When it comes to polling numbers, I’m not really into spin. I want real numbers uncluttered by bias or poor poll design.
Some of the issues that continue to be of concern are:
1. Over-sampling of democrats. That’s grossly wrong, but it is not a problem with Rasmussen.
2. The 9% - apparently only 9% of the population will bother answering pollsters any more. One has to wonder if the oddities that make them willing to respond to pollsters also makes them unusual regarding their choice of candidates.
3. The cellphone/landline problem. Pollsters claim to have solved this, but it is nonetheless true that senior americans are far less likely to be a “cell phone only” household.
4. The time of day/week concern: Calling on weekends gets a certain type of person and calling during normal working hours gets a different type of person.
These are all concerns in polling that could explain the wide disparity between actual results and exit polling results from the last few elections.
The easiest answer (and legitimate, in my view) is that polling organizations manipulate polls for personal/political reasons. Some will attack that response and ask why in the world they’d do that. The answers are too obvious: good old fashion being bought off, or a strongly held political view, or publicity for the firm, etc.
Re Balance has the Rasmussen poll at
48.4- 45.4 using a +0.44% D sampling model.
http://polls2012.blogspot.com/
Gallup at the same number with the correction applied to the 0.44% D sampling model.