Electoral E(X)
*This is a post I initially made on April 5th. I’m planning on updating the probabilities periodically until the election to maintain an accurate expected value for the electoral college. Following the electoral college analysis is my analysis of the 2008 Senate Elections.
Last Update: April 5th
Looking ahead to the general election between John McCain and Barack Obama, we’re likely going to be looking at an electoral map that is moderately different from that of the last two elections. Sure Florida, Ohio and Pennsylvania will be key, but other states, traditionally red and blue, will be in play and add to the dynamics of the race.
In this post I evaluate each state independently, not necessarily based on current polls, but rather on the demographics of each state and the results from the exit polls from the primaries. I give each state an arbitrary value between 0 and 1 that is the probability Obama will win the state. I then multiply that value by the number of electoral college votes from the state and then added up the final value of each state to get the expected number of electoral college votes for each candidate. The equation looks like this… E(X)(Electoral Votes State 1) + E(X)(Electoral Votes State 2)… = Total number of expected electoral college votes for Obama, where (E)X is the probability Obama wins the state. I also broke break down all of the states by region.
The final total for the electoral college, based on raw winner-take-all probability, is McCain 258 to Obama’s 253 electoral votes with 27 (Ohio and Iowa), which are .5 probability tossups. However, when I calculated the expected values, the total was 290.1 electoral votes for Obama to 247.9 electoral votes for McCain. Read on to see how I got these values and for additional analysis.
Northeast:
Obama will be weaker in the northeast than Kerry was and Clinton likely would have been, however, it will be in no way devastating to the campaign. Republicans are a rare breed in the northeast, however, John McCain is the candidate that will likely do better than most Republicans in the northeast. He is the country club-type, moderate, practical Republican that, unlike George W. Bush, still appeals to the GOP in the northeast. But that probably still won’t be enough to deliver him any Kerry states besides New Hampshire. The following are my probability values for Obama wining northeastern states.
Maine: 4 electoral votes, .8 probability
Vermont: 3 electoral votes, 1
New Hampshire: 4 electoral votes, .45
Massachusetts: 12 electoral votes, .95
Rhode Island: 4 electoral, 1
Connecticut: 7 electoral votes, .95
New York: 31 electoral votes, .95
New Jersey: 15 electoral votes, .85
Total electoral votes: 80 (Obama 76, McCain 4)
Expected Values (Obama 72.5, McCain, 7.75)
This scenario assumes the Bush map from 2000 in which Gore easily won the Northeast, but Bush barely won New Hampshire. McCain will likely be closer than Bush most or all of these states, but I don’t think he wins any except New Hampshire, which is decidedly the most partisan of the bunch.
Mid Atlantic:
Based on the 2004 outcome, the mid Atlantic region is slightly more pro-Obama than it was pro-Kerry, but it probably won’t have a huge impact here. I know Obama did well in the Virginia primary, but I think the talk of Virginia going blue in this election ended when McCain won the nomination. Due to the high military population in Virginia McCain can probably keep it in the GOP fold at least this time around. North Carolina on the other hand is a possible sleeper for Obama.
Delaware: 3 electoral votes, 1 probability
Washington D.C.: 3 electoral votes, 1
Maryland: 10 electoral votes, 1
Virginia: 13 electoral votes, .2
West Virginia: 5 electoral votes, .1
Kentucky: 8 electoral votes, .1
Tennessee: 11 electoral votes, .15
North Carolina: 15electoral votes, .3
Total electoral votes: 68(Obama 16, McCain 52)
Expected Values (Obama 25.6, McCain 42.4)
Deep South:
Like North Carolina, some states in the South could be real wild-cards in this election. Whether or not Obama can bring out a massive number of new black voters in states like Georgia, South Carolina and Mississippi will determine if he can shake things up in this region. Either way this disadvantages the GOP because they want to win every one of these states without spending much money on the presidential or in congressional races and no matter what, that will not be the case in 2008.
South Carolina: 8 electoral votes, .35 probability
Georgia: 15 electoral votes, .4
Florida: 27 electoral votes, .4
Alabama: 9 electoral votes, .1
Mississippi: 6 electoral votes, .35
Louisiana: 9 electoral votes, .25
Arkansas: 6 electoral votes, .15
Total electoral votes: 80 (Obama 0, McCain 80)
Expected Values (Obama 25.75, McCain 54.25)
Likely one will be able to tell the result of the election just by looking at the Deep South. Evaluated as random events, the odds that McCain runs the table in the South is very small, at .087 percent. However, it is likely that the results of these states won’t be independent of each other and that Obama doing well in South Carolina will likely mean that he also does well in Georgia and Mississippi.
Midwest
Besides the northeast, where Obama will likely win anyway, the midwest is the one area where Hillary is decidedly stronger than Obama. As the hardest hit region in the country, economically, Hillary is the strongest here because of her appeal to working-class, typically Democratic voters. This matchup between McCain and Obama is particularly interesting because it is these lower-income Dems that have been most against Obama in the primary campaign, however, they are much more in agreement with Obama on labor issues and NAFTA than McCain.
Going into this election it was literally impossible to see a Republican winning Ohio due to the state of the Ohio GOP, however, if a Republican is going to win in Ohio, McCain versus Obama is the perfect matchup for it to happen. Minnesota is a wildcard depending on whether or not popular governor Tim Pawlenty is McCain’s running mate. I also added Pennsylvania to the midwest because I wasn’t sure where else to put it and it matches best with the midwest geographically. I think Obama ultimately will win Pennsylvania but it will probably be close throughout. So here are my probabilities for the midwest.
Ohio: 20 electoral votes, .5 probability
Pennsylvania: 21 electoral votes, .7
Michigan: 17 electoral votes, .7
Indiana: 11 electoral votes, .2
Illinois: 21 electoral votes, 1
Missouri: 11 electoral votes, .4
Wisconsin: 10 electoral votes, .85
Iowa: 7 electoral votes, .5
Minnesota: 10 electoral votes, .6
Total electoral votes: 128 (Obama 79, McCain 22, tossup 27)
Expected Values (Obama 82.2, McCain 45.8)
Plains States
There’s not a lot of electoral votes here and they likely won’t change much from 2004, however, Obama has distinct chance of taking Montana, which has been trending blue for a while.
North Dakota: 3 electoral votes, .05 probability
South Dakota: 3 electoral votes, .05
Nebraska: 5 electoral votes, .1
Kansas: 6 electoral votes, .1
Oklahoma: 7 electoral votes, 0
Wyoming: 3 electoral votes, 0
Montana: 3 electoral votes, .25
Idaho: 4 electoral votes, 0
Total electoral votes: 34 (Obama 0, McCain 34)
Expected Values (Obama 2.15- McCain 31.85)
South West
While Arizona, Utah and Texas are out of reach, I think it’s very possible that Obama wins the other three and probable that he wins two of the three. But the South West is the region I’m most unsure about because I don’t know what the Hispanic turnout will look like. I think Richardson definitely helps Obama in New Mexico, and he probably wins there, but I don’t know how much that support will translate to Colorado and Nevada.
Texas 34 electoral votes, .05 probability
New Mexico: 5 electoral votes, .75
Colorado: 9 electoral votes, .45
Utah: 5 electoral votes, 0
Arizona: 10 electoral votes, 0
Nevada: 5 electoral votes, .4
Total electoral votes: 68 (Obama 5, McCain 63)
Expected Values (Obama 11.5, McCain 56.5)
Pacific Coast
I’m reasonably sure Obama will win all three of these without a lot of money being spent on either side. Like 2000 and 2004, the major battles will be elsewhere.
California: 55 electoral votes, .95 probability
Oregon: 7 electoral votes, .8
Washington: 11 electoral votes, .8
Total electoral votes: 73 (Obama 68-McCain 0)
Expected values (Obama 66.65, McCain 6.35)
Other
Hawaii: 4 electoral votes, 1 probability
Alaska: 3 electoral votes, 0
Total electoral votes: 7 (Obama 4, McCain 3)
Expected Values (Obama 4, McCain 3)
Summary
So what this leaves us is a map that looks exactly like 2000, at least considering the raw probabilities in each state. Again, if Obama wins in every state where his probability exceeds .5 and looses where it is below .5, the total are Obama 253, McCain 258 with 27 still tied up in Ohio and Iowa. However, the numbers look much better for Obama when we consider the expected values. Obama easily wins with the expected values 290.1 to 247.9 electoral votes. This is due to the fact that Obama will be competitive in states where Gore and Kerry were not.
This fact is illustrated by the fact that Obama scores 29.75 electoral votes in his five highest expected value states that he doesn’t win (Florida, Georgia, North Carolina, Missouri and Nevada). McCain only scores 19.85 electoral votes from his five highest losing states (Pennsylvania, Michigan, Minnesota, New Jersey and Washington). Also, based on the probabilities and viewing the states as independent events, there is a 91.6 percent chance Obama wins one of these five McCain states and only an 80.0 percent chance that McCain wins one of the close Obama states.
But ultimately this race comes down to the mid west region, there’s really no way around it. There’s been a lot of talk about the south west being important, and maybe Obama can win Colorado and Mississippi or some other similar state but the election will no doubt be about the economy and whomever wins Ohio will almost surely be the next president.
I also want to point out that the individual probabilities for each state are not based on any scientific investigation, but rather my own perception of how the campaign will play in each state. However, I think that when evaluated as a whole, the values do represent an accurate depiction of the campaign and thus do not undermine the basis of this experiment.
Senate elections expected Values
Having already done an expected values calculation for the 2008 electoral college, it only made since to calculate the expected composition of the Senate following the November election.
To do this I gave each race a value between -1 and 1. A negative value represents the probability that the Democrats lose a particular seat, while a positive value is the probability that they gain the seat. A value of 0 means that the seat is “safe” in that there is a zero probability that it will change parties.
I list the probabilities for each election by the degree of competition that will categorize the race. My four categories are safe, leaning, marginal, and toss-up.
Using this method I got an expected value of 4.2 for the number of seats the Democrats will likely gain. Read on to see how I got this figure and if I think it’s an accurate representation of the election.
Safe Seats (0 probability of changing hands)
Alabama- Jeff Sessions (R)
Arkansas- Mark Pryor (D)
Delaware- Joe Biden (D)
Georgia- Saxby Chamblis (R)
Illinois- Dick Durbin (D)
Kansas- Pat Roberts (R)
Massachusetts- John Kerry (D)
Michigan- Carl Levin (D)
Mississippi – Thad Cochren (R)
Mississippi – Roger Wicker (R)
Montana- Max Baucus (D)
New Jersey – Frank Lautenberg (D)
North Carolina – Elizabeth Dole (R)
Oklahoma – Jim Inhofe (R)
Rhode Island- Jack Reed (D)
South Carolina Lindsey Graham (R)
Tennessee- Lamar Alexander (R)
West Virginia- Jay Rockefeller (D)
Wyoming- Mike Enzi (R)
Wyoming- John Barrasso (R)
Safe Seats Round-Up: 20 seats, 9 DEM, 11 GOP
Seats Leaning Significantly
Alaska- Ted Stevens (R) Probability = .15
Stevens is 85 years old and being investigated for some shady real estate deals and facing strong primary opposition from, of all people, a real estate developer. He is also polling about even with his most prominant Democratic challenger Anchorage Mayor Mark Begich. But still this is Alaska and I don’t think this seat will be as competitive as some think. McCain will also help here.
Kentucky- Mitch McConnell (R) Probability = .05
In 2004, Republicans heavily targeted and defeated Senate Minority Leader Tom Daschle. It’s possible the DEMs could get their payback in ‘08, but unlikely. Unfortunately, the best Democratic candidate Rep. Ben Chandler passed on the race almost assuring that McConnell wins reelection.
Idaho- Larry Craig (R) Probability = .1
Following his mens room escapades, Craig is retiring from the Senate. Running for his seat is Lt. Governor Jim Risch. Risch easily beat his opponent Larry LaRocco in the 2006 Lt. Governor race. Look for the same in ‘08 despite the large bankroll LaRocco will have.
Nebraska- Chuck Hagel (R) Probability = .2
GOP “Maverick” Hagel is retiring from the Senate. Former governor and US Ag. Secretary Mike Johanns is the GOP nominee against former Congressman Scott Kleeb for the Dems. Kleeb seems to have some cross-over appeal in conservative Nebraska and this might be one of the more competitive seats in this category.
Iowa- Tom Harkin (D) Probability = -.05
Harkin should win easily, but I’m still not completely comfortable putting this is the safe category considering it is Iowa in a Presidential year. Harkin should be benefited by Obama’s name on the ballot despite the fact that he endorsed Hillary.
Maine- Susan Collins (R) probability .1
If Olympia Snow’s 74% total in in 2006 is any indicator, Susan Collins shouldn’t have much of a problem in Maine, especially with the Democrats unable to get a top-tier opponent.
New Mexico Pete Domenici (R) probability = .80
The retirement of Domenici, who was being investigated for ethics violations surrounding the firings of the US attorneys, has left the door wide opened for Dem Rep. Tom Udall to be strongly favored. With the support of very popular governor Bill Richardson, Udall should be able to handle Steve Pierce or Heather Wilson, who are Republicans Reps vying for the seat. Each of New Mexico’s three house reps are currently running for the senate seat.
Oregon- Gordon Smith (R) probability = .15
Like Collins, it seems like the least vulnerable Republicans are the more moderate senators in generally Democratic states who have been somewhat questionable of the war. Smith has a stronger opponent than Collins, the Speaker of the state house Jeff Merkley, but Smith’s approval has been hovering around 50%, which shows that he is pretty strong going into the election.
Texas- John Cornyn (R) probability = .05
Cornyn would likely be vulnerable in most states, but he is still going to be pretty tough going into Texas. But at the same time I’m definitely not comfortable putting this seat into the safe category. Cornyn won in 2002 with over 80 percent, but his opponent, State Rep. Rick Noriega is a solid candidate and might break the 40 point mark. However, it will take a couple ‘maccaca’ moments for Noriega to win.
Virginia – John Warner (R) probability = .95
I really can’t think of a reason why former Gov. Jim Gilmore or State Del. Bob Marshall could beat popular former Gov. Mark Warner, but in Virginia I don’t think anyone can consider a Democrat as a lock to win a state-wide election. I also think McCain’s name on the ballot definitely doesn’t help Warner, but he’ll still probably win by at least 15 points.
Safe Seats Round-Up: 8 Seats, 1 DEM, 9 GOP E(X)= 2.5
Excluding Virginia and New Mexico the Probability that the Republicans hold all of the other six GOP seats = .445. Thus, the odds at greater than .5 that one of the Democrats other than Warner or Tom Udall wins one of these seats.
Marginal Advantage
Colorado- Wayne Allard (R) Probability = .75
With Senator Allard retiring, Dem Rep. Mark Udall has a strong chance of picking up a seat for the Democrats. His opponent is former Rep. Bob Shaffer who is polling slightly behind Udall. Udall will also benefit from the Democratic Convention in Denver and Obama’s name on the Ballot. If Obama even gets close here, Udall will win. I don’t see a lot of people splitting between Obama and Shaffer.
Louisiana – Mary Landrieu (D) Probability = -.25
If the Democrats lose a seat in ‘08 it will surely be Louisiana. Democrats have lost some reliable voters due to Katrina, but Obama on the ballot should help Landrieu. State Treasurer and party switcher, John Kennedy will challenge Landrieu and would appear to be a very strong candidate, however, he is polling behind Landrieu at the moment.
New Hampshire – John Sununu (R) Probability = .85
Following the “New Hampshire Miracle” of ‘06, in which Democrats took back both state houses for the first time since 1874 and each congressional district (they actually made a net gain of over 100 seats in the House), it would only seem natural for that trend to continue into 2008. The Democrats also got he best possible candidate, former Governor Jeane Shaheen to challenge Sununu. Besides a random ARG poll in December that showed Sununu up 11, Shaheen has been at least 8 points ahead in most polls. McCain probably has to win by about 7 points here for Sununu to have a shot.
South Dakota- Tim Johnson (D) Probability = -.15
Johnson won by only a few hundred votes in ‘02, but his popularity has increased after recovery from health problems, not to mention this year is a lot better for Dems than 2002. The Republicans also haven’t been able to get a top-tier candidate.
Marginal Round-Up: Four total candidates, 2 Dem 2, GOP. E(X)= 1.2
There is a .406 chance that the Democrats win all four seats.
Total Toss-up
Minnesota- Norm Coleman = .55
I wrote a little about some poll results in Minnesota about a month ago. What is interesting about this election is that Minnesota is really the only true toss-up in which it is very difficult to say that either candidate is necessarily ahead. But I think if Minnesota can elect Jesse Ventura as governor they can certainly pick Al Franken over Norm Coleman. Whichever party wins the electoral votes in Minnesota will also win the Senate seat.
E(X) = .55
Final Results
Overall there are 35 Senate races (12 DEM ,23 GOP), 20 safe seats (9 DEM, 11 GOP), 10 leaning seats (1 DEM, 9 GOP), 4 marginal seats (2 DEM, 2 GOP) and one tossup that is a Republican seat. Again, the expected Democratic gains in each category are 2.5 for the leaning seats, 1.2 for the marginal seats and .55 for the toss-up. The ads up to a grand total of 4.25, which would put the makeup of the Senate at 55-45 assuming Lieberman stays with the Dems.
Based on raw probability, that is if a candidate’s probability exceeds .5 he or she wins, the Dems would pick up 5 seats giving them 56. No Republicans would win using this method.
Thinking about the election as a whole the E(X) of 4.2 seems rather low, and I think this is due to my reluctance to go out on a limb for any of the leaning GOP seats. The Democrats will probably win one of the leaning seats and protect all of their seats, so, ignoring all the work I just did, I think the Dems will probably gain 6 seats in the Senate. I think as I alter these probabilities as the election gets closer the E(X) will better reflect this line of thinking.
Additionally, I know there is a way to run a large number of simulations using these probabilities and use the normal model to find the probability that the Democrats achieve a filibuster-proof, 60-seat majority. However, I’m not exactly sure how to do that, so I think that might be a project that will have to wait until this summer. I’m also planning on doing this state-by-state for the House elections, but that will be a heck of a thing and will also have to wait until the summer.
[...] Electoral College E(X) jump to navigation [...]