Bush needs 85% of undecided voters to have a 50% prob. of winning

[TruthIsAll]

Only problem: The challenger gets 60-80% of the undecideds.
Just a little nugget I found by playing "what-if" in the election model.

All you DUers feel better now?

http://www.geocities.com/electionmodel/

Undecided allocation to Kerry
....... 14% 20% 40% 50% 60%

Out of 5000 election trials:
Wins 2589 3411 4778 4942 4994

Win Prob 51.8% 68.2% 95.6% 98.8% 99.9%

National % 48.30 48.69 49.96 50.60 51.88

EV Avg 268 281 313 323 339

Median 271 283 314 324 340

EV Max 350 361 386 386 396

EV Min 147 153 204 208 234

States 22 24 29 29 31

[Zo Zig]

1. TIA question

I know that you enjoy doing these types of models, have you looked at trends on the national polls regarding the weighting of voters in them over time? Has the split shifted from one party to another in any or all, second are the pools of those polled a rolling inventory of sorts, example groups that are re-polled A-Z at various times?
Just wondering.
Any insight you might have would be great.

[TruthIsAll]

3. No, I use unweighted national polls for trend/win prob. analysis...

of the 11-poll and 15-poll groups and for the individual polls.

I haven't looked into the underlying samples. I only consolidate and analyze monthly data from pollingreport.com. If available, I use polls of registered voters (excluding Nader); if not, polls of likely-voters.

I don't know if it's a coincidence, but pollingeport.com decided to include poll summaries (of likely and registered voters). I've been doing in Excel manually for months and then copied and pasted the results on DU.

I only started the electionmodel site two weeks ago. I have looked at over 35 sites and none that I know of provide anything approaching my stats and analysis. Only a couple even calculate win probabilities.

tia

[starroute]

2. And they say the incumbent rarely gets very many of the undecided

That is, most of the undecided are people who've already turned against Bush but are waiting to see if they can trust Kerry or if they should just stay at home on Election Day.

[TruthIsAll]

4. Notice that Kerry can win with 49% of the national vote. Bush can't.

That's because Kerry wins the big EV states.