Journal of Sports Sciences, October 2006; 24(10): 1065 – 1073

When ‘‘where’’ is more important than ‘‘when’’: Birthplace

and birthdate effects on the achievement of sporting expertise

JEAN CÔTÉ1, DANY J. MACDONALD1, JOSEPH BAKER2, & BRUCE ABERNETHY3

1
School of Kinesiology and Health Studies, Queen’s University, Kingston, Ontario, Canada, 2School of Kinesiology and
Health Science, York University, Toronto, Canada and 3Institute of Human Performance, The University of Hong Kong,
Hong Kong and School of Human Movement Studies, The University of Queensland, Brisbane, QLD, Australia

(Accepted 20 October 2005)

Abstract
In this study, we assessed whether contextual factors related to where or when an athlete is born influence their likelihood of
playing professional sport. The birthplace and birth month of all American players in the National Hockey League, National
Basketball Association, Major League Baseball, and Professional Golfer’s Association, and all Canadian players in the
National Hockey League were collected from official websites. Monte Carlo simulations were used to verify if the birthplace
of these professional athletes deviated in any systematic way from the official census population distribution, and chi-square
analyses were conducted to determine whether the players’ birth months were evenly distributed throughout the year.
Results showed a birthplace bias towards smaller cities, with professional athletes being over-represented in cities of less than
500,000 and under-represented in cities of 500,000 and over. A birth month/relative age effect (in the form of a distinct bias
towards elite athletes being relatively older than their peers) was found for hockey and baseball but not for basketball and
golf. Comparative analyses suggested that contextual factors associated with place of birth contribute more influentially to
the achievement of an elite level of sport performance than does relative age and that these factors are essentially independent
in their influences on expertise development.

Keywords: Elite athletes, children in sport, city size, relative age, athlete development

was very successful, there would never have been a

Introduction
middle or later period’’ (p. 141). Furthermore,
Individual sport development can vary because of evidence suggests that the making of an expert
different learning opportunities and the psychosocial athlete usually begins in an environment where
environment in which learning takes place (Côté, children are exposed early and regularly to sporting
Baker, & Abernethy, 2003). Differences in unique activities (Baker, Côté, & Abernethy, 2003;
environmental experiences during childhood could Kalinowski, 1985; Monsaas, 1985; Soberlak & Côté,
lead to disparities among elite and less elite athletes, 2003). Accordingly, some children might benefit
in motivation to practise, and in the type of skills from situations that provide them with more
acquired, as well as in how and when exceptional opportunities to become involved in sports. The
abilities are developed (Bloom, 1985; Côté et al., relative age effect (Musch & Grondin, 2001) and, to
2003). Retrospective studies of elite athletes show a lesser extent, the size of the city from which an
that critical incidents that promote a child investing athlete comes from (Curtis & Birch, 1987) are two
in one sporting activity over others include positive variables that have been associated with increased
experiences with a coach, encouragement from an early exposure to sport and the achievement of
older sibling, early success, and/or simple enjoy- expertise.
ment of the activity (Carlson, 1988; Côté, 1999; The relative age effect shows that the older one is
Kalinowski, 1985; Monsaas, 1985). Kalinowski relative to one’s peers in the same grouping or junior
(1985) described the early years of 21 elite swimmers sport team (i.e. the greater one’s ‘‘relative age’’), the
as critical for later achievement in swimming as greater the probability of eventually becoming an elite
follows: ‘‘Had there been no excitement during the athlete (Baxter-Jones & Helms, 1994, Dundink, 1994;
early years, and no sense that the young swimmer Helsen, Hodges, Van Winckel, & Starkes, 2000).

Correspondence: J. Côté, School of Kinesiology and Health Studies, Queen’s University, Kingston, Ontario K7L 3N6, Canada. E-mail: jc46@[Link]
ISSN 0264-0414 print/ISSN 1466-447X online Ó 2006 Taylor & Francis
DOI: 10.1080/02640410500432490
1066 J. Côté et al.

Studies examining birthdates of professional athletes (1988) concluded that elite players predominantly
in baseball (Thompson, Barnsley, & Steblelsky, 1991), came from rural areas, and that these areas provided
ice hockey (Barnsley & Thompson, 1988; Boucher & the athletes unlimited opportunities to participate in
Mutimer, 1994), soccer (Barnsley, Thompson, & sports. Carlson also suggested that coaches in rural
Legault, 1992; Dundink, 1994), and basketball areas were more likely to take great care in main-
(Hoare, 2000) have shown a skewed birthdate dis- taining the player – coach relationship even if they
tribution favouring players that were born in the first did not have the technical tennis knowledge of the
quarter of each sport-year. In an extensive review, coaches in urban centres. In another study, Curtis
Musch and Grondin (2001) proposed mechanisms and Birch (1987) examined the city size of the
that could be responsible for the relative age effect in birthplace of Canadian and US Olympic hockey
sport. Competition, physical development, psycho- players and Canadian National Hockey League
logical factors, and experience were discussed as (NHL) players. They found that for Canadian
factors related to the relative age effect that would players, regions with a population of less than 1000
alter the environment in which young children prac- inhabitants and those with a population greater than
tised sport. The most compelling hypothesis about 500,000 inhabitants were under-represented in rela-
the relative age effect suggests that older children tion to the expected proportions of the population in
within a group will be provided with environments the same age range. The remaining values in each
that facilitate the improvement of their skills early in of the census subdivisions of population yielded
their development. For example, a coach could values that closely resembled the expected propor-
identify children as being more mature or physically tions. Based on these findings, Curtis and Birch
larger and, accordingly, give them more practice or (1987) suggested that ‘‘top players are more likely
opportunities for learning, thereby facilitating their to come from communities large enough to build
development. rinks, but not so large that the demand for ice time
Another environmental variable that has received outweighs opportunities to skate’’ (p. 239). Unfortu-
little attention in sport expertise research is the size of nately, the qualitative nature of Carlson’s (1988)
the city where elite athletes gain their formative study and the focus of Curtis and Birch’s (1987)
experiences. This variable could have a significant study on ice hockey did not permit the identification
influence on how athletes will first be exposed to of optimal city sizes for sport development across
sports, which, like the relative age effect, can limit or different sports and different sport systems.
benefit performance. It is apparent that many The primary purpose of this study was to examine
children who live in smaller cities have access to whether the size of the city in which an athlete is born
facilities that introduce them to sport in different ways (i.e. the birthplace effect) influences the likelihood of
than children from urban centres. Children from a playing professional sport. A secondary purpose was
larger urban centre have potential access to a larger to examine the relative age effects on the same
number of resources compared with their counter- sample of professional athletes and to compare the
parts from smaller cities (e.g. arenas, specialized magnitude of any observed influences of birthplace
coaching). Urban athletes are also more likely to with the well-documented birthdate (relative age)
practise their sport in a structured setting such as a effects on the probability of becoming a professional
league, which is monitored by coaches with specific athlete. To maximize generalizability and identify
practice times and games, whereas individuals in effects that may be sport specific, athletes from
smaller cities are more likely to engage in games several professional sports were surveyed. Further-
without the structure of the urban setting. There more, a comparison for the same sport across two
might also be greater diversity in player size and different countries was made to clarify whether any
ability in small cities, since all the children from the observed effects were due to the sport demands or to
neighbourhood gather to play together independent the variations in sport development systems of
of age and ability. Urban athletes, who live within a different countries.
more densely populated and structured environment,
usually find themselves playing opponents and having
team-mates who are all relatively the same age, size, Methods
and ability. It has been suggested (Côté et al., 2003;
Participants
Soberlak & Côté, 2003) that more opportunities to
play with older children and adults and experiment A total of 2240 male athletes were evaluated.
with different types of sport and physical activity, The birthplace and birthdate of American players
such as those found in rural settings, might facilitate in the National Hockey League (NHL, n ¼ 151;
the development of sport expertise. 2003 – 2004 roster), National Basketball Association
Data on the ‘‘urban – rural’’ debate are limited. In (NBA, n ¼ 436; 2002 – 2003 roster), Major Lea-
one study of 10 Swedish elite tennis players, Carlson gue Baseball (MLB, n ¼ 907; 2002 – 2003 roster),
 Birthplace effects 1067

