The Sports Report

A Statistical Take on Sports and Politics

Fuzzy Election Math

It has been a while since I reported on the ongoing events in the political polling world.. So today I have decided to finally announce one of my biggest spreadsheet innovations in a very long time. For the last week, I have been editing a small program published on Nate Silver’s amazing Five Thirty Eight blog that automatically computes the winner of the popular vote based on input information for party identification, and retention rates.

If Obama wins Democratic ID voters by a margin of 73.5... What would McCain have to do?

If Obama wins Democratic ID voters by a margin of 73.5... What would McCain have to do?

Without any further ado, take a look at the graph on the right side of this page. Taking given information from Rasmussen and Gallup in terms of independent voter preference and party identification breakdown, I have analyzed what a typical election situation could look like. The key problem to understand in this graph, is that if Obama wins the possible democratic electorate (minus given points to Ralph Nader, the independent candidate,) by a certain margin, and with everything else held constant, what would McCain have to do in order to tie? (Recent Gallup weekly averages have Obama winning the democratic electorate by about a margin of 73.5, and that is why I have that in my this breakdown.) My conclusion to this problem is that since the democrats are expected to have such a huge advantage in terms of identification, and since Obama is currently winning independent voters, McCain would have to win the republican base by a nearly impossible margin of 82.33%.

Of course, I completely understand that I am taking a whole lot of statistical liberties by holding so many different things constant. The key concept to understand, however, is that under the current political environment, I find it particularly unlikely for John McCain to be leading in the popular vote count. Yes, I know that ever since the Republican National Convention and the announcement of Sarah Palin as his running mate, McCain has been solidly ahead of Obama in most polls. The problem I identify in these polls, however, is that a radical and sudden shift in party identification, and independent voter support is not likely under any situation. Just because John McCain decided to pick only the second-woman ever to run for on a national presidential ticket, I do not see reasoning for so much change to occur so suddenly in all of the polls.

My logic behind this theory follows suit with the work of Silver. He previously had adjusted for the fact that most conventions hold a specific bounce for their respective candidate. He recently changed that adjustment within his graphs and calculations, because there had never been such a peculiar situation in history whereas the two conventions were held back-to-back in late August and early September. I strongly disagree with his decision, based on a poll he took on his website, to eliminate this adjustment from his system. Yes, there is no historical precedence to this year’s election in that fashion, but this election was already incredibly ground-breaking in the fact it will be the first national test of the supposed Bradley Effect. I believe that in order to fairly historically average in the most recent polling, we must have a way to account for the convention bounce that Silver himself projected just a few weeks ago, would last up until October.

If Obama wins Democratic ID voters by a margin of 73.5... What would McCain have to do?

If Obama wins Democratic ID voters by a margin of 73.5... What would McCain have to do? (Part Two)

Now take a look at the notes on the side here in connection to my graph up above. In the standard situation that I have created, based on averages from two accredited polling companies, about a half million more registered democrats are voting than republicans. In addition, these notes breakdown my calculation of the so-called “Magical Number” that McCain would have to attain in order to tie Obama in terms of the overall popular vote. As you can also see on the bottom of these notes, I also accounted for the expected increased turnout in this presidential election in comparison to the previous two Bush elections in 2004 and 2000 in terms of the total turnout of the United States voter-eligible population.

If McCain wins Republican ID voters by a margin of 79.5... What would Obama have to do?

If McCain wins Republican ID voters by a margin of 79.5... What would Obama have to do?

Keep in mind now, that until this point I have not specifically stated what Gallup currently states for McCain’s actual numbers among republican registered voters. He is currently averaging a margin of victory in the republican electorate by about 79.5 percent. Thus, I can hold his numbers constant just like I did for Obama’s and determine who exactly comes out on top. In this situation, with Obama’s numbers as the variable information, and McCain’s retention rate being held constant along with all of the other given information in my notes, I can see what Obama’s magical number would be. In this situation, in order to tie McCain in the popular vote, Obama would actually only need to win the democratic electorate by a margin of about 70.96%. Do you notice anything about the changes to McCain’s and Obama’s numbers that might be the cause of the 500,000 extra democratic voters?

In the 2008 political environment, I believe that Barack Obama is the clear favorite to win the popular vote. If I were to switch away from my automatic spreadsheet that calculates “Magical Numbers” and adjusts to create a tie in the popular vote, to a system much like Silver’s original spreadsheet, I could see how big this win might be. With all of the current conditions and averages, I have Obama leading by about 1.06% nationally. Why is this not in line with many of the major most-recent national polls? The most important answer to that question may be that this is just a computer model, and not the actual opinions of American voters. However, I fundamentally believe that those polls are outliers mostly caused by the recent conventions, and thus should not be treated accurate predictors for the election in November. I believe that if the trends in the political environment, highlighted by the party identification information on Rasmussen hold true to a certain degree, then Obama will be in a great situation to win by the time the results actually do matter.

Additional notes: In my calculations, I actually included the November 2004 identification numbers, because I thought that those would help ground my equation in a more realistic environment, than just the numbers from the last year or so…. Just in case you were interested enough to think about what Obama and McCain would need under many different situations, I also included a table I created in my spreadsheets that shows the changes in “Magical Numbers” based on my given conditions. Check out that image at the bottom of this article… Just to finish up, I wanted to let everyone know that if you really like my calculations here, or just want to see the spreadsheet yourself for any reason, e-mail me at I will attach my spreadsheet to you in my reply, and thoroughly explain how the program works.

Here is how the numbers between Obama and McCain match up in terms of winning their respective bases.

Here is how the numbers between Obama and McCain match up in terms of winning their respective bases.


September 11, 2008 - Posted by | Politics | , ,


  1. Interesting. Some of your explanations are a little hard to follow for my non-statistics-graduate-student-self, but I appreciate what you’re saying here. As a strong, yellow-dog Democrat, I badly want your calculations to give me hope. They aren’t though, because we know that Obama is going to win the most populous states (i.e. CA, NY, MA) by a lot, and because they are big states, a lot there means a large margin for the convincing-yet-sadly-meaningless national popular vote (though I actually appreciate that the Electoral College means that we don’t do nation-wide recounts).

    Not that you would have anything better to do with your time than this, but have you thought about, or attempted to apply your same statistical machete to the electoral thickets of OH, PA, IN, CO, NM and any other tied or candidate =/- <5% state you can manage to get to?

    I really appreciate this, because I know that I’m not representative of the average voter (though I cannot understand why). As a person who sacrificed to bring a Boys and Girls Club to a rural, depressed area, I was insulted and incensed, for example, by Gov. Pallin’s insults of “community organizers” in her convention speech, and was convinced that others would find her statements to be as untruthful and divisive as I did, only to find that people liked her moxie (I respect the moxie too, if you make me be truthful. I just wish she was truthful) and we have this maddening bump, even though it is clear through media reports that much of the substantive statements of “fact” she has made have been de-bunked.

    I’m geeking out on your stuff like Chuck Todd on too much Adderall.

    Comment by Kathie | September 13, 2008 | Reply

  2. It would be extremely difficult for anyone to come up with the registration numbers for these states. I completely agree with your argument to my work, but it just might not be possible for anyone to do.

    These past two weeks have been by far the dirtiest of the election. Sarah Palin’s presence has excited the base, and the Obama campaign is struggling to react to this change in the political environment. The purpose of this post, however, was to show that Obama has so much more room to grow off of his base than do the Republicans. It will be next to impossible to win by a margin of greater than 80 or so percent, and thus Obama retains the national popular vote edge. I am confident that this standard will hold true in many key states as well because there are simply more registered Democrats out there, and all polls (up until this past week) indicate that indepdents prefer Obama.

    Haha I appreciate the joke.

    Comment by Jacob Rosen | September 14, 2008 | Reply

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: