Calculating the probability of a Kerry win using Simulation

[TruthIsAll]

Assuming Kerry gets 70% of the undecided/3rd party vote, I
have calculated the probability he will win the election based
on current state polls.

Using simulation, the probability of Kerry getting more than
269 electoral votes, given the following assumptions, is
approximately 99%. If Bush were to win under these
assumptions, it would be fair to say that Kerry was Diebolded.

I first approximated the probability of a Kerry win in each
state based on current polling numbers, assuming a polling
margin of error (MOE) of +/-4%. 

I assume there is a 95% probability that the election result
in each state is within the polling MOE. I also assume that
the results are equally distributed within the MOE.		

For example, if Kerry has 50%, then he has a 50% probability
of winning the state.

If Kerry has 54%, Bush 46%, then Kerry has a 97.5% chance of
winning the state.

I ran many random number simulations in Excel, 10 at a time.
A typical run produced the following:
SIM	EV	Kerry		
1	287	Win		
2	301	Win		
3	341	Win		
4	326	Win		
5	326	Win
6	338	Win
7	335	Win
8	337	Win
9	329	Win
10	333	Win
AVG	325	

	
Adjk= Kerry + .70* (100-Kerry-Bush)				

State	EV	AdjK	Diff	Prob(win)
AL	9	42.7	-14.6	 0.001 
AK	3	40.7	-18.6	 0.001 
AZ*	10	47.6	-4.8	 0.221 
AR*	6	52.6	5.2	 0.827 
CA	55	59.0	18.0	 0.990 
CO	9	49.3	-1.4	 0.414 
CT	7	62.6	25.2	 0.990 
DE	3	57.1	14.2	 0.990 
DC	3	90.4	80.8	 0.990 
FL*	27	54.3	8.6	 0.975 
GA	15	45.3	-9.4	 0.010 
HI	4	60.9	21.8	 0.990 
ID	4	39.0	-22.0	 0.001 
IL	21	58.3	16.6	 0.990 
IN	11	44.4	-11.2	 0.001 
IA*	7	53.6	7.2	 0.942 
KS	6	41.6	-16.8	 0.001 
KY	8	45.3	-9.4	 0.010 
LA	9	49.0	-2.0	 0.375 
ME	4	52.3	4.6	 0.788 
MD	10	59.0	18.0	 0.990 
MA	12	66.3	32.6	 0.990 
MI*	17	53.9	7.8	 0.975 
MN*	10	53.5	7.0	 0.923 
MS	6	36.3	-27.4	 0.001 
MO*	11	50.2	0.4	 0.538 
MT	3	42.8	-14.4	 0.001 
NE	5	36.5	-27.0	 0.001 
NV*	5	53.3	6.6	 0.904 
NH*	4	53.7	7.4	 0.975 
NJ	15	56.6	13.2	 0.990 
NM*	5	55.3	10.6	 0.990 
NY	31	61.3	22.6	 0.990 
NC	15	48.3	-3.4	 0.298 
ND	3	37.2	-25.6	 0.001 
OH*	20	49.6	-0.8	 0.452 
OK	7	43.1	-13.8	 0.001 
OR*	7	54.5	9.0	 0.990 
PA*	21	54.3	8.6	 0.975 
RI	4	64.2	28.4	 0.990 
SC	8	43.5	-13.0	 0.001 
SD	3	44.8	-10.4	 0.001 
TN*	11	41.8	-16.4	 0.001 
TX	34	42.6	-14.8	 0.001 
UT	5	29.7	-40.6	 0.001 
VT	3	60.1	20.2	 0.990 
VA	13	48.5	-3.0	 0.317 
WA*	11	56.0	12.0	 0.990 
WV*	5	48.6	-2.8	 0.337 
WI*	10	50.4	0.8	 0.558 
WY	3	30.8	-38.4	 0.001 
	538			
				
				

[qazplm]

1. I think 70% is too high

what are your numbers with 60%?

[TruthIsAll]

2. Getting 60% of the undecided/Nader vote, Kerry has about a 95% probability

SIM	EV	
1	300	Win
2	336	Win
3	304	Win
4	313	Win
5	313	Win
6	303	Win
7	315	Win
8	317	Win
9	256	Lose
10	322	Win
AVG	308	
Win	9	

[Suzi Creamcheese]

9. How does Kerry get 60% of the Nader vote?

Is Nader going to quit?

