Study population and sports league characteristics

We analyzed the dates of birth of 38,381 youth sports participants aged 9–14 years: 13,467 girls (34.9%) and 24,914 boys (65.1%). This included a total of 37 sports: 36 sports for boys and 34 for girls. Participants trained during the week and had competitions on weekends. In team sports, the competitions ran in a league format. Individual sports had competitions along the season, with a final championship at the end of the season. All sports had regional levels. Most sports were organized in two-year age bands; however, in sports with many participants, some teams were based on one-year age bands.

Football was divided in outdoor and indoor football for both sexes. Multisport was a discipline in which athletes participated in a wide variety of sports in the same season one after another, including handball, basketball, football, athletics and swimming.

Most sports had only one league, although a few sports with a large number of participants had more than one league. For boys, outdoor football had three leagues: 1^st league, 2^nd league, and 3^rd league (named “rendimiento”, “competición”, and “participación”, respectively). The 1^st league was for 13–14-year-olds only.

In basketball, in both sexes there was a unique league for 9–11-year-olds, and older ages had two leagues: 1^st league and 2^nd league (“rendimiento” and “competición”, respectively). These leagues were defined by competitiveness of the teams, with the 1^st league being the most competitive and the 2^nd and 3^rd leagues the least competitive. Basque pelota had four leagues for boys: A (least competitive), B, C, and D (most competitive). The complete list of sports and number of participants by age group are shown in S1 Table for boys and S2 Table for girls.

Due to the low numbers of boys participating in some sports [i.e., rhythmic gymnastics (n = 27), artistic gymnastics (n = 15), trampolining (n = 15), artistic skating (n = 13), aerobics (n = 1), and synchronized swimming (n = 1)], these sports were considered together as a single group titled “gymnastics” (n = 72).

The Ethics Committee of the University of the Basque Country (UPV/EHU) for Research on Human Subjects approved this study. All data were anonymized for the research study.

All sports participants in Bizkaia county must have a license, for which they must supply basic information including date of birth. Thus, we used these data to compare the dates of birth of sports participants to the distribution of general births occurring in Bizkaia in the same year using information from the Eustat Institute for Statistics [32]. Participant dates of birth were first analyzed according to the month of birth. The cut-off date for all sports was January 1. Thus, the year was divided into four quarters for all sports (Q1: January 1–March 31, Q2: April 1–June 30, Q3: July 1–September 30, and Q4: October 1– December 31).

Statistical analyses

A χ² goodness-of-fit test was used to assess differences between the distributions of observed and expected birth dates. The distribution of expected birth dates was based on the distribution of live births in Bizkaia in the same year [32], following previous studies [13]. The effect size Cramer’s V was measured, considering 0.06, 0.17, and 0.29 as small, medium, and large effect sizes (ES), respectively [33]. Standardized residuals (SR) were determined, with a value of ≥1.96 indicating an overrepresentation and a value of ≤-1.96 indicating an underrepresentation of participants compared to the general population (p < 0.05) [28].

Odds ratios (OR) and 95% confidence intervals were calculated to determine the odds of a boy/girl born in January playing in the highest leagues of football (boys) or basketball (girls) compared to the odds of a boy/girl born in each remaining month.

To analyze age as a continuous variable [34], the number of days between the day of birth and December 31 was calculated. This new variable was named “relative age”, and the medians and interquartile ranges (IQR; 25^th percentile–75^th percentile) were calculated. Relative age was compared among the different leagues as well as different age groups using Wilcoxon and Mann–Whitney U tests.

To give a wide picture of the RAE in all youth competitive sports, we compared the dates of births of all athletes with the actual births in the complete sample (Table 1 for boys, Table 4 for girls). Then analysis was performed within each age group (S3–S8 Tables for boys, S9–S14 Tables for girls), of which statistically significant differences are shown in Table 2 for boys and Table 5 for girls. Additionally, comparison of the median age between leagues and age groups of boys participating in football is shown in Table 3.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics of the birth distribution of all male sport participants (9–14 years old) and the general population.

https://doi.org/10.1371/journal.pone.0254687.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics of the birth distribution of male sport participants divided by age group (only statistically significant results are shown).

https://doi.org/10.1371/journal.pone.0254687.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Relative age of male football participants according to age group and league [medians and interquartile ranges (25th– 75th percentiles)].

https://doi.org/10.1371/journal.pone.0254687.t003

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Descriptive statistics of the birth distribution of female sport participants and the general population.

https://doi.org/10.1371/journal.pone.0254687.t004

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Descriptive statistics of the birth distribution of female sport participants divided by age group (only statistically significant results are shown).

https://doi.org/10.1371/journal.pone.0254687.t005

Statistical analysis was performed using IBM SPSS statistics software (v 22.0). The level of significance was set at p < 0.05.

Analysis of the RAE in grassroots male athletes

In the sport-by-sport analysis, the RAE was almost exclusively present for males in football. In addition, in all males in Basque pelota and handball, there was a small effect RAE with no significant standardized residuals and a large effect in volleyball (Q1) that could not be corroborated by year analysis. In contrast, in two large studies undertaken in the London Youth Games [30] and the Swiss Talent Development Program RAE was confirmed in most analyzed sports [31]. These differences among the studies may be due to the selection processes for sport participation, as the participants of the two last-mentioned studies were either selected to represent their school [30] or were part of a national talent program [31]. The selection hypothesis is the most mentioned cause for the RAE [1–16]. Interestingly, our study could not corroborate the presence of a clear RAE in other sports with more than one league (basketball and Basque pelota), which would a priori involve selection of players. Therefore, it is reasonable to think that the underlying reasons for the RAE not only encompass the selection processes of athletes, but are more complex and probably multifactorial, involving a combination of factors in addition to the selection processes.

As mentioned before, physical growth and better functional capacities of relatively older children have been claimed to be the main causes of the RAE [35, 36], particularly in sports in which a large body size and strength are advantageous [1, 2]. Supporting this notion, studies on sports with weight categories [17–20] have not detected the RAE, similar to our results. However, our study did not observe RAE in other sports in which physicality is relevant, such as basketball, handball, or rugby. Therefore, physical attributes and related performance indicators are undeniable factors, but other contributors seem to be just as necessary for RAE to occur.

In this regard, our results indicate that the popularity of a sport (football) is a strong moderator of the RAE, as demonstrated in culturally popular sports in other countries, including rugby in the UK [7], ice hockey in Canada [37], skiing in Austria [12], and handball in Germany [8]. Participation in these sports was lower in our study, 2.5% and 1% of the male participants for handball and rugby, respectively, and therefore it could be assumed that there would not be selection constraints for boys willing to participate in these sports and as a consequence, the birth dates of their players were evenly distributed, and thus no RAE was detected. In addition, particularly in team sports, due to the limited number of spaces on each team, coaches must select a certain number of players and the depth of competition (as represented by the number of participants and number of players selected for a higher competition) may be an additional moderator [2, 8, 30].

Maturity should also be considered when exploring the RAE, specifically around puberty [38, 39]. A closer look at the performance football league demonstrates that the oldest relative age players were those in the 12-year-old group, where 42% were born in Q1, but this percentage decreased in subsequent years. The drop the in relative age may be due to the influence of maturation on the selection of players around puberty, which may be more important than the relative age at other ages [39]. Thus, advanced maturity status may offset some of the disadvantages associated with being younger [40].

Analysis of the RAE in grassroots female athletes

We found some similarities and discrepancies in the distribution of birth dates of females compared to males. While the RAE almost exclusively occurred in male football participants, particularly at the highest performance levels, in female participants it occurred in three of the four most popular sports (basketball, handball, and football), all of which are team sports. Similarly, RAE has been described in female basketball [6] and football [25] players as young as 7 and 8 years old, respectively. Nonetheless, we did not detect the RAE in female participants in rhythmic gymnastics, the second most popular sport, in agreement with results from Van Rosum [41] in ballet dancers and Baker et al. [42] in figure skaters. Rhythmic gymnastics is an individual sport (no need for coaches and technical staff to identify the best athletes) and an aesthetic sport, where technical staff search for females with small, lean, thin bodies and good flexibility and coordination. Therefore, not merely popularity but its combination with depth of competition and also the physicality of these sports may be responsible for the RAE. Further, other related factors also may be associated.

