Kaosium
Philosopher
- Joined
- Oct 12, 2010
- Messages
- 6,695
The final Rasmussen poll in '08 was within the margin of error, that's a 'hit' not a miss. Their 'miss' was in the overall '10 congressional race which my guess would be was a result of underestimating the Dem turnout. By and large the states that have become reliably Dem generally have a higher turnout tradition, e.g. California turns out to vote moreso than Texas, the middling (population) states like Maine, Washington, Oregon, MA, MN, have stronger voting habits than the ones more generally Republican like the deep south, Nebraska etc. So even in a low-enthusiasm year for the Dems they're gonna get a certain minimum from those Dems who always vote as it's cultural, and when they suggest they might not vote it means less and thus actually should be considered 'likely' voters even if they suggest they're not. They're just pissed off at whatever and despite what they tell the pollster are more likely to actually vote because it's (moreso) ingrained in them.
So despite the GOP base being all worked up and with momentum as shown in his latter polling, that advantage was mitigated by fact the Dems will get a turnout minimum from the one who consider the exercise of their franchise as their duty in a participatory democracy. Dems have a higher percentage of both higher turnout voters and lower turnout voters on each end of the scale, kinda like a 'U' shaped curve with the mass majority in the middle. At least that's the way it used to be, and a cursory glance appears to indicate that hasn't changed. That's the sort of thing that's very hard to catch when trying to differentiate between 'likely' and the 'less likely' to vote category which in a mid-term election year like 2010 (where turnout is always down ~10% of the eligible population) might skew someone looking specifically for 'likely' voters as Rasmussen does.
In this election cycle I'm starting to suspect that there might be a systematic error cell-phone only biasing many polls in favor of Dems as a result of the last election and the changing habits of the population. That election was historic in that so many these days use only cell-phones and in '08 that population was heavily Dem; many people outside the most partisan would be displeased were they to receive constant polling attempts on their cellphones. Having endured the past two years in Wisconsin, where it was determined that one vote for governor wasn't enough and we had to do a replay, (with primaries and everything!) as well as recall elections in our state legislature and thus am a 'survivor' of being constantly assaulted with phone messages, I've noted a couple of times when unwarily just answering the phone and getting
If those cell-phone only populations are polluted with partisans that might well bias many polls in favor of Dems, which can be seen in many of these polls I looked for the particulars on, which show both more Dems than ought to be responding, and that's even borne out moreso by the cell-phone demographic. I've also noticed the larger samples such as the recent Gallup and other polls tend to be more in favor of Romney than the other smaller sample polls which might indicate a cellphone partisan bias being mitigated.
What I'm getting at is the number of people interested in being polled regularly through their cell-phone is more likely to be dominated by partisans, and that population in '08 was heavily for Obama. While I'm sure they make every effort to expand their potential sample base, with cellphone-only voters that's not as easy to do as with others who can be more conveniently called randomly, therefore it might just be that many of the polls that appear to be more favorable to Obama are due to a part of the sample being Obama partisans being polled over and over again, as they have been since '08.
How many times have you guys been polled this cycle? Granted circumstances were and are different here in Wisconsin, but despite trying to actively avoid those calls most of the time, I must have responded to a half-dozen since shortly before the conventions. That's probably only a small percent of the attempts as I'm catching on to how to differentiate them on caller ID and recognize now just what ones were probably pollsters and which ones are/were robo-calls. What I'm wondering is that if there's a population that skewed heavily Democratic in '08, and is more difficult to expand, that too much of the cell-phone only population is carried over from the '08 election and polling the same partisans over and over isn't actually sampling a random population as good as it might. Thus the larger sample polls and Rassmussen who uses a different methodology to account for the cell-phone bias are more likely to be accurate this election than many of the other ones.
Just a thought, I could be wrong of course.
Last edited:
