Monday, October 06, 2008

Tackling The Old Republican "Perot The Spoiler" Story

In 1992. Bill Clinton had just won his first term, defeating the incumbent, George H.W. Bush. Eight years later, After George W. Bush beat Gore in Florida by just a few hundred votes which gave him the presidency after some Supreme Court assistance, Ralph Nader was widely seen as the cause of Gore's defeat. It didn't take a degree in math or statistics to figure out that if even a fraction of Nader's voters voted for someone other than Nader and voted even slightly more for Gore, then Gore would have won. The point of this post is not to re-hash all of that though. It is to challenge the now common assumption among Republicans that the same thing happened to H.W. Bush in 1992 because of Ross Perot being in the race.

This picture is a lot more complex, because Clinton won by a much greater margin in the electoral college than Bush, but Perot also had a lot more support than Nader. First let's start with the states where Clinton won, but his margin was smaller than Perot's support. These states are: CA, CO, CT, DE, GA, HI, IL, IA, KY, LA, ME, MA, MI, MN, MO, MT, NV, NH, NJ, NM, OH, OR, PA, RI, TN, VT, WA, WV, and WI. These are all of the states Clinton won except AR, DC, MD, and NY. Sounds like this could be a big difference-maker when you look at those lists, right? Maybe, maybe not. Consider also that the only state where the Perot votes were smaller than the margin for Bush was in MS. Let's see how close those margins were and what the Perot voters would have had to do to tip the states and the electoral college to Bush.

I created a spreadsheet that could take into account both the percentage of Perot voters that would vote at all, and the percentage that would vote for either candidate out of that pool. It turns out that if every single Perot candidate voted, and voted for Bush or Clinton instead of a third party candidate, Bush would have needed over 66% of those votes. When switching the percentage from 66% to 67%, Iowa flips to Bush, giving him a 274-264 electoral vote victory. Perot voters were clearly a pretty broad coalition, as evidenced by his fairly even support across the country. Other analyses show that Perot was drawing as many or more votes from Clinton than from Bush, and I have yet to see a poll that says otherwise, much less a two to one advantage for Bush among his voters.

Consider also that Perot's voters, as with all third party voters, have a fair probability of voting third party or not voting at all in the absence of their candidate. Personally, I think 80% is a generous figure to determine how many would have voted for Bush or Clinton, and that participation percentage would have required Bush to get more than a whopping 71% of those remaining Perot votes. So, it is very clear that the Perot Factor in 1992 was not nearly the determining factor that the Nader Factor was in 2000. I don't even know why Republicans would even wish for this to be true other than making it seem like it's "even" because they had bad luck too. For anyone that would like to check my work, I would be happy to send you the spreadsheet. Just email me at egrow and then at the domain (to confuse the spammer bots).


Rightwingsnarkle said...

If you're into this kind of number crunching, I hope you've been spending time at Nate Silver's site.

Erik Grow said...

Hm, I've not seen it, but I will look now, thanks.

Erik Grow said...

Oh I HAVE seen that site! I just didn't recognize the name. I like myself.