[TruthIsAll]

10. Suzi, I assume by the election Nader will have been reduced to under 1%,

with at least 70% of his defectors voting for Kerry.

[POed_Ex_Repub]

3. Anyone know what the number is historically?

Or is that an unanswerable question?

[Lex]

4. TIA has said between 60 to 80, I think.

60 to 80 percent of the undecideds break for the newcomer, NOT the incumbent--historically.

[POed_Ex_Repub]

5. Looks good for us, thanks!

:D

[louis c]

11. Challenger gets between 60% and 80% of undecided vote

In 1972, McGovern received the low of just over 60% of undecided by challenger.

Reagan received the high of nearly 80% in 1980.

All other challengers since 1948 have received a total of undecideds on election day that falls between these two totals.

In 1988, Perot and Clinton received 75% combined.

The reasoning is that if a voter has not committed to the incumbent President after watching nearly 4 years of his performance, the chances are that voter will not vote for him.

I also agree that Nader's vote will dissipate as election day draws closer. It did so in 2000, and knowing that history as well as what's at stake this time, a 2% total for Nader is generous by anyone's estimation.

[TruthIsAll]

6. For those who enjoy math...kick

tia

[TruthIsAll]

7. Simulations for 80,75,70,65,60,55,50,45% of undecideds to Kerry

Undecided to Kerry: 0.8
SIM	EV		
1	305	Win	
2	346	Win	
3	378	Win	
4	351	Win	
5	351	Win	
6	356	Win	
7	340	Win	
8	386	Win	
9	330	Win	
10	306	Win	
AVG	344.9		
Win	10		


Undecided to Kerry: 0.75
SIM	EV		
1	316	Win	
2	327	Win	
3	312	Win	
4	365	Win	
5	365	Win	
6	338	Win	
7	350	Win	
8	324	Win	
9	337	Win	
10	318	Win	
AVG	335.2
Win	10		

Undecided to Kerry: 0.7
SIM	EV		
1	310	Win	
2	325	Win	
3	311	Win	
4	333	Win	
5	333	Win	
6	357	Win	
7	347	Win	
8	304	Win	
9	324	Win	
10	343	Win	
AVG	328.7
Win	10		

Undecided to Kerry: 0.65
SIM	EV		
1	306	Win	
2	310	Win	
3	307	Win	
4	302	Win	
5	302	Win	
6	336	Win	
7	315	Win	
8	306	Win	
9	331	Win	
10	334	Win	
AVG	314.9		
Win	10		

Undecided to Kerry: 0.6
SIM	EV		
1	273	Win	
2	329	Win	
3	343	Win	
4	290	Win	
5	290	Win	
6	349	Win	
7	305	Win	
8	268	Lose	
9	318	Win	
10	259	Lose	
AVG	302.4		
Win	8		

Undecided to Kerry: 0.55
SIM	EV		
1	288	Win	
2	325	Win	
3	312	Win	
4	276	Win	
5	276	Win	
6	272	Win	
7	311	Win	
8	298	Win	
9	253	Lose	
10	269	Tie	
AVG	288		
Win	8
Tie   1

Undecided to Kerry: 0.5
SIM	EV		
1	295	Win	
2	287	Win	
3	294	Win	
4	285	Win	
5	285	Win	
6	299	Win	
7	260	Lose	
8	290	Win	
9	288	Win	
10	319	Win	
AVG	290.2
Win	9		

Undecided to Kerry: 0.45
SIM	EV		
1	275	Win	
2	231	Lose	
3	285	Win	
4	253	Lose	
5	253	Lose	
6	256	Lose	
7	270	Win	
8	272	Win	
9	282	Win	
10	314	Win	
AVG	269.1		
Win	6

[TruthIsAll]

16. Monday, Monday

tia

[TruthIsAll]

8. However you slice it, these scenarios show why Bush MUST cancel

He can never get the public to believe he won if he Diebolds his way into reselection, when his poll numbers are so bad.