In the yearly analysis by sport, RAE occurred in two distinct scenarios. On one hand, this phenomenon was linked to basketball, when this sport started to be organized in two leagues (i.e., 12-year-old 1^st league). This indicates that popularity of the sport and the selection processes are responsible for the biased distribution of players’ birth dates, similar to what happens with males. On the other hand, a large effect size RAE was observed in female 9-year-old basketball players and also in 11- and 13-year-old handball players. These results are interesting because they cannot be attributed to coach selection or depth of competition. Instead, it is possible that social influences might be involved. One possibility is that parents do not encourage/enroll their relatively younger children in sports with high physical demands, as may be the case for handball, for fear of injury or failure [5]. Additionally, relatively younger players with smaller bodies and lower performances might eschew participation because they perceive that they are not as capable as older players. This effect is referred to as internal or self-selection [1, 2]. Additionally, in the most popular sports, younger players are more likely than older players to drop out, reinforcing presence of the RAE [35, 43, 44].

Further analysis: The case of chess and the inverse RAE

Chess is of particular interest because it is a mind sport, relying on perception and cognitive skills rather than physical strength [45]. Studies on this discipline are scarce, but the RAE has been described in expert adult chess players [46] and top junior male and female players [47]. Helsen et al. [45] confirmed the presence of the RAE in U8–U20 chess players in the Belgian Championship but failed to demonstrate it in non-elite U12–U20 chess players, supporting the uneven distribution of birth dates observed in the small group of male 13-year-old males in this study—which incidentally indicates an inverse RAE.

Athletes born in the last quarters of the year may take up other less popular sports, resulting in an overrepresentation of athletes born in the last months before the cut-off date, known as inverse RAE [20, 21, 29], observed in the present study in male athletics, traditional sport, chess, taekwondo, and gymnastics across 11-, 12-, 13-, and 14-year-old groups, respectively; which partially corroborates our hypothesis. This inverse RAE has been described in 15–17 years old male French shooters [20] and female table tennis players, fencers and snowboarders of a Swiss talent development program [29]. These sporting activities require high technical skills or aesthetics for performance [29] and physical attributes may not be that important [20], thus relatively younger, smaller and less strong athletes may have an advantage and are more likely to participate and be selected [29]. Moreover, in sports with less attendance, all children who are willing to participate have the opportunity to do so, independent of their relative age, while in more popular sports, there are some obstacles that are difficult for relatively younger players to overcome.

General implications of the RAE

The overrepresentation of relatively older children involved in the RAE is not the major concern—what is worrisome is the significant and systematic underrepresentation of relatively young participants. If children are not able to participate and enjoy sports from a young age, it will be more difficult for them to take up a sport in adolescence and even harder in adulthood. While birth date was not a strong constraint to participation in most sports in Bizkaia, there were 360 fewer boys and 207 fewer girls born in the last quarter of the year participating in sports relative to the numbers expected from births in the entire population. This underrepresentation was especially concerning in some particular sport contexts, and the implications of these findings are twofold. First, and most importantly, there is a clear discrimination against relatively young children solely on the basis of having been born in certain months of the year. As an illustration, a boy born in January was 1.8-times more likely than a boy born in December to play in the competition football league, and this difference was even larger (3.2-times) in the performance league. The odds of a girl born in January playing basketball were significantly larger than those of a girl born later than July. Therefore, special effort should be made to ensure that all children have the same opportunities to participate in a sport, particularly at young ages (i.e., 9 years old). Second, as talent is born homogeneously across the year, talent may be missed each time a relatively young athlete is not selected, discarded, or drops out of a sport.

Measures to mitigate the RAE

In the literature, several recommendations have been proposed to avoid the RAE. Our results suggest the RAE occurs because of a combination of different factors and is closely linked to a particular sports context, so all steps should be taken accordingly. In the most popular sports, selection processes should be avoided at the grassroots level and delayed as much as possible, preferentially until post-maturation [1, 48]. If selection procedures are indispensable, they should be improved to prevent the RAE [48]. In this line, changing and rotating the cutoff date has been proposed as a strategy to remove particular selection time points and give all athletes the possibility of being relatively older at some point [1, 2, 4, 26]. Thus, it has been proposed a rotation of the cut-off dates by 3 months between seasons of competition, to ensure players have experience in each quartile position [49]. However, despite being a largely mentioned solution its implementation would be rather complicated in the actual sporting competitions. Also, “birth-day banding”, where athletes move up to their next birthdate group on their birthday, has been successfully applied in the England squash Talent Pathway [50].