Professional Golfer’s Association (PGA, n ¼ 197; Although US golf does not adhere to a strict calendar
2003 – 2004 roster), and Canadian players in the year and age-restricted categories in the way other
NHL (n ¼ 549; 2002 – 2003 roster) were collected sports do, players are classified as junior if they are
from official websites (ice hockey: [Link] under 18 years at the time of the national junior
com; basketball: [Link] baseball: championship. This championship is usually held
[Link] golf: [Link] during the last two weeks of July. To calculate the
The total number of athletes displayed on these calendar year for golf, we used the same categoriza-
websites was 2291; however, the birthplace city of 51 tion as baseball (i.e. from 1 August to 31 July).
athletes could not be matched with official census
data and, as a result, these athletes were dropped
Statistical analyses
from further analyses.
To assess differences between population and league
distributions, Monte Carlo simulations were con-
Procedure
ducted based on methods discussed by Press,
The distributions of athletes’ birthplaces across Flannery, Teukolsky and Vetterling (1986). Monte
various city sizes were compared with the distribu- Carlo simulation is a bootstrapping technique that
tion of similar aged individuals in the general involves drawing samples from a well-defined popu-
population using census statistics. Because our lation (i.e. the census distribution; Hoyle, 1999;
examination involved the birthplace of the profes- Mooney & Duval, 1993). The simulations yielded
sional athletes, census statistics from the 1976 census estimates of the expected standard deviations for
for Canada (Statistics Canada, 1979) and the 1980 randomly and unbiased samples using the same
census for the United States (US Bureau of the numbers of cases (i.e. athletes) represented in each
Census, 1981) were used, since these years more sport. These standard deviations were then used to
accurately represented the Canadian and American determine the probability of the deviations of cases in
statistics during the players’ birth year. The census each sport from the general population across the
statistics provided the number of males under the age different city sizes. For example, using MLB players,
of 14 who lived in each of the population subdivi- we randomly selected 907 cases (the same number of
sions. The birthplace of athletes provides a proxy for athletes in the MLB portion) to create one sample,
the location in which children spent their develop- and then determined how this sample corresponded
mental years. It is important to recognize that the to the actual population. By repeating this re-
place of birth does not always coincide with the place sampling process 10,000 times, we obtained a
of development. For example, athletes born in large sampling distribution to use as the basis for
urban centres might have moved to smaller commu- comparisons to the sport under examination (in this
nities during their development or, conversely, case baseball). From our data, we were able to
athletes born in small towns might have moved to compare the sport distribution to the sampling
larger cities. Although migration of some individuals distribution obtained from the 10,000 re-samples
between small towns and larger urban centres is and determine the likelihood that the sport distribu-
probable within our sample, the net movements tions were due to chance. Using the sampling
between the two are likely to be essentially equal and distribution and standard deviations obtained from
opposite. the Monte Carlo simulation, z-scores and probabil-
To test the relative age effect, birthdates for all ities were calculated for each sport and city size.
players were collected from the same websites. Birth Alpha levels were adjusted using the Bonferroni
month of each player was compiled into quarters method and set at P 5 0.001.
(Q), which reflected the calendar year of each sport Odds ratios were also calculated across the
at the time that these athletes were involved in youth different city sizes for the US and Canadian data.
sport. The calendar year of US and Canadian hockey The odds ratios were calculated by dividing the odds
is from 1 January to 31 December (Q1 ¼ January, of becoming a professional athlete in each sport by
February, March; Q2 ¼ April, May, June; Q3 ¼ July, the odds of being born in a city of a specific size. A
August, September; Q4 ¼ October, November, 95% confidence interval (CI) was calculated. An
December). The calendar year of US baseball is odds ratio greater than 1 (with upper and lower limits
from 1 August to 31 July (Q1 ¼ August, September, higher than 1) implies that an athlete born in a city of
October; Q2 ¼ November, December, January; a given size is more likely to become a professional
Q3 ¼ February, March, April; Q4 ¼ May, June, July). athlete than if born in a city of any other size. An
The calendar year of US basketball is from 1 odds ratio less than 1 (with upper and lower limits
September to 31 August (Q1 ¼ September, October, less than 1) implies that an athlete born in a city of a
November; Q2 ¼ December, January, February; given size is less likely to become a professional
Q3 ¼ March, April, May; Q4 ¼ June, July, August). athlete than if born in a city of any other size.
1068 J. Côté et al.

(11.18, 11.18)
(0.16, 70.08)

(0.18, 70.13)
Any odds ratios with a CI range that contains the null

(0.14, 0.01)

(0.88, 0.87)
(1.64, 1.63)
(1.47, 1.46)

(2.35, 2.34)
value of 1 are considered not to be statistically

OR (CI)
significant.
Chi-square tests were conducted on the birthdates

PGA

0.04
0.08
0.02
0.88
1.64
1.46
11.18
2.34
of each player according to the four quarters of their
sport calendar year to determine the significance of
deviations for the expected number of births in each
quarter. Similar to other studies on relative age (e.g.

0.5
1.0
0.5
11.1
16.8
13.5
11.1
45.7
%
Barnsley & Thompson, 1988), the expected values
were calculated based on the assumption of an even
distribution of birth throughout each quarter of the

(20.82, 20.82)
year.

(1.24, 1.24)
(2.04, 2.04)

(1.69, 1.69)
Table I. Representation of the US population, professional athletes, and odds ratios across cities of different sizes.
Sport-specific and overall effect sizes (Cohen’s d)

(.18, .15)
(.23, .21)
(.15, .12)
(.54, .54)
OR (CI)
were calculated to evaluate the magnitude of the
relative age effect and the birthplace effect. For

MLB
the relative age effect, Cohen’s d was calculated as

20.82
.17
.22
.14
.54
1.24
2.04

1.69
the difference between the number of players that
were born in the first 6 months of a given sport-year

US professional athletesb
and the number of players that were born in the last 6

16.8
1.8
2.8
2.9
7.1
13.3
17.8

37.7
%
months divided by the standard deviation of the
sample. For birthplace, Cohen’s d was calculated as
the difference between the number of players that

(10.86, 10.86)
were born in large cities (500,000 and more) and the

(0.38, 0.36)
(0.56, 0.55)
(0.34, 0.33)
(0.96, 0.95)
(1.50, 1.49)
(1.80, 1.80)

(1.10, 1.09)
number of players born in small cities (less than

OR (CI)
500,000) divided by the standard deviation of the
sample. [This effectively created a ‘‘top half/bottom

Percentage of professional athletes in 2002 – 2004 born in each of the subdivisions of the 1980 US census.
NBA
half ’’ comparison similar to that undertaken for

10.86
0.37
0.55
0.33
0.96
1.50
1.80

1.10
birthdate, as in 1980 some 51.8% of the US
population resided in cities with a population in
excess of 500,000, with the balance (*48.2%) in

Percentage of males under the age of 14 in each of the subdivisions of the 1980 US census.
3.9

10.8
6.7
6.9
11.9
15.6
16.1

28.2
smaller cities.] For the Canadian data, residents
%

in areas of less than 1000 were not included in the

‘‘small cities’’ bracket for the calculation of the effect
(18.70, 18.70)
(0.16, 70.04)

size. The 1976 Canadian census classified areas of

(0.24, 0.19)
(0.18, 0.13)
(0.51, 0.49)
(1.17, 1.16)
(2.05, 2.05)

less than 1000 as ‘‘rural’’ and as lacking any type of (1.79, 1.79)
OR (CI)

infrastructure that is common to a city.

Finally, some analyses were conducted to deter-
NHL

mine if the relative age effect and the birthplace were

0.06

18.70
0.21
0.15
0.50
1.16
2.05

1.79

independent from each other in predicting elite

performance in sport. First, Pearson correlations
Abbreviations: OR ¼ odds ratio, CI ¼ confidence interval.

were calculated between the players’ birth month

0.7

17.2
2.6
3.3
6.6
12.6
17.9

39.1

(according to their sport’s calendar year) and city size

%

of birthplace. Second, chi-square analyses were

conducted to examine the relationship between the
US population

ratios of the players born in small cities (less than

500,000) and large cities (500,000 and more) in each
(%)a

9.9
11.4
18.1
12.4
11.0
9.6
1.1
26.4

of the sport-year’s quarters to determine if place of

birth in any way moderated any relative age effect.

Results
2,500,000 – 4,999,999
1,000,000 – 2,499,999
500,000 – 999,999
250,000 – 499,999
100,000 – 249,999

Birthplace
50,000 – 99,999
45,000,000

Table I contains data from the 1980 US census (i.e.

550,000

the census that most closely reflects the birth year of

City size

the professional athletes) on the percentage of boys

under the age of 14 that lived in cities of different
b
a
 Birthplace effects 1069

sizes (i.e. expected values) and the percentage of between expected and observed values. Table III also
players in 2002 – 2004 from the NHL, NBA, MLB, contains the odds ratios of becoming a professional
and PGA that were born in these different areas (i.e. hockey player when born in cities of different sizes in
observed values). Table I also contains the odds Canada. The Canadian data show that cities with
ratios of becoming a professional athlete when born populations larger than 500,000 produced signifi-
in cities of different sizes in the USA. Table II cantly less ice hockey players than expected. The
presents mean differences between expected and Canadian data also suggest that rural areas with
observed values for US hockey, baseball, basketball, populations of less than 1000 produced significantly
and golf together with the corresponding z-scores. less professional players than expected.
Odds ratios (Table I) and Monte Carlo simulation
(Table II) show that cities over 500,000 are con-
Relative age effect
sistently under-represented in terms of producing
professional athletes, while cities under 500,000 are Table IV illustrates the frequency and percentage
of expected proportions or over-represented. Cities distributions of the players’ birth months and the
with populations between 50,000 and 100,000 results of the chi-square analysis. These results show
present the best odds of producing elite athletes in a relative age effect for professional ice hockey
hockey, basketball, baseball, and golf. players from both Canada and the USA, and US
Table III contains data from the 1976 Canadian baseball players. Generally, players born in the first
census on the percentage of boys under the age of 14 quarter of the sport-year were over-represented
that lived in cities of different sizes, the percentage of compared with players born in the fourth quarter.
players in 2002 – 2003 from the NHL that were born No relative age effects were found for US golf and
in these different areas, and the mean differences basketball.

Table II. Difference between US population and US NHL, NBA, MLB, and PGA players across cities of different sizes.

NHL NBA MLB PGA

a b a b a b a
City size Diff. s z Diff. s z Diff. s z Diff. sb z

45,000,000 79.3 2.4 73.9* 76.0 1.4 74.2* 77.9 1.0 77.9* 79.5 2.1 74.6*
2,500,000 – 4,999,999 78.4 2.5 73.3* 74.8 1.5 73.1 78.4 1.0 78.2* 710.0 2.2 74.6*
1,000,000 – 2,499,999 714.7 3.2 74.6* 711.8 1.9 76.0* 715.1 1.3 711.9* 717.5 2.7 76.6*
500,000 – 999,999 75.4 2.7 72.0 70.5 1.6 70.3 75.4 1.1 75.0* 70.9 2.2 70.4
250,000 – 499,999 1.6 2.6 0.6 4.5 1.5 3.0 2.0 1.1 1.9 5.8 2.2 2.7
100,000 – 249,999 7.9 2.4 3.2* 6.4 1.4 4.5* 8.4 1.0 8.4* 3.5 2.1 1.7
50,000 – 99,999 16.2 0.8 20.0* 9.7 0.5 19.9* 15.9 0.3 48.3* 10.1 0.7 14.8*
550,000 13.1 3.5 3.7* 1.8 2.1 0.8 11.6 1.5 7.9* 19.7 3.1 6.4*
a
Difference between percent of US population and percent of professional athletes in each city size (from Table 1).
b
Standard deviations determined by Monte Carlo simulations.
*z-score significant at P 5 0.001.

