Thursday, February 21, 2008

2008 Battlefield States

Unless something very, very strange happens in the next few weeks, we know the contest for the fall election; Barack Obama versus John McCain, Vapid Hope matched against Angry Moderation. While both campaigns will guard against premature assumptions, both will also have begun planning strategy for the fall’s General Election.

I have said before that the Presidential election is not a single national contest, but a set of 51 regional contests, which produce the Electoral Vote tallies which determine the winner of the office. Even with the enormous amounts of money spent on the election, a candidate must still plan carefully where to allocate resources, especially since time and personal appearances remain limited. Some states are regarded as “safe”, states which will support the candidate in every forseeable condition, while others are considered too unlikely to be worth much effort. The states in between, the states which “belong” to the candidate but which may be lost, or the states which “belong” to the opponent but which may be won, and the states which seem undecided, these are targeted for varying degrees of attention, promises, and displays of appreciation, and are called the battleground states. The question early on for candidates, is knowing which states will fall into this category, because that varies from election to election. More than a few candidates have lost elections because they did not correctly identify their weak spots and opportunities. And it is not as easy as some people might think to work out the territory which matters.

Take California, for example. With 55 Electoral Votes, California is the biggest of all prizes, and so would seem to be a smart place to invest time and money. But is it? After all, the last four elections saw California choose the Democrat, and by 10 or more points of the popular vote each time. That would seem to suggest that California is locked up for the Democrats, but then again, maybe not. Kerry took California by 10 points in 2004, but that was 4 points closer than Clinton’s margin of victory in 1996. And for comparison, Reagan took California by a 17 point margin in both 1980 and 1984. Even GHW Bush won California in 1988, so there is historical evidence to suggest that the right conditions could put it into play. How does someone properly weigh the benefits and costs of spending resources in a state? One way is to examine the historical behavior of a state, and from that determine the probability of a positive or negative reaction to unusual focus on that state. This essay examines that historical structure. It does not project a winner, nor does it weigh the specific positions of any particular candidate, because its purpose is to suggest areas of interest to each of the two major political parties.

I started by noting the popular vote results per state for every Presidential election from 1948 to 2004. I chose that range for two reasons; the world has changed significantly over time, and it is not reasonable to assume that election conditions in, say, 1880, are really comparable to the modern condition. World War 2 was the Rubicon in many ways to American election development, and so the 15 Presidential elections after that war stand as the best relative indicators as to the political identity of each state. Also, the 60-year time frame covers the majority of voter demographics, and so is a reasonable representative of how voters as a whole have behaved. I took the popular vote collected by each party and rounded it up or down, then compared votes for a whole integer margin. This means that ‘0’ margins exist, which are treated as statistical ties for this purpose. That information provided averages, largest and smallest margins of victory and defeat, and of course the streak of victories by a party. But that is just the start. The margin from one election could also be compared to the margin for the next, and the change in results represents variance. Delaware, for example, supported Reagan by 2% in 1980, but 20% in 1984, which is an 18% gain for the Republicans. Bush won Delaware in 1988, but only by 12%, which was an 8% gain for Democrats, and so on. Taking each variance and squaring it, summing up those squared-variance tallies and then finding the square root give the Standard Deviation of each state’s Popular Vote margin. The smallest SD was 6.254 for the District of Columbia, and the largest SD was 42.838 for Mississippi. This is because D.C.’s margin of victory has never changed by more than 9 points, while Mississippi’s margin of victory changed by 60 points or more on four occasions. The ten most volatile SD’s are as follows:

1. Mississippi – 42.838
2. Alabama – 33.388
3. Georgia – 30.988
4. Hawaii – 29.281
5. Arkansas – 27.575
6. Oklahoma – 24.751
7. Louisiana – 24.648
8. Maine – 24.150
9. Rhode Island – 23.469
10. South Carolina – 23.266



The SD is an important indicator of historical volatility, but it may not be a valid indicator of current trends. Of the top 10 states for volatility, five are from the Deep South which rebelled at one time against both the Democrats and Republicans; Mississippi’s volatility, for example, ranged between 60 and 85 points between 1964 and 1976, but the SD for 1980-2004 is only 11.661. Reweighting the SDs to increase proportionately as the time frame approaches the present, then dividing that result by the consecutive victories by the party which last won the state produced the following revised 10 most and least volatile SD’s:

