Out-Liar: What Malcolm Gladwell Gets Wrong About the Relative Age Effect in Pro Hockey

In trying to bring more coverage to Bloody Elbow regarding sports and science, I thought it would be fitting to get interesting contributors to explore the relationship between the two. And I can think of none more interesting than Jon Levey who was kind enough to allow me to post his article addressing the relative age effect. A brief paragraph from the original piece was published in Sports Illustrated in its August issue as part of a small 'Mythbusters' series. Below is the article in its entirety. Jon Levey can be found on twitter @72unforced. You can find more of his work here at his website.

After lording over some of the most heralded teams in college basketball history, the great John Wooden concluded: "I'd rather have a lot of talent and a little experience than a lot of experience and a little talent." The debate over whether great athletes are born or made has become fertile ground for research and speculation. Lately, there has been increasing support given to the school of thought that what we see as a unique talent on the playing field is actually a product of circumstance. In his recent best-seller, Outliers, native son Malcolm Gladwell claims that if a Canadian wants a chance to skate for Lord Stanley's Cup it's a huge help to be old. Not Martin Brodeur old, but old relative to his peers.

It's what's known as the relative-age effect, the results of which suggest that being born in a particular month can greatly influence an athlete's prospects for long-term success; circumstance matters just as much, if not more, than natural talent. In Gladwell's analysis it leads to what he calls an "iron law of Canadian hockey: in any elite group of hockey players - the very best of the best - 40 percent of the players will have been born between January and March, 30 percent between April and June, 20 percent between July and September, and 10 percent between October and December." It's a compelling theory as to why some hockey players become stars, and others don't. It's also not true.

In Canada the eligibility cutoff for age-class hockey is January 1. Someone born in February therefore has an extra 10 months of growth over a December baby. While insignificant to an adult, to a 9-year old hockey player, that could mean an additional sweater size. The relatively older players are bigger, stronger, more coordinated, and thus get more pats on the back. Accurate or not, these kids merit the "gifted" label. They play in more games in better leagues while receiving superior coaching. It's what sociologists commonly refer to as accumulative advantage; the initial gap in perceived ability gets widened even further.

On the other hand, relatively younger players develop slower, get relegated to lesser-skilled teams, grow frustrated, and perhaps prematurely hang up their skates. Only obvious talents - August baby Sydney Crosby - can overcome the disadvantages of their birth months. Gladwell points to this uneven playing field as the reason why so many relatively older players have achieved success in professional hockey. Reorganizing this shortsighted approach to development could lead to more kids like Syd.

Except for this problem: Being relatively young in Canada is not a death sentence for a professional hockey career. Far from it. In fact, "younger" players cash fatter paychecks and more of them make the country's Olympic team. For a mite in Moose Jaw with a Christmas birthday, those are some pretty good job prospects.

One of the first to research relative age in Canadian hockey players was Dr. Roger Barnsley. In the mid-1980s he studied players in the Western Hockey League (WHL) and found that if the calendar is divided into quarters, the first three months of the year will have given birth to the largest percentage of players, and the numbers will decrease with each subsequent quarter. Indeed the team Gladwell uses to illustrate the point, the 2007 Medicine Hat Tigers of the WHL, was heavily skewed with players born in the first quarter of the year.

Now here's where the relative-age effect gets more complicated. Barnsley's research was done primarily on 16-to-20 year-old prospects. While he hasn't monitored the numbers in many years, he's confident the effect still holds true in the WHL. If the Medicine Hat team is any indication, he's correct. But as you move up the food chain to more elite teams, the case for talent starts to rear its innate head. For example, take any Canadian Olympic team with professionals - the epitome of the very best of the best. Here's how the birthdates of players selected to those rosters break down by time of year:

Team # born Jan-March # born after March # born in the 2^nd half of year

2010 3 (13%) 20 13

2006 4 (14%) 24 12

2002 5 (22%) 18 10

1998 4 (17%) 19 12

Incidentally, the 1994 Olympic hockey team - the last to use prospects instead of pros - had a roster with 12 of 23 players (52%) born in the first quarter. Yet the only star player to come out of the group was Paul Kariya: an October baby. Barnsley admits that it's at the lower levels that the relative-age effect really bears fruit. "The classic early numbers - the 40, 30, 20, 10 - comes out of the developmental league," he says. "The NHL is not as predominant as that."