Table III. Canadian population, Canadian NHL players, and odd ratios across different city sizes.

City size Canadian populationa (%) % NHLb Diff. sc z OR (CI)

4500,000 33.2 15.7 717.5 2.0 78.7* 0.37 (0.38, 0.37)

100,000 – 499,999 13.3 33.2 19.9 1.5 13.6* 3.24 (3.24, 3.24)
30,000 – 99,999 7.6 15.8 8.2 1.1 7.3* 2.28 (2.28, 2.28)
10,000 – 29,999 7.3 10.4 3.1 1.1 2.8 1.47 (1.48, 1.47)
5000 – 9999 3.4 7.7 4.3 0.8 5.6* 2.37 (2.37, 2.37)
2500 – 4999 3.4 6.0 2.6 0.8 3.4* 1.81 (1.82, 1.81)
1000 – 2499 3.3 6.2 2.9 0.8 3.8* 1.94 (1.94, 1.93)
51000 28.5 5.1 723.4 1.9 712.2* 0.13 (0.15, 0.12)
a
Percentage of males under the age of 14 in each of the subdivisions of the 1976 Canadian census.
b
Percentage of players in 2002 – 2003 born in each of the subdivisions of the Canadian census.
c
Standard deviations determined by Monte Carlo simulation.
*z-score significant at P 5 0.001.
Abbreviations: OR ¼ odds ratio, CI ¼ confidence interval.
1070 J. Côté et al.

Table IV. Chi-square values and related probabilities between observed frequencies and expected frequencies.

Number and % of players per quarter

Q1 (%) Q2 (%) Q3 (%) Q4 (%) Total w2 P

US hockey 52 (34.4) 33 (21.9) 37 (24.5) 29 (19.2) 151 8.02 0.045*

US basketball 118 (27.7) 102 (23.4) 119 (27.3) 97 (22.2) 436 3.43 0.330
US baseball 276 (30.4) 231 (25.5) 200 (22.1) 200 (22.1) 907 17.09 0.001*
US golf 55 (27.9) 45 (22.8) 48 (24.4) 49 (24.9) 197 1.07 0.784
Canadian hockey 162 (29.5) 164 (29.9) 118 (21.5) 105 (19.1) 549 19.95 0.000*

Table V. Cohen’s d effect sizes for birthplace effect and relative age
Effect size effect across all sports.
Effect sizes (Cohen’s d) of the relative age effect and
Effect size (d)
the birthplace effect are presented in Table V. The
birthplace effect sizes varied between 1.98 for Birthplace Relative age
Canadian hockey to 4.22 for US hockey. The relative
US hockey 4.22 0.74
age effect sizes for Canadian hockey and US hockey US basketball 4.16 0.04
were the largest at 0.90 and 0.74, respectively. The US baseball 3.94 0.44
relative age effect sizes for US baseball, golf, and US golf 3.27 0.09
basketball were all under 0.5. Canadian hockey 1.98 0.90
Means (all sports) 3.51 0.44

Interaction between birthplace and relative age

Results of the correlation between the players’ the NBA, 15% of the players in MLB, and 13% of
birth months (according to their specific sport’s players in the PGA. The Canadian ice hockey
calendar year) and the players’ birth city size data corroborate the US data, showing that cities
show low and non-significant correlations for basket- with populations larger than 500,000 are under-
ball (r ¼ 70.02), baseball (r ¼ 0.04), US hockey represented in terms of players in the NHL. The
(r ¼ 0.02), Canadian hockey (r ¼ 0.04), and golf Canadian data suggest that there is also a critical
(r ¼ 70.01). Chi-square results for the ratios of the value for minimum community size (i.e. 41000).
players born in small cities (less than 500,000) and The smallest rural areas (51000) produced signifi-
big cities (500,000 or more) for each of the sport- cantly fewer professional players than expected. The
year’s quarters showed no significant values for any 1976 Canadian census characterizes populated areas
of the sports calculated. Overall, these results show of less than 1000 as lacking the infrastructures that
that the effect of birthdates and birthplace are are usually common to a city, which may result in
independent of each other. fewer opportunities to invest in physical activities and
sports. The lack of facilities and playing partners,
typical in small rural areas, appears to be detrimental
Discussion
to the early development of sport skills (Curtis &
The results of this study provide strong support for Birch, 1987). This minimum populated areas effect
the view that smaller cities provide early opportu- was not seen in the US data and may be due to the
nities for talent development in sport that are not 1980 US census not providing the number of people
matched by larger cities. This study also provides living in rural areas in the USA. Therefore, based on
additional evidence for relative age as a factor that the collective Canadian and US data, it appears that
can affect elite performance in ice hockey and the optimal community size for the early develop-
baseball, although no support was provided for a ment of professional athletes is greater than 1000 but
relative age effect in PGA golfers and NBA basketball less than 500,000. Furthermore, the US census
players. subdivision allows us to show that the odds ratio of
The US data indicate that children born in becoming a professional athlete increase linearly as
communities of 500,000 or more are significantly the population decreased, indicating better odds in
disadvantaged in terms of their likelihood of becom- cities that have a population smaller than 250,000.
ing an elite athlete compared with children from On theoretical grounds, we can posit that the
communities of less than 500,000. While nearly 52% physical environment and psychosocial climate of big
of the US population reside in cities with populations urban centers and smaller cities (excluding rural
over 500,000, such cities produce approximately areas of less than 1000) are different and accordingly
13% of the players in the NHL, 29% of the players in provide children with different sport experiences at
 Birthplace effects 1071

