posted at 6:30 pm on November 3, 2012 by Ed Morrissey
Duane, I have to say I’m surprised that pollsters like Marist have not adjusted their likely-voter screens as Election Day nears, and I’m not sure that polling criticism isn’t the reason. In elections past, sample data didn’t get as much scrutiny, perhaps because in 2008 the outcome wasn’t in that much doubt. I’m wondering if pollsters like Marist simply are doubling down on earlier bad calls, afraid to move because it would undermine the long-term trending in their poll series.
In cycles past, we would see media polls like the WaPo/ABC and CBS/NYT suddenly adjust their samples in the final week in order to have bragging rights after the election. Either these pollsters are convinced beyond any doubt that the turnout model is going to be similar to 2008, or they are more reluctant to shift samples at the end with all of the increased scrutiny. We’ll see how that works out on Tuesday, but at least we can be sure that the final assessments in 2012 will apply to the totality of their polling, and not just the final week.
Recently in the Green Room: