Tuesday, November 4, 2008

Your Election Watching Guide

I know, I swore I'd only write one piece on the election . . . but I couldn't help myself.

If you're anything like me then tonight you'll find yourself in front of a TV with friends or family watching the election returns come in. Since you're reading this blog I'll assume that you're smart and well-informed, but many of those friends and family that are watching with you might not be. Rather than succumbing to information overload, here are a few things to look out for, in the form of answers to the following key questions:

1.) What are the odds that Obama/McCain will win?
2.) What are the key states to watch?
3.) When will we know it's over?
4.) What's going on with the Senate; will the Dems win 60 seats?
5.) How reliable are the polls?
6.) Anything else I should keep in mind?
7.) Couldn't you have said that all in one paragraph? (a.k.a the summary)


What are the odds that Obama/McCain will win?

According to FiveThirtyEight.com, there's a 98.9% chance that Obama will win the election. Why so high? Read on . . .

Depending on which national tracking poll you follow, you might think that the race is close. But candidates don't win based on the popular vote; they win based on victories within states. Here are how the polls indicate the states are leaning right now. Following each state is the number of electoral votes that state has (in parentheses) and four indicators of how far ahead the favored candidate is in that state:

1. The average of the recent polls, computed by RealClearPolitics
2. The average of the polls, using a regression line, computed by Pollster.com
3. The projected vote, using poll averages, historical trends, and demographic data, as computed by FiveThirtyEight.com
4. The odds of victory, as computed by 538

McCain Will Win in:
South Carolina (8)
Texas (34)
Alabama (9)
Mississippi (6)
Tennessee (11)
Kentucky (8)
Oklahoma (7)
Arkansas (6)
West Virginia (5)
Kansas (6)
Nebraska (5)
Utah (5)
Idaho (4)
Wyoming (3)
Alaska (3)
Total: 120

McCain is definitively ahead in:
Louisiana (9): N/A|10.6|9.9|99%
South Dakota (3): 8.3|8.0|8.8|99%
Arizona (10): 3.5|4.9|4.9|94%
Georgia (15): 4.0|2.9|3.7|90%
Total: 37

McCain is slightly ahead in:
Montana (3): 3.8|2.2|2.7|81%
North Dakota (3): N/A|0.7|2.7|76%
Indiana (11): 1.4|1.2|1.5|70%
Total: 17

The following states are true toss-ups:
Missouri (11): M0.7|O1.1|M0.2|M53%
North Carolina (15): M0.4|O0.4|M0.2|M53%%
Total: 26

Obama is slightly ahead in:
Florida (27): O1.8|O1.7|O1.7|O73%
Ohio (20): 2.5|3.1|3.4|88%
Total: 47

Obama is definitively ahead in:
Nevada (5): 6.8|7.1|4.9|95%
Virginia (13): 4.4|5.6|5.6|97%
Colorado (9): 5.5|7.6|6.6|98%
Pennsylvania (21): 7.3|7.2|8.1|100%
Total: 48

Obama will win in:
New Hampshire (4)
Minnesota (10)
New Mexico (5)
Massachusetts (12)
Connecticut (7)
Maine (4)
Rhode Island (4)
Vermont (3)
New York (31)
New Jersey (15)
Maryland (10)
D.C. (3)
Delaware (3)
Michigan (17)
Illinois (21)
Wisconsin (10)
Iowa (7)
California (55)
Washington (11)
Oregon (7)
Hawaii (4)
Total: 243

McCain will win 120 electoral votes. He's definitively ahead in states worth another 37 and slightly ahead in states worth another 17. That brings him up to 174. If he wins all the states in which the race is a toss-up (26) or where Obama's slightly ahead (47) that would give him 247 electoral votes. 270 are needed to win. That means he needs to win 23 electoral votes from among NV, VA, CO, and PA -- all of which are long-shots for him.


What are the key states to watch?

McCain has a very small margin of error, so if he loses any close state it's a big deal. In other words, there an awful lot of outcomes that would prevent him from winning 270. If you scroll up and read through the chart again, you'll see that Pennsylvania would put him close if he also wins all the closest states. Since it would take a miracle to win either PA, however, the more likely way that he could go over the top is to win Virginia, Colorado, and Nevada. That won't be easy, but it's certainly something to keep an eye on. In other words, if McCain loses Virginia it's pretty much over -- and if he loses both PA and VA then he should call Obama and concede.

Virginia is the state with the earliest poll closing time that is likely to serve as a firewall for Obama. But it's not the only state to watch. Barring a miracle in PA, McCain has to win virtually every single state in which Obama's not already ensured victory. All of his safe states + the close states only gives him 247 electoral votes. Adding VA, CO, and NV would give him 274. That means he only has four electoral votes to spare. In other words, if he loses any of the following states (in order of likelihood of losing): CO, VA, NV, OH, FL, NC, MO, IN, GA, AZ, LA, SC, WV, TX, AR, MS, TN, KY, KS, NE, AL, OK, or UT or two of: ND, MT, SD, AK, ID, and WY then he's virtually assured to lose (again, barring a miracle in PA).


When will we know it's over?

Depends on what your definition of "know" is. January 20th is the safest bet. When the networks put a check mark next to one of the candidate's names is the second safest bet. If and when McCain loses Virginia, however, you'll pretty much know. When might that be? Here's a list of the time that polls close in every state (a second time listed in parentheses means that part of the state closes earlier -- results will start trickling in then, but after the Florida debacle in 2000 no network in their right mind will call the state yet). If you're a graphical person, here's a cool map of all the closing times. All times are Eastern.

7pm
KY & IN (6pm)
SC, GA, VA, VT

7:30pm
WV, NC, OH

8pm
FL & NH (7pm)
MS, TN, AL, OK, MO, PA, ME, NJ, CT, MA, DE, IL, MD, DC

8:30pm
AR

9:00pm
TX, KS, SD, LA, ND, MI (8pm)
NE, WY, AZ, CO, NM, MN, WI, RI, NY

10pm
UT, MT, NV, IA

11pm
ID & OR (10pm)
WA, CA

midnight
Alaska

1am
Hawaii


What's going on with the Senate; will the Dems win 60 seats?

The senate will be much more heavily Democratic than it has been in a long time. A lot of people are speculating about the possibility that Democrats will control 60 seats in the Senate (60 is an important number b/c it takes 60 votes to override a filibuster). 60 seats is a reach, but 59 is not. Since the Presidential race will likely be over early (regardless of how long it takes the networks to actually put that check next to somebody's name), the true political junkies will be watching the Senate races. Here's what to watch for:

The Democrats currently control 39 seats (including the two Independents) and the Republicans 26. 35 seats are up for grabs; of those, 17 are virtually assured to go Democratic and 12 Republican. That gives the Dems a minimum of 56 seats (a week ago that number was 55, but then Ted Stevens (R-AK) was convicted on felony charges and the race went from a toss-up to a sure win for the Dems). That leaves 3 races that are eminently winnable for the Dems:

-Jeff Merkley (D) is all but a shoe-in over Gordon Smith (R) in Oregon -- giving them 57 seats.
-Kay Hagan has a bit closer race in North Carolina, but will likely beat Elizabeth Dole -- which brings them to 58.
-Al Franken (D) is currently a toss-up against Norm Coleman (R) in Minnesota (polls have ranged from Franken (D) +5 to Coleman +6 in the last week) -- meaning that the Democrats have a legitimate shot at 59.

Winning a 60th seat, however, will be tough. There are three races that will likely be won by Republicans but that are not quite automatic victories. If the Democrats can win one of those three, it could mean that they will control a filibuster-proof majority in the Senate. Those three are:

-Minority leader Mitch McConnell (R) will likely beat Bruce Lunsford (D) in Kentucky
-Saxby Chambliss (R) will probably beat Jim Martin (D) in Georgia, but some polls have this race very close
-Roger Wicker (R) is the favorite to defeat Ronnie Musgrove (D) in the special election in Mississippi

For a more thorough breakdown, see 538's final senate projections.


How reliable are the polls?

Well, I guess we'll find out pretty soon.

Overall, we should expect the polls to provide a rough picture of what will happen. Some individual polls will certainly be far off, but it would be a surprise if the general trends are wildly out of line. That said, it's not impossible. Polls could be wrong for any number of reasons, but three key ones this year are:

1.) Cellphones. A large number of Americans use cellphones instead of landlines most or all of the time. People under 30 -- a demographic that heavily favors Obama -- are especially likely to do so. By federal law, however, pollsters cannot use automatic dialing machines to call cellphones. This means that in order to sample cellphone users a pollster has to pay a human to dial those numbers by hand -- a task that is many times more expensive. As a result, many of the polls do not sample cellphone users. Theoretically, correctly weighting responses can still yield accurate estimates -- but that may or may not be happening. A recent glance at the national polls found that those who sampled cellphones had Obama up 9.4 points and those that didn't had him up 5.1.

2.) Who will vote? In deciding whether or not their sample is representative of the population pollsters rely on their estimates of who will vote -- in other words, their "turnout models." Normally, pollsters can predict with some precision what percentage of the voters will be of a certain age, race, or party affiliation. They factor in those demographics, along with questions asked of individuals (for example, about whether they've voted in the past, how likely they are to vote this year, etc.) in order to decide who is likely to vote and what the entire population of voters will look like. This year, however, may prove harder to predict. Huge numbers of people registered to vote for the first time during the primaries and over the summer and nobody is quite sure how many of those people will vote. Furthermore, many are speculating that larger than average numbers of young voters and African-American voters (both groups that support Obama by large margins) will turn out on or before election day.

3.) The "Bradley Effect." Named for California Gubernatorial candidate Tom Bradley, the Bradley Effect is the notion that White voters will tell pollsters they're voting for the Black candidate so that they don't appear racist but, ultimately, will pull the lever for the White candidate when nobody is looking. Empirical evidence suggests that this happened with some regularity in the 80's and early 90's but is no longer a factor. Interestingly, it may not have actually happened to Tom Bradley (both his aide and his opponent's aide argue that it's a myth).

In short, polls could be off-target for a large number of reasons but there are a couple of large issues that could mean that polls are regularly underestimating Obama's support.


Anything else I should keep in mind?

Other than not taunting your friends (too much) that were rooting for the other guy if your candidate wins or leaving the country if your candidate wins here are some things to keep in mind:

1.) Early voting. As you can see in more detail on this map, a lot of people are voting early -- and in key states. Poll results have indicated that early voters are leaning more toward Obama -- often by huge amounts -- than are people who have yet to vote. In some states the majority of the electorate has already voted, and Obama is up by quite a bit. The table below shows the results reported in a few polls within the last week (click on the table to see it in readable quality).

CNN's website has a map where you can see the number of votes cast early by state; sometimes broken down by party ID.

Assuming the polls are accurate, these results could mean any number of things. It could indicate that Obama's supporters voted early and McCain's supporters are voting later (which wouldn't affect the results). Or it could mean that Obama's supporters are more enthusiastic and better organized. Or it could mean that the current voter turnout models the pollsters are using are significantly underestimating Obama's support. It's tough to imagine, however, that there's any way that these numbers could help McCain. Indeed, starting election night down about 20 points with about 2/3 of the votes cast, as he is in CO, NV, and NM, is not a position in which I'd want to find myself. Georgia, by all rights, should vote for McCain, but Obama is ahead with over half the votes counted -- if that result carries through it's a sign that pollsters significantly underestimated the turnout of Obama voters, that these early voting results matter, and that Obama will win in a rout.


Couldn't you have said that all in one paragraph?
(a.k.a the summary)

Yes, I could've. For those of us who like brevity, here's the condensed version:

The latest polls give Obama heavy advantage, but don't preclude any possibility of a McCain victory. McCain essentially has to win every single state in which the race is close except for Nevada in order to win -- or pull out a miracle in Pennsylvania. If he loses Virginia it's highly unlikely that he'll win. If he loses both VA and PA, it's over. The senate race essentially starts out with the Dems controlling 57 seats. A Hagan victory in NC and Franken victory in MN would give them 59. If they steal a seat in MS, KY, or GA they'll have a filibuster-proof majority. This all assumes, of course, that the polls are reasonably accurate. Whether or not pollsters accurately estimated who will vote will likely determine their accuracy. The fact that so many people (most of whom were Obama supporters) voted early is something to keep your eye on -- it might mean nothing, but it could signal an impending Obama rout. If Obama wins Georgia, it's a sign of just that.


In case you were wondering, here are my predictions:

1.) Obama will win 364 electoral votes and win the popular vote 52-45
2.) We'll know he's won when Virginia is called for Obama -- at, I'll say, 8:30pm Eastern
3.) The Dems will control 59 seats in the Senate when all is said and done

Update: My follow-up post contains two things you might find helpful while watching the election: a spreadsheet with all the information you want to know about each state (how many electoral votes, who's up in the polls by how much, and what time(s) the polls close) in addition to a scorecard that will help you keep track of who's winning.

Enjoy your election watching.

No comments: