Gender Effect on the Relationship between Talent Identification Tests and Later World Triathlon Series Performance

ground: We examined the explanatory power of the Spanish triathlon talent identi ca- tion (TID) tests for later World Triathlon Series (WTS)-level racing performance as a function of gender. Methods: Youth TID (100 m and 1000 m swimming and 400 m and 1000 m running) test performance times for when they were 14–19 years old, and WTS performance data up tothe end of 2017,were obtained for 29 female and 24 male “successful” Spanish triathletes. The relationships between the athletes' test performances and their later best WTS ranking positions and performance times were modeled using multiple linear regression. Results: The swimming and running TID test data had greater explanatory power for best WTS ranking in the females and for best WTS position in the males (R 2 a = 0.34 and 0.37, respectively,p 0.009). The swimming TID times were better related to later race performance than were the running TID times. The predictive power of the TID tests for WTS performance was, however, low, irrespective

data had greater explanatory power for best WTS ranking in the females and for best WTS position in the males (R 2 a = 0.34 and 0.37, respectively,p 0.009). The swimming TID times were better related to later race performance than were the running TID times. The predictive power of the TID tests for WTS performance was, however, low, irrespective of exercise mode and athlete gender. Conclusions: These results con rm that triathlon TID tests should not be based solely on swimming and running performance. Moreover, the predictive value of the individual tests within the Spanish TID battery is gender speci c. Keywords:elite; testing; prediction; triathlete; talent; gender 1. Introduction Triathlons involve sequential swimming, cycling and running. Only athletes with around a top 150 world ranking may compete in the World Triathlon Series (WTS), i.e., the highest level of competition below the Olympic Games. The annual WTS circuit involves up to nine races over the Olympic (OD) (1.5 km swim, 40 km bike, 10 km run) and Sprint (0.75 km swim, 20 km bike and 5 km run) distances, plus a (more highly scored) Grand Final over the OD. The nal WTS season ranking equates to a world championship ranking. Climatic conditions permitting, at any given event, both sexes compete over the same race distances. Within-competition analyses [1–3] have, however, demonstrated that gender differences exist in the relative importance of individual swimming, cycling and running performance to the overall race result in elite triathlon. The running discipline makes the most decisive contribution to nishing position in World Cups [3,4], World Championship/WTS Grand Finals and the Olympic Games, but more so for males. Swimming performance affects nal race position less [1,3]. Vleck et al. [4] reported World Cup performance to relate both to average swimming speed and position at the swim exit. Slower swimmers must reduce their time gap to the leading bike pack(s) by the run start. Swimming speed more strongly affects nishing position in females, however [1,3]. Because of their differences in speed and performance density, cy- cling performance can become more important in females [2]. Females form more,

both to average swimming speed and position at the swim exit. Slower swimmers must reduce their time gap to the leading bike pack(s) by the run start. Swimming speed more strongly affects nishing position in females, however [1,3]. Because of their differences in speed and performance density, cy- cling performance can become more important in females [2]. Females form more, smaller, cycle packs and are likely less able to bridge gaps between such packs [3]. Sports2021,9, 164.

Sports2021,9, 164 2 of 11 Pacing strategy also affects elite performance. In both genders, speed up to the rst buoy of a one-lap World Cup swim was associated with nishing position. The top 50% of males swam this faster than the bottom 50%. Thereafter, swimming speeds were similar [3]. Elites also reportedly adopt a positive or a reverse J-shaped pacing strategy out of the bike–run transition (T2), running faster within the rst kilometer and, when faced with direct opponents, over the nal 400 m or less of the run [3–6]. Because they generally exit T2 in larger groups, and have similar 10 km times, this ability to do an “end spurt” may prove to be especially important in males. To date, only one race analysis [2] has exclusively focused on WTS events. It examined the relative in uence of the three triathlon disciplines on WTS performance, across two Olympic cycles, in 1670 males and 1706 females. Competitors were grouped by nishing position (G1: 1st–3rd place; G2: 4th–8th place; G3: 9th–16th place and G4: 17th place). The main effects of years and rank groups were compared. For females, swim and bike segment differences existed only between G4 and the other groups (p= 0.001–0.029). Each group differed from the other for the run (p< 0.001). For males, swimming performance differed only between G4 and the other groups (p= 0.001–0.039). Although running was where differences existed between all the groups (p< 0.001), it was apparently important for success that a good runner be positioned with the rst cycling pack. Bike splits did not differ, however, between the different male groups, for whom “the bike leg seemed to be a smooth transition towards running” [2]. Only the rst 16 women had similar bike splits, however. Even at the WTS level, females likely divide up into more bike groups, further apart, and may therefore be more affected by residual fatigue at the run start than males [2]. These gender differences in the relative extent to which performance in its component disciplines in uences triathlon performance [2,3] may have important consequences for tal- ent identi cation

however. Even at the WTS level, females likely divide up into more bike groups, further apart, and may therefore be more affected by residual fatigue at the run start than males [2]. These gender differences in the relative extent to which performance in its component disciplines in uences triathlon performance [2,3] may have important consequences for tal- ent identi cation (ID). The extent to which talent ID test results relate to adult performance is used to justify the resources that are allocated to it. This then can markedly impact the selected athletes' likelihood of sporting success. To date, however, most triathlon talent ID research has either involved mixed-gender groups or just males. “While. . .we know very little about predictors of talent in elite sport, we know even less about predicting talent in female athletes. Given the often-unique development systems for high-performance female athletes, this discrepancy might limit our ability to gain a deeper understanding of talent, and . . . lead to potentially harmful consequences for the female athlete population” [7]. We do not know the extent to which the individual predictive capacities of the tests within the Spanish Triathlon Federation (FETRI) battery differ with gender. Although they were previously reported to have little predictive capacity [8] (in that case for National Championship-level performance), the study sample was of mixed gender. The extent to which FETRI talent ID test results separately relate to male and female WTS performance is unknown. Therefore, the aim of this study was to explore, as a function of gender, the extent to which the FETRI talent ID test results predicted the later WTS performance of Spanish triathletes. 2. Materials and Methods 2.1. Study Design An explanatory transverse study design was used to establish the relationship between the FETRI talent ID test results (Table), and later WTS performance, of both males and females (Table). Two independent analyses were carried out: (1) for best end-season WTS ranking (RankWTS) and (2) for best WTS individual event position (PositionWTS). 2.2. Participants Our subjects were considered “the successful products” of the FETRI talent ID process because they either obtained

relationship between the FETRI talent ID test results (Table), and later WTS performance, of both males and females (Table). Two independent analyses were carried out: (1) for best end-season WTS ranking (RankWTS) and (2) for best WTS individual event position (PositionWTS). 2.2. Participants Our subjects were considered “the successful products” of the FETRI talent ID process because they either obtained a nal WTS ranking or raced at the WTS level in 2009–2017.

Sports2021,9, 164 3 of 11 Table 1. FETRI scoring system for performance of the individual components of the Spanish triathlon ID test battery. Age Points Females Males R400 R1000 S100 S100 R400 R1000 S100 S1000 14 years 3 1:19.0 3:55.0 1:17.0 14:10.0 1:07.0 3:14.0 1:09.0 13:30.0 10 1:12.0 3:20.0 1:10.0 13:00.0 1:00.0 2:46.0 1:02.0 12:20.0 12 1:10.0 3:10.0 1:08.0 12:40.0 0:58.0 2:40.0 1:00.0 12:00.0 16–17 years 2 1:19.0 3:55.0 1:17.0 14:10.0 1:07.0 3:14.0 1:09.0 13:30.0 10 1:11.0 3:15.0 1:09.0 12:50.0 0:59.0 2:43.0 1:01.0 12:10.0 11 1:10.0 3:10.0 1:08.0 12:40.0 0:58.0 2:40.0 1:00.0 12:00.0 18–19 years 1 1:19.0 3:55.0 1:17.0 14:10.0 1:07.0 3:14.0 1:09.0 13:30.0 10 1:10.0 3:10.0 1:08.0 12:40.0 0:58.0 2:40.0 1:00.0 12:00.0 Athlete's age was determined by their age on 31 December in the year of the tests. Performance times are given in mm:ss.s. S100: 100 m freestyle swimming test, S1000: 1000 m freestyle swimming test. R400: 400 m running test; R1000: 1000 m running test. Table 2. Performance times achieved within the FETRI talent identi cation test battery; best sea- sonal WTS rankings, best seasonal nishing positions within an individual WTS event achieved by “successful” Spanish triathletes (mean SD). Talent ID Test Performances Females Males N Time (min:ss.s s) 4R (%) N Time (min:ss.s s) 4R (%) S100 29 01:05.82 2.56 93.28 4.15 24 01:00.81 3.52 90.04 5.30 S1000 29 12:53.07 42.01 89.64 16.46 24 12:27.38 55.54 89.38 6.71 R400 28 01:10.96 3.99 88.78 7.41 22 00:58.51 2.92 92.95 4.27 R1000 29 03:17.72 9.64 93.90 5.80 23 02:46.26 9.98 93.57 4.23 WTS Results Females Males N Mean SD Min-Max N Mean SD Min-Max Rank 26 100 44 44–159 24 83 53 1–165 Position 21 36 14 18–58 19 24 18 1–55 N: number of triathletes; F: females, M: males, 4R: percentage of the best ever times within the talent identi cation test, S100: 100 m freestyle swimming test, S1000: 1000 m freestyle swimming test. R400: 400 m running test; R1000; 1000 m running test. WTS: World Triathlon Series; Rank: best ranking position obtained by the triathletes within the WTS (i.e., RankWTS), Position: best position obtained by the triathlete