[TruthIsAll]

12. Just ran some more simulations...

Undecided to Kerry:0.8
SIM	EV		
1	322	Win	
2	305	Win	
3	366	Win	
4	336	Win	
5	310	Win	
6	330	Win	
7	334	Win	
8	335	Win	
9	330	Win	
10	373	Win	
AVG	334.1		
Win	10		
			
			
Undecided to Kerry:0.75
SIM	EV		
1	343	Win	
2	341	Win	
3	322	Win	
4	334	Win	
5	336	Win	
6	342	Win	
7	334	Win	
8	316	Win	
9	303	Win	
10	337	Win	
AVG	330.8		
Win	10		
Undecided to Kerry:0.7
SIM	EV		
1	296	Win	
2	350	Win	
3	304	Win	
4	327	Win	
5	327	Win	
6	351	Win	
7	331	Win	
8	327	Win	
9	316	Win	
10	335	Win	
AVG	326.4		
Win	10		
Undecided to Kerry:0.65
SIM	EV		
1	323	Win	
2	326	Win	
3	333	Win	
4	336	Win	
5	327	Win	
6	312	Win	
7	248	Lose	
8	311	Win	
9	322	Win	
10	282	Win	
AVG	312		
Win	9		

Undecided to Kerry:0.6
SIM	EV		
1	328	Win	
2	316	Win	
3	306	Win	
4	291	Win	
5	293	Win	
6	332	Win	
7	342	Win	
8	313	Win	
9	317	Win	
10	312	Win	
AVG	315		
Win	10		

Undecided to Kerry:0.55
SIM	EV		
1	301	Win	
2	334	Win	
3	323	Win	
4	328	Win	
5	307	Win	
6	302	Win	
7	293	Win	
8	310	Win	
9	305	Win	
10	289	Win	
AVG	309.2		
Win	10		
Undecided to Kerry:0.5
SIM	EV		
1	303	Win	
2	275	Win	
3	294	Win	
4	319	Win	
5	270	Win	
6	311	Win	
7	202	Lose	
8	238	Lose	
9	274	Win	
10	301	Win	
AVG	278.7		
Win	8		
Undecided to Kerry:0.45
SIM	EV		
1	269	Tie	
2	208	Lose	
3	291	Win	
4	252	Lose	
5	297	Win	
6	253	Lose	
7	265	Lose	
8	276	Win	
9	278	Win	
10	290	Win	
AVG	267.9		
Win	5		

[ISUGRADIA]

13. Polls at this time of the year mean nothing as to an ultimate predictor

of actual election returns. They are only a snapshot in time.

If the election were held today and you did exit polling of your actual sampling universe, there might be some credence. Using your logic, One probably could have said there was an overwhelming chance Dewey would win come November and the only way he could lose in through fraud.

You cannot control for all the variables that effect voter opinion between now and November. Voters opinions do change. You are being dishonest with statistics if you try and make any xx% of a Kerry win based on current polls. Ask any statistician or pollsters if current polls mean anything more than THAT exact peiod in time, especially 4 months before an election.

[TruthIsAll]

14. There is nothing "dishonest" about this analysis.

Tell me something I don't know.

I KNOW I cannot control voter opinion.

I KNOW there are 4 months to go.

I also KNOW that the polls UNDERSTATE Kerry's strength. Do you know that? If current polls mean nothing, why is Bush in near-panic mode?

I said the analysis uses the latest polling data. I will continue to run the model up to the election, as polls are updated..

Too bad you see no value in this.

[ISUGRADIA]

15. I do think analysis of polls at this stage of the election IS useful.

I just argued with the ability to put any statistical interpretation saying X candidate has Y% of winning the election based upon them as if that is written in stone. Statistics in an inexact science.

If you are acknowledging that variables could change the final election result, then that is all I am asking for. I can recall a poll from this time or maybe later in the summer in 1988 showing Dukakis was ahead in Utah over Bush Sr. I knew there was no way in hell a Democrat was going to carry Utah. Dukakis ran a terrible campaign and let Lee Atwater demonize him which blew his chances to win.

Polls at this time of the election year showing an incumbent president with these bad of numbers in so many states do not forbode a good result for the incumbent. That is, as you properly say, "why Bush is in near panic mode". He's trying to shore his fundi base with calls for an anti-gay constitutional amendment and other actions.

There are a lot of things that Kerry could do that would blow his chances that we cannot anticipate. Yet, given Bush's mediocrity I would say a good dabate performance by Kerry could help swing this election decicively as it did for Reagan in 1980. If you reevaluate your analysis with new polls and use multiple polling sources for your analysis I think it will be useful to show a timeline of support.