Abstract
The relative age effect associated with cut-off dates for hockey eligibility has been an ongoing debate in certain academic circles and in the popular media. The effect is primarily found in Canadian Major Junior Hockey, where a disproportionate share of birthdays fall in the first three months of the year. But when the National Hockey League rosters of Canadian-born players are examined, the pattern is less pronounced. Using publically available data of hockey players from 2000–2009, we find that the relative age effect, as described by Nolan and Howell (2010) and Gladwell (2008), is moderate for the average Canadian National Hockey League player and reverses when examining the most elite professional players (i.e. All-Star and Olympic Team rosters). We also find that the average career duration is longer for players born later in the year. In sum, there is a surprising ‘relative age effect reversal’ that occurs from the junior leagues to the most elite level of hockey play. This supports an ‘underdog’ hypothesis, where the relatively younger players are thought to benefit by more competitive play with their older counterparts.
What is hockey success? Is it simply making a professional roster or is it actually excelling on one? This is a critical distinction Nolan and Howell do not explore in their recent article in the July 2010 issue of the International Journal of Sociology of Sport. They were motivated by the first chapter in the best-selling book Outliers (Gladwell, 2008). Malcolm Gladwell’s Iron Law of Canadian Hockey – citing the work of Barnsley et al. (1985) – states that birth months are destiny. The ‘Iron Law’ claims that at least 40 percent of players on minor league rosters across Canada were born in the first three months of the year. Also known as the relative age effect (RAE) scholars argue that this occurs for two reasons. First, because of the 1 January cutoffs for hockey play, players born in January, February, and March are simply bigger and, therefore, better players than those born later in the year – labeled primary effects. Second, these players simply have more exposure to competitive play and training – secondary effects – which contribute to more time for deliberate practice (Baker and Logan, 2007; Barnsley et al., 1985; Ericsson, 2007; Musch and Gondin, 2001).
Nolan and Howell (2010) found, like Barnsley et al. (1985), that age bias is not only prevalent in the minor leagues but also carries over into the National Hockey League (NHL). Whereas, Barnsley et al. reported that 32 percent of NHL players were born in the first quarter in the 1982–1983 season, Nolan and Howell found that 30 percent of NHL players, 25 years later, were born in the first quarter. While this is not the large imbalance of 40 percent – the standard to meet Gladwell’s ‘Iron Law’ – it still supports a relative age effect.
Surely being on an NHL roster is an achievement, but who is the best in the NHL? Unfortunately, Nolan and Howell (2010) ignore studies that move beyond a static measure of NHL rosters done since Barnsley et al. (1985). When examined, these studies complicate our understanding of RAE and its lasting impact on the career trajectories of players. Although the relative age effect is present in the NHL, the effect fades across draft rounds. And for more advanced play, RAE has no effect on statistical measures of skill and performance (Wattie et al., 2007). 1 The fading effect has also been found in handball (Schorer et al., 2009) and other sports at the elite level (for a meta-analysis, see Cobley et al., 2009).
Without this literature cited, Nolan and Howell (2010) are able to ignore the growing question for the RAE literature; why does the relative age effect fade across advancing levels of hockey play? Although we do not resolve this important question here, our results take this question one step further. We suggest that the effect may not only fade (as Wattie et al., 2007 suggest) but reverse. Although a reversal has been found in soccer where players born at the end of the year earn more than their teammates born at the start of the year (Ashworth and Heyndels, 2007) there is little collaborating evidence that suggests this pattern extends beyond soccer. As we will show, the relative age effect appears to reverse for players selected onto All-Star and Olympic teams – the most elite circles of hockey play. 2 When hockey success is specified this way, we argue that an underexamined phenomenon may be occurring – a reverse relative age effect.
Data and methods
To examine birth distribution for elite hockey, we compiled three sets of data, all focused on the examination of RAE in recent years. 3 First, we focused on minor league rosters (Major Junior Hockey). We examined the rosters of the Western Hockey League’s Medicine Hat Tigers and Vancouver Giants in 2007, as reported in Gladwell’s Outliers as well as their rosters in 2010. Although this is not a critical analysis for the literature, it served as a starting point for the popular media’s coverage of the effect.
Second, we examined the distribution of birth months of first round draft picks of Canadian players in the NHL for the years 2007, 2008, 2009, and 2010 (NHL.com, 2010). These data represent only minor-league accomplishments and serve as a projection of their future value. Their success in the top levels of professional hockey has yet to be determined at this stage of their careers. The total number of Canadian-born first round draft picks were n = 15 in 2007, n = 17 in 2008, n = 17 in 2009, and n = 15 in 2010.
Third, we focused on major league rosters (the National Hockey League). We examined all Canadian hockey players who played from 2000 to 2009 – a total of 1109 players. We collected data on the players age of entry and exit, career length (entry minus exit) and birth month (www.databaseHockey.com, 2010).
Finally, we examined the All-Star and Olympic hockey rosters. We compiled information on Canadian hockey players from All-Star rosters in 2007 (n = 25), 2008 (n = 26), and 2009 (n = 24) (ESPN.com, 2007, 2008; NHL.com, 2009) and the Canadian National Olympic Team rosters 1998 (n = 23), 2002 (n = 23), 2006 (n = 28), and 2010 (n = 23) (HockeyCanada.com, 2010). The All-Star starting line is chosen by the fans while the remainder of the team is chosen by the NHL’s Hockey Operations Department. 4 This is in contrast to Olympic team selection that is at the discretion of Canadian administrators and general managers.
Our analysis is straightforward. We tallied the number of Canadian hockey players born in each of the four quarters of the year: January, February, March; April, May, June; July, August, September; and October, November, and December (n = 1109). As another way to understand hockey success, we also examined career length by subtracting time of league exit by entry to create a career-length variable. 5 For this measure, we looked at all careers that ended between 2000 and 2009. Because some players’ careers were not yet complete in the data, we chose to exclude players who, at the time of the collection of the data, were registered as still active in 2010. Thus, only those players who had completed their careers by 2009 were included in the sample. The career-length variable restricted the sample of 1109 Canadian hockey players to 1003.
Results
In our analyses, we found a strong relative age effect that eventually fades, then reverses across levels of hockey play among Canadian-born players. In our first data, early birth-month advantage is apparent in the Medicine Hat Tigers championship roster of 2007 (56%) and for their opponents the Vancouver Giants (44%), but it is less true of the same teams three years later (33% and 39% respectively). The effect is also apparent among Canadian-born first round draft picks, with 40 percent, 41 percent, 47 percent, and 33 percent born in the first quarters of 2007, 2008, 2009, and 2010 respectively) (see Figure 1). But for the average player in the NHL, the effect seems to fade. Although the first round draft picks confirm Gladwell’s law (33–47 percent across 2007–2010) – a reflection of their Major Junior Hockey performance – the percent of all Canadian hockey players in the NHL born in the first three months is a modest 28 percent (see Figure 1).
Distribution of Canadian hockey players born in January, February, or March across various rosters.
Looking at the most elite levels of play in hockey, the RAE reverses. Of NHL All-Star rosters in 2007, 2008, and 2009 respectively, only 20 percent, 15 percent, and 13 percent consist of Canadian-born players with birthdays in the first three months of the year. This pattern can also be found among Canadian Olympiads. The 2010 gold medal-winning Canadian Olympic hockey team had a mere 13 percent of its players born in the first three months. Previous years confirm the trend with 17 percent in 2006, 26 percent in 2002, and 14 percent in 1998 (see Figure 1).
It appears that being born at the start of the year reduces the chance of elite play. Consider the average distribution of players born in the first quarter of the year for the NHL Canadian-born players – 28 percent. The combined average of the All-stars and Olympic rosters is 17 percent. This represents a 40 percent reduction in the distribution of players born in the first three months of the year. If birth month had no effect on elite play, the percentage would remain 28 percent. 6
Perhaps the most surprising finding is that the career duration of all Canadian-born NFL hockey players who were born in the first quarter of the year was one season shorter than those players born in the last quarter of the year (see Figure 2). Whereas Wattie et al. (2007) found no relative age effect on a related measure of career-length – career number of wins – we find a reverse relative age effect when examining the actual length of a player’s career.
NHL career duration of Canadian hockey players by birth month, 2000–2009.
Conclusion
Hockey success is not simply joining the NHL, but it is participating in elite play. When elite play is considered, it appears that it is better to be born in any month beside January, February, or March. While popular accounts of RAE may be persuasive, the phenomenon is complex. Our findings illustrate how critical it is to define hockey success. When hockey success is defined as playing Major Junior Hockey, the effect is strong, as Gladwell reported in the popular press. But the effect diminishes when success is defined as making the NHL (Howell and Nolan, 2010), and fades when performance and skill are considered (Baker and Logan, 2007). When hockey success is defined as the most elite levels of play, the relative age effect reverses.
Our findings lead to two unresolved questions. What advantages would a relatively younger player accrue in the junior leagues that would lead to greater representation on All-Star and Olympic rosters? Why would early disadvantage lead to longer careers? Although we leave this puzzle for future work, we think one explanation is persuasive. As Schorer et al. (2009) suggest, relatively younger players are challenged by their more advanced – slightly older – peers. This ‘underdog’ hypothesis flips the relative age effect on its head. What truly makes an elite player may simply be overcoming the odds, including the relative effect of age.
Footnotes
Acknowledgements
We are grateful for the assistance from Aaron Woodall and Shawn Meiners.