Friday, November 16, 2007

Even More on the Pittsburgh Mayoral Race

My thesis on this series of post is that the Pittsburgh progressive community, or at least the activists who swing on the same grapevines that I am affiliated with, constitute one of the major swing blocs within the city limits. Normally these people will vote Democratic, and tend to support City Councilor Bill Peduto and his allies, but I believe that they, or at least the precincts which they heavily dominate will often vote for technocratic, non-asshole, non-partisan Republicans. And this ability to swing has rendered the progressive activist demographic moot at the city wide level, although they can have significant impact on Democratic candidates at the sub-city level (Dowd, Kraus, Arnet etc) and higher than city levels. The last post in this series was an attempt to build a baseline of Republican support for different candidates and campaigns. We learned a few things.

  • Democrats dominate in high turnout elections against self-identified conservatives


  • Republicans have tended to do better, but still poorly, with technocratic, non-social conservative candidates in low to medium turnout elections.


  • The East End will flip, and Chris Briem fills in some more back-story at his place.


  • Mark DeSantis' vote share was most highly correlated with the former GOP County Executive Jim Roddey vote shares, .87 with the 1999 Roddey campaign, and .82 with the losing 2003 Roddey campaign.


  • DeSantis had decent but weaker vote share correlations with the two previous GOP nominees, Weinroth @.72, and Carmine @.52.
  • Bill Peduto ran as a liberal/progressive technocratic reformer with a good governance bent in the 2005 Democratic mayoral primary. He lost, and he lost big time as he was unable to expand outside of his local base in the East End which is where he dominated, but he got crushed everywhere else in the primary. The Angry Drunk Bureaucrat mapped this out nicely in 2005, and explained the Pittsburgh Democratic political scene at the same time.

    The obvious remarks are that Bob O'Connor won big and broadly across the city neighborhoods, picking up local presinct wins across all segments of the city... except for a big blue blob in Shadyside, Friendship, North Point Breeze, Point Breeze, and Squirrel Hill. That portion of the City, The People's Republic of the 14th Ward, fell to the armed ruffians of Bill Peduto. Similarly, Bill picked up support in segments of Highland Park, Southside, and the Mexican War Streets. As I said before, Bill's wins were not broad based, but rather reflective of a bourgeoise-bohemian ethic, which is popular in certain segments of the City... namely the ones listed above. Lamb, by contrast, clung to victories in equally concentrated areas: Duquesne Heights, Mt. Washington, Brookline, and Beechview. I would characterize these neighborhoods as post-war suburban Pittsburgh ethic; upper working & middle class.
    Richard Florida has a strong following in the Peduto areas of the city. At some points in 2003 and 2004 you could not go to a house party in Shadyside without tripping over his book. Peduto is a Florida favored candidate and so is Mark DeSantis, as Florida wrote on his blog last week the following passage on DeSantis:

    DeSantis worked for John Heinz and is one of a dying breed of sensible John Heinz-style, moderate Republicans. He is non-partisan and open-minded on most every issue. He believes in markets, innovation and diversity - a true 3T thinker and doer......Wouldn't it be interesting if he [DeSantis] ran for county executive, where the political odds are far more in his favor. And even more so, if after four[ed note: 2 years is the term for this election] more years of the current mayor continuing to shoot himself in the foot, if Bill Peduto could take the mayor's spot. Now that would be a one-two punch worth watching! Maybe it's wishful thinking [emphasis mine]
    I think Florida is onto something here on who is voting for Peduto and DeSantis, so I ran a couple of quick checks. The first one is a basic correlation between Peduto 2005 Dem. Primary vote share by Precinct compared to DeSantis general election correlation. This is the same exact correlation I've been running with other candidates.

    This correlation is higher than the correlation between DeSantis vote shares by precincts and previous Republican mayoral candidates. Interesting.....

    Next I ran a single variable regression without taking into account precinct characteristics or nesting [I know this is quick and dirty!] to see if Peduto's vote share was a strong predictor of DeSantis vote share. It is, and here is the result:

    Okay, there are some methodological problems with this type of analysis as each precinct is not indepedent of each other, for example, the polling place I used when I lived in Pittsburgh was in the basement of a local synagogue. It served two divisions/precincts. My precint boundary was the north side of a street and its population was a mix of graduate students, young professionals, Russian immigrants and assisted living Jewish retirees. The other precinct served was on the south side of the street, and its population was a mix of graduate students, young professionals, Russian immigrants and assisted living Jewish retirees. There is some covariance (<3%) here which is not being accounted for. That makes the standard error appear smaller than it actually is.

    However even with this proviso, it is safe to say that the Peduto Dem primary vote share is a good predictor of DeSantis vote share, which allows one to imply that Peduto voters are rather likely to be DeSantis voters also....

    No comments: