Abstract
urpose of this study was to investigate the multivariate pro le of different types of Brazilian runners and to identify the discriminant pattern of the distinct types of runners, as a runners' ability to self-classify well. The sample comprised 1235 Brazilian runners of both sexes (492 women; 743 men), with a mean age of 37.94 9.46 years. Individual characteristics were obtained through an online questionnaire: Sex, age, body height (m) and body mass (kg), socioeconomic status, and training information (i.e.,
the discriminant pattern of the distinct types of runners, as a runners' ability to self-classify well. The sample comprised 1235 Brazilian runners of both sexes (492 women; 743 men), with a mean age of 37.94 9.46 years. Individual characteristics were obtained through an online questionnaire: Sex, age, body height (m) and body mass (kg), socioeconomic status, and training information (i.e., self-classi cation, practice time, practice motivation, running pace, frequency and training volume/week). Multivariate analysis of variance was conducted by sex and the discriminant analysis was used to identify which among running pace, practice time, body mass index and volume/training could differentiate groups such as professional athletes, amateur athletes and recreational athletes. For both sexes, running pace was the variable that better discriminated the groups, followed by BMI and volume/week. The practice time is not a good indicator to differentiate runner's types. In both sexes, semi-professional runners were those that better self-classify themselves, with amateur runners presenting the highest classi cation error. This information can be used to guide the long-term training, athlete's selection programs, and to identify the strengths and weaknesses of athletes. Keywords:runners; discriminant analysis; performance 1. Introduction There is no one size ts all strategy to determine sport performance, given that performance differs across modalities and speci c abilities [1]. Performance is multi- factorial and the identi cation of variables that allow us to describe and differentiate an athlete's athletic ability poses a unique challenge [2]. Recently, interest in these variables has grown among amateur and non-professional athletes, especially in activities that are practiced on a large scale [3,4]. Over the past 10 years there has been a growth of 57% in the number of runners participating in marathon and endurance events, with a notable decrease of the gender gap of participants [5]. In 2019, a total of 459,029 marathon nishers were recorded, with most of the events held in the United States (61.6%), the United Kingdom (10.7%) and Canada (10.0%) [6]. In addition, considering 19,614,975 marathon results from 20082018 across the globe, there was an increase in the number of participants from India, Portugal and Ireland, while
gender gap of participants [5]. In 2019, a total of 459,029 marathon nishers were recorded, with most of the events held in the United States (61.6%), the United Kingdom (10.7%) and Canada (10.0%) [6]. In addition, considering 19,614,975 marathon results from 20082018 across the globe, there was an increase in the number of participants from India, Portugal and Ireland, while the most representative countries were the United Int. J. Environ. Res. Public Health2021,18, 4248.
Int. J. Environ. Res. Public Health2021,18, 4248 2 of 9 States (456,700), the United Kingdom (97,254) and Germany (86,032) [7]. It is interesting to note that the increase in participation has not resulted in faster marathon completion times. Over the last twenty years the average nish time has increased by approximately forty minutes [7]. However, this is a positive sign as it does indicate that more amateur, recreational and non-professional runners are participating in these endurance events. In the Brazilian context, there are approximately 4 million non-professional runners [8] training in various capacities to improve their performance in organised competitions and events [9]. However, there is considerable variation within this group of athletes, speci - cally relating to the amount of time spent training and overall running performance [10]. In this context, running pace is one of the main variables used to differentiate athletes at the various levels of competition [11]. Running pace is determined by the time taken to cover one kilometre or mile, and is expressed as time per distance covered (min/km or mile) [12]. At the international and elite level, this index has been used as a cut-off point to stratify athletes into different competitive levels during an event or to determine eligibility to participate in the competition (e.g., six majors marathon) [13]. Running pace determination is particularly effective when implemented as an expected running pace (calculated to achieve a speci c time) versus actual running pace (pace being achieved) in order to track performance whether in training or competition [14]. In addition to running pace calculations, various anthropometric components [15] and training variables [16] have been used to determine an athlete's competitive level. For example, Thuany et al. [10] found, in a study with Brazilian runners, that amateur runners who completed the highest volume and running frequency/week were four times more likely to produce a higher performance compared to recreational runners. In marathoners, the training velocity and body fat explained approximately 44% of the variance in perfor- mance [17]. Interestingly, in male half-marathoners, practice time, training volume, sum of skin folds and body mass index (BMI)
