with Kirk's margin in Cook enough to overcome the fact that more votes were cast in Lake, but is the geogrpahic divide really this harsh? The portions of Cook Coutny outside the 10th are radically different, but Lake is entirely suburban, stretching from conservative Schamburg to moderate Barrington to the home bases of liberal lions Garret and Link in the 10th. (In an example of dramatic irony like no other, Lake is Kirk's home base.) However, voters in Lake County were twice as likely as their Cook County counterparts to offer "his stance on the issues" as the reason they had voted for Kirk in 2004. It seems that among his supporters, being represented by Democrats in the state legislature has no tempering effect on ardence. It could also be that "Kirk" voters in Lake County were more likely to be Democrats, who wouldn't, of course, have decided on party label.
   Other inconsistences lay in the gender tabulation. In the section labeled "Ballot by Democratic Candidate" female voters were 15 percentage points more likely than their male counterparts to favor potential Demcoratic candidate Barry Bradford. (They were 11 percentage points more likely to favor Footlik. The gender divide was least apparent with Smith.) However, in the "Candidate Profile" section, female voters and Democrats were more likely to favor Zane Smith than any other candidate. (The "Ballot by Democratic Candidate" was conducted after the "Candidate Profile one.") One wonders about their methodology: were questions asked about the profiles after each individual one had been read, or after all of them? (The total "likely[to vote for the Democratic candidate]" and "unlikely" boxes for each candidate sum to 100, which would suggest that they were not mutually exclusive.) Smith's was listed first; perhaps sheer sequence is to blame for this inconsistency. However, that would be a serious indictment on Penn's tactics.


Post a Comment

<< Home