Relative age effect: beyond the youth phenomenon.

INTRODUCTION: Initially described in a sports context in ice hockey in 1985, the relative age effect (RAE) refers to the performance advantages of youth born in the first quarter of the birth year when trying-out for select, age-restricted sports. The competitive advantage bestowed to the relatively older athlete in their age band is the result of the older athlete being more physically and emotionally mature. These more mature players will likely go on to be exposed to better coaching, competition, teammates and facilities in their respective sport.
OBJECTIVES: Our study sought to characterise the ubiquity of this effect by examining the birth distribution of some of the world's most elite athletes, Olympians.
METHODS: We extended the exploration of the RAE beyond specific sports by examining the birth quarter of over 44 000 Olympic athlete's birthdates, born between 1964 and 1996. Our hypothesis was that the RAE would be prominent in both Olympic athletes as a whole and in selected subcategories of athletes. RESULTS AND
CONCLUSION: The fractions of births in the first versus the fourth quarter were significantly different (p<0.001) from each other for the summer and winter Olympians, ball and non-ball sports, and team as well as individual sports. This significant difference was not gender specific. We found the general existence of the RAE in Olympic athletes regardless of global classification. Our findings suggest that coaching staff should be cognisant of the RAE when working with young athletes and should take relative age into consideration when evaluating a burgeoning athlete's abilities. © Author(s) (or their employer(s)) 2020. Re-use permitted under CC BY. Published by BMJ.

INTRODUCTION

Multiple literary sources have provided evidence there are secondary factors, which have an indirect effect on the success of an athlete.[1] Helsen et al in 2000 demonstrated a positive linear relationship between excellence on the field and practice, showing there are more factors that contribute to athletic success than solely genetics.[2] One of these indirect effects is called the relative age effect (RAE), and multiple literature sources demonstrate this effect in sports such as soccer and hockey.[1 3 4]

In an attempt to equalise competition among players in youth sports, governing bodies have applied cut-off dates to groups of players according to their age, which appear to be arbitrary and without an international standard.[4] It has been proposed that athletes born earlier in the year, soon after the beginning cut-off date, would have an indirect advantage over those born just before the ending date when trying to advance to elite status in their sport. The postulation is that the individual born ‘early’ in the defined birth year is older (ie, more mature) than the other athletes born later in their age cohort. When the try-out process for select teams favours the more mature athlete, a substantial number of players lose access to more advanced play and training because they are less mature, not necessarily because they have less athletic potential. The participation and performance advantage of players born early in the birth year is collectively known as the ‘relative age effect’.

Initial descriptions of RAE were not limited to sports. Grondin in 1984 and Barnsley in 1985 discussed a relationship between relative age and scholastic achievement.[5-7] This pattern soon became apparent in the sports world as well in Canadian hockey in 1988.[8] The most accepted hypothesis for the RAE is that older children will be more physically and emotionally mature than their relatively younger counterparts early in their athletic career, and this early success would ultimately give the older athlete access to a better environment (eg, coaching, competition) for developing his/her skills, which would ultimately result in greater athletic achievement long-term.[1 9]

Barnesley found in 1988 that hockey players born in the first half of the year played minor league hockey to a later age than players born in the first half of the year. The RAE was further demonstrated by his finding that nearly 70% of the minor league hockey players in his study were born in the first half of the year (ie, January through June). The RAE has been demonstrated in professional baseball, American football, the first round of the NHL draft, professional soccer and professional basketball.[1 3 4 10–13]

In a 2009 meta-analysis, it was demonstrated that the RAE exists in numerous team and ball sports.[14] This paper examined 42 separate papers.[14] One provided evidence of the RAE in a non-ball sport (ie, gymnastics), while two provided evidence of the RAE in individual sports (eg, gymnastics and golf).[4] We propose the RAE is not only present in a few select sports but can be evident in all sports and levels of play.

At the top of the sport hierarchy is the Olympic movement that welcomes International Olympic Committee member nations to send their elite to compete in dozens of sports. The massive number of athletes who have participated in the Olympics since their inception in 1896 present a valuable opportunity to evaluate the widespread effect of the RAE at an elite level. Our study sought to ascertain if the RAE was present at the highest level of sports, and to characterise that RAE in terms of its appearance or absence in ball sports, team or individual sports, and across gender and season of sport. We hypothesised all Olympic sports would demonstrate this RAE. Thus, we proposed more Olympians would be born in the first quarter of the year (ie, January through March) than in the last quarter of the year (ie, October through December). Additionally, we hypothesised this effect would be present across different divisions within the Olympic Games (ie, the summer and winter games), gender and sport classification (team and individual; ball and non-ball sport).

