By using this site, you agree to our Privacy Policy and our Terms of Use. Close
HappySqurriel on 25 October 2012
Edit Thread
binary solo said:
HappySqurriel said:
binary solo said:
Funny thing is the Ppundits are saying Ohio or Florida are key battleground states, but if the states that look fairly locked in for Obama all go his way then of the toss up states Obama only needs Colorado and New Hampshire to get over the line. This means he can lose Ohio and Florida and still become president.

There are really 5 toss up States. Obama can win with one or 2 of those states falling to him. Romney needs 4 of those 5 to fall for him. Iowa is pretty marginal for Obama, so if that's too close to call then the number of toss up states each candidate needs evens out a bit. But Romney still needs more of those too ups than Obama.

Interestingly the so-called Missouri bellwether didn't go with Obama last time and it's definitely not going with Obama this time, so if Obama wins then it's barometer reputation is severely weakened.


In my opinion, realclearpolitics has a better electoral college map than fivethirtyeight because of how it handles toss-up states; any state where the two candidates are closer than 5% is considered a toss up, and this works better because in this range poll methodology, voter enthusiasm, and likely-vs-unlikely voters make a huge difference.

At the moment there are 10 toss up states, 8 of which both candidates are within 3% of eachoter, and 5 of which the candidates are within 2% of eachother.

Basically, if party turnout matched 2008 it is likely that Obama will win, if it is closer to 2004 it is likely that Obama will lose ...

If you have a series of polls, over several weeks, from various polling organisations, including through the debates, and they all fall for the same candidate and it averages out to 4-5% then the state is probably not really a toss up and the 4-5% is likely to be a true difference. Most polls have a 3% MoE, so an individual poll with a 5% margin is a toss up result. But if you get an ongoing trend of ~5% for the same candidate then the meta-MoE across all the polls drops away and moves closer to <1%. A poll of polls that gives a 3-5% lead to one candidate is outside the toss-up zone, but it's not yet in the locked-in zone. Huffington Post calls it leaning, and it also gives a confidence in the lead of the candidate. So a 3% leaning state with a 75% confidence in the lead is still a pretty fragile state, but a betting man would put money on the currently leading candidate. A true toss-up state, within 3%, with a 60% confidence in the lead would not be a good bet for the leading candidate, and it would be a pretty poor bet for the trailing candidate.

It does tend to be the case that low turn out elections favour the right / conservatives. Heck, wet weather favours the conservatives (because it lowers turn out). But because of the ECV it's not actually national turn out that would indicate the likely winner but swing state turn out. So those fragile 7-8 states are the ones to watch in terms of turn out.

Except 8 of 10 of those toss up states are within the margin of error of those polls, a large portion of which are (probably) over-estimating democrat turnout and under-estimating independent voter turnout.

Now, I'm not suggesting that there will be a massive shift in these states, but anyone who pays attention to politics (especially on an international level) will admit that these polls indicate that the race is far too close in these states to predict who will win.