an early age. From a physical environment pers- cities may facilitate the achievement of elite sport
pective, smaller communities provide children with performance by providing a more supportive
more space for physical activities such as cycling, and facilitative psychosocial environment for early
running, skating, and playing sports with peers development.
(Kytta, 2002). Moreover, Kytta showed that smaller The results of this study support the relative age
communities provide a natural environment that is effect in ice hockey in both Canada and the USA as
safer for children to move around independently, well as US baseball; however, no relative age effect
without adults’ supervision. On the other hand, was found in golf and basketball. The results from
physical activity for young children in larger com- hockey and baseball replicate past studies that have
munities is more structured, taking place largely also noted a relative age effect with NHL and MLB
within the school system and organized leagues players (Boucher & Mutimer, 1994; Thompson
(Kristjansdottir & Vilhjalmsson, 2001). Such orga- et al., 1991). As noted by Musch and Grondin
nized sport programmes require a lot of human (2001) in their review of studies of the relative age
resources, such as parental involvement, adult super- effect in sport, it is likely that the causes of the
vision, and coaching, that could limit the time relative age effect in sports such as hockey and
children spend playing sports. Smaller cities’ less baseball are multiple, including physical, cognitive,
structured, more natural, more spacious, and safer and social elements. The findings for basketball are
physical environment might facilitate various types of consistent with the results of Daniel and Janssen
sport involvement and longer hours of involvement (1987), who also found no relative age effect with
in sports at a young age, a characteristic that has been NBA players for the 1984 – 85 season. On the other
associated with later investment in sport (Baker et al., hand, Hoare (2000) reported a relative age effect
2003; Soberlak & Côté, 2003). with professional male basketball players in Australia.
When discussing the geographical nature of The fact that the high school developmental sys-
modern sport, Bale (2003) argues that many small tem for basketball in the USA has a ‘‘grade fail
American towns use their sport success to proclaim exemption’’ could partially explain the absence of a
their pride. This is often reflected on the welcoming relative age effect in US basketball. The ‘‘grade fail
signs of US towns and cities highlighting local exemption’’ allows three players per team to play with
sporting heroes (Bale, 2003). Although bigger cities players of the same grade if they fail a grade. This
offer more alternatives for children to engage in exemption allows older players to play with younger
organized sports (Curtis & McPherson, 1987) and players and could eventually eliminate the relative
other structured leisure activities such as arts and age effect in professional basketball. Studies of
music, a study of leisure-time use in adolescents relative age with primary school elite basketball
demonstrated that urban adolescents reported de- players could shed light on the function that the
creased amounts of satisfaction with their leisure- ‘‘grade fail exemption’’ rule plays in reducing the
time use compared with their rural counterparts magnitude of the relative age effect in US basketball.
(Gordon & Caltabiano, 1996). Therefore, the re- No evidence of a relative age effect in golf was found
duced number of leisure activities and the for the US PGA players. Golf is a sport where age-
psychosocial environment of smaller cities could be related factors such as size and body mass might be
important factors for future investment in sports. less likely to influence performance. In addition, the
Sport participation studies suggest that organized structure of youth golf in the USA does not have as
sport programmes have often been unsuccessful in strict age groupings as other organized youth sports,
retaining children because of the psychosocial including ice hockey and baseball. Finally, the
environment in which sport is practised (De Knop, beginning of organized play in golf is also likely to
Engström, & Skirstad, 1996; Kristjansdottir & occur at an older age, which would limit the amount
Vilhjalmsson, 2001). Individuals from smaller com- of time that golfers can benefit from a possible relative
munities receive more social support, have higher age advantage.
self-efficacy, and experience less conflict with Table V displays the Cohen’s d effect sizes for
others than those from larger cities (Elgar, Arlett, & birthplace and relative age across all sports. It is
Groves, 2003). These psychosocial characteristics noteworthy that the birthplace effect, which has to
have been associated with sustained involvement in date received little attention, is considerably and
sport (Pelletier, Fortier, Vallerand, & Brière, 2001; consistently stronger than the well-documented
Robinson & Carron, 1982). It is probable that the relative age effect. The birthplace effect shows effect
more intimate and informal environment of smaller sizes that are consistent across sports and that are
cities is more conducive to experiencing early well above the 0.80 value that Cohen (1988)
success, which, in turn, intrinsically drives propen- suggested as constituting a ‘‘large’’ effect. On the
sity for more training. As a result, during the other hand, the effect sizes for the relative age effect
early years of a child’s involvement in sport, smaller are not consistent across sports, with large effects for
1072 J. Côté et al.

