Washington Post distributes fake running news

I hate to beat an already battered meme into further unrecognizability, especially since the Washington Post has done some yeoman (pronounced "Yo, man!") work recently. And on the whole, it's unfair to refer to the column that triggered this post as nonsense. It does, however, feature some readily identified veracity issues.

The piece is by 1968 Boston Marathon winner and former Runner's World editor-in-chief Ambrose Burfoot, who has the sort of name I would love to see more often. Amby explores the reasons for the progressive slowing of average finishing times in U.S. marathons, operating gamely on the shaky premise that the reasons for this are shrouded in cultural or mathematical mysteries.

A new report, which calls itself the largest-yet analysis of U.S. road race results, has concluded that “American runners have never been slower.”

The report, based on more than 34 million individual race results from 1996 to 2016 and published on the Dutch website RunRepeat.com, focuses on marathon results. It finds that the average American marathon time has slowed from about 4:15 to about 4:40 over the 20-year period.

Amazingly, nowhere in this column does Amby report that between 1996 and 2016, the number of finishers of American marathons increased by about 67 percent, from around 300,000 to over 500,000. This obviously implies, or at strongly suggests, that the demographics of the category "marathon runner" have shifted significantly in the past 20 years.

Intuition is clearly not always a reliable guide to experiencing the world, but let's put it to use here: If you were told that marathon running has moved away from being a domain of people focused chiefly on running fast and toward a domain of people focused chiefly on finishing, which would you expect -- faster median times or slower median times?

Something else important to note about the worsening times for both men and women: That trend actually ended back in 2005, with 2016 being the slowest for both sexes in 11 years. (The Competitor article unfortunately uses the nonsense term "average median," so it's not clear whether 4:22:07 and 4:47:40 represent average times or median times, which aren't quite the same thing.) So it's not fair to start yammering about marathoners getting slower on the basis of what may be just a statistical burp (even if it's clearly still valid to note that the typical finisher is a lot slower today than he or she was in the mid-1990s).

American marathoning has changed dramatically over the past three or four decades, thanks in large part to the women’s running boom. In 1980, Running USA estimates that only 10 percent of marathon finishers were women. Last year, that figure reached 44 percent. Because women are, on average, 10 percent slower than men, more women participants will necessarily slow the average times. (Most marathon experts have tied slower marathon times to the increase in women and in less-serious racers.)

The RunRepeat report, however, says that slowing men have contributed more to the decline (54 percent) than increased participation by women (46 percent).

It also asserts that casual, back-of-the-pack runners have slowed only slightly more than those in the front of the pack.

Ken Young, dean of worldwide road race statisticians, says those casual racers share more of the blame than the RunRepeat report shows...

This passage introduces a misleading idea. It suggests, however passively, that the effects on average marathon times of front-of-the-pack runners slowing down are as pronounced as the effects of midpackers and slower runners getting slower. This is clearly not the case, as there are simply so many more slower and average runners than there are fast ones. If 2,000 people in a town of 5,000 are within 10 percent of the median income of $40,000, and all 2,000 suffer a 10 percent pay cut, this will hurt the town GDP more than its ten $1 million-a-year-earners moving away altogether will. In fact, it will hurt it more than those same ten wealthy people getting a 10 percent raise will.

Which introduces the next misleading bits.

While average times are getting slower, American elites have continued to improve. In 1996, the American marathon records stood at 2:10:04 (men) and 2:21:21 (women). Today they are 2:05:38 and 2:19:36.

While it may be accurate in some senses to say that elite American marathoners have improved over the past two decades -- in absolute terms, if not necessarily in relation to the rest of the world -- this passage does nothing to establish such a fact. Joan Benoit Samuelson set the American record of 2:21:21 in Chicago in October 1985, when only Norwegian Ingrid Kristiansen's 2:21:06 from six months earlier was faster. Since then, only Deena Kastor has beaten Joanie's time (2:21:16 in 2003 and 2:19:36 in 2006, both in London). The women's AR has therefore dropped by one minute, 45 seconds in almost 32 years.

The men's AR is held by Khalid Khannouchi. This isn't to spark another "real American" argument, but Khannouchi is a Moroccan native who ran 2:05:42 before becoming a U.S. citizen after years ol living in New York, so it's safe to say he is not entirely representative of the runners who setting AR's back in the day. In any event, Pat Petersen ran 2:10:04 in 1989; Jerry Lawson matched that time in 1996 and broke it a year later with a 2:09:35, and David Morris ran 2:09:32 in 1999. Khannouchi's mark has stood since 2002.

The preceding two paragraphs suggest that referring to American records in the marathon is a lazy-at-best way to establish that fast Americans are getting faster. The women's mark is over 11 years old and then men's is approaching 15.

Then there's this mess:

Many of the best Americans enter the Boston Marathon each April. Due to Boston’s tough qualifying standards, rare among big marathons, only about 10 percent of all U.S. marathoners run times fast enough to enter Boston. Yet it has also gotten slower, on average, over the past two decades. In 2000, according to MarathonGuide.com, the average finish time at Boston was 3:41:39. Three months ago, in the 2017 Boston Marathon, the average finisher crossed the line in 3:58:03.

It is brazenly irresponsible at this point to not mention two key facts: The Boston Marathon, with the help of logistical improvements, has greatly expanded its entry fields in the past two decades; and the number of non-qualified (mostly charity-bib) runners has reached about 20 percent of total entries. (This doesn't even get into how slow the conditions were in 2017 compared to 2000 -- I was standing at the 23-mile mark in both years.)

The column goes on to observe that although obesity in the U.S. has risen in concert with slower marathon times, the former is most likely not responsible for the latter. This seems obvious enough; the population of people who are running marathons doesn't overlap greatly with the set of Americans who are obese (at least at the time they record their marathon finishes). This is rather like observing that even though the number of Americans who fluently speak Spanish continues to rise, my own facility with the language hasn't gotten much better over time.

Amby quotes a few people who mention something else that is obvious enough to most runners -- not everyone's in it or the kill these days. And that's find, but it has drastically altered the complexion of the road-race landscape. When I started running as a teenager in the mid-1980s, no one cared about anything other than an accurately marked course with water on hot days, and maybe a T-shirt for those who didn't already have about 50 of them from past events. The whole phenomenon of live bands on the course, massages at the finish, lavish post-race spreads instead of just bagels and bananas, and other niceties that pack runners tend to demand (the sort of stuff that drives up entry fees, of course) was years away.

Anyway, at least my buddy Matt Fitzgerald, tireless author and scion, gets a mention. It's interesting how many levels of running on which we have intersected over the years, but I think I will save that for a post that is more generally positive in its slant, as Matt is a very positive fellow.