Trump’s victory was not a landslide

In the 2016 U.S. Presidential election, Donald Trump defeated Hillary Clinton by a projected electoral vote margin of 306 to 232. When the Electoral College voted on December 19, the final tally was 304 to 227. Given the widespread predictions of a Clinton victory, it is safe to call Trump’s win an upset, and the final electoral margin was not close. However, does this make it a landslide?

In late November, after Michigan became the last state to certify its election results, Trump and his campaign manager, Kellyanne Conway, were widely criticized for tweets1 in which they called Trump’s victory a landslide:trump-conway-tweets

The most obvious way to start evaluating the landslide claim is to compare Trump’s share of the electoral vote with those in previous Presidential elections. At least two well-known websites, FactCheck and FiveThirtyEight, took this approach and reached the same conclusion: Trump’s percentage of the electoral vote (56.9%)2 is on the low side by historical standards. FiveThirtyEight took the slightly better approach of excluding elections before 1804 because those elections occurred before the Twelfth Amendment and therefore did not distinguish between Presidential and Vice Presidential electoral votes.

However, such a comparative approach is only the first step because it does not address how the word “landslide” has been used.  It is at least theoretically possible that many or most Presidential elections could be lopsided enough to qualify as landslides, and this is arguably true in practice.  For example, if we consider the seemingly conservative benchmark that a winning percentage of 70% or more is a landslide, and if we limit our discussion to two-candidate races,3 then a bare majority of U.S. Presidential elections since the Twelfth Amendment—24 out of 47, or 51%–would still be considered landslides.

Dictionary definitions4 are too vague to be applied numerically, and there is no standard definition in political science. We therefore chose six major news sources, reflecting an ideological range from left-leaning to right-leaning, and reviewed their coverage of all Presidential elections in the past 100 years to see which were described as landslides.

Specifically, we searched for the keyword “landslide” in post-election coverage in the archives of the New York Times, the Washington Post, Time magazine, the Christian Science Monitor, the Chicago Tribune, and the Dallas Morning News. We then reviewed each occurrence to determine whether a given Presidential election was described as a landslide. We counted only occurrences when this appeared in a headline or the narrative of an article. We did not include letters to the editor; quotations, because these were often made by candidates’ surrogates or other interested parties; predictions made before full results were available; or descriptions that referred to only part of a candidate’s win as a landslide, such as having won a landslide in a given region or among a given demographic. If the news source had even one instance of describing a given election as a landslide, we counted that as a Yes; otherwise, it was a No. We then tabulated these, sorted in decreasing order of the winner’s vote percentage, to see if a pattern emerged.5landslide-yes-no-tableThis shows the expected trend: in general, the higher the vote percentage, the more likely it is to be described as a landslide. There is no clear-cut boundary, but the transition appears to occur around the mid-to-upper 60s.  In the area that we would expect to be solid green, three years (1940, 1944, and 1988) have exceptions, but these make sense when we compare them against preceding elections: each was a lopsided victory but less lopsided than the two elections immediately preceding it, so observers declining to call them landslides may have become accustomed to previous years’ margins.

It would have been unreasonable to expect an ironclad numerical threshold to hold across different news outlets in the span of a century—or even more locally. For example, readers of the Christian Science Monitor on November 5, 1992, might have been puzzled to see these conflicting descriptions in the same edition:

  • “The Republican era launched 12 years ago by Ronald Reagan ended abruptly with a Democratic electoral landslide as America’s voters gave Gov. Bill Clinton a decisive coast-to-coast victory over President Bush.”
  • “It was a big win for the Democrats, but hardly a landslide.  Millions stayed with the known Republican philosophy of George Bush.  Ross Perot’s feisty, independent candidacy garnered nearly 20 percent of the vote.”

Even the same writer might use the term inconsistently over time: Robert Novak of the Chicago Sun-Times6 described Obama’s 2008 win as “the first Democratic Electoral College landslide in decades,” despite having written twelve years earlier that Clinton “won an Electoral College landslide” in 1996.

For Trump’s margin of victory, the trend in the table points to an answer of No. (The fact that none of these six sources called his win a landslide is secondary; we care about the historical trend.)

In light of this, is there any alternative sense in which Trump’s victory could be considered a landslide? Given that he was widely expected to lose but instead won by a comfortable margin, exceeding expectations so dramatically is likely to be the reason that Trump and Conway perceived it as a landslide.  Indeed, the Washington Post later reported that Conway sent them an email confirming that this was the sense in which she meant it: “In an email, Conway said her tweet was referring to how Trump did ‘compared to what was predicted.’”  While psychologically understandable, this sense of beating expectations is more conventionally conveyed by “upset” than “landslide.”

What about the two outlets that called Harry Truman’s 1948 win a landslide?  Truman’s share of the electoral vote was nearly identical to Trump’s, and Truman was likewise an underdog who upset pollsters’ expectations by winning by a comfortable margin.  The relative sense of exceeding expectations is likely to be a contributing factor for why his win was sometimes called a landslide. However, another factor in Truman’s case was that electoral votes were split among three candidates (Truman, Thomas Dewey, and Strom Thurmond), so Truman (56.6%) was far ahead of second-place Dewey (35.3%).

Of course, in a race with three or more major candidates, even a minority share could be considered a landslide. As a non-Presidential example, the wide-open race to replace California’s governor in 2003 saw Arnold Schwarzenegger win with 49% of the popular vote, which was described as a landslide in outlets such as Time magazine and the Ocean County Register. However, this is not relevant to electoral votes in the 2016 Presidential race because no third-party candidate won, or came close to winning, any electoral votes.

Admittedly, beyond Presidential general elections, there are recent instances of two-way races in which smaller winning percentages than Trump’s have been called landslides.  For example, Grace Wyler, writing for Time magazine in 2013, described Scott Walker’s win in the 2012 Wisconsin governor recall election as a landslide victory, although Walker defeated Tom Barrett by a margin of only 53.1% to 46.3%.  Likewise, H. A. Goodman, writing for the Huffington Post blog in 2016, said Bernie Sanders won a landslide over Clinton among women voters in the New Hampshire primary, although Goodman’s article cited Sanders as having lead of 53% to 46% among New Hampshire women voters. Both of these usages were likely influenced by the consideration noted earlier, namely the dramatic upset of expectations.  However, such usage appeared to be uncommon, though a thorough review of media coverage of non-Presidential elections would be impractically time-consuming.

This analysis, like those of FactCheck and FiveThirtyEight, has focused solely on electoral votes in Presidential elections because those determine the winner.  It would not be reasonable to define the threshold for a Presidential landslide in terms of the popular vote because the popular vote plays no direct role in the outcome; moreover, it is theoretically possible for a candidate to win 537 out of 538 electoral votes without having a majority, or even a plurality, of the popular vote, yet surely such a lopsided result would be called a landslide. Five of the six news outlets described Bill Clinton’s 1992 victory as a landslide, though his popular plurality of 43.3% was the lowest popular vote by a winning candidate since 1912. Nonetheless, the fact that Trump did not receive a plurality of the popular vote seems likely to influence perception and usage, even if it has no more practical bearing on the outcome than the similarly perception-influencing fact that Trump’s electoral victory exceeded expectations by a large margin.

Conclusions:

1. Given his electoral vote percentage as compared to those of previous Presidential victors, and given the news media’s use and non-use of the word “landslide” for Presidential elections over the past century, Trump’s victory should not be described as a landslide.
2. There is no clear-cut threshold for the electoral vote percentage above which a Presidential victory is considered a landslide, but a rough estimate is that the transition occurs in the mid-to-upper 60s.
3. Victories are more likely to be described as landslides if they are upsets, and they are less likely to be described as landslides if immediately preceding elections had even larger margins of victory.

Footnotes:

1. Trump made a similar statement on Fox News Sunday on December 11, 2016:  “We had a massive landslide victory, as you know, in the Electoral College.”  As one would expect, the most vocal criticism came from left-of-center news outlets such as CNN, the New York Times, the Los Angeles Times, and PBS NewsHour, but even the National Review agreed Trump’s victory was not a landslide.

2. His share of the projected electoral vote was 56.9%, namely 306 divided by 538. This was the projected count at the time of Trump’s and Conway’s tweets because the Electoral College did not meet until a few weeks later, so this is the most appropriate figure for evaluating the validity of their landslide claim. If the number is updated by deducting his two faithless electors, his percentage of the Electoral College drops to 56.5% (304/538). Of course, his losses were not Clinton’s gains because, once faithless electors were factored in, her percentage of the electoral vote dropped even more, from 43.1% (232/538) to 42.2% (227/538). If faithless electors are disregarded for both candidates in order to re-simulate a two-candidate race, the percentages become 57.3% (304/531) for Trump and 42.7% (227/531) for Clinton. In practice, neither 56.5% nor 57.3% is different enough from 56.9% to affect whether Trump’s win should be considered a landslide, although the 57.3% figure would place Trump one notch higher in the historical standings, rising above Truman’s 57.1% in 1948.

3. It seems like common sense that the threshold for a landslide would be lower in a multi-way race than in a two-way race, and later we will cite examples of this. There have been seven Presidential elections in which three or more candidates received electoral votes: 1824, 1856, 1860, 1892, 1912, 1948, and 1968.

4. For example, Webster’s Ninth New Collegiate Dictionary (1985) defines it as “a great majority of votes for one side” and “an overwhelming victory,” while the online Oxford English Dictionary (accessed 2017) defines it as “a sweeping electoral victory.”  None of this helps us establish a quantitative threshold.  Interestingly, the OED’s earliest citation for “landslide” in the political sense is from the November 4, 1888, edition of the New York Times, referring to “a veritable landslide in Mr. Hewitt’s favor.”  However, retrieving that article in full from the New York Times archive, we found the OED’s citation did not refer to an electoral victory at all, but rather to a groundswell of endorsements for Abram Hewitt for his reelection bid as mayor of New York; two days later, he placed third in the election itself, winning only 26% of the vote.

5. The “N/A” in the table reflects that Time magazine was founded in 1923, so it had no coverage of the 1912 election. However, we marked it “Yes” for the 1920 election because it referenced that as a landslide during its subsequent coverage of the 1924 election. Unlike FactCheck and FiveThirtyEight, our winning percentages for past years did not deduct any subsequent faithless electors; Trump and Conway made their statements before 2016’s faithless electors were known, and the news articles covering past elections were likewise written before any faithless electors came into play, so ignoring faithless electors across the board makes the more appropriate comparison. In practice, this difference is immaterial because only one past winner’s percentage changed when faithless electors were ignored: Nixon in 1972 rose from 96.7% to 96.8%.

6. After some cursory research, the Sun-Times had to be removed from the survey group when it was discovered that results before 1988 could not be retrieved.  Unfortunately, for similar reasons, no West Coast newspaper archive was accessible enough for the survey; however, for this survey, geographical diversity is less relevant than ideological diversity.

Sources:

Presidential vote counts were taken from the 2017 World Almanac, pages 10 and 533.  Portions of the periodical archives are paywalled, but readers without paid access can reproduce most of the findings from the following public websites:

New York Times:  http://www.nytimes.com/ref/membercenter/nytarchive.html

Washington Post:  http://pqasb.pqarchiver.com/washingtonpost/search.html

Time:  http://content.time.com/time/magazine/archives/

Christian Science Monitor:  http://pqasb.pqarchiver.com/csmonitor_historic/advancedsearch.html (until 1980); http://www.csmonitor.com/content/search (since 1980)

Chicago Tribune:  http://pqasb.pqarchiver.com/chicagotribune/advancedsearch.html

Dallas Morning News:  http://ece.dallasnews.com/archive/

© 2017 by Quiznox.com

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s