From Vol. 28, Issue No. 2 of The Polling Report, January 30, 2012
In the great, late-era western The Man Who Shot Liberty Valance, an aged United States senator comes home to attend the burial of a little known man who played a seminal role in the events that launched the senator’s career. The senator tells a newspaper reporter the real story of those events, which is at considerable variation with the legend, and the reporter effectively says, “When the facts collide with the legend, print the legend.”
So it is with the 1992 presidential election, as even today many reporters state as fact that independent Ross Perot cost George H. W. Bush the presidency. It is, after all, a great narrative: Vengeful billionaire sinks President’s reelection bid. Much sexier than: Billionaire runs for president, draws votes equally from both candidates, doesn’t change outcome. The Perot-as-Bush-destroyer narrative is also highly favored by Republicans; it makes them blameless for the objective conditions in the country extant at the time of the election: “It was that crazy Perot’s fault, not ours.”
How this canard gained such wide purchase among reporters, particularly given the wealth of easily available empirical evidence to debunk it, is a bit of a mystery. Most recently, Politico reporters John Harris and Jonathan Martin noted in a January 11, 2012, article that Perot “gravely damaged then-President George Bush’s reelection prospects.” They cite none other than Rush Limbaugh as a source for this. Mr. Limbaugh certainly carries considerable heft as a radio entertainer; as a serious political analyst, not so much.
We have pre-election polls and exit polling to make an assessment of what impact Perot actually had on the outcome. In a three-way match-up nationally, in early June 1992, Perot led with 39%, Bush was second with 31%, while Bill Clinton trailed with 25%, according to Gallup. Perot exited the race during the Democratic convention in mid-July. In the immediate aftermath of the convention, Gallup had Clinton leading Bush 56% to 34%, clearly a post-convention bounce. But a month later, Clinton still led—by between 17 and 25 points—in half a dozen national media polls, with President Bush not exceeding 37% of the vote in any of them. In mid-September, with Perot still out of the race, an ABC News/Washington Post poll gave Clinton a commanding 58%, with the incumbent still stuck at a very familiar 37%.
Then, on October 1st, Perot re-entered the race. An October 8-11 poll—done by the Times Mirror Center for The People & The Press, directed by the outstanding Andrew Kohut—found that Clinton had dropped to 48%, with Bush at 35%, and Perot at 8% (in mid-September, they had found Clinton leading Bush 53%-38%). An October 20-22 follow-up poll of the same 1,153 voters surveyed earlier in the month found that Clinton had slipped to 44%, while Bush held at 34%, and Perot had jumped to 19%. The very first sentence of the extensive press release, dated October 26, noted that, “Ross Perot's surge in the polls is drawing somewhat more support from Bill Clinton than from George Bush, and the third party candidate seems poised to make more gains that might further narrow Bill Clinton's nationwide margin.”
That press release came out the same morning that Perot’s bizarre charges that Republicans had conspired to ruin his daughter’s wedding floated into the general political consciousness, and that was the end of the Perot surge. Nonetheless, he still drew 19% on Election Day, to Clinton’s 43% and Bush’s 37.5%.
What do we learn in sum from these pre-election polls? First, we learn that from mid-July through September of 1992, Clinton’s lead over Bush was at its greatest, nationally, and consistently ranged at or in excess of 15 points, except for a very brief time after the Republican convention. Second, we learn that the race began to narrow as Perot picked up support after re-entering the race. Clinton ended up winning by about 5.5 percentage points, far below his peak margins of the summer.
Third, and most important, we learn that the vote share of President Bush stayed within a narrow range in all of the polls. In the polls cited above, his vote share ranged from a low of 31% to a high of 38%, and he ended with 37.5%. The “change” vote oscillated between the two challengers—when Perot was up, Clinton was down, and vice versa.
Beyond the polls I’ve cited, the vast majority of other public polls taken between mid-July and the election showed Bush in the same range, and Clinton with the same kind of lead in a two-way match-up. Bush’s numbers just did not move very much, regardless of whether Perot was in the race or out of it.