Since the origin of the RAE lies on the date of birth of the children and youth, some authors [2, 4] have recommended a different classification system based on the biological age of the participants instead of the chronological age, similar to bio-banding. In order to equalize the advantages of maturity, in bio-banding athletes are grouped according to their maturity status [51]. In younger immature children, classification could be based on body height, weight and size.

Organizing leagues and championships among children of a similar age (i.e., teams made up of children born a maximum of six months apart, or less if the magnitude of the sport allows it) could be possible for very popular sports [50]. Also, different squads may be established for those athletes who are technically at high standard, but who are lacking physical development, in order to give them a better chance for fair competition [2]. Further, a mean compulsory age could be implemented so that coaches equally incorporate relatively younger and older players to teams to provide similar opportunities [1].

Additionally, providing scouts who are responsible for finding talent with information about the relative age of the players being observed helps to avoid the RAE on player selection [1]. Mann et al. (2016) demonstrated that the selection bias was eliminated when scouts were provided with information about the relative age of the players and players wore shirts with numbers corresponding to their relative age while training [52].

In individual sports in which performance is measured by numeric data (e.g., time, distance), individual data can be compared using corrective adjustment procedures, as proposed in athletics [52, 53] and swimming [15, 54]. Interestingly, Abbot et al. [15] found RAEs in 12–14 years old Australian female 100m and 200m Breaststroke swimmers which increased at the highest levels. However, when corrective adjustments procedures were applied the relative age inequalities were removed. Further, these corrective adjustments could also be used for data related to other types of physical condition tests used in team sports, for instance (e.g., yo-yo intermittent test).

Technical staff of the clubs, coaches, and scouts must be aware of relative age and its negative effects [55]. Coaches should avoid instant progress and short time success [56]; instead, they should adopt a long-term vision, giving more emphasis to technical and tactical competences [5, 48] rather than strength, speed, power, and body size, which are tightly related to the growth process.

Further, other factors beyond coaches’ selection seem to be involved in the RAE, so solutions should extend in a broader context as well. Hancock et al. [5] observed a significant overrepresentation of children born in Q1 and underrepresentation in Q4 in children aged 5 years old engaged in ice hockey and suggested that parents might contribute to these early RAEs. As parents may be involved in the enrollment of athletes in sports, they should be provided with appropriate information and education on the sports. To our knowledge no studies have analysed the reasons and the influence of parental decisions to enroll their children in one sport or another on the RAEs, which would be of interest [5].

This study shows that, in addition to all the above measures, it is important to offer an extensive range of sports with various characteristics to promote participation of children and adolescents in physical activity, allowing all children the opportunity to find their preferred sport.

More studies are still necessary to diminish the deleterious effect of relative age. Particularly, programs implementing the proposed measures are crucial. Moreover, broad longitudinal studies should be conducted to discover shifts of athletes between sports in relation to their relative age. In addition, the physical, physiological, social, and psychological mechanisms involved in the continuation and drop-out of relatively young participants should be examined. In this vein, it is interesting the concept of the Relative Age Effect Reversal, which describes the higher sporting level achieved by the relatively younger athletes. The players born at the end of the year and that tended to be neglected in the younger categories, being less often selected for teams, arrive at the adult category with advantageous attributes, rather than the players born in other months [57]. In fact, once a relatively younger player overcomes the selection process and is recruited to play in a high-level club, the chances to become professional increase [55]. Further, relatively younger professional athletes play more matches [58] and have better performances [59–61], higher salaries [62], and longer careers [58–63], which can be explained by a psychological benefit of the relatively younger competing against the relatively older an bigger peers [59], the “underdog hypothesis” [64, 65] and the superior ability [59] and genuine talent of the relatively younger who draw the scouts’ attention to be selected [55].

One limitation of this study is that we examined RAE in a county with a specific sport organization, which is not necessarily a reflection of other places. Nevertheless, sports often are organized in a similar way within Spain and other European countries. In addition, we compared the birth dates of the athletes to the total birth dates in the same county, which does not account for immigrants and emigrants that may alter the distribution of birth dates. However, the impact of this should be low. Thirdly, it is possible that a number of children participated in ballet, dance, and other non-competitive sports in private gyms and would not be included in the present study. Therefore, the conclusions of this study must be considered with this in mind.