MATERIAL AND METHODS

This study was approved by our Institutional Review Board. Data were collected from the International Olympic Committee’s database on Olympic athletes. The data range from 1896 to 1996 and compiles a total of 112 152 athletes and their corresponding birthdates. Each athlete was entered into our database according to their first Olympic participation and their respective birthdate, gender and sport were recorded. Olympians were recorded into our database as only participating in their first Olympics and for one event; participation in subsequent Olympiads or multiple events were not recorded. Our database included every registered Olympian dating back to the first Olympics in 1896 through 1996. Of note, the exact historical date of when children began to be grouped by age cohort for sports competition is not well documented and is likely different for most sports. Furthermore, there is no guarantee age cut-off dates were used for sports participation in 1896. However in 1971, Barrow and McGee[15] demonstrated that separating children by age for sports competition helped to facilitate instruction, promote programme continuity, and promote safety. Given this data and the fact more athletes were competing with each Olympiad, we did not include every athlete’s birthdate, as it likely would have skewed our sample. As a result, we included the last 30 years of our sample data and had a total of 44 087 birthdates in our data set. This population was comprised of 27 372 male and 16 761 female Olympians. Of the total study population, 36 030 and 8057 Olympians competed in the summer games and winter games, respectively. Olympians competing in ball sports totalled 11 411 and non-ball sports totalled 32 676. Whereas, team sports totalled 10 169 and individual sports totalled 33 918.

For the purpose of this study, we defined the birth year as beginning on 1 January, as was used by the majority of the sports documented in the literature.[11 12 14–16] Some literature divides birthdates into the first and the last half of the year.[11] However, the majority separates and compares the birthdates by quartiles.[14] Thus, we used quartiles for our athlete’s birthdate evaluation. The first, second, third and fourth quartiles were composed of athletes born between 1 January through 31 March, 1 April through 30 June, 1 July through 30 September, and 1 October through 31 December, respectively. Although quartiles are equally distributed by month, they are not equally distributed by day when using a standard 365.25-day calendar (0.25 day allows for leap year athletes), as was done in this study. Because each quartile would not be equally distributed by day, we calculated the percentage of the year each quartile represented, which was as follows: first quartile—24.2%; second quartile—24.4%; third quartile—25.7%; and fourth quartile—25.7%. These percentages represented our control and what we expected our birthdate percentages to be if all athelete’s birthdates were equally distributed throughout the year. After division into quartiles, we got rid of all athletes born on the first of each month due to the arbitrary assignment of the first day of the month as the default birthdate in many lesser developed countries.

The data were summarised by birth quarter (eg, 1 January through 31 March) and presented as percentages and 95% CIs. The observed distribution was tested against the theoretical distribution (number days in each quarter/365.25) and tested using χ2 goodness of fit, where a p value ≤0.05 was considered to be significant (JMP, SAS Institute; Cary, NC, USA). Comparisons between birth quarters of significant distributions were made by overlapping CIs. The main comparison in this paper was made between the first quarter of the year and the last, which is consistent with the literature.

Patient and public involvement

Patients were not involved in any part of this study. They were not involved in the writing of the manuscript or in the acquisition of data and were not considered when examining this paper for readability or accuracy.

RESULTS

Our primary interest was the comparison of athletes born in the first quarter versus the last quarter of the year. For the entire dataset, every paired comparison of birth quarters was significantly different (p<0.001).

As demonstrated in table 1, participant births were not distributed equally throughout the year (p<0.001). The highest percentage of these elite athletes were born in the first quarter of the year with a decreasing number of athletes in each subsequent quarter of the year.

Table 1

Fractional distribution (%, upper/lower 95% CI) of birth month by category

			Q1			Q2			Q3			Q4
Category	n	Lower	%	Upper	Lower	%	Upper	Lower	%	Upper	Lower	%	Upper
Expected			24.2			24.4			25.7			25.7
Olympic	44 087	27.2	27.6	28	25.3	25.7	26.1	24.3	24.3	24.8	22.3	22.3	22.7
Summer	36 030	27.1	27.6	28.1	25.3	25.7	26.2	24.3	24.3	24.8	22.3	22.3	22.8
Winter	8057	26.4	27.4	28.3	24.9	25.8	26.8	23.7	24.6	25.6	21.3	22.2	23.1
Ball	11 411	27.4	28.2	29.1	25	25.7	26.6	23.7	24.5	25.3	20.8	21.6	22.3
Non-ball	32 676	26.9	27.4	27.9	25.3	25.7	26.2	23.9	24.3	24.8	22.1	22.6	23
Team	10 169	27.5	28.3	29.2	25	25.8	26.7	23.6	24.4	25.3	20.6	21.4	22.2
Individual	33 918	26.9	27.4	27.9	25.2	25.7	26.2	23.9	24.3	24.8	22.1	22.6	23

All distributions (Q1%, Q2%, Q3% and Q4%) were significantly different from expected (all p<0.001).

