October 1996 Volume 23 Number 5
'Quantitative historian' Allan Lichtman employs classic operations research to predict presidential elections.
As you read this, the presidential campaign is in full swing, and the pundits are telling a new story each week, whenever a new poll appears. To at least one analyst, however, the result was already pretty much a sure thing months ago: a win for President Clinton.
Meet Allan J. Lichtman, professor of history at The American University in Washington, D.C., consultant to the 1992 Gore campaign (among others), successful early predictor of the last three presidential elections, and author of the popular book, "The Keys to the White House, 1996." His forecast of a Clinton victory is on record at least as early as June, 1996, when he was the featured speaker at the Washington OR/MS Council (WORMSC) annual banquet. WORMSC is the D.C.-area chapter of INFORMS.
Professor Lichtman's success conveys good news and bad news for OR/MS analysts. The good news is, his method is a classic example of good OR/MS &emdash; an application of mathematical and statistical methods from another discipline to a complicated problem. The bad news is that the profession can't claim credit for him: he didn't even know what OR/MS was when he was invited to the WORMSC banquet. He describes himself as a quantitative historian; his Ph.D. is in history, from Harvard.
Professor Lichtman's predictions are based on 13 questions (see box), each with a "yes" or "no" answer. "Yes" answers favor the incumbent party. If five or fewer answers are "no," the incumbent party retains the presidency; if six or more are "no," the challenger wins.
The Thirteen Keys to the Presidency
1. The incumbent party holds more seats in the U.S. House of Representatives after the midterm election than after the preceding midterm election.
2. There is no serious contest for the incumbent-party nomination.
3. The incumbent-party candidate is the current president.
4. There is no significant third-party or independent candidacy.
5. The economy is not in recession during the campaign.
6. Real (constant-dollar) per capita economic growth during the term equals or exceeds mean growth for the preceding two terms.
7. The administration has effected major policy changes during the term.
8. There has been no major social unrest during the term.
9. The incumbent administration is untainted by major scandal.
10. There has been no major military or foreign-policy failure during the term.
11. There has been a major military or foreign-policy success during the term.
12. The incumbent is charismatic or is a national hero.
13. The challenger is not charismatic and is not a national hero.
Note: If six or more of these statements are false, the incumbent party loses.
The keys do not predict the margin of victory. For example, in 1972, with four keys against him, Richard Nixon won by a huge margin, 60.7 percent to 37.5 percent for George McGovern. In 1880, another four-key election, James Garfield won by less than 2,000 votes, the closest presidential election in history.
The method is based on, of all things, a statistical pattern recognition algorithm from seismology. The highest plurality of the popular vote, not the electoral vote which actually decides the presidency, is the criterion &emdash; which means that in two elections, 1876 and 1888, the "winner" as defined in this method did not end up as president.
Out of nearly 200 questions, which were all binary ("yes" or "no") variables, the algorithm picked those which displayed the greatest difference between the proportion of the time the variable was "yes" for years when the incumbent party won and the corresponding proportion for years when the challenging party won, using all U. S. elections starting with 1860 as the training set.
In English-language terminology, the technique most closely resembles kernel discriminant function analysis. The references are somewhat unclear, as the key one is from a 1980 Russian-language seismology journal that apparently refers in turn to an earlier source which Lichtman and his co-author, Volodia Keilis-Borok, did not cite.
The method does not include an algorithm for selecting variables.
"We started with about 200 variables we considered possible, narrowed
it down fairly quickly to about 30, and then just kept trying
different combinations until we got one that worked well," Professor
Lichtman explains. "Of course, subject matter knowledge and political
theory had a lot to do with selecting variables &emdash; it wasn't
just random trial and error. But it wasn't done by the computer."
The whole idea started with a chance encounter. In 1980, based on his work on 19th-century voting blocs, Dr. Lichtman won a one-year research fellowship to Cal Tech. At the welcoming dinner for new fellows, he happened to be seated next to Russian seismologist Keilis-Borok, another of that year's visitors. It was Keilis-Borok who insisted on the collaboration, claiming, "We do the same thing! We both look at patterns of many variables to see whether there will be stability or upheaval. We MUST work together."
Although skeptical, Professor Lichtman agreed, and the two published their conclusions in the Proceedings of the National Academy of Sciences, in 1981.
The original study had 12 variables, not 13, and picked one election (1912) wrong, with another (1908) too close to call. Additional experimentation changed a couple of political keys; the foreign policy and military keys (10 and 11) were the last ones added. The method now calls correctly every election since 1860, and &emdash; more important &emdash; called the 1984, 1988 and 1992 elections correctly several months in advance.
"1984 was easy," Lichtman concedes, "but calling the election of George Bush in early 1988, when Bush still trailed Michael Dukakis by 20 percentage points in the polls, looks like a real accomplishment to me."
Even the term "keys" has a story behind it. Lichtman and Keilis-Borok referred to their predictors only as "conditions." It was a newspaper science reporter who picked up the story from their journal article who first referred to the predictors as "keys."
The following year, 1982, Professor Lichtman began a collaboration with Ken deCell, a senior editor of Washingtonian magazine, which led to several joint articles and, in 1990, a book. In 1992, however, they parted company, as deCell called the election for Bush in July, while Lichtman insisted that Ross Perot would re-enter the race and tip the sixth and decisive key to Clinton.
"Despite our efforts to pick keys that are clearly defined, interpreting the keys can take considerable judgment," he explains. "The third-party key turns if the candidate gets five percent or more of the vote. With the money and publicity he had, Perot was a good bet to do that if he came back in. I told Ken it was just too early to make a call, but he jumped &emdash; and we argued."
Lichtman adds, "I guess I ought to be happy that you have to know
some political science and history to interpret the keys. If anybody
can do it, no one has to ask me."
The "Thirteen Keys" theory has several important implications about the presidential election process:
"It was not the 'selling' of Nixon that propelled him into office in 1968," Lichtman asserts, "it was the cumulative failures of Lyndon Johnson's second term. For all the public-relations expertise he assembled, Nixon couldn't sell his first-term domestic policy initiatives as changes comparable to the New Deal or the Great Society. His administration, which appeared headed for defeat at the time, won a second term only through such genuine election-year accomplishments as his breakthroughs in relations with China and Russia and the igniting of an economic boom &emdash; events that Americans would have perceived as significant irrespective of their packaging."
He is quick to add, however, that this conclusion applies only to the range of campaign activities that have actually occurred. Every campaign raises money, mobilizes party organizations, deploys volunteers and advertises widely in the communications media. "If one party decided not to show up for the campaign," he points out, "the nature of presidential campaigns would be changed, and there is no guarantee that the verdict of the keys would still apply."
What does seem to follow is that the exhausting schedules of personal appearances and the hordes of consultants, strategists, advertising people and "spin doctors" make less difference than they would like us to believe.
"I'm surprised to hear that the Clinton people are pushing to include Perot in the debates," Lichtman says in response to a question about a Washington Post story which appeared the first week in September. "That's definitely not what I would have advised them to do."
He adds, "The effect negative campaigns do have is to turn people against politicians in general and to make it more difficult for the president to get bipartisan support for policy initiatives. Bush's 1988 tactics hurt his ability to perform effectively in office, which helped to defeat him in 1992."
"Barring a foreign-policy disaster or dramatic new revelations of scandal, Clinton's re-election is assured," Professor Lichtman claims. "And I don't see either as likely to happen in the couple of months we have left."
You can't say he's being evasive. In a few weeks we will know
whether he was right.
1. Lichtman, Allan J., "The Keys to the White House, 1996," Lanham, Maryland: Madison Books, 1996.
2. Lichtman, Allan J., and deCell, Kenneth, "The Thirteen Keys to the Presidency," Lanham, Maryland: Madison Books, 1990.
3. Lichtman, A. J., and Keilis-Borok, V. I., "Pattern Recognition
Applied to Presidential Elections in the United States, 1860-1980:
Role of Integral Social, Economic and Political Traits,"
Proceedings of the National Academy of Science, Vol. 78, No.
11, November, 1981, pp. 7230-7234.
Doug Samuelson, a frequent contributor to OR/MS Today, is president of InfoLogix, Inc., a consulting company in Annandale, Va., and current president (1996-97) of WORMSC. He is also a political pro himself, having worked as a campaign staffer in a U. S. Senate campaign in Nevada in 1970 and as a county coordinator in a gubernatorial campaign in California in 1974.
For more information, put the number 2 in the
appropriate space on the
Reader Service Form
OR/MS Today copyright © 1996 by the Institute for Operations Research and the Management Sciences. All rights reserved.