A Dead Heat? Maybe. But Probably Not A Tie

Annoying overuse of "dead heat" in reports on election campaigns

Although "dead heat" is an accepted term in statistics, reporters and editors, often less than quantitatively gifted, are tossing it into headlines as a synonym for "tie" -- which it is not!

Yes, saying Mr. Red and Ms Blue are "in a dead heat" suggests an exciting athletic race. (Think "neck and neck" and "qualifying heat.") But it rarely means what it actually means: that polls show Red and Blue separated by less than the margin of error. Indeed, if the margin of error is 4%, Blue could be leading Red by 3 percentage points.

Experts have been explaining that for years. (See the accompanying box, "Experts scorn media use of "dead heat.")

The New York Times rarely uses the term "dead heat" in reporting on candidates for election. It was used -- properly -- in one instance by the paper's then statistical superstar Nate Silver to describe an early October "statistical dead heat" between 2012 presidential candidates Barack Obama and Mitt Romney. Reading Silver's analysis of multiple polls to reach that conclusion is instructive. Silver is careful to point out that the presidential race was not tied.

Annoyance with the off-the-mark use of "dead heat" prompted the launch of this website in September 2014. But there is a serious consideration too. If you look at the news stories scrolling on this page, you are likely to find one or more that rely on a single, internal campaign poll to proclaim the race a "dead heat."

Kudos to the Providence Journal, which on September 15th called out a candidate for Rhode Island governor for using a mostly undisclosed internal poll to declare a dead heat. "Fung poll: Race is 'dead-heat,' but Fung better-liked; Take our poll," said the paper's headline.

The sports motif also seems incompatible with any meaningful discussion of policy. Open a few of the scrolling reports and see if they contain any mention of candidates' positions on the issues. With luck, you may learn about the candidate's fundraising and spending.

It's worth considering that some of the reports on "dead heat" electoral races are from television and radio stations which are most likely cashing in on those incessant campaign ads (yes, they're much more annoying than the term "dead heat"). Could it be that they have a motive for portraying electoral campaigns as exciting contests rather than serious undertakings with consequential impact on our lives?

Experts scorn media use of "dead heat"

In a paper for the 2001 session of the International Statistical Institute, Prof. Douglas Lonnstrom detailed four errors made by "the media" in using the term "statistical dead heat." According to Lonnstrom: "There certainly can be a tie if each candidate has the same percentage, i.e. 45% to 45%. In any other case the candidate with the higher vote is more likely to be ahead, even if the difference is only one percentage point."

In his February 2012 Junk Charts blog posting entitled "A statistical dead heat is not a virtual tie," Kaiser Fung notes that journalists "frequently abuse language as they try to explain what the polls mean." In particular, he writes, they "frequently ignore the margin of error," even though in polls with small samples, these margins could be five percentage points or more.

A 2006 blog post, "Statistical dead heat is a dumb term," explains (somewhat technically) how, with even a small difference in percentages and a large margin of error, the probability is that the candidate with the larger percentage is leading. The media's use of "statistical dead heat," wrote author cyberianpan, "appears to show a fundamental misunderstanding of probability."

"Dead heat" in the news

    Definitions and Meanings

    Two dictionaries define dead heat as a tie of the althetic variety. Merriam-Webster defines it as "a contest in which two or more competitors earn the same score or finish at the same time."

    The Oxford Dictionary of American English defines dead heat as "A situation in or result of a race in which two or more competitors are exactly even." At the end of a list of examples the Oxford dictionary gives this political example: "Coming out of the Democratic Convention, the race was a dead heat." It does not directly state that this race was also a tie, but that conclusion would not be illogical.

    According to the Oxford Dictionary, the earliest use of dead heat was in 1796 by Sporting Magazine: "The whole race was run head and head, terminating in a dead heat."

    There is also a 1988 movie entitled Dead Heat -- about an LAPD officer killed by reaniminated zombies.

    Wikipedia makes no mention of statistical dead heats, but lists a number of books and other works entitled Dead Heat