| |
|
|
|
|
| KERRY ELECTORAL VOTE PROJECTIONS |
| from State-Level Poll Data |
| 1-Nov-2004 |
| |
number of states |
Total Electoral Votes |
|
|
| Bush > 55% |
17 |
133 |
Bush total: |
|
| Bush 50-55% |
13 |
145 |
278 |
|
| Kerry 50-55% |
9 |
92 |
Kerry total: |
|
| Kerry > 55% |
12 |
168 |
260 |
|
| |
51 |
538 |
|
|
| polls used in estimate: |
beginning |
ending |
|
| 10/27/2004 |
11/1/2004 |
| |
|
|
|
| |
|
|
| |
|
Estimated
Kerry Vote Share |
|
estimated
probability of Kerry win if election held now* |
| |
|
|
| |
|
|
| |
|
|
| state |
EV |
stdev |
| |
|
|
|
|
| AL |
9 |
43.1% |
2.2% |
1.0% |
| AK |
3 |
34.0% |
5.2% |
1.0% |
| AZ |
10 |
46.8% |
1.8% |
1.0% |
| AR |
6 |
49.6% |
7.3% |
46.1% |
| CA |
55 |
56.6% |
0.9% |
99.0% |
| CO |
9 |
46.5% |
3.0% |
1.0% |
| CT |
7 |
58.6% |
2.1% |
99.0% |
| DE |
3 |
56.6% |
0.7% |
99.0% |
| DC |
3 |
90.5% |
0.2% |
99.0% |
| FL |
27 |
50.0% |
1.1% |
47.2% |
| GA |
15 |
45.3% |
3.3% |
1.0% |
| HI |
4 |
59.5% |
1.6% |
99.0% |
| ID |
4 |
31.4% |
6.1% |
1.0% |
| IL |
21 |
56.5% |
1.7% |
99.0% |
| IN |
11 |
42.2% |
2.7% |
1.0% |
| IA |
7 |
50.6% |
2.7% |
67.3% |
| KS |
6 |
40.1% |
2.6% |
1.0% |
| KY |
8 |
44.2% |
5.1% |
1.2% |
| LA |
9 |
46.8% |
4.9% |
9.9% |
| ME |
4 |
53.0% |
3.8% |
94.5% |
| MD |
10 |
58.5% |
0.1% |
99.0% |
| MA |
12 |
64.5% |
1.4% |
99.0% |
| MI |
17 |
52.8% |
1.8% |
99.0% |
| MN |
10 |
51.9% |
4.1% |
81.7% |
| MS |
6 |
41.8% |
2.7% |
1.0% |
| MO |
11 |
49.6% |
3.7% |
41.5% |
| MT |
3 |
39.6% |
7.8% |
1.0% |
| NE |
5 |
35.1% |
2.5% |
1.0% |
| NV |
5 |
48.8% |
1.7% |
7.4% |
| NH |
4 |
49.5% |
2.3% |
31.7% |
| NJ |
15 |
57.4% |
2.7% |
99.0% |
| NM |
5 |
51.1% |
2.6% |
80.4% |
| NY |
31 |
62.9% |
1.3% |
99.0% |
| NC |
15 |
44.5% |
2.7% |
1.0% |
| ND |
3 |
36.1% |
5.1% |
1.0% |
| OH |
20 |
48.5% |
2.3% |
9.3% |
| OK |
7 |
39.3% |
3.5% |
1.0% |
| OR |
7 |
51.5% |
3.0% |
83.3% |
| PA |
21 |
52.8% |
1.8% |
99.0% |
| RI |
4 |
65.5% |
1.5% |
99.0% |
| SC |
8 |
42.3% |
2.4% |
1.0% |
| SD |
3 |
40.1% |
5.4% |
1.0% |
| TN |
11 |
48.9% |
2.2% |
16.8% |
| TX |
34 |
40.7% |
5.0% |
1.0% |
| UT |
5 |
28.9% |
5.1% |
1.0% |
| VT |
3 |
56.0% |
3.6% |
99.0% |
| VA |
13 |
46.0% |
1.3% |
1.0% |
| WA |
11 |
53.9% |
2.6% |
99.0% |
| WV |
5 |
48.7% |
6.4% |
34.6% |
| WI |
10 |
50.3% |
2.3% |
59.4% |
| WY |
3 |
32.6% |
8.7% |
1.0% |
| |
|
|
|
|
These projections are calculated from
state-level poll data.
Methodology: Throughout, I work
with the ratio Dem/(Rep+Dem), ignoring fringe and third-party
candidates. For the elections of 1980-2000, 1992-2000 (a/o
21oct), I calculated the mean, correlation and covariance of each state's
vote with every other state's vote.
This permitted me to compute "alpha" and "beta"
for all 2550 (51^2 - 51) pairs of states, as if I had regressed each state's
vote on each other state's vote. Then
I calculated the ratio Kerry/(Bush+Kerry) for the most recent state-level
poll as reported on www.electoral-vote.com.
Plugging the poll data ratio into the equation alpha + beta*ratio
yields a projection of one state's vote given another state's vote.
Regression theory provides that the projections are the
"best" (i.e. minimum variance within the class of unbiased
estimators) that the data will allow.
(Details can be found in any econometrics text. I recommend Pindyck and Rubinfeld.) |
| For example, if the most recent poll in
Wisconsin yields Kerry 45, Bush 45, others 10, then the relevant fraction for
Kerry is 1/2. To project Iowa's
popular margin given the assumption that that is how Wisconsin votes, I find
the "alpha" and "beta" corresponding to "Iowa vote
given Wisconsin vote." Plugging
the value 1/2 into the equation gives a projection of the Iowa vote. I do this for each poll published during
the most recent five days for which I have data (ordinarily the data is one
or two days behind today's date). I use
the median for each as my estimate, and estimate a standard deviation from
the interquartile range. *I plug
these numbers into the cumulative normal distribution to generate the
probabilities (except that I limit them to the interval [0.01,0.99]). |
| posting date: |
11/6/2004 |
|
|
|
| data source: |
http://www.electoral-vote.com/ |
|
| Copyright (c) 2004 by Stuart Eugene Thiel. All Rights Reserved. |
|
|
|
|
|