Not by a slapshot from the blue line. According to nhl.com, at the 2010 Olympic break there were 499 Canadians on NHL rosters. That's about 55% of the players in the entire league. If you broke their birthdates down by quarters of the year you get the following:

Canadians Non-Canadians (as of the end of the 09-10 season)

Jan-Mar: 25.7% 34.2%

Apr-June: 28.5% 23%

July-Sept: 25.5% 21.3%

Oct-Dec: 20.3% 21.5%

As you can see, if there's a country with an "old" hockey workforce, it's not Canada. There were actually more Canadian NHL players born in September (43) than January (41), and June was the most populous month (50). True, there are more players born in the first half of the year, but the notion that Canada is only producing successful players from a small portion of the calendar seems to be, at best, somewhat of an overstatement.

So what happened? If the predominance of Canadian prospects are relatively older, how come that's not represented on the country's most elite team (Olympic) or its highest professional level (NHL)? Let's put on our Gladwell hat and (blindly) speculate.

Perhaps the best place to start is back in the 1980s when Barnsley was conducting his research. During the 1984-85 NHL season, 512 Canadians occupied roster spots, essentially the same number as today. But back then, with nearly 240 fewer players, that number comprised 76% of the league. Here's what the birthdate breakdown looked like that season:

Canadians Non-Canadians

Jan-Mar: 30.3% 36.4%

Apr-June: 29.9% 23.5%

July-Sept: 22.2% 24.7%

Oct-Dec: 17.6% 15.4%

While they fall short of the 40-30-20-10 pattern, the birthdays that season definitely trended older than the current configuration. Why? One possibility is with more roster space open to Canadians, teams had to fill those spots with whatever talent was available. "Where you get the age effect is on the tail of the competency to make it in," says Barnsley. "What you're getting are players that would not have made it on talent alone but got all that other benefit of the age effect."

Fast-forward 15 years to the 1999-2000 season. Over that time the NHL underwent significant changes. Most notably it expanded by a third, from 21 to 28 teams. (We won't debate the questionable logic of putting franchises in cities like Nashville, Phoenix, and Atlanta). Even though this meant openings for more players, the 532 Canadians constituted only 57.5% of the league. Thanks to the fall of the Soviet Union, a flood of talented Eastern European players entered the league and gobbled up jobs. It's freshman econ: the skill level had gone up and opportunities for Canadian players had gone down. Here is the birthdate breakdown for that season:

Canadians Non-Canadians

Jan-Mar: 28.2% 35%

Apr-June: 29.3% 23.7%

July-Sept: 22.2% 23.4%

Oct-Dec: 20.3% 17.9%

The numbers still reflect a bias towards players born in the first half of the year. But you begin to see the trends, especially among the Canadians. And while Europeans have the option of coming to North America to play hockey, there are also many professional leagues in their native countries. Canadians are only going to play in the NHL, which, like all marketplaces, has gotten more competitive. As a result, only the most skilled Canadians are finding work. Those who once earned spots as role players in the big leagues are topping out in the minors. "As you shorten the number of spots, you can be more selective in who you pull in," says Barnsley. "My bet is that talent wins then."

This is in no way meant to disparage Canadian hockey. As the Olympics proved, the country is pound-for-pound the sport's undisputed champ. Perhaps recognizing the relative-age effect and more effectively mining talent from its younger population is a reason why. Mike Pelino was the men's hockey coach at Brock University in Ontario when he participated in a relative-age study that was published in 2000. He has also served a four-year stint as a coach with Hockey Canada at various levels and worked as an assistant coach with the Florida Panthers, New York Rangers, and most recently in the Phoenix Coyotes organization. It wouldn't be inappropriate to say he's a hockey lifer. And over that lifespan he's seen the game and the teaching of it evolve. "When I was playing juniors back in the 50s and 60s you played hockey in the fall and winter and then other sports the rest of the year," says Pelino. "Now kids play all year round. It's very specialized. If somebody falls behind his peers, there's more opportunity to make up the gap."

