Outliers -- a quick fact check.

On the recommendation of several people, I have started reading Outliers by Malcolm Gladwell. It’s interesting so far, and he makes some thought-provoking arguments about the role of opportunities in youth and hard work (“the 10,000 hour rule” — how much it takes to master anything) as opposed to innate ability as drivers of achievement. In the opening chapter, he uses the example of Canadian hockey players to illustrate how a subtle bias in opportunity could have major consequences for who succeeds and who doesn’t. Specifically, he notes that at various junior and professional levels, players born between January and March are greatly overrepresented. The explanation he offers is that these players are up to almost a year older than other players in each division because the cutoff in Canadian minor hockey is January 1. So, if you’re born on January 1 you will be the oldest in your division, and if you’re born on December 31st you will be the youngest. This difference in age translates into a difference in size, coordination, and other features associated with early athletic development. Thus, the older/larger boys are chosen for the “rep” teams while the younger/smaller boys remain in “house” leagues. This leads to differences in coaching, practice time, number of games, quality of teammates and opponents, and encouragement. Therefore, being born in the first few months of the year is, according to this argument, one of the major predictors of how far a player will go in the sport.

I think this is an interesting observation, and the interpretation seems reasonable enough.  However, something just didn’t sit right and my inclination as a scientist was to test the robustness of the reported pattern.  So I spent about 10 minutes looking up some of the best Canadian players in the NHL in recent times. Specifically, the gold medal Olympic teams from 2002 and 2010.  Here are the birth dates of the world champion players:

2002 Olympics
Belfour, Ed Apr 21
Blake, Rob Dec 10
Brewer, Eric Apr 17
Brodeur, Martin May 6
Fleury, Theoren June 29
Foote, Adam July 10
Gagné, Simon Feb 29
Iginla, Jarome July 1
Joseph, Curtis Apr 29
Jovanovski, Ed June 26
Kariya, Paul Oct 16
Lemieux, Mario Oct 5
Lindros, Eric Feb 28
MacInnis, Al July 11
Niedermayer, Scott Aug 31
Nieuwendyk, Joe Sept 10
Nolan, Owen Feb 12
Peca, Mike Mar 26
Pronger, Chris Oct 10
Sakic, Joe July 7
Shanahan, Brendan Jan 23
Smyth, Ryan Feb 21
Yzerman, Steve May 9
2010 Olympics
Bergeron, Patrice July 24
Boyle, Dan July 12
Brodeur, Martin May 6
Crosby, Sidney Aug 7
Doughty, Drew Dec 8
Fleury, Marc-André Nov 28
Getzlaf, Ryan May 10
Heatley, Dany Jan 21
Iginla, Jarome July 1
Keith, Duncan July 16
Luongo, Roberto Apr 4
Marleau, Patrick Sept 15
Morrow, Brenden Jan 16
Nash, Rick June 16
Niedermayer, Scott Aug 31
Perry, Corey May 16
Pronger, Chris Oct 10
Richards, Mike Feb 11
Seabrook, Brent Apr 20
Staal, Eric Oct 29
Thornton, Joe July 2
Toews, Jonathan Apr 29
Weber, Shea Aug 14

Kids — if you’re born in the summer or fall, don’t worry about it.


7 comments to Outliers — a quick fact check.

  • That’s the thing in general with Malcolm Gladwell’s books and similar books (such as Freakonomics) — they present a lot of interesting, plausible hypotheses without the actual data involved (which may in fact never have been collected). The general public probably thinks that’s how science works — that there’s no need to show that a plausible trend is actually supported by data if you can come up with a compelling argument for why it should be.

      (Quote)

  • Perhaps for the outlier outliers, birthday is not a significant variable.  You chose the best of the best.  I know you probably did this because it was a quick test…but you used a different data model.
    We need evidence-based arguments to be backed up by actual evidence, not just hand-waving in the direction of possible evidence.  Still, in this case, I think the critique only works with a data set of all hockey players in the NHL or a random selection – not a set of players selected for being the best of the best.
    I tried to find some birthday data sets, but didn’t have a lot of luck.
    http://www.nhldigest.com/nhl-player-birthday-calendar/
    This looks quite interesting though:
    http://www.socialproblemindex.ualberta.ca/relage.htm

      (Quote)

  • I haven’t read <i>Outliers</i> yet, but its on my reading list. But regarding your criticism, aren’t NHL players already selected for high achievement?  They’re playing the sport professionally, after all!  After you reach the elite status of professional athlete, I’d imagine inborn ability or work ethic would play a much larger role that how big you are.
    And did you actually run any stats on this?  As you well know, patterns may not be obvious to the causal observer.  Its seems like there should be more of a linear correlation between time spent maturing and achievement, anyway.  Eight on your list were born Jan-Feb, but only three were born Nov-Dec.  I image overall birthrate may vary over the year, with higher rates appearing ~9 months after certain events or holidays.  Maybe the effect is only seen after you normalize the data.  It is quite frustrating when authors neglect any sort of citation when they make theses sorts of claims.
     

      (Quote)

    • So… in the book, he gives one major example in detail, a Memorial Cup winning team. In that sense, my quick check is totally in line with his approach in the book. Would I accept my “analysis” as convincing? No — and I didn’t think his was either, which was my point. If he’s right, and the NHL is disproportionately populated with people born Jan-Mar, then why aren’t teams chosen from within the NHL disproportionately biased in the same way even if sampling is random? The fact that no bias toward early births is apparent on these best-of-the-best teams should, I think, raise some flags.

        (Quote)

  • Marcus Freeman

    Interesting, but I think the point that you`re cherry picking the best-of-the-best is valid.  I just did a quick check of 5 local OHL teams, which is where the successful juniors end up, and I think is more the meat of Gladwell’s argument.  A quick summary of the month of birth for 100 players shows that more than 50% were born in April or earlier, and almost 75% were born in June or earlier.  I think if you were to do this across the league, you’d find it holds fairly true.
     

    Month
    Frequency
    Cumulative %

    1
    18
    17.82%

    2
    15
    32.67%

    3
    7
    39.60%

    4
    14
    53.47%

    5
    9
    62.38%

    6
    12
    74.26%

    7
    8
    82.18%

    8
    4
    86.14%

    9
    3
    89.11%

    10
    5
    94.06%

    11
    4
    98.02%

    12
    2
    100.00%

      (Quote)

  • The book is about OUTLIERS — by definition, extremes that do not fall within the main distribution.  He confuses “above average” or “successful” with a proper meaning of the term. Yes, I cherry-picked the best of the best — because that’s what outliers are. He cherry-picked too (a Memorial Cup winning team — good players, but hardly “outliers”, likewise with your OHL teams). If he called the book “ABOVE AVERAGERS” (admittedly not as catchy) then this would be ok. And, anyway, that’s not how the age divisions work in minor hockey. Every kid, regardless of birthday, will be playing against older kids when joining a new division.
     

      (Quote)

Leave a Reply

 

 

 

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

*