that amateur runners who completed the highest volume and running frequency/week were four times more likely to produce a higher performance compared to recreational runners. In marathoners, the training velocity and body fat explained approximately 44% of the variance in perfor- mance [17]. Interestingly, in male half-marathoners, practice time, training volume, sum of skin folds and body mass index (BMI) accounted for approximately 90% of performance variance [18]. Besides the relevance of these studies, generally, these attempts are usually focused on a univariate competitive pro le, leading to the presentation of different classi - cations to describe/classify runners, such as amateur runners, recreational runners, competitive level runners [9,10]. Considering the relationship between variables of different characteristics, information about multivariate predictors of performance and a discriminant analysis, encompassing classes of runners, is necessary. Therefore, the purposes of this study are (1) to describe the multivariate pro le of different types of Brazilian runners, (2) to identify the discriminant pattern of the distinct types of runners, and (3) to verify the runners' ability to self-classify. This information may assist the guidance of long-term training, athlete's selection programs and identify the strengths and weaknesses of athletes. 2. Materials and Methods 2.1. Design and Sample The data came from the Intrack project (https://intrackproject.wixsite.com/website), a cross-sectional research project conducted to identify the predictors of running perfor- mance based on an ecological approach [19]. The sample comprised 1235 runners of both sexes (492 women; 743 men), with a mean age of 37.9 9.4 years (ranging from 18 to 72 years), from the ve Brazilian regions (Southeast = 453 (36.7%); Northeast = 441 (35.7%); South = 145 (11.7%); Midwest = 104 (8.4%); North = 89 (7.2%); Missing data = 3 (0.2%)). To be considered eligible for the study, runners should have answered the online ques- tionnaire; those aged below 18 years, and those that did not answer all the mandatory questions from the applied questionnaire were excluded during data analysis. This study was conducted in accordance with the Declaration of Helsinki, and was approved by the Ethics Committee of the Federal University of Sergipe, Brazil (protocol n
study, runners should have answered the online ques- tionnaire; those aged below 18 years, and those that did not answer all the mandatory questions from the applied questionnaire were excluded during data analysis. This study was conducted in accordance with the Declaration of Helsinki, and was approved by the Ethics Committee of the Federal University of Sergipe, Brazil (protocol n 3.558.630).
Int. J. Environ. Res. Public Health2021,18, 4248 3 of 9 2.2. Procedures and Data Collection Information collected was self-reported through the questionnaire section Pro le characterization and associated factors for runner's performance [20]. The instrument was available through an online social media platform (Facebook, Instagram, WhatsApp). The study was conducted between September 2019 and March 2020. The following information was obtained: 2.2.1. Individual Characteristics Sex, age, body weight (kg), and height (m) were self-reported. Body mass index (BMI) was computed using the standardized formula (weight (kg)/height (m) 2 ). 2.2.2. Demographic Information Educational level was dichotomized as ungraduated and graduated. The socioe- conomic status (SES) was categorized based on the Brazilian minimum wage in 2019 [21] as 3 minimum wages and >3 minimum wages. The state of residence was given by the runners, which allowed the identi cation of the regions where the states belong. This information was used for group characterization. 2.2.3. Training Information Running pace: expressed in minutes/km, was self-reported by runners, taking into ac- count their preferred distance. Practice time: runners reported their practice time in months. Frequency: this was reported in counts (17 session/week), and was further dichotomized into at least 3 sessions/week and more than 3 sessions/week. Volume/week: the mean value was reported (in kilometres) for runners, considering the weekly amount. 2.2.4. Self-Classi cation Runners were invited to answer the following question: Regarding the race, you consider yourself as: Professional athlete (has some employment relationship with sports companies/racing clubs); Amateur athlete (has no employment relationship with sports companies/racing clubs, but seeks to improve performance and participation in competi- tions); Recreational athlete (has no competitive interest with road racing). This classi ca- tion was used during analysis. 2.3. Statistical Analysis Descriptive statistics are presented as mean, standard deviation (SD), frequencies (%). Univariate normality was tested for BMI, running pace, practice years, and volume/week, by self-classi cation groups. For graphical representation, variables were standardized (running pace was multiplied by 1). The presence of multivariate outliers was tested by the Mahalanobis distance. To identify pro le differences between runners' classes, multivariate analysis of variance was conducted by
mean, standard deviation (SD), frequencies (%). Univariate normality was tested for BMI, running pace, practice years, and volume/week, by self-classi cation groups. For graphical representation, variables were standardized (running pace was multiplied by 1). The presence of multivariate outliers was tested by the Mahalanobis distance. To identify pro le differences between runners' classes, multivariate analysis of variance was conducted by sex, and Pillai's trace values were considered, given that variance and covariance homogeneity were not observed. Eta squared (n 2 ) was used as a measurement of the effect size. The discriminant analysis [22] was used to identify variables (BMI, running pace, practice time and volume/week) that could differentiate groups of professional athletes, amateur athletes and recreational athletes. The software IBM SPSS Statistics (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp) was used for the analysis. Signi cance was accepted atp< 0.05. 3. Results Descriptive information is presented in Table. More than 50% of runners classi ed themselves as amateur runners (68.5% and 77.4%, among women and men, respectively) for the studied sample. For both sexes, the runners self-classi ed as professional athletes were the youngest ones, and reported a higher frequency of training (i.e., more than three sessions/week). Regarding the socio-economic information, this last group also presented
Int. J. Environ. Res. Public Health2021,18, 4248 4 of 9 the highest frequency of ungraduated and as having 3 minimum wages for both sexes. Further, the majority of runners self-classi ed as recreational runners with a training frequency 3/week, and high educational and economic levels. Table 1.Descriptive information for different types of runners, by sexes. Women Men Amateur Runners (n= 337) Recreational Runners (n= 151) Professional Athlete (n= 4) Amateur Runners (n= 575) Recreational Runners (n= 150) Professional Athlete (n= 18) Variables Mean (SD) or Frequency (%) Mean (SD) or Frequency (%) Age (years) 37.8 (8.5) 39.4 (9.0) 29.1 (12.2) 37.4 (39.8) 39.7 (10.5) 29.1 (9.3) Regions Midwest 28 (8.3%) 18 (11.9%) 1 (25%) 45 (7.8%) 9 (6%) 3 (16.7%) Northeast 99 (39.4%) 63 (41.7%) 1 (25%) 204 (35.5% 66 (44%) 8 (44.4%) North 31 (9.2%) 8 (5.3%) 2 (50%) 44 (7.7%) 3 (2%) 1 (5.6%) Southeast 128 (38%) 48 (31.8%) 0 218 (37.9%) 54 (36%) 5 (27.8%) South 49 (14.5%) 14 (9.3%) 2 (50%) 63 (11%) 18 (12%) 1 (5.6%) Frequency/Week 3 train/week 222 (65.9%) 119 (78.8%) 1 (25%) 278 (48.3%) 113 (75.3%) 4 (22.2%) >3 train/week 115 (34.1%) 32 (21.2%) 3 (75%) 297 (51.7%) 37 (24.7%) 14 (77.8%) School Level Ungraduated 75 (22.3%) 24 (16%) 3 (75%) 195 (34.2%) 35 (23.6%) 15 (83.3%) Graduated 262 (77.7%) 126 (84%) 1 (25%) 376 (65.8%) 113 (76.4%) 3 (16.7%) SES 3 minimum wage 104 (31.2%) 39 (26.4%) 2 (50%) 197 (34.7%) 49 (32.7%) 15 (83.3%) >3 minimum wage 229 (68.8%) 109 (72.2%) 2 (50%) 371 (64.5%) 100 (66.7%) 3 (16.7%) Figure semi-professional runners. In both sexes, the same pattern was observed for the variables used to differentiate the groups. In both, the practice time was the variable that, visually, presented the lowest discrepancies among runners. Figure 1. Multivariate pro le for runners considering BMI, practice time, volume/week and running pace (A) Female runners; (B) Male runners. All variables were standardized.
the lowest discrepancies among runners. Figure 1. Multivariate pro le for runners considering BMI, practice time, volume/week and running pace (A) Female runners; (B) Male runners. All variables were standardized.
Int. J. Environ. Res. Public Health2021,18, 4248 5 of 9 In both sexes, the multivariate variance analysis identi ed differences for variables among runners' self-classi cation (Table). For women, a macro analysis indicated that 7.2% of the total variance was explained by belonging to the group; while for men, this group effect explains 9%. A mid-level analysis indicated that, for both sexes, only the practice time did not differ between runners. Moreover, the Bonferroni post-hoc showed, at a micro-level, signi cant differences between recreational and amateur runners for BMI, volume/week, and running pace for women, while among men, these variables differed between all groups. Table 2.Results for multivariate analysis of variance for both sexes. Women Men Amateur Runners Recreational Runners Professional Athlete p-Valuen 2 Amateur Runners Recreational Runners Professional Athlete p-Value n 2 Mean (SD) Mean (SD) BMI (kg/m 2 ) 23.4 (3.2) 24.6 (2.9) * 21.47 (4.6) <0.001 0.044 24.3 (2.7) 25.7 (2.7) *, 20.6 (2.9) <0.001 0.078 Practice time (months) 49.6 (37.5) 45.2 (35.9) 29 (16.0) 0.371 0.005 66.6 (69.3) 68.4 (71.1) 