males, 4R: percentage of the best ever times within the talent identi cation test, S100: 100 m freestyle swimming test, S1000: 1000 m freestyle swimming test. R400: 400 m running test; R1000; 1000 m running test. WTS: World Triathlon Series; Rank: best ranking position obtained by the triathletes within the WTS (i.e., RankWTS), Position: best position obtained by the triathlete within an individual WTS race (i.e., PositionWTS). 2.3. Procedures From 2009 to 2016, 3502 fourteen to nineteen year-olds underwent the FETRI test battery. This comprised two freestyle swimming tests (i.e., S100: a 100 m time trial; S1000: a 1000 m time trial) in a 25 m pool, and two running track tests (i.e., R400: a 400 m time trial; R1000: a 1000 m time trial). Each test performance time (in seconds, T) was scored, up to a maximum of 12 points, using a proprietary FETRI age- and sex-speci c scale (Table) [8–10]. Those who scored 8 or more points in each of at least three tests were then separated into age- and gender-speci c subgroups. One-year age groups, as opposed to category (e.g., “junior” or “cadet”) groupings, were used to offset relative age effect(s) [11]. Each individual's total test performance time was then expressed as a percentage of the fastest ever summated four (swim and run) test times for their age subgroup (variable 4R). The World Triathlon results database (see was then used to identify the “successful” 24 males and 29 females to which this study pertains before their talent ID data were obtained from FETRI. The research protocol was both in accordance with the Declaration of Helsinki and approved by the local University Ethics committee. 2.4. Statistical Analysis Suf cient sample size was calculated using G * Power v3.1.9.4 for Windows (Heinrich- Heine-Universität Düsseldorf, GER), resulting in an N of 48 being considered appropriate

Sports2021,9, 164 4 of 11 (effect size = 0.36; error probability = 0.05; power = 0.95). Sample normality, linearity and homoscedasticity were assumed after carrying out the Kolmogorov–Smirnov test. Pearson's bivariate correlation coef cient was used to determine the inter-relationships between test times. The relationships between the successful athletes' talent ID test results and their RankWTS and PositionWTS data were modeled using step-by-step multiple linear regression. The degree of data independence was calculated using the Durbin– Watson test (and assuming independence of values between 1.5 and 2.5). Variance in ation factor (VIF) values above 10 were taken to indicate multicollinearity. The 95% con dence level was considered statistically signi cant. All the analyses were performed with the Statistics Package for the Social Sciences (SPSS version 19.0 for Windows, SPSS Inc., Chicago, IL, USA). 3. Results The athletes' swimming and running test times (Table) were positively intercorre- lated in the males. The correlation coef cients were large for between S100 and S1000 (r = 0.853,p= 0.001), and moderate for between S100 and R400 (r = 0.431,p= 0.045), S100 and R1000 (r = 0.431,p= 0.045), S1000 and R1000 (r = 0.552,p= 0.006) and R1000 and R400 (r = 0.742,p= 0.001). In the females, only R400 and R1000 (r = 0.836,p= 0.001), and S100 and S1000 (r = 0.750,p= 0.001) were signi cantly intercorrelated. Table 3. Summary of the linear regression models for the best seasonal WTS ranking position and best seasonal nishing positions within an individual WTS event achieved by “successful” Spanish triathletes. WTS Performance Predictors Predictors R 2 R 2 a R Error Sig D-W Rank All 4R R1000, T S100, 4R S1000 0.346 0.303 0.58840.7100.0001.16 F 4R R1000, TS100, 4R S10000.415 0.336 0.64535.4790.0071.73 M T R400, 4R S1000 0.391 0.326 0.62544.8630.0091.56 Position All 4R R1000, T S100, T S1000 0.415 0.365 0.64413.5770.0002.10 F T S1000, 4R S100 0.342 0.268 0.58411.9770.0231.13 M T S1000, T S100 0.442 0.372 0.66414.3510.0091.69 R: Multiple linear regression; R 2 : R Square; R 2 a : adjusted R2; Error: standard error; D-W: Durbin–Watson test. F: females, M: males, 4R: percentage of the best ever times

0.62544.8630.0091.56 Position All 4R R1000, T S100, T S1000 0.415 0.365 0.64413.5770.0002.10 F T S1000, 4R S100 0.342 0.268 0.58411.9770.0231.13 M T S1000, T S100 0.442 0.372 0.66414.3510.0091.69 R: Multiple linear regression; R 2 : R Square; R 2 a : adjusted R2; Error: standard error; D-W: Durbin–Watson test. F: females, M: males, 4R: percentage of the best ever times within the talent identi cation test, S100: 100 m freestyle swimming test, S1000: 1000 m freestyle swimming test. R400: 400 m running test; R1000; 1000 m running test. WTS: World Triathlon Series; Rank: best ranking position obtained by the triathletes within the WTS(i.e., RankWTS), Position: best position obtained by the triathlete within an individual WTS race (i.e., PositionWTS). Table ID test results and both RankWTS and PositionWTS. For the values obtained with the Durbin–Watson test (with the exception of “all cases” in the response to RankWTS and “females” in the response to PositionWTS) independence of the residuals was assumed. As no VIF values exceeded 3.5, multicollinearity was not considered to be a problem. The females' talent ID test results best explained their best WTS ranking. According to the value of the adjusted coef cient of determination (R 2 a ) (p 0.007), 33.6% of the total variance in best female ranking at the end of the season was explained by 4RR1000, TS100 and 4RS1000. The regression equation was: Female RankWTS = 960.306 + 7.060 4RS1000 2.852 4RR1000+ 10.285 TS100(1) The males' talent ID test results, however, better explained best individual WTS race position than best male season-end WTS rankings. The corresponding R 2 a (p 0.009) indicated that 37.2% of the total variance in male PositionWTS was explained by S100 and S1000 performance times: Male PositionWTS = 101.692 0.315 TS1000+ 5.933 TS100 (2)

Sports2021,9, 164 5 of 11 4. Discussion Few data relating to the accuracy of early talent decisions exist [7]. “High-quality scienti c research is needed in order to (a) determine the reliability and validity of tal- ent identi cation and selection initiatives, (b) inform evidence-based models of athlete development, and (c) identify gaps in current understanding and directions for future work. Ineffective or inaccurate decisions have important repercussions for all stakeholders involved (e.g., dropout, decreased motivation, misplaced resources, and investment)” [12]. Baker et al. [12] stated that it is “imperative to better understand factors related to female-speci c talent development.” Although their review of the talent-related literature indicated over thirty such triathlon studies to have taken place thus far, we believe this to be the rst one to examine the accuracy of talent decisions for expert male and female triathletes. FETRI test performance poorly predicted WTS performance in both genders. In our “successful” females, talent ID results explained 33.6% of the variance in best end- season WTS ranking (i.e., the more important of the two variables) and 26.8% of the variance in best individual WTS race placing. In “successful” males, the corresponding values were 32.6% and 37.2%. In our results, when both genders were analyzed together (Table), the explanatory power of the tests dropped (from 33.65% in females and 32.6% in males) to 30.3% overall for best end-season WTS ranking and to 36.5% for best individual WTS event position. This is both unsurprising, given that the constraints and developmental models of females differ from those of males, and con rms that the predictive capacity of the battery FETRI talent ID test is gender speci c. The explanatory power of theindividualFETRI tests for best WTS performance also differed with gender. Again, this nding, given the gender differences in the relative importance of performance within each triathlon discipline that exists at the WTS level, was expected, since the “disciplines that precede the triathlon run appear to have more impact on overall race performance in females than they do in males. In males, where the performance density is better, the ability to complete

this nding, given the gender differences in the relative importance of performance within each triathlon discipline that exists at the WTS level, was expected, since the “disciplines that precede the triathlon run appear to have more impact on overall race performance in females than they do in males. In males, where the performance density is better, the ability to complete a fast, sprint type, run nish can be de nitive” [2]. However, we did not set out to predict WTS performanceper se.Rather, we explored how much of the variance in male and female WTS performance could be explained by performance in each of the FETRI swim and run tests. The prognostic validity of these predictors for draft-legal OD triathlon performance is uncon rmed, nor are the optimal pacing strategies within the WTS competition yet known. However, the (44 race) analysis that was conducted by Piacentini et al. [2] found differences in swim times, bike times and run times between podium (G1), 4th and 8th place (G2), 9th and 16th place (G3), and 17th place (G4) female WTS nishers. Within males, these differences occurred only for swimming and running. No difference in swimming segment times was noted, in both sexes, between the rst three such groups. It was clearly important to overall WTS performance that good runners were able to position themselves within the rst cycling packs to reach T2. Piacentini's study population would have included our males and females, classing them as G1–G4 and G4 triathletes, respectively. In males, therefore, we expected to see signi cant relationships between the swim and run FETRI test results and performance. Piacentini et al. [2] observed that for males, exiting the water and exiting T2 close to the leader, with a fast running split, appeared to be major determinants of success. In females, both the T1 and T2 exits were important, as was a very fast run split. In males, G1 also differed from G2 to G4 as regards entry into T1 and exit from T2. Entry into T1 was less important than exit from T2, and run sprinting ability was likely more important,

This study explores how talent identification tests predict triathlon performance based on gender.