Canadian hockey, medium effects for US hockey and cities. The smaller cities could offer increased
baseball, and small effects for US basketball and golf. opportunities to experience early success in sport,
Despite these varying effect sizes across sports, it is which in turn would increase self-efficacy and the
generally considered that the relative age effect has motivational drive to play and practise more. In other
a strong influence on talent development in sport words, smaller cities might present more opportu-
(e.g. Helsen, et al., 2000; Musch & Grondin, 2001). nities for the type of developmental experiences
Based on the results of this study, birthplace has a and practice known to be associated with expert
considerably stronger influence on talent develop- performance.
ment than relative age in major US and Canadian The birthplace effect found in this study is
sports. However, the birthplace effect requires exami- strikingly consistent across sports (baseball, basket-
nation in other countries and other sports to establish ball, ice hockey, golf), countries (Canada and USA),
how generalizable this effect is. and sport development systems. On the other hand,
The possible interactive relationship between although a relative age effect was found for hockey
birthdate (relative age) and birthplace of the athletes and baseball, the same effect was not observed across
was examined in our sample using Pearson correla- all sports. This study showed that birthplace is a
tions and chi-squares analyses. These analyses were much stronger and consistent factor than relative age
conducted to determine if birthdate and birthplace in determining elite performance. Our data are
were associated. Correlations between a sport-year’s limited by possible discrepancies between registered
birth months and birthplace were uniformly low for place of birth and the location where athletes
every sport analysed. To further examine the possible spent the bulk of their developing years. However,
relationship between birthdate and birthplace, we US census data show a relative stability in population
assessed whether the ratios of players born in small growth between cities of different sizes (Schachter,
cities and big cities were evenly distributed between Franklin, & Perry, 2003), supporting the use of
the different quarters of a sport-year. Results of this ‘‘birthplace’’ as a reliable indicator of where athletes
analysis show a symmetrical distribution of players spend their early years in sport. The birthplace
from small cities and big cities born in the different effect found in this study reinforces the conclusion
quarters of their sport-year. Overall, these analyses that contextual factors play a significant role in
show that birthplace has an effect on athletes’ determining who achieves the highest level of
development that is independent of the relative age sporting skill.
effect, and presumably therefore has different origins
of causation.
Acknowledgements
Financial support for this work was provided by the
Conclusion
Social Sciences and Humanities Research Council of
Our results show that children who live in smaller Canada (SSHRC grant #410-02-0325 and grant
cities have more opportunities for the development #410-04-1207). We are grateful to Bill Pearce for his
of sport expertise. Drawing on the existing evidence help with the statistical analyses.
about factors known to be important to expert
development in sports, we can propose possible
factors and mechanisms that contribute to the
birthplace effect (Ericsson et al., 1993; Côté et al., References
2003). The effect could be mainly due to skill Baker, J., Côté, J., & Abernethy, B. (2003). Sport specific training,
acquisition factors related to the quality and quantity deliberate practice and the development of expertise in
of play and practice afforded by the physical team ball sports. Journal of Applied Sport Psychology, 15, 12 –
25.
environment of smaller cities. The quality of play Bale, J. (2003). Sports geography (2nd edn.). New York: Routledge.
and practice could be a key factor because the Barnsley, R. H., & Thompson, A. H. (1988). Birthdate and
physical environment of smaller cities is more success in minor hockey: The key to the NHL. Canadian
conducive to unstructured play activities between Journal of Behavioral Science, 20, 167 – 176.
children and adults of different ages and to experi- Barnsley, R. H., Thompson, A. H., & Legault, P. (1992). Family
planning: Football style. The relative age effect in football.
mentation with various forms of sporting activities. International Review for Sociology of Sport, 27, 77 – 87.
The quantity of play and practice in smaller cities Baxter-Jones, A., & Helms, P. (1994). Born too late to win.
could also be a factor because smaller cities present Nature, 370, 186.
fewer safety concerns, better access to open spaces, Bloom, B. S. (1985). Developing talent in young people. New York:
and less competing sources of leisure-time use by Ballantine.
Boucher, J. L., & Mutimer, B. T. P. (1994). The relative age
children. Another factor that could contribute to the phenomenon in sport: A replication and extension with ice-
birthplace effect is the more intimate and, possibly, hockey players. Research Quarterly for Exercise and Sport, 65,
less competitive psychosocial environment of smaller 377 – 381.
 Birthplace effects 1073