Table 2 further stratifies the categories presented in table 1 according to gender.

Table 2

Fractional distribution (%, upper/lower 95% CI) of birth month by gender and category

				Q1			Q2			Q3			Q4
Category	Gender	n	Lower	%	Upper	Lower	%	Upper	Lower	%	Upper	Lower	%	Upper
	Female	16 761	26.6	27.4	28.2	24.7	25.4	26.2	23.6	24.3	25.1	22.2	22.9	23.6
	Male	27 372	27.2	27.7	28.4	25.3	25.9	26.5	23.8	24.4	25	21.4	22	22.5
Summer	Female	13 782	26.9	27.8	28.7	24.5	25.4	26.2	23.5	24.4	25.3	21.7	22.5	23.3
Winter	Female	2979	23.7	25.5	27.3	23.9	25.8	27.6	22.4	24.1	25.9	22.9	24.6	26.4
Summer	Male	22 294	26.9	27.6	28.3	25.3	25.9	26.6	23.7	24.3	25	21.6	22.2	22.9
Winter	Male	5078	27.1	28.5	30	24.5	25.9	27.3	23.5	24.9	26.4	19.5	20.7	22
Ball	Female	4737	26.6	28.1	29.6	24.6	26.1	27.5	22.2	23.5	25	21	22.3	23.7
Non-ball	Female	12 024	26.2	27.1	28	24.3	25.2	26.1	23.7	24.6	25.5	22.2	23.1	24
Ball	Male	6720	27.1	28.3	29.6	24.3	25.5	26.7	23.9	25.1	26.4	19.9	21	22.1
Non-ball	Male	20 652	26.8	27.6	28.3	25.3	26	26.7	23.5	24.2	24.9	21.6	22.3	22.9
Individual	Female	12 527	26.3	27.2	28.1	24.3	25.1	26.1	23.8	24.7	25.6	22.2	23	23.8
Team	Female	4234	26.3	27.9	29.5	24.7	26.2	27.8	21.9	23.3	24.8	21.1	22.6	24
Individual	Male	21 437	26.8	27.5	28.2	25.3	26	26.7	23.5	24.2	24.8	21.7	22.3	23
Team	Male	5935	27.3	28.7	30	24.2	25.5	26.8	24	25.2	26.6	19.5	20.6	21.8

All distributions (Q1%, Q2%, Q3% and Q4%) were significantly different from expected (all p<0.001) except for Female Winter (p=0.3414).

The trend demonstrated in table 1 was also present in table 2, which controls for gender and type of sport (p<0.001). This implies the effects of the RAE likely apply to all athletes, rather than being limited to specific groups of people. However, the relationship was not statistically significant in female Winter Olympians (p>0.05).

DISCUSSION

Broadly, our analysis of Olympians’ birthdates demonstrates that the RAE exists at the most elite level and transcends gender and type of sport.

Team sports

Our data support current literature when considering team sports participating in the Summer and Winter Olympic Games. Our data also supported the literature as well for male and female team sports.[14] The RAE is not only present in our entire Olympic population but also present for both male and female Olympians.

Our data show 28.3% of the Olympic team sport athletes were born in the first quartile vs 21.4% who were born in the fourth quartile. Male and female team sports demonstrate a significant increase in births during the first quarter of the year compared to the final quarter of the year (28.7% vs 20.6%, and 27.9% vs 22.6%, respectively).

Ball and non-ball sports

Prior to 1997, the cut-off date for youth soccer in Belgium was 1 August, and after 1997 the cut-off date for youth soccer was 1 January.[4] Vaeyens et al demonstrated the presence of the RAE.[17] This study showed that prior to 1997, 29% of the national senior soccer team was born within the months of August through October.[17] After 1997, 32% of the national senior soccer team was born in the months of January through March.[17]

This literature represents team sports and sports that incorporate a ball. Our data for ball sports are consistent with previously established data, with 28.2% of our Olympic ball athletes born in the first quartile vs 21.6% born in the last quartile.[4 8 14] Additional examination demonstrated that 28.3% of our entire Olympic male ball and 28.1% of all Olympic female ball sport athletes were born in the first quartile, while only 21% of male and 22.3% of female ball sport athletes were born in the last quartile.

