- Joined
- May 19, 2011
- Messages
- 4,811
- Reaction score
- 2,647
- Gender
- Undisclosed
- Political Leaning
- Undisclosed
First, let me say that I'm the first in line to call a conspiracy theory a conspiracy theory. People who think that polls are biased against one candidate or another, or provide useless information are quite frankly, deluded. Polls are a great resource that provide invaluable information about the direction of the race.
However, I'm not sure that the polls are currently doing a great job of accurately predicting what's going to happen come election day.
Likely voter models may be double, or triple counting voter intensity.
If you interview 1000 randomly selected people, you're going to have an extremely accurate measure of the race. The problem is selecting random people. It's fairly easy to randomly select people to contact, but only 10% or fewer people actually respond to surveys. Moreover, I have a suspicion that this number is dropping as the race drags on; as people are inundated with campaign adds and robo-calls.
That means that the people who you're surveying are the people who are fired up enough about the race to waste time responding to a poll. Granted it's not a scientific analysis, but I'm fairly engaged in the race and I wouldn't give a pollster two seconds of my time. That means that you're only polling the likeliest of likely voters. Applying likely voter models to this data, could actually be making it less accurate.
What are we seeing in the polls? In likely voter models, we're seeing a tightening of the race in battleground states, on the national stage, and among gender and ethnic groups. However, in the registered voter models we're seeing a very consistent race.
If there's one thing that has been abundantly clear from the beginning of the race it's that there have been a relatively low number of undecideds. This has been true in all polls as well as anecdotally among people I know. I simply don't think it's possible that Obama has closed the gap among men while at the same time Romney erased a 16 point gap among women and a 40+ point gap among Hispanics.
For this to be true, then we'd be looking at an electorate with 7-8% undecideds in which 8-10% of the people changed their minds and supported the other candidate while reducing the undecideds by no more than half.
I suspect that very few individuals have actually changed their minds. Obama probably has the same 5 point lead he's had among registered voters since the beginning of the campaign.
That said, a registered voter isn't the same thing as an actual voter. Republicans generally have better turnout than Democrats and it's likely to be even more true this year. However, Obama clearly has a better GotV effort in battleground states than Romney.
What does any of this mean?
I think it means that we don't really know who's going to win. Sure, the average of the polls may be close to the truth, but that may be more dumb luck than anything else. I'd suspect that Obama's a 3:1-4:1 favorite, but so would any incumbent in a reasonably good, but not great economy.
Bottom line... I'm not sure the poll are going to mean all that much over the next two weeks and your vote may or may not make a difference. But you won't know until after the election. So vote your conscience. If nothing else, it gives you the right to gripe for the next four years.
However, I'm not sure that the polls are currently doing a great job of accurately predicting what's going to happen come election day.
Likely voter models may be double, or triple counting voter intensity.
If you interview 1000 randomly selected people, you're going to have an extremely accurate measure of the race. The problem is selecting random people. It's fairly easy to randomly select people to contact, but only 10% or fewer people actually respond to surveys. Moreover, I have a suspicion that this number is dropping as the race drags on; as people are inundated with campaign adds and robo-calls.
That means that the people who you're surveying are the people who are fired up enough about the race to waste time responding to a poll. Granted it's not a scientific analysis, but I'm fairly engaged in the race and I wouldn't give a pollster two seconds of my time. That means that you're only polling the likeliest of likely voters. Applying likely voter models to this data, could actually be making it less accurate.
What are we seeing in the polls? In likely voter models, we're seeing a tightening of the race in battleground states, on the national stage, and among gender and ethnic groups. However, in the registered voter models we're seeing a very consistent race.
If there's one thing that has been abundantly clear from the beginning of the race it's that there have been a relatively low number of undecideds. This has been true in all polls as well as anecdotally among people I know. I simply don't think it's possible that Obama has closed the gap among men while at the same time Romney erased a 16 point gap among women and a 40+ point gap among Hispanics.
For this to be true, then we'd be looking at an electorate with 7-8% undecideds in which 8-10% of the people changed their minds and supported the other candidate while reducing the undecideds by no more than half.
I suspect that very few individuals have actually changed their minds. Obama probably has the same 5 point lead he's had among registered voters since the beginning of the campaign.
That said, a registered voter isn't the same thing as an actual voter. Republicans generally have better turnout than Democrats and it's likely to be even more true this year. However, Obama clearly has a better GotV effort in battleground states than Romney.
What does any of this mean?
I think it means that we don't really know who's going to win. Sure, the average of the polls may be close to the truth, but that may be more dumb luck than anything else. I'd suspect that Obama's a 3:1-4:1 favorite, but so would any incumbent in a reasonably good, but not great economy.
Bottom line... I'm not sure the poll are going to mean all that much over the next two weeks and your vote may or may not make a difference. But you won't know until after the election. So vote your conscience. If nothing else, it gives you the right to gripe for the next four years.