Which begs the question: Is it such a disadvantage to be born relatively young? A study done in 2007 by Joseph Baker and A. Jane Logan at York University in Canada looked at the birthdate and birthplace effects in North American NHL draftees from 2000-05. While their findings upheld that more relatively older players were drafted, the researchers were surprised that it was the younger players who were chosen earlier in the draft. They concluded that these players must have exhibited superior performance or shown better long-term potential to warrant the higher selections. While fewer in numbers, on a whole, the younger players were the class of their classes.

Which was also seen in the results of a study conducted by John Ashworth and Bruno Heyndels examining the relative-age effect on the earnings of German-born soccer players in the Bundesliga. Again, they found an overrepresentation of relatively older players, but the younger players had higher average salaries. One of their hypotheses for the discrepancy was that the latter group represented a more selective subgroup of players that survived a system that discriminated against them. That which didn't kill them only made them stronger, and eventually wealthier.

A comparison of the top salaries in the NHL for the 2009-10 season showed a similar pattern. According to nhlnumbers.com, 72 of the top 150 salaries in the league belonged to Canadian players, roughly half (35) of which were born in the second half of the year. Only 11, or 15%, of these top earners were born in the first quarter of the year, while 13 were born in the last three months. The numbers reflected a similar scenario with the non-Canadian players. More of them had birthdays in November and December (15) than January and February (12) and a majority (43) had birthdays after June.

While salary level isn't necessarily an accurate indication of talent, it's certainly a fair barometer of a professional athlete's worth and - to Gladwell's point - success. In this regard, there seems to be almost no relative-age bias in the NHL. "If Canada had a second hockey league for those children born in the last half of the year," Gladwell proposes, "it would today have twice as many adult hockey stars." For that to be true, there would have to be twice as many stars born in the first half of the year. Given the equality of top-earners across the entire calendar, that's clearly not the case. At best, a second league for "younger" players might result in a few more Canadian players earning the league minimum. It might be arbitrary which players start out with the most advantages, but having broad shoulders, superior hand-eye coordination, and a wicked wrister is what makes a player a star. Not a winter birthday.

Some have even suggested that relatively young players are better off for having started behind the eight-ball. Another study explored different variables - competition level, gender, nationality, career stage, and position - that may moderate the relative-age effect of German handball players (Schorer et al., 2009). In addition to finding that relatively younger players enjoyed longer careers, the researchers determined that the long term effect of relative-age is questionable. They noticed a significant spike in relatively older players in the early stages of competition, but the trend dissipated at the higher levels. Here's their theory as to why:

"It might be beneficial for relatively younger players to have the opportunity and necessity to develop the specific technical or tactical skills needed to be able to compete successfully against their older, more mature opponents. In the long run, this may result in a larger repertoire of skills, making them superior performers."

Think of it as the "Big Brother" effect. How many hall-of-famers, in any sport, credit an older sibling for providing their first meaningful measuring stick? They were forced to compete against a bigger, more advanced opponent, and it significantly enhanced their own abilities in the process. Once they matured and lack of size was removed from the equation, overall skill became the determining factor for their continued success.

Perhaps the clearest answer can be offered by Gladwell's Medicine Hat team. It's been four years since they lost in the Memorial Cup final, the top prize in the top junior hockey league in the world. The players are now at an age and experience level where most should have graduated from prospect status. But of the 23 Canadians on the team (there was a Slovak and an American), 17 have never suited up for an NHL team. And of the six who have stepped on NHL ice (one for nine minutes as an emergency goalie), not one has made a significant impact in the league. There's not an All-Star in the bunch. The benefits of the relative-age effect took them as far as they could go: the top of the juniors. Even with the all the extra early coaching and ice time, their comparative lack of skill kept them from achieving the highest of success in the sport. To be fair, one of the players, Tyler Ennis, is only 22 years-old and a highly touted winger for the Buffalo Sabres. There's still time for him to blossom into a star.

Of course, he was born in October.