| Earlier, KT, Matt and I compared and contrasted the "old media voodoo predictions and new media facts and analysis" styles of making predictions on Texas House races. Our fundamental problem was that the old media -- and the old consultant class, which is apparently who Burka was talking to -- use terrible, terrible logic and analysis at looking at House races. They are thinking about things as if we still lived in the 1980's. Many Democrats, though, believe that research and facts should dictate analysis and predictions. The Lone Star Project has, for the past few years, done excellent work on fact-based research and analysis. Matt Angle, director of the Lone Star Project, shared with me how they analyze House races -- and it is really quite interesting. Lone Star Project's Advanced Analysis Structures for Analyzing Races First of all, I must impress on our readers that this is just the "top line" analysis numbers. There is a much, much, much more sophisticated voter identification processes that goes into all of this -- but that's not we're talking about. This is the "top line", or the basics, as to how the Lone Star Project uses fact-based research and analysis to predict House races, and Angle shares it with us as a learning tool for our community: From Matt Angle, in an e-mail he sent me earlier today (edited for easier reading): Our straightforward, “top line” analysis starts with - An expected vote calculated by the National Committee for an Effective Congress (NCEC), after consultation with the Lone Star Project. NCEC uses advanced statistical models to predict the expected vote down to the precinct level. Some obvious key elements of their calculation are the number of eligible voters, the number of registered voters, demographics and historical turnout patterns.
- The Lone Star Project takes the NCEC projected vote to calculate our 2008 win number.
- We build toward a win number by stacking the votes Kerry received in 2004 with the 2008 Democratic primary voters who did not vote in 2004. This calculation combines a realistic and conservative base performance (Kerry in 04) with the likely benefit a Democratic candidate will receive from the improved political environment and new Democratic participation in 2008.
Compare that with the models that Burka is repeating, which I guess is being spread by Republicans. Not even close, huh? Angle also shared four PDF models for the four races Burka listed in his post. To see how much better of an analysis the LSP (and even BOR) has in comparison to what Burka shared, here are links to the relevant analysis from all three of us: HD 96 - Democrat Chris Turner vs. Republican Rep. Bill Zedler
HD 97 - Democrat Rep. Dan Barrett vs. Republican Mark Shelton HD 101 - Democrat Robert Miklos vs. Republican Mike Anderson
HD 102 - Democrat Carol Kent vs. Republican Tony GoolsbyWe post this to show how much better it is to use real numbers and real statistics to determine predictions -- and how being public and accountable, as Angle has done today and as we at BOR always strive to do -- is better than using ridiculous models from Austin insiders, like Burka has done. All the more reason to believe Democrats are much, much more organized and prepared to win this year. P.S. If you like the analysis of the Lone Star Project -- or are at least thankful for them openly sharing their top-line analysis like this -- feel free to donate some dollars to their cause. |
