I got a little nervous just now when I checked the latest poll numbers. Obama had an almost 8 point lead about a week ago and now the spread is down to 5.9. Why is he slipping? I heard on the news that they expected the race to tighten nearer to the election, but why would that be?
http://www.fivethirtyeight.com/2008/10/todays-polls-1030.html gives a pretty nice explanation about shifting findings from various polls, what’s probably real vs. what’s not.
Lots of other discussion there too about the reliability of polls and polling.
Keep in mind that RCP sometimes picks polls that favor McCain and excludes polls that favor Obama.
Didn’t know that. I thought they just took everyone’s polls and averaged them. Who do you think they are biased?
Since it’s the electoral votes that count, it’s better to see the individual state polls and how the electoral votes are allocated.
I’ve been watching three: www.fivethirtyeight.com, www.electoral-vote.com (both liberal leaning) and rove.com (quite the opposite). None have shown any changes in the electoral votes in the past week.
A change from 8 to 5.9 is within the margin of error and does not necessarily indicate any real change.
RCPs no toss up electoral mapjust changed too. A few days ago, they had Obama at 380-something and McCain was 150-something. Now they have Obama at 353 and McCain at 185.
I can’t speak to whether or not they are biased, but the fact that they don’t clearly disclose their methodology or weighting is a bit suspicious. At least, I couldn’t find it.
When someone comes up with the factual answer to this question, I hope he or she will also tell us why the stock market does whatever it’s going to do today. 
Seriously, there are lots of ideas people have about why you see these tiny fluctuations in the polls, but no one really knows why.
http://www.realclearpolitics.com/epolls/maps/obama_vs_mccain/ shows that overall Obama is still doing very well and has a large lead in the electoral race.
Obama/Biden 311
238 Solid 73 Leaning
McCain/Palin 142
127 Solid 15 Leaning
Toss Up 85
85 Toss Up
Intrade still claims Obama has 364 EV to McCain’s 174.
Here is a discussion at fivethirtyeight.com about Clear Politics dropping and then picking up polls depending on which candidate they favor.
It’s not scientific, but I’ve noticed the same thing in many other elections. The spread between the candidates gets smaller as you near election day.
If you’re still nervous, you might want to knock on wood and read the following article from openleft.com
http://openleft.com/showDiary.do;jsessionid=A784DE4A1330AEAD784ED48B9440540B?diaryId=9555
The numbers aren’t really changing past the MOE, and if you look all the way back to when Obama won the nomination, you’ll see that the graphs are pretty noisy. Obama’s had a big climb lately, it’s only natural that it’s going to settle back into equilibrium.
While we’re in GQ, can somebody please tell me the difference between the projection and the trend lines on 538? I’m not sure I understand the distinction.
Roughly speaking, the trend is the way the polls have been moving, and the forecast is the way they expected the polls to move.
Intrade is not a reliable source. See this recent thread for discussion.
The commentary at FiveThirtyEight may be pro-left, but I’ve never seen any reason to think their model is biased. Have you?
I find it odd that guy focuses exclusively on Iowa, New Mexico and Colorado. It may or may not be the safest guess according to the polls; I’ve seen similar point spreads in Ohio and Virginia, and I think both of those states’ polls close before any of the others. However, hesitance to rely on Ohio is certainly understandable.
All guys want to get laid, so they were hanging out with the cool chicks and promising to vote obama in hopes of getting some.
Now its come down to the wire and the phrase cold day in hell has probably been uttered by the obamettes and the guys are thinkin, screw this , I’m voting McCain.
declan
I am not too concerned- I haven’t been polled by anyone, and I am registered independant- but I am hella for obama. I also know several others in the same situation.
To be a bit more precise, the trend line is the expected result if the election were held today (roughly the same as the RCP average). The projection line is what will happen on Nov 4. There is a projected tightening as the race gets closer based on historical trends. This is why the projection line typically shows a lower margin of victory than the trend line (5.5% vs. 6.5% as of today).
There is some debate as to whether it even makes sense to have a projection line. The only thing you can really model with polling is what would happen if they election were held today (in fact that is typically how the questions are worded in the polls). Just because on historical average there is tightening doesn’t mean it makes sense to incorporate it into the model, IMO.
I haven’t, but the fact that the model was constructed by a left-leaning human, and that humans are fallible, means that one should at least consider the possibility that the model is biased. It’s probably safe to assume that the truth lies somewhere in between the results shown by 538 and Rove.com (though I would suspect closer to the 538 end).