Carlson, R. C. (1988). The socialization of elite tennis players in Kalinowski, A. G. (1985). The development of Olympic swim-
Sweden: An analysis of the players’ backgrounds and develop- mers. In B. S. Bloom (Ed.), Developing talent in young people
ment. Sociology of Sport Journal, 5, 241 – 256. (pp. 139 – 192). New York: Ballantine.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences Kristjansdottir, G., & Vilhjalmsson, R. (2001). Sociodemographic
(2nd edn.). Hillsdale, NJ: Lawrence Erlbaum. differences in patterns of sedentary and physically active
Côté, J. (1999). The influence of the family in the development of behaviour in older children and adolescents. Acta Paediatrica,
talent in sports. The Sports Psychologist, 13, 395 – 417. 90, 429 – 435.
Côté, J., Baker, J., & Abernethy, B. (2003). From play to practice: Kytta, M. (2002). Affordances of children’s environments in the
A developmental framework for the acquisition of expertise in context of cities, small towns, suburbs, and rural villages in
team sports. In J. Starkes & K. A. Ericsson (Eds.), Expert Finland and Belarus. Journal of Environmental Psychology, 22,
performance in sports: Advances in research on sport expertise 109 – 123.
(pp. 89 – 110). Champaign, IL: Human Kinetics. Monsaas, J. A. (1985). Learning to be a world-class tennis player.
Curtis, J. E., & Birch, J. S. (1987). Size of community of origin In B. S. Bloom (Ed.), Developing talent in young people
and recruitment to professional and Olympic hockey in North (pp. 211 – 269). New York: Ballantine.
America. Sociology of Sport Journal, 4, 229 – 244. Mooney, C. Z., & Duval, R. D. (1993). Bootstrapping: A nonpara-
Curtis, J. E., & McPherson, B. D. (1987). Regional differences in metric approach to statistical inference. Thousand Oaks, CA: Sage.
leisure physical activity of Canadians: Testing some alternative Musch, J., & Grondin, S. (2001). Unequal competition as an
interpretations. Sociology of Sport Journal, 4, 363 – 375. impediment to personal development: A review of the relative
Daniel, T. E., & Janssen, C. T. L. (1987). More on the relative age age effect in sport. Developmental Review, 21, 147 – 167.
effect. Canadian Association for Health, Physical Education, and Pelletier, L. G., Fortier, M. S., Vallerand, R. J., & Brière, N. M.
Recreation, 53, 21 – 24. (2001). Associations among perceived autonomy, support,
De Knop, P., Engström, L.-M., & Skirstad, B. (1996). Worldwide forms of self-regulation, and persistence: A prospective study.
trends in youth sport. In P. De Knop, L.-M. Engström, Motivation and Emotion, 25, 279 – 306.
B. Skirstad, & M. Weiss (Eds.), Worldwide trends in youth sport. Press, W. H., Flannery, B. P., Teukolsky, S. A., & Vetterling,
(pp. 276 – 281). Champaign, IL: Human Kinetics. W. T. (1986). Numerical recipes. New York: Cambridge
Dundink, A. (1994). Birthdate and sporting success. Nature, 368, University Press.
592. Robinson, T. T., & Carron, A. V. (1982). Personal and situational
Elgar, F. J., Arlett, C., & Groves, R. (2003). Stress, coping, and factors associated with dropping out versus maintaining
behavioural problems among rural and urban adolescents. participation in competitive sport. Journal of Sport Psychology,
Journal of Adolescence, 26, 574 – 585. 4, 364 – 378.
Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The Schachter, J. S., Franklin, R. S., & Perry, M. J. (2003). Migration
role of deliberate practice in the acquisition of expert and geographic mobility in metropolitan and nonmetropolitan America:
performance. Psychological Review, 100(3), 363 – 406. 1995 – 2000 (Census 2000 special reports). Washington, DC: US
Gordon, W. R., & Caltabiano, M. L. (1996). Urban – rural Department of Commerce.
differences in adolescent self-esteem, leisure boredom, and Soberlak, P., & Côté, J. (2003). The developmental activities of
sensation-seeking as predictors of leisure-time usage and elite ice hockey players. Journal of Applied Sport Psychology, 15,
satisfaction. Adolescence, 31, 883 – 901. 41 – 49.
Helsen, W. F., Hodges, N. J., Van Winckel, J., & Starkes, J. Statistics Canada (1979). Census of Canada 1976, Vol. 2 Population:
(2000). The roles of talent, physical precocity and practice in Demographic Characteristics. Ottawa: Statistics Canada.
the development of soccer expertise. Journal of Sports Sciences, Thompson, A. H., Barnsley, R. H., & Steblelsky, G. (1991).
18, 727 – 736. ‘‘Born to play ball’’: The relative age effect and Major League
Hoare, D. (2000, September). Birthdate and basketball success: Baseball. Sociology of Sport Journal, 8, 146 – 151.
Is there a relative age effect? Paper presented at the 2000 Pre- US Bureau of the Census (1981). Base data source, Tables 1 and 10:
Olympic Congress, Brisbane, Australia. Census of Population and Housing, 1980: Summary Tape File
Hoyle, R. H. (1999). Statistical strategies for small sample research. 1C [machine-readable data file]. Washington, DC: The
Thousand Oaks, CA: Sage. Bureau.