In non-ball sports, our data show that 27.4% of all the athletes, 27.6% of male athletes and 27.1% of female athletes were born between 1 January and 31 March, whereas 22.6% of all non-ball athletes, 22.3% of male athletes and 23.1% female were born between 1 October and 31 December. These data imply that an RAE exists, in general, for team sports. Furthermore, relatively older children for their age cohort may be more likely to be selected to better teams, glean better coaching and be challenged by better competition as a result.

Individual sports

Our individual sports data found nearly an identical trend with 27.4% of the athletes being born during the first 3 months of the year, which included 27.5% of male and 27.2% of female athletes. Additionally, 22.6% were born between 1 October and 31 December; of which, 22.3% were male and 23% were female. One important factor to consider with individual, non-ball sports, such as wrestling and swimming, is the presence of age cut-offs for competition. If an athlete is successful early on, they are more likely to continue with the sport, and it can be inferred from our data that children who are more physically and emotionally mature for their age are also more apt to have success in individual sports.

Olympians in general

Musch and Grondin discuss four mechanisms to explain the RAE, ‘competition as a necessary condition’, ‘physical development’, ‘psychological factors’ and ‘experience’. The first principle applies directly to the Olympic population, as a whole, as the larger the pool of potential athletes for a specific sport the stronger the RAE that should result.

A majority of the literature is focused on male competitors. A study examining French basketball players, aged 7–17 years, born in the first versus the second half of the year demonstrated the RAE in their athletes.[12] A study examining Brazilian female volleyball players, under the age of 14 years, showed 74% of their athletes were born in the first half of the year.[18] The Brazilian study dealt with prepubescent females and demonstrated the RAE. However, many other studies which have not demonstrated the RAE in female sports dealt with post-pubescent females, particularly soccer players.[13] Their hypothesis to explain this phenomenon was when a post-pubescent woman’s body habitus changes to a more endomorphic form, it can be disadvantageous for certain sports. We demonstrated the RAE in all female sports, including ball, non-ball, team and individual (table 2). The only exception was in regards to female winter Olympians, who did show statistically significant evidence of the RAE.

An explanation for this finding could be that female summer Olympic sports (eg, gymnastics, diving, swimming) are more conducive to pre-pubescent females than winter Olympic sports (eg, figure skating). When analysing the female Olympic pool as a whole, we find that there is a demonstrated RAE, which could mean that the RAE present in the summer and winter Olympic sports may be masking the lack of the RAE present in female winter Olympic sports (table 1). This delineation could be elucidated with further research of each summer and winter Olympic sport.

Our data do not differentiate between pre-pubescent and post-pubescent females, but does demonstrate a significant RAE between the first quartile (27.3%) compared to the last quartile (22.9%) among all female Olympians, including ball and non-ball female athletes, as well as female team and individual athletes. Although our study does not concern itself with any specific sport and determine the presence of the RAE for each one, it does provide insight into global questions that have not been reported in the literature with such a large study population. Other limitations include the majority of our background data is from primarily western nations, and therefore, our conclusions may be more well suited to addressing issues in those nations. We purposefully did not account for differences in sport season and hemispheric seasonal differences, as we hoped to provide a more global picture of how athletic training may be affected by issues outside of genetic ability and hard work. Further studies would target more specific geographical areas and could be tailored to more specific sports to account for some of the seasonal differences and cut-off dates unique to each nation and sport.

We demonstrate the RAE in individual and non-ball sports, as well as in an overall large population of female sports. All male sports, regardless of classification, demonstrate a statistically significant RAE. Further research is certainly warranted to elucidate the finer points concerning which specific sports demonstrate the RAE and which, if any, determine if sports may demonstrate a reverse RAE.

Our findings regarding the RAE at the highest level of athletic achievement may indicate that there may be more to an athlete’s success than hard work and a genetic predisposition.

We demonstrated the RAE in all but one subcategory of our study, female winter Olympians.

These findings can play an integral role in shaping how sports organisations rethink their classifications of young athletes based solely on rigid birthdates and age ranges to provide a more equitable playing field for all.