MOST VOLATILE
1. Florida, 18.745
2. New Hampshire, 18.045
3. Arkansas, 13.622
4. New Mexico, 13.368
5. Iowa, 12.819
6. Oregon, 12.065
7. Wisconsin, 10.971
8. Louisiana, 10.473
9. West Virginia, 9.744
10. Tennessee, 9.744

LEAST VOLATILE
1. District of Columbia, 1.200
2. Indiana, 1.295
3. Delaware, 1.344
4. Nebraska, 1.405
5. South Dakota, 1.432
6. Virginia, 1.500
7. Idaho, 1.544
8. Kansas, 1.566
9. Utah, 1.639
10. North Dakota, 1.712

It’s hardly a surprise to see Florida at the top, but it may surprise some folks to see Oregon, Wisconsin, Louisiana, and Tennessee in play. It is also interesting, to me at least, to see that the numbers indicate that states like Virginia and Indiana are not budging.

But there is also another indicator, a simpler one, about which states are likely to switch. I took the SD and the weighted SD, and divided each by the margin from the 2004 election. If the margin was greater than the SD or weighted SD, it is very unlikely that the state would switch in 2008. That produced these top 10 lists:

STATES MOST LIKELY TO SWITCH IN 2008
SD
1. New Hampshire, 21.092
2. Alaska, 21.047
3. Iowa, 18.787
4. New Mexico, 16.404
5. Wisconsin, 15.672
6. Arizona, 13.06
7. Nevada, 9.474
8. Ohio, 8.676
9. Michigan, 5.235
10. Pennsylvania, 5.157

Weighted SD
1. New Hampshire, 18.045
2. Alaska, 17.901
3. New Mexico, 13.368
4. Iowa, 12.819
5. Wisconsin, 10.971
6. Nevada, 8.182
7. Ohio, 7.083
8. Michigan, 4.323
9. Pennsylvania, 3.885
10. Florida, 3.749

STATES LEAST LIKELY TO SWITCH IN 2008
SD
1. District of Columbia, 0.301
2. Utah, 0.406
3. Idaho, 0.461
4. Wyoming, 0.501
5. Nebraska, 0.569
6. North Dakota, 0.757
7. Indiana, 0.760
8. Oklahoma, 0.773
9. Kansas, 0.780
10. Massachusetts, 0.782

Weighted SD
1. District of Columbia, 0.059
2. Utah, 0.356
3. Idaho, 0.406
4. Nebraska, 0.426
5. Wyoming, 0.473
6. Massachusetts, 0.568
7. North Dakota, 0.612
8. Kansas, 0.626
9. Oklahoma, 0.642
10. Indiana, 0.648

Twelve states have a less than 1% chance of switching in 2008, by the SD calculation, and by the weighted SD eighteen states have less than a 1% chance of switching in 2008.

But that still leaves the question of where candidates should focus their resources. The final answer I found, took the average of the Standard Deviation of popular vote margins and the Weighted Standard Deviation, and multiplied that number by the portion each state has of the total 538 Electoral Votes. That produced the following target list:

1 Ohio 29.292
2 Wisconsin 24.761
3 Iowa 20.562
4 Florida 19.916
5 Pennsylvania 17.647
6 Michigan 15.101
7 New Hampshire 14.549
8 New Mexico 13.835
9 Arizona 13.219
10 California 11.981
11 Alaska 10.859
12 Nevada 8.204
13 Minnesota 8.091
14 New Jersey 7.202
15 Texas 5.587
16 Colorado 5.473
17 New York 5.333
18 Georgia 5.145
19 Illinois 4.922
20 Missouri 4.769
21 Oregon 4.603
22 North Carolina 4.578
23 Virginia 4.318
24 Washington 3.821
25 Arkansas 3.056
26 Tennessee 2.860
27 Louisiana 2.542
28 Maryland 2.329
29 Connecticut 2.235
30 Mississippi 2.126
31 Hawaii 2.003
32 South Carolina 1.914
33 Alabama 1.851
34 Massachusetts 1.506
35 West Virginia 1.493
36 Indiana 1.439
37 Kentucky 1.432
38 Maine 1.251
39 Oklahoma 0.921
40 Kansas 0.784
41 Rhode Island 0.758
42 District Columbia 0.599
43 Delaware 0.559
44 Vermont 0.559
45 Nebraska 0.462
46 Montana 0.436
47 South Dakota 0.416
48 North Dakota 0.382
49 Utah 0.354
50 Idaho 0.322
51 Wyoming 0.272

Both parties, therefore, should be expected to focus primarily on Ohio, followed by Wisconsin, Iowa, Florida, Pennsylvania, and Michigan.

No comments: