By In Stuff

Fun With Electoral Votes

It really is no wonder that Nate Silver got so into this election thing. Playing with the electoral numbers is really just about as much fun as playing with baseball statistics. Even more, it’s a whole lot like playing with NFL playoff possibilities at the end of the season (which, of course, is awesome).

The other day, I was trying to explain the election to my daughter, and we were playing with the CNN election map. (Warning: Do not click on that link unless you want to spend the next several hours coming up with your own electoral scenarios). She didn’t really understand when I explained that there are 42 states that are more or less decided already. I tried to explain to her polling and state’s histories, but she had trouble with the concept. Her thought: How can you know who will win the election until you actually have the election?

Still, I showed her that according to CNN — and I think this is probably about right based on the little I know — there are eight states really in play in the 2012 presidential election. And we clicked each state red and blue to see how the numbers added up. It was great — the first time time I was really able to share with her how much fun statistics and numbers can be. And the great thing is that it really was all about numbers — we didn’t talk about  the issues, the candidates, the contentiousness or anything like that. Not this time. It was an escape from all that. It was Candidate A and Candidate B. And it was a math puzzle.

Repeat: A math puzzle.


With 42 states more or less locked in, the numbers look like this:

President Obama 237, Mitt Romney 209.

There is some wiggle room in those numbers, of course. Those numbers assume Romney wins North Carolina, and assumes Obama wins Pennsylvania and Minnesota and Michigan. It’s possible those could go another way.

But, again, we’re playing a stats game here. Let’s go with the assumption that there are eight states in play. Here they are, their electoral vote total and (for fun) Nate Silver’s percentages on who likely will win the state.

New Hampshire: 4 (Obama 75.4%)

Nevada: 6 (Obama 82.8%)

Iowa: 6 (Obama 74.4%)

Colorado: 9 (Obama 60.7%)

Wisconsin: 10 (Obama 88.1%)

Virginia: 13 (Obama 61.8%)

Ohio: 18 (Obama 77.6%)

Florida: 29 (Romney 59.3%)

It should noted here that I started this post on Tuesday … by Wednesday, Obama’s percentages had gone up in all eight states (including Florida, where Romney was a 64% favorite before). If Obama wins the seven states where Nate has made him the favorite, he will win the election rather handily (303 to 248). This might help explain why Nate now has Obama 77.4% likely to win the election.

But you really can play all sorts of games with those eight states to determine the winner. Colorado and Virginia are basically toss-ups at this point (Nate’s percentages for Obama were in the mid 50s on Tuesday, so you can see how volatile the polling is right now). Win those two and Florida and Romney is an Ohio away from the Presidency.

Some questions we tried to answer:

Question 1: Is there a realistic way for Romney to win without winning Florida?

Answer: I don’t think so, no. I mean, we are assuming there is no big surprise in Michigan or Pennsylvania, but assuming those go blue (and it’s hard to imagine Romney winning Pennsylvania but losing Florida), Romney simply cannot lose Florida and win the election. There just aren’t enough electoral votes in play. Even assuming he wins Florida, he seems to me to be facing a pretty severe uphill electoral fight. Without it, Obama essentially would wrap things up by winning ANY of the other seven states in play, including Wisconsin which looks less and less like a toss-up state and more and more like an Obama certainty.

Question 2: Is there a a way for Romney to win without Ohio?

Answer: Yes, but it’s a longshot. It goes without saying that Romney would have to win Florida and sweep the toss-up states of Virginia and Colorado. Without Ohio, though (or an upset in Pennsylvania or Michigan), Romney would have no choice but to win Wisconsin and either Iowa or New Hampshire (Wisconsin would be a pretty big long shot at this point) OR he would have to win win Nevada, Iowa and and New Hampshire (Nevada would be a tough one). It’s become an election cliche, but it really does seem true: Romney’s hopes are in winning Ohio.

Question 3: Is there a way for Obama to win without winning Ohio?

Answer: Yes. It’s tricky, but certainly less tricky than Romney’s map. Obama will probably win Nevada and Wisconsin. That would leave him 17 votes shy of the 270 needed. He could get those 17 any number of ways even without Ohio. Virginia and any of the other three states would get him to 270. And if he lost Virginia, then a sweep of Colorado, Iowa and New Hampshire would get him there. But if Obama lost Ohio, that would probably be a bad sign for those toss-up states. And without Virginia or Colorado, he would could not reach 270.

Question 4: What would be the biggest surprise?

1. Obama wins Utah. Republicans have won the last three elections by 28, 45 and 40 points … and Romney obviously has deep connections to the state.

2. Romney wins Hawaii. Not only has Hawaii gone hugely Democrat the last three years, Obama was born there and won by 45 points in 2008. Hawaii has gone blue every year except during Nixon’s landslide in 1972 and Reagan’s landslide in 1984.

3. Obama wins Wyoming. Republicans have won last three by an average of 48 percentage points. Last time Wyoming voted blue? That was Lyndon Johnson in his 1964 landslide.

4. Romney wins Illinois. The state has gone double-digit blue the last three elections and Obama won his home state by 25 points in 2008. Illinois was a reliably red state until 1992, and has voted Democrat every election since.

5. Romney wins New York. Democrats have won by 27, 18 and 25 points the last three elections. Obama is ahead by about 25 points in the polls. The last time New York went Republican? Ronald Reagan in 1984.

6. Obama wins Alaska. Republicans won last three by 21, 27 and 31 points and Alaska has gone blue only one time, for Lyndon Johnson in 1964. As far as I know, there has not been a significant poll done in Alaska. None is really needed.

7. Romney wins Massachusetts. His time as governor would make this only slightly surprising; Democrats have won by an average of 26 points the last three elections. Other than the Reagan elections, Massachusetts has gone blue every year since Ike.

8. Obama wins Oklahoma. Republicans have won the the last three elections there by 31, 31 and 21 points. And the state has voted blue exactly one time since 1950.

9. Obama wins Idaho. Republicans have won the the last three elections by 25, 38 and 39 points.

10. Romney wins California. The state was once pretty reliably Republican, but has not gone red since 1988 and had gone double-digit blue each of the last three elections.

Others receiving votes: Obama winning Alabama, Mississippi, Lousiana, Kansas; Romney winning Maryland, Connecticut, Washington.

Question 5: What is the most likely scenario for a Romney win?

The most likely scenario is he wins Florida and Ohio and finds himself 17 electoral votes shy of 270. He can get those any number of ways, but I would say a victory in Virginia is the best one. That would leave him four shy, and he gets those four in Colorado or Iowa or New Hampshire.

Question 6: What is the most likely scenario for an Obama win? 

It’s likely Obama will win Nevada and Wisconsin, leaving him 17 votes shy. His most likely victory comes in winning Ohio. There are other opportunities, but Ohio would get it done.

Bonus question: For brilliant reader Owen … can we get a tie?

Yep. There are probably several ways to do it, but here’s the easiest: If Romney wins Florida, Colorado, Nevada, Iowa and Virginia, that gets him to 269. Obama would win Wisconsin, Ohio and New Hampshire, getting him 269. At that point, the House of Representatives would pick the next president, which would certainly be Mitt Romney. However, the Senate would choose the Vice President, which would probably be Joe Biden. What fun for everyone.

24 Responses to Fun With Electoral Votes

  1. Wes says:

    Your lede reminds me of that quote from the Simpsons episode with Bill James: “I’ve made baseball as much fun as doing your taxes!”

    (And I love doing my taxes).

  2. Owen Ranger says:

    Did you work out/explain the scenarios for a 269-269 tie? Because that would make the math puzzle even more fun.

  3. ARR says:

    You forgot what would be the biggest surprise- Romney winning the District of Columbia.

    Since receiving 3 electoral votes in the 1964 election, D.C. has never voted for a Republican. The recent breakdowns:
    2008: Obama 93% McCain 7%
    2004: Kerry 89% Bush 9%
    2000: Gore 85% Bush 9%
    1996: Clinton 85% Dole 9%

    D.C. is over 50% African American, a demographic which generally is supporting the President at more than 90% in most polls. In Ronald Reagan’s historic 1984 blowout, D.C.’s electoral votes were the only ones he lost, along with his rival’s home state of Minnesota.

    A Romney win there- heck even Romney reaching double digits in his voting percentage- would be historic.

  4. Old Joe says:

    Thanks for the breakdown, Joe. I’ve been following Nate’s fivethirtyeight column for several weeks and was amazed at how much it felt like reading fantasy baseball/football writings. Your daughter is a lucky young lady to have a dad who takes that kind of time with her!

  5. Chris says:

    “Not only has Hawaii gone hugely Democrat the last three years, Obama was born there”

    Really, Joe? Was he?

    Sorry, just doing it before somebody else comes on and says the same thing (but in earnest).

  6. For a little more math fun on how the actual vote split relates to percentage chance of winning you can read this:

    The point of it is that a half a percentage point of actual vote, say from 51% to 50.5% would have a 10 point swing on likelihood of winning. So the whole election is too close to call and yet what fun is that for predicting.

  7. Jeff Harris says:

    I’m really hoping for a tie.

  8. BobDD says:

    For statistical models to work, the data has to be good. If the data is fudged (climate control hockey stick), then it’s a joke. Nate’s polls appear to have the Dem-GOV breakdown at 2008 levels rather than 2010 or a combination thereof, so I see his models as ‘garbage-in-garbage-out’ wishful thinking. Now I know you tried to stay away from the politics of it and make it just statistical and so did I, but intentionally using bad breakdowns in polling is political chicanery, so my break with you on this is statistical, but I see it as coming about because of politics.

    • bluwood says:

      If you truly have been following Nate’s 538 blog for several weeks, you would understand and acknowledge the massive effort he makes to remove ANY leaning, whether it be left or right.

    • BobDD says:

      Hoping that our own biases are not making either of us stupid, I see him doing that somewhat in his comments but not in the numbers where he still has not adjusted for 2010 to my way of thinking. The election will tell us (and him) if 2010 should be ignored or not.

      The thing is that I see unrealistic partisan sampling as anti-stat savvy and that is the part that angers me about Nate right now. He always says that the NYT does not lean on him, but then he always leans left. It made him look very smart in 2008, look average in 2010, and for this year we most likely have differing expectations about how his model will fare. If he is just a leftist his predictions this election will tarnish his reputation, but if he really has the superior system he will be good again and I’ll have to admit I got it all wrong.

    • drunyon says:

      bluwood: Nate makes adjustments if he thinks a poll typically “leans 1 point Democratic” or “leans 2 points Republican”. He also weights different polls subjectively. What he does NOT do, to my knowledge, is adjust for the fact that the sample may be off. The sample in a lot of the state polls he is using has turnout around a D+8 or so, which is in line with 2008, but not really realistic in 2012. So the polls themselves, if they use the wrong sample, are simply inaccurate.

    • Rob Smith says:

      I love how the partisans attack the poll takers if they don’t match up with their desired results. I also think many Republicans look at the “National Poll” and think Romney is looking good, when it’s the Electoral College is all that matters (as Joe points out). The election will be close, and who know what will happen in the next few days, but it does look like Obama has a clearer path to victory through the electoral college than Romney. Romney just must have Ohio and Florida to win, and even those states don’t get him all the way there.

    • bluwood says:

      drunyon: Nate will take the poll results, and compare them to the historical results of that polling company, THEN compare all of those poll results with the ACTUAL results to see how far off they were, and then he calculates that into how to weigh the current poll produced by them. So he does adjust each sample based on their historic ability to be correct (as well as adjusting for obvious outliers).

    • Dinky says:

      BobDD, Nate makes a living from doing this stuff accurately. He explains his methods, and I’ve read him state that his adjustments are only based on historical biases in the poll takers. But don’t trust 538; go to their conservative counterpart, RealClearPolitics. There, Obama is given 201 electoral votes, Romney 191, with the rest toss-up. But if you give Obama every state where RCP’s analysis gives Obama a 2% lead or more, you wind up with Obama at 277 votes (without New Hampshire or Colorado, but still enough to win). If you ask for the no toss-up, the electoral count is 290 for Obama. So the only difference right now between the major liberal and conservative poll crunchers is differing on how Virginia will vote. That hardly seems like Silver is badly biased.

      It is extremely common for mid-term elections to go against the president’s party. The Republicans took Congress in 1994, but Clinton was still reelected. It is possible that 2010 represents a sea change, but it’s also quite possible (history would say almost certain) that people stung by the results in the last presidential election are likelier to go vote in the midterm election. Given the national polling numbers so hostile to the Tea Party, I suspect that history has it right, and you don’t.

    • Dinky says:

      It now appears that Nate Silver was a mathematician, as opposed to exhibiting any bias. Perhaps someBobDD owes Nate an apology, or the polling companies that provided the data Nate used.

  9. Frank says:

    Counting electoral college votes is much like match play golf with different point values assigned to each hole.

  10. AMusingFool says:

    Interesting playing around with the numbers. And a cool idea to do it with your daughter. I’ll have to remember that for next time, when my daughter will probably be old enough to comprehend it.

    One note, though: A tie is not certain to produce a Romney presidency, because the House would vote with each state delegation getting one vote. And if the state is deadlocked? Yeah, it could get really weird. In fact, I heard one explanation that could result in a Biden presidency (if the House deadlocks, the Senate decides, although I forget the exact mechanism).

    • Dinky says:

      There’s little hope of a House deadlock. Romney wins more states. Right now, without the 12 swing states at RealClearPolitics, Romney leads 22 states to 16, and if you give Obama nine of the twelve swing states (so there would be a House tie in states) then Obama wins the electoral college before it gets to the House. Yes, it’s possible that some of the more gerrymandered states could actually have equal numbers of electors from each party and deadlock themselves, but the likeliest examples would need to have Democrats gerrymandering a state that would otherwise go for Obama for that to happen, and the most blatant gerrymandering is going the other way. So I think it is certain that the Senate will not decide the next president.

      That mechanism is: if the House deadlocks 25 to 25 for each presidential candidate, each state getting one vote, then no president is chosen. The Senate’s vote is separate to pick the vice-president. If the Senate picks as VP Joe Biden (seems likely, with 52 Senators seeming to be Democrats) then newly chosen VP Biden (by the Senate) is VP with no President, and thus immediately becomes President. But it’s not going to happen. Refer to the 12th Amendment to see all the convolution. I think there’s also some added weirdness (never tested) that if the House deadlocks, AND the Senate deadlocks, then the old VP casts the tie-breaking vote in the Senate, which could lead to VP Biden choosing President Biden. But others think it could be interpreted differently, leading to an unbreakable tie.

      We really need statehood for Puerto Rico or Washington DC to prevent that 50-50 tie.

      There are also two states where the presidential electoral votes match their Representatives (Nebraska and Maine) which could lead to one vote going the other way. That opens up several more possibilities of an electoral college tie. Right now I think the only elector that does not look certain (merely likely) is in Maine, but that’s the likelier way it would need to swing to make a tie.

  11. bigsteveno says:

    Joe, you forgot to mention that Obama’s WAR number is much higher than Obama’s.

  12. My inner math geek loves this! It’s so much fun to play with the scenarios.

    How big a surprise will it be if Missouri goes blue? Just curious, as I live here and all and miss the days of being a swing state. (Except not really, since political ads make me thankful to the inventor of the DVR and the mute button).

    Also tangentially relevant and to add a history aspect to you discussion with Elizabeth, see this recent XKCD strip:

Leave a Reply

Your email address will not be published. Required fields are marked *