queenzboulevard – my favorite site which has a pro Obama slant (but which is not 24/7 Obama propaganda) is www.fivethirtyeight.com.
The site has two purposes. The first, and main reason I visit it several times daily, is that it aggregates the polling data with the intention of trying to essentially handicap the Presidential race (and Senate races this year as well). Essentially there are several firms that are doing polling, and these 2 guys, one of whom is a baseball statistician who has a keen interest in politics and is a firm Obama supporter, decided to try to come up with metrics to take some of the margin of error out of polling.
In other words, if you have one poll, it will say, there is a margin of error of plus or minus 4 percent. But let’s say you have 100 polls, well, you can aggregate them and kind of average them out to make some conclusions. But just averaging them alone doesn’t necessarily give you the most accurate picture, so one thing they also do is look at historical accuracy rates of various pollsters and adjust for that. Another thing they do is to look at known bias from certain pollsters and adjust for that (and by bias, I mean how it shows up in methodology, for example they may undersample the youth vote and oversample the 65+ vote because they’re looking at 2004 numbers, but the ranks of the youth voter registratios have swelled). What they do is they assign each individual poll a weighting, based on the reliability of the pollster and bias of the pollster, the sample size of the poll, and the age of the poll (so a small poll conducted yesterday might carry the same weight as a much larger poll conducted on Monday). They put this all together and have one trendline which says “if the election were held today, what would happen.”
Then they apply a statistical concept called “regression analysis” to see where they think the numbers will regress to the mean. In other words, they’re trying to predict, based on what polls looked like that many days before an election in previous elections, and try to predict what will happen in November. They take all this data and then they use it to run 10,000 simulations of the election every day. In other words, with polling right here, this many times out of 10,000 we see this result, this many times we see that result. The whole thing is then reported in terms of 1) statistical probability that Obama will win, 2) what the mean/average electoral votes each candidate would get based on the simulations, and 3) what percentage of the popular vote each candidate will end up with. Then they have a chart showing the two trendlines (if the election were held today what would happen over time, and what was the prediction of what would happen on Nov 4 over time). There is also a bell graph showing where the simulations end up.
And the whole thing is done on a state by state basis as well, they do not just look at national numbers (they factor this in but mostly they’re looking at probability on a state by state basis), and they calculate daily what the real swing states, or tipping point states are. In other words, “if I were Obama, these would be the top 15 states you should spend money in.” There’s just a myriad of data. And right now, they think Obama has about an 85% chance of winning the election. But the numbers themselves are not biased…they do their job, they know how to eliminate bias, noise, etc. from the polls…a couple weeks back, they had McCain with an almost 60% chance of winning…they won’t massage the numbers to make it look better for their candidate.
But then the other part, the part you asked about, is the blog. There are daily blog updates with analysis about what is going on. Some of the blog posts are about the media interviews the guy who runs the site is giving, but usually it’s an analysis of what’s going on in the election. If there’s a controversy, they’re all over it with what they think it means. They liveblog from the events like the debates and the conventions. They have a series where they’re travelling around the country to check out he field operations on the ground. There is a LOT of discussion about field organizers…or how pollsters adjust for swelling voter registrations, or how they adjust for the problem that a lot of people only have cell phones these days. They’ll discuss the hot rumors like “Will Palin leave the ticket?” And the blogs though I feel they are very fair and balanced, are definitely Obama leaning. They make no attempt to hide their unfettered support of Obama, and any good news they will attempt to make sure you know it, and any bad news, they’ll tell you why you should or should not be worried. I’ve found it a great place of comfort as an Obama supporter, but also to be one of the best sources of comprehensive information, which when it counts is unbiased.