Response to MIT/Caltech

The MIT/CalTech study suggests that the difference between the exit polls and the vote tally can be explained

by volatility of the exit polls early in the day and that the accuracy of the final exit poll with the complete sample

is the one to be trusted.

Below are three tables of data from exit polls displayed on CNN’s website on Nov 2 and early morning Nov 3.

As you can see the swings from late night to early morning are dramatic and can not be explained by the indicated

increase in the sample size. In other words, the changes in the poll can not be attributed to more accuracy due a

more complete sample as the MIT/CalTech study suggests.

I believe that the changes below can only be made by manipulation of the database, either by moving data around

or by changing the weightings used throughout the day. The former is fraudulent, the later could have an explanation

but to my knowledge none has been offered. This indicates a fraudulent manipulation of the final exit poll update, but

is not conclusive.

Ohio CNN exit poll data
		Proportion		Bush			Kerry
Time	Respondents	Male%	Female%	Male%	Female%	Total	Male%	Female%	Total	Margin
12:21am		47.0%	53.0%	49.0%	47.0%	47.9%	51.0%	53.0%	52.1%	-4.1%
	1963	923	1040	452	489	941	471	551	1022	-81
1:41am		47.0%	53.0%	52.0%	50.0%	50.9%	47.0%	50.0%	48.6%	2.4%
	2020	949	1071	494	535	1029	446	535	982	47
1 hour 20 min	57			42	46	88	-24	-16	-40

National CNN exit poll data
		Proportion		Bush			Kerry
Time	Respondents	Male%	Female%	Male%	Female%	Total	Male%	Female%	Total	Margin
9:06:00 PM		46.0%	54.0%	51.0%	45.0%	47.8%	47.0%	54.0%	50.8%	-3.0%
	13000	5980	7020	3050	3159	6209	2811	3791	6601	-393
1:36:00 AM		46.0%	54.0%	52.0%	50.0%	50.9%	47.0%	50.0%	48.6%	2.3%
	13531	6224	7307	3237	3653	6890	2925	3653	6579	311
4 hr 30 min	531			187	494	681	115	-137	-23

Florida CNN exit poll data
		Proportion		Bush			Kerry
Time	Respondents	Male%	Female%	Male%	Female%	Total	Male%	Female%	Total	Margin
8:40:00 PM		46.0%	54.0%	52.0%	48.0%	49.8%	47.0%	52.0%	49.7%	0.1%
	2846	1309	1537	681	738	1418	615	799	1414	4
1:01:00 AM		46.0%	54.0%	53.0%	50.0%	51.4%	46.0%	49.0%	47.6%	3.8%
	2862	1317	1545	698	773	1470	606	757	1363	108
3 hr 21 min	16			17	35	52	-10	-42	-52

The MIT/CalTech charts and the table below appear to agree only in the discrepancy in NY. Their charts show

exit polls agreeing with vote tallies in most states except NY. My data which I believe to be correct and taken

from late night exit polls before the final update show a consistent and strong swing from Kerry to Bush.

		Actual Vote			Early Exit Poll			Actual Vote - Exit poll			Total
State	EV	Kerry	Bush	Nader	Kerry	Bush	Nader	Kerry	Bush	Nader	Swing
Alabama	9	37	63	0
Alaska	3	34	62	2
Arizona	10	44	55	1	45	55	0	-1	0	1	-1
Arkansas	6	45	54	1	45	54	0	0	0	1	0
California	55	54	45	0
Colorado	9	46	53	1	49	50	0	-3	3	1	-6
Connecticut	7	54	44	1
Delaware	3	53	46	1
D.C.	3	90	9	1
Florida	27	47	52	1	51	49	0	-4	3	1	-7
Georgia	15	41	59	0
Hawaii	4	55	44	1
Idaho	4	29	69	0
Illinois	21	55	44	1
Indiana	11	39	60	0
Iowa	7	49	50	1	50	46	0	-1	4	1	-5
Kansas	6	36	63	1
Kentucky	8	40	60	0
Louisiana	9	42	57	0	42	57	0	0	0	0	0
Maine	4	53	45	1
Maryland	10	56	43	1
Massachusetts	12	62	37	0
Michigan	17	51	48	1	52	46	1	-1	2	0	-3
Minnesota	10	52	47	1	52	48	2	0	-1	-1	1
Mississippi	6	40	60	0
Missouri	11	46	54	0	47	52	1	-1	2	-1	-3
Montana	3	39	59	1
Nebraska	5	32	67	0
Nevada	5	48	51	1	49	48	1	-1	3	0	-4
New Hampshire	4	50	49	1	54	44	1	-4	5	0	-9
New Jersey	15	53	46	1	54	44	1	-1	2	0	-3
New Mexico	5	49	50	1	50	48	0	-1	2	1	-3
New York	31	58	40	2	62	36	2	-4	4	0	-8
North Carolina	15	43	57	0
North Dakota	3	36	63	1
Ohio	20	49	51	0	51	49	0	-2	2	0	-4
Oklahoma	7	34	66	0
Oregon	7	53	47	0
Pennsylvania	21	51	49	0	53	46	0	-2	3	0	-5
Rhode Island	4	60	39	1
South Carolina	8	41	58	1
South Dakota	3	38	61	1
Tennessee	11	43	57	0
Texas	34	38	61	0
Utah	5	26	72	1
Vermont	3	59	39	2
Virginia	13	46	54	0
Washington	11	53	46	1
West Virginia	5	43	56	1	45	54	1	-2	2	0	-4
Wisconsin	10	50	49	1	51	48	1	-1	1	0	-2
Wyoming	3	29	69	1
	10	46	54	1	50	49	1	-5	5	0	-10

By: David Dodge, BSMME (Manufacturing and Management Engineering), November 12, 2004