Bear with me as I natter on again about polls and their flawed statistics
I know, I know me and my love of stats. I love them because it is through statistics that we can get a really clear picture of what is going on in the world, how humans will or won’t really behave under what kind of conditions and so much more. If you don’t understand statistics, you can be easily hood-winked into believing the statistics that are thrown at you even when the actual statistics, when you delve into the study, reveal something quite different.
I’ve been yelling for awhile now about how polls have been over-sampling Democrats by either assuming that Democrats will turn out in the same historically high numbers as they did in ’08 or, even more egregiously, assuming that they will break the record-breaker and turn out in even higher numbers this time around by anywhere from 3-9% more Dems pulling the levers than during the history-breaking year. Given so many obvious factors, the first assumption is absurd and the second so far out there in la la land it isn’t even worth comment. Except. Now it is clear how far out in la la land that is and just how skewed these polls really are.
I’ve been giving Rasmussen kudos because their model is the only one based on the assumption that the D/R/I break-down will be between the turn-outs seen in 2004 and 2008 and their polls consistently show a dead heat. Turns out, even they are over-sampling Dems and under-sampling both Republicans and Independents by a far margin. Here’s why.
There have been changes in the number of people registered to vote and what their voter registration reflects as far as party identification and we are talking BIG changes. Take the number of people registered as Democrats to vote in little New Hampshire, worth only 4 electoral college votes but still a linchpin state: There are now nearly 20% fewer folks registered as Democrats today than in 2008. It goes deeper. The number of registered Democrats in that State is less today by approximately 9% than it was in 2004 (and the nearly 20 years preceding). These folks have done one of two things: either dropped off the voter rolls entirely or changed their voter affiliation to R or I. The make-up of registered Republicans has also changed –in the other direction: There are more registered Republicans this year than in 2008 although their gains are far smaller than the Democrat’s losses (about 3% point gain). By a 10-1 margin, more Democrats have changed to either Republican or Independent-leaning-Republican than have gone from Republican to Democrat or Independent-leaning-Democrat. Independents, overall, lean Republican by an 11% margin.
New Hampshire is not alone, it is a pattern across the country and in the other swing states. So the assumption that a significantly newly shrunken base of Democrats is going to turn out in beyond record numbers, that a newly expanded base of Republicans is going to turn out in far fewer numbers than history pre-’08 suggests, and that ballooning number of Independents will turn out in the same numbers as before is…quite simply…laughable. Most of these polls are as meaningful as tipping over a cup of tea and reading the election results from the leaves left over — with only one thing clear and that is that the Obama-lead/dead-heat they are reporting is actually an Obama deficit and a big one. A Carter-Reagan big one.