It's the Data, Stupid

 Peter Muhly/Getty Images

Peter Muhly/Getty Images

Recently, FiveThirtyEight.com teamed up with producers at ESPN to release a video about the “myth” of Ross Perot’s effect on the 1992 presidential election.

Well, we watched it. And while we give it an A for production and entertainment, and it does include many clips that will make anyone laugh at the weirdness that was that election cycle – their conclusions are borderline indefensible. To say that Ross Perot did not have an effect on George H. W. Bush’s loss of a second term ignores the data the website claims to love so much. They point to the Exit Polls from that year as proof, and while you can make an argument using that data point alone, the only way it works is to ignore all the other available data.

Let’s take a look at voting trends by both race and party support from the years 1976 to 2008, with a special focus on 1988 and 1992 – the years Bush 41 ran at the top of the ticket.

Going back to 1976, Hispanic voter support for the two parties has remained relatively consistent. There are a few noticeable shifts, such as peaks in 1980 and 2004 when the GOP captured noticeably high numbers of Hispanic voters. However, from 1980 to 1996, there was a steady decline in Hispanic support for the GOP candidate from a high of 37% in 1980 and a low of 21% in 1996. On the Democratic side there was a little bit of movement over the years and 1992 was a low year for Democrats compared to the election before and after.

In 1992, Ross Perot received the support of 15% of Hispanic voters, according to exit polls. However, given the fact that they made up 2% of the overall electorate, this shift did not have a big effect on the overall results. In general, for Hispanic voters, we would call the Ross Perot effect to be minimal based mostly on their small proportion of the electorate.

There is even less to say for African-American voter turnout when it comes to identifying trends by political party – steady is the name of the game. When compared with Hispanic and White voters, African-American voters have the most consistent partisan levels of support from the general elections of 1976 through 2008. Seventeen (17) percent of African-American voters supported the Republican candidate in 1976, and that number dropped to 14% in 1980, 9% in 1984, and then went up to 11% in 1988. In 1992, the number was 10%. On the Democratic side, support ranged from 83% in 1976 and 1980, and went up to 91% and 89% respectively in 1984 and 1988. In 1992, support for the Democratic candidate dropped to 83%. It is hard to look at these numbers, and the graph below, and say that there was any real difference during the Perot elections.

Not much interesting data so far, and you might be saying, "See, Perot did not have an effect." Well, let’s check out white voters, shall we?

You don’t have to be a data scientist to notice a trend here in 1992 and 1996. In the elections leading up to 1992, white voter support for the Republican candidate were: 52%, 56%, 66%, and 60%. In 2000, it was 55% and in 2004, it was 58%. During the Perot years it was 41% and 46%. In 1992, the support for the GOP candidate among white voters was 17.5 points lower than the average of the previous 4 elections and 9 points lower than the lowest level of support the GOP enjoyed among this cohort in 1976.

Just to go back over the history a little here, George H. W. Bush was a President that peaked at an 89% approval rating (which is unheard of this century), oversaw the end of the Cold War, and won the war in the Persian Gulf swiftly, among other things – all of which is mentioned in this video. So what explains his nearly 20-point loss in support in 1992? Did white people just really love Bill Clinton that much?

The answer to that last question is no, and we can look to this same data to prove that. Support for Clinton wasn’t any higher among white voters in 1992 than it was for Michael Dukakis in 1988 – in fact, it was technically lower (39% to 40%) – so we know that the 1 in 5 white voters that walked away from Republicans on the ballot in 1992 didn’t just hop over and vote Blue.

So, let’s get this straight. Even with this simple turnout data showing that trends with white voters were disrupted in this specific election year and that fact that the white voters that caused this disruption were Republicans who moved, not into the Democrat’s vote share, but into the “Other” category (which can heavily be implied as Perot’s vote share), Nate Silver still says Ross Perot’s effect on this election is a myth? All because of one question in the Exit Polls?

We’re going to call BS on this one. Bill Clinton was not going to beat Bush in 1992, no Democrat was. Ross Perot absolutely had a significant impact on the outcome of the 1992 presidential election and the data is there and points to all of this being the case – see, we weren’t lazy about checking it.