How often do voters choose the best candidate?

To answer that question, I looked at 16 different U.S. elections -- those in which two or more of the major candidates sooner or later served as president. With both candidates' performance as president a matter of public record, we can attempt to judge whether or not the voters made the "right" choice.

As a measure of the relative merit of the different presidents, I used the data collected in the Wikipedia article Historical rankings of presidents of the United States, which tabulates the results of 20 different surveys in which historians ranked the past presidents of the U.S. (There are fewer than 20 data points for some of the more recent presidents, since the earliest surveys were conducted in the mid-20th century; for each president, I also excluded as premature any results obtained before he had completed his full term in office.)

I was surprised at how little controversy there is about the relative ranking of most presidents. I looked at 12 sets of presidents (11 pairs and 1 threesome); for 7 out of the 12, the surveys were in 100% agreement as to which president in the set was the best, and for most of the others the rate of agreement was also very high. The only pair regarding whom there was no consensus was Carter and Ford, each of whom was ranked higher by exactly half of the surveys in the data set.

Here are the relevant data. The numbers in parentheses represent what percentage of the popular vote each candidate won. The name of the candidate judged by historians to have made a better president is in boldface. An asterisk after a candidate's name indicates that he had already served as president prior to the election in question and was thus a known quantity.

• 1796: J. Adams (53.4%), Jefferson (46.6%)  -- 100% rank Jefferson higher
• 1800: Jefferson (61.4%) , J. Adams* (38.6%) -- 100% rank Jefferson higher
• 1824: Jackson (41.4%), J. Q. Adams (30.9%) -- 100% rank Jackson higher
• 1828: Jackson (56.0%), J. Q. Adams* (43.6%)  -- 100% rank Jackson higher
• 1836: Van Buren (50.8%), W. H. Harrison (36.6%) -- 100% rank Van Buren higher
• 1840: W. H. Harrison (52.9%), Van Buren* (46.8%) -- 100% rank Van Buren higher
• 1848: Taylor (47.3%), Van Buren* (10.13%) -- 85% rank Van Buren higher
• 1856: Buchanan (45.3%), Fillmore* (21.5%) -- 100% rank Fillmore higher
• 1888: Cleveland* (48.6%), B. Harrison (47.8%)-- 100% rank Cleveland higher
• 1892: Cleveland* (46.0%), B. Harrison* (43.0%) -- 100% rank Cleveland higher
• 1912: Wilson (41.8%), T. Roosevelt* (27.4%), Taft* (23.2%) -- 90% rank Roosevelt highest, 10% Wilson, 0% Taft
• 1932: F. D. Roosevelt (57.4%), Hoover* (39.7%) -- 100% rank Roosevelt higher
• 1960: Kennedy (49.72%), Nixon (49.55%) -- 100% rank Kennedy higher
• 1976: Carter (50.1%), Ford* (48.0%) -- no consensus
• 1980: Reagan (50.7%), Carter* (41.0%) -- 93% rank Reagan higher
• 1992: Clinton (43.0%), G. H. W. Bush* (37.4%) -- 89% rank Clinton higher

Excluding the 1976 election from consideration, we can see that the populace chose the "right" candidate two-thirds of the time (i.e., in 10 out of the 15 elections being considered). This is considerably higher than a democracy skeptic like myself would have predicted.