These poll results are consistent with the dynamic of an electorate that has rendered a negative judgment on an incumbent that is very unlikely to change. Bush’s approval rating hit 40% in March of 1992 (before Perot was a major factor) and effectively stayed at or below that level until the election. By the summer and early fall of 1992, his job approval ratings were consistently in the upper 30s, and up to 75% of the voters said the country was off on the wrong track. This is a prescription for incumbent defeat; and Bush’s problems helped create Perot, not the other way around.
Exit Poll Data and the Perot Vote
Now, let’s briefly consider the 1992 exit poll data and the actual composition of the Perot vote. According to the exit poll data, 38% of the Perot voters said they would have voted for Clinton in a two way race, 38% would have voted for Bush, 24% would not have voted. Perot won 30% of independents, 17% of Republicans, and 13% of Democrats. Put another way, of his 19% popular vote share, 8 percentage points came from independents, 6 from Republicans, and 5 from Democrats. Fully 53% of Perot’s vote came from self-defined moderates, 27% from conservatives and 20% from liberals; so about 10 points of his 19% came from self-described moderates, with 5 points coming from conservatives and 4 points from liberals. We also know from the exit polls that the Perot voters were angrier at the political system than supporters of the other candidates. Do these Perot supporters really look like voters that would have gone heavily to incumbent Bush in a two-candidate race?
It is just possible that Perot cost Bush a state here or there where Clinton squeezed out a very narrow plurality (Colorado, Montana, Ohio and Georgia come to mind as possibilities), but there is no empirical evidence that documents this that I am aware of. Even if true for all four states (a very unlikely probability), it merely reduces Clinton’s electoral vote majority from a near landslide to very comfortable.
When confronted with hard data on the Perot voters, some proponents of the Perot-damaged-Bush theory will fall back on the argument that “Perot changed the dynamics of the election; he helped Clinton by ganging up on Bush and making it two against one,” or that “Perot launched himself on some kind of kamikaze mission against Bush” (Limbaugh’s fact-challenged theory) and concentrated his fire exclusively, or nearly so, on the President. Both conjectures are false. Perot directed his fire at the political system and Washington, D.C., climate more than at Bush, and he found a responsive national audience for that anti-status quo message.
Character and Communism
The second part of this argument—if Bush had been able to go one on one with Clinton from July onward they could have made his character the defining issue—also lacks credibility. It ignores the fact that Bush did have a two-way race with Clinton for well over two-thirds of the time between mid-July and the election (from mid-July to the end of September), including the Republican convention and its aftermath.
The Bush team made little or no progress closing the margin, despite the fact that Perot was out of the way for all of this time. Even a cursory examination of media reporting illustrates the strenuous efforts the Bush people made against Clinton’s character during that time frame. Despite attacking Clinton for pot smoking, draft dodging, adultery, and a trip to Moscow in which they accused him of possibly meeting with communists for nefarious purposes, they still trailed by 15 points or so in late September. It is absurd to believe that another month of the same would have produced a Bush victory.
Like 1980 and 2008, 1992 was a big-issue election held in very cranky times. The reality of 1992 was that by the early fall (at the latest) voters had decided they wanted a change, Clinton had passed the presidential threshold test with swing voters, and they had made an uneasy peace with his personal foibles. Absent a bombshell in the last month, more character attacks were not going to get Bush the win in a one-on-one race with Clinton.
The Clinton people had it right, it was the economy; he got that the country was unhappy, and the swing voters got that he got it. Those who spin the argument that campaign process or dynamics would have changed the outcome in 1992 ignore the much bigger picture of voter discontent that was in play. Absent Perot in the final month, the Bush campaign likely would have made some progress in closing the enormous gap he faced, but in the end he would have lost, and handily.
It would be nice if reporters, in particular, would not do a variation of the Liberty Valance story when assessing the impact of Ross Perot on the 1992 election. It is unfortunate that, while substantial amounts of easily accessible data are there to refute it, the legend of Bush/Perot still retains currency among those who ought to know better. ï®
Tim Hibbitts is a Principal and Founder of DHM Research.