A President’s Approval and Simple Math
In my American Government classes, there is a lecture on Public Opinion, in which I briefly address how the U.S. Presidents’ job approval ratings generally decline over the course of their time in office. This consistently downward trend can be shown with the following table compiled from Gallup polls (www.gallup.com/poll/124922/presidential-approval-center.aspx).
President Starting approval rating Ending approval rating
Truman 87% 32%
Eisenhower 68 59
Kennedy 72 58
Johnson 78 49
Nixon 60 24
Ford 66 53
Carter 66 34
Reagan 51 63
Bush I 63 56
Clinton 51 65
Bush II 57 34
Obama 65 59
Trump 45 34
Of all the Presidents since Truman, only Reagan and Clinton finished their terms with job approval ratings higher than when they started (despite dealing with second-term scandals). One reason for their deviance from the norm is that they both started at a significantly lower rating than the average starting rate. Otherwise we can speculate that they had generally “likeable” personalities, or that Reagan benefitted from the softening Soviets, or that Clinton benefitted from the globalizing economy. But beyond these two outliers, the downward trend in approval ratings is remarkably consistent. It cannot simply be blamed on bad luck or timing.
The textbook explanation for this describes how Presidents begin their term with a heightened sense of optimism, particularly when they are replacing the opposition party’s administration that has gone stale. In addition, the President-elect has made many promises to various groups in the course of the campaign. These elevated expectations, combined with the realities of the office that leave many promises unfulfilled, inevitably results in the gradual decline in public support. But there is another consistent factor that must be considered, with the help of some simple math.
First we can look at the typical voter turnout for the Presidential elections. Since 1948, percent turnout of the voting age population (VAP) has ranged from 49% to 63%, but until 2020 had not been close to 60% since 1968 (International Institute for Democracy and Electoral Assistance). The average turnout over this period had been 55%, and was about 56% in the 2016 election. The 2020 election saw a significant spike in VAP turnout, to about 61.5%, due to several factors that could be discussed later.
Now let’s look at the percent of the popular vote that the winning candidate has obtained in each of these elections (Roper Center, Cornell University):
Year Winner’s % of popular vote Winner’s Electoral votes
1948 49.6% 303 of 531
1952 54.9 442 “
1956 57.4 457 “
1960 49.7 303 of 537
1964 61.1 486 of 538 (270 to win)
1968 43.4 301 "
1972 60.2 520 "
1976 50 297 "
1980 50.5 489 "
1984 58.5 525 "
1988 53.1 426 "
1992 42.9 370 "
1996 49.2 379 "
2000 47.9 271 "
2004 50.7 286 "
2008 52.9 365 "
2012 51.1 332 "
2016 46.1 304 "
2020 51.3 306 "
Note that about a third of the winners received less than a majority of the popular vote. This is primarily due to the presence of “3rd party” candidates, as well as the Electoral College process, which will be discussed in a future post. The average winning percentage over this period was 52%. But let’s be generous here and note that 55% would be a convincing victory; literally a blowout by Electoral vote counts.
Therefore, even an “easily” elected President is directly supported by 55% of 55% of the voters, based on our averages. And according to the math: 55% of 55% is just 30%! What this means is that at any given time about 70% of the adult population did not vote for the person who serves as President. A large majority of citizens can therefore deny any ownership to our leadership, and true to our nature, be quick to lavish our derision and dissatisfaction. Apathy appears to be a close relative to negativity. And partisan loyalty among non-voters can be just as loud as that of voters.
The polling industry has been facing some challenges in recent years: response rates have fallen dramatically due to the increase in cell phones and decrease in willingness to do interviews. It would be wise to pay attention to the confidence interval (or margin of error) as well. But polls and surveys remain useful, providing a fuzzy snapshot of what people are thinking about at the moment. With this exercise we can compare polls with some real numbers of turnout and vote count, using a little math to help us understand what is happening with human behavior.
RKL for Maximatician.com