How to (accurately) predict a presidential election


ALISON STEWART, PBS NEWSHOUR WEEKEND ANCHOR: In all nine presidential elections since 1984, American University history professor Allan Lichtman has correctly predicted the outcome not by using opinions polls but by using a system he helped create, 13 simple true/false statements testing whether the incumbent party will retain or lose the White House.

Professor Lichtman joins me now via Skype from Doha, Qatar. Professor, how did you arrive at these 13 true/false statements?  And what is the rule that determines the outcome?

ALLAN LICHTMAN, AMERICAN UNIVERSITY HISTORY PROFESSOR: I came across the keys to the White House totally by accident.  In 1981, I met the world's leading authority in earthquake prediction, Vladimir Keilis-Borok.  And it was Keilis-Borok who suggested we collaborate using his mathematical modeling to predict American presidential elections.

So, we studied every American election from 1860 to 1980.  This was in 1981, guided by the thesis that presidential elections are primarily judgments on the performance and strength of the party holding the White House.  And from that study, we came up with 13 simple, true/false questions, where an answer of "true" always favors the re-election of the White House party.  And we came up with a really simple decision rule: if six or more of the keys are false, that bodes defeat for the party holding the White House.

ALISON STEWART: Well, let's look at those first four rules, those keys, because they're basically about political climate, as you said, something that you can, you know, objectively take a look at.  They were false for the incumbent party, for the Democratic Party.

So, what were the other two that sent it over the edge for the Republicans?

ALLAN LICHTMAN: Critically, the party holding the White House did not achieve major policy change in the second Obama term.  So, they didn't have a big domestic accomplishment to run on.  In addition, they didn't achieve a big splashy foreign policy success.

ALISON STEWART: Your model takes into account the generic Republican candidate and the generic Democratic candidate, but we really didn't have a generic Republican candidate this time around.  So, why do you think it works still?

ALLAN LICHTMAN: Well, the force of history is very powerful.  This is a very robust model because it goes retrospectively back to 1860 and prospectively ahead to the present.  So, it takes into account enormous changes in our politics.

ALISON STEWART: You've made a point of saying, "polls are not predictions."  Everyone this week has been talking about how wrong the polls were.  Can you explain that statement?

ALLAN LICHTMAN: First of all, polls are snapshots.  They give you sentiment at a particular point in time, and it does not necessarily follow that that's going to hold at a future point in time.  In addition, the polls are entirely dependent on predicting who is going to be a likely voter, and they really don't know very well who the likely voters are.

ALISON STEWART: Do your keys, or your factors, your statements, take into account October surprises at all or emotions or passions?

ALLAN LICHTMAN: I initially made my prediction for a Donald Trump victory in late September before the women coming out alleging sexual assault by Donald Trump, before the Comey letter and the Comey retraction.  And I doubled down on that prediction on October 28.  So, my predictions were not turning on these campaign events.

ALISON STEWART: Professor Allan Lichtman, thanks so much for sharing all your information.

ALLAN LICHTMAN: My pleasure.

Recently in Politics