Relative age effects in Japanese athletes: A 25-year historical analysis

Background: The mechanisms underlying the relative age effect (RAE), a biased distribution of birth dates, in sport events have been investigated for more than two decades. The present study comprised an historical analysis involving the most recent quarter-century (1993-2018) on RAEs among Japanese male professional athletes (soccer, baseball, basketball, and volleyball) to clarify how the RAEs changed over time. Methods: Birth data were obtained from 7,805 Japanese male professional athletes registered in 1993, 2001, 2010, and 2018. The athletes were divided into four groups based on their month of birth: quartiles Q1 (April–June), Q2 (July–September), Q3 (October–December), and Q4 (January–March of the following year). In addition, based on the data in 1993 for soccer and baseball and in 2010 for basketball and volleyball, the expected numbers of players were calculated in 2001, 2010, and 2018 for soccer and baseball, and 2018 for basketball and volleyball. Results: Significant RAEs were observed among soccer and baseball players in 1993, 2001, 2010, and 2018, and strong tendencies of RAEs were found among basketball and volleyball players in 2010 and 2018. The magnitudes of the RAEs in soccer, baseball, and volleyball decreased over time, but not in basketball. Conclusion: The exact reasons for the decreasing or unchanging RAEs among these professional players remains unclear, but socio-culture factors, such as low birthrates and the popularity of sports in Japan, might be related to the changing RAEs.

to achieve sporting success, compared with relatively younger children. Within the same age category, there can be a difference of almost a full year between the oldest and youngest children.
In addition to physiological and psychological factors, the competition principle has also been considered as an important factor affecting RAEs. According to Musch & Grondin [16], 'Competition will come from the number of players available for the places, and this number will depend on the popularity of a given sport in a given country (p. 154)'. Thus, the level of competition is associated with the popularity of the sport. For example, in Japan, soccer, baseball, basketball, and volleyball are very popular among male elementary and junior high school students [17]; however, handball, rugby, badminton, and American football are not so popular. Indeed, Nakata et al. [11] showed significant RAEs among Japanese players in soccer, baseball, basketball, and volleyball, which were major sports in Japan, but no significant RAEs were observed in handball, rugby, badminton, American football, or golf. Consequently, if a sport does not need a physical advantage and is not so popular in a given country, RAEs may not be observed. In addition, it takes several years or decades for a sport to gain popularity in a given country. Therefore, historical analysis is needed to know the beginning of RAEs in a country and compare differences in the magnitude of RAEs among generations, and to be considered based on socio-cultural factors. To date, there have been several studies examining RAEs from a historical perspective [18][19][20][21][22]. For example, Nakata & Sakamoto [22] investigated the existence of RAEs in Japanese professional baseball players born in 1911-1980. They reported that significant RAEs were observed among Japanese professional baseball players born in the 1910s and onward, and the magnitude of RAEs increased with time. These studies suggest that the magnitude of media coverage may have markedly contributed to this. However, after a thorough literature search, we do not know of any study conducting historical analysis of recent generations of RAEs in a given country, rather than the beginning of RAEs [18][19][20][21][22].
Moreover, previous historical studies only examined RAEs on one sporting event. We consider that RAEs on some sports should be evaluated simultaneously in a given country to clarify the sociocultural factors relating to RAEs. In fact, demographics have clearly changed compared with 50 years ago. For example, in Japan, the number of children is markedly decreasing (Ministry of Internal Affairs and Communications, the Statistics Bureau and the Director-General for Policy Planning of Japan: https://www.stat.go.jp/data/jinsui/), and recent RAEs on some sports might change over time. Based on this research background, the objective of the present study was to investigate the characteristics of RAEs over a recent quarter-century (1993-2018) among professional soccer, baseball, basketball, and volleyball players simultaneously in Japan. Japan has applied a unique annual-age grouping for education since 1886, basing group assignment on student' birthdates between April 1 (the "new" year) and March 31 of the following year among elementary, junior high, senior high, and university (college) students and in government and company employment. Sports calendars also follow this system, giving Japanese children and adolescents born between April and June a relative age advantage over those born between January and March. We investigated whether the magnitude of RAEs in Japanese professional soccer, baseball, basketball, and volleyball, which are major maledominated sports in Japan, changed over time.

Method Samples
Data on professional Japanese soccer players (N = 3,490 males) who played in the Japan Professional  [27][28][29][30]. Data on professional Japanese basketball players (N = 586 males) who played in the Japan Professional Basketball League (B-league) registered in 2010 and 2018 were extracted from an official publication [31] and a previous study [11]. Data on professional Japanese volleyball players (N = 633 males) who played in the Japan Volleyball League (V-league) registered in 2010 and 2018 were extracted from an official website (https://www.vleague.jp/) and a previous study [11]. Each league is the highest level in Japan.

Data analysis
Professional players were divided into four groups based on their month of birth: Q1 (April-June), Q2 (July-September), Q3 (October-December), and Q4 (January-March of the following year). Foreign players in the Japanese professional leagues were excluded because they had not passed through the Japanese school system. Chi-squared tests were applied to each group according to the four quarters to assess the significance of deviations from the expected number of births in each quarter. In line with previous studies [22,32,33], the expected distribution was calculated based on the general population of males in national birth statistics from 1962-1999 in Japan (Ministry of Internal Affairs and Communications, the Statistics Bureau and the Director-General for Policy Planning of Japan).
According to a previous study [11], the age range for calculating the general population was set at 12 years. For example, the age of players registered in 1993 were considered to be 18 to 30 years old, and we calculated the general population born between 1962 (i.e., 30 years old) and 1974 (i.e., Table 1). The effect size of Chi-square tests was also calculated in each group. To measure the effect size, Phi (φ) was defined as:

years old) (Supplementary
where n = the number of observations. The odds ratio (OR) and 95% confidence intervals were then calculated to provide additional information for both quartile and half-year distributions, following a previous review article [15]. The ORs for the Q1, Q2, and Q3 vs. Q4 comparisons were interpreted as follows: OR < 1.22, 1.22 ≤ OR < 1.86, 1.86 ≤ OR < 3.00, and OR ≥ 3.00, indicating negligible, small, medium, and large effects, respectively [3,34], which was recommended by sample size calculation [35]. In addition, based on Statistical tests were performed using computer software (SPSS for windows ver. 22.0, SPSS).
Significance was set at p < 0.05, and the p-values are shown as p < 0.05, p < 0.01, and p < 0.001. Table 1 presents the results of chi-square tests for each sport and year. Significant RAEs of soccer and baseball were observed in 1993, 2001, 2010, and 2018. The effect size (Phi) in both sports gradually decreased over time. No significant RAEs of basketball and volleyball were observed at any ages, but the percentage of Q1 was the highest, and that of Q4 was the lowest. The effect size in basketball increased slightly over time, while that in volleyball decreased over time. The number in the second row is the predicted number of players obtained using the chi-square test. n = number of players; X 2 = Chi squared value; * p < 0.001.      [38] reported a significant selection bias for the scouts of youth soccer in a no-age information group, and that bias remained when scouts knew the players'

Results
dates-of-birth. The selection bias was eliminated when scouts watched the games knowing the shirt numbers corresponded to the relative ages of the players. However, with regard to the decrease in RAEs among professional players in Japan, it is unlikely that efforts to reduce the RAEs were made nationwide. Therefore, other factors need to be considered.
One of the possibilities is that the popularity of sports has decreased annually. Table 4 shows changes in the number of youth baseball teams, players, and players per team. Data on baseball were from official websites of the Nippon Junior High School Physical Culture Association (http://njpa.sakura.ne.jp) and Japan High School Baseball Foundation (http://www.jhbf.or.jp/). Since data from 1993 were not available, we used data from 1994. As a result, the number of youth players per team at junior high school clearly decreased in 2018. The numbers of teams at high school (hardball and rubber ball) were almost the same for 25 years. Therefore, we considered that the decrease in the OR for professional baseball players was partly related to the decrease in the popularity of baseball, especially in junior players. In contrast, youth baseball players who played at junior high school might tend to continue to play at high school. Katsumata et al. [3], focusing on Japanese youth baseball players in recreational levels at elementary school, junior high school, and high school, showed that significant RAEs were noted in recreational junior high school and high school players, but not in elementary school players, and the effect size became larger with increasing grade (0.063 in elementary school players, 0.151 in junior high school players, and 0.227 in high school players). Their data suggest that relatively younger players may gradually drop out from playing baseball even as recreational players. Delorme et al. [5] reported that the relative age was associated with sport dropout. They showed that relatively younger players in categories of 9-10 years old, 11-12 years old, and 13-14 years old tended to drop out from basketball, compared with relatively older players. Indeed, the present study did not directly elucidate the relationships between these backgrounds and effect sizes in professional baseball players; further studies are necessary to clarify the detailed mechanisms. As a limitation of the present study, our finding might be specific to Japan, because many activities, sports-related or academic, are based on a unique cutoff date (April 1st), which is not the case in other countries. Data on RAEs with an historical analysis should be examined in other countries.

Conclusion
The present study comprised an historical analysis to clarify the characteristics of RAEs among Japanese professional soccer, baseball, basketball, and volleyball players over time. Significant RAEs were observed among soccer and baseball players in 1993, 2001, 2010, and 2018, and strong tends of RAEs were found among basketball and volleyball players in 2010 and 2018. In addition, the magnitudes of RAEs among soccer, baseball, and volleyball decreased over time, but those of basketball did not. The exact reasons for these remains unclear, but socio-culture factors, such as low birthrates and the popularity of sports in Japan, might be related to the changing RAEs.
Abbreviations RAE relative age effect Declarations Declarations Nao Sasano, Yoichi Katsumata, and Hiroki Nakata declare that they have no competing interests.

Ethical Approval and Consent
Not applicable

Consent for publication
No individual person's data.

Availability of data and materials
Not applicable

Competing Interests
The authors declare that they have no competing interests.

Funding
This study was supported by a Japan Society for the Promotion of Science KAKENHI Grant-in-Aid for Scientific Research C (19K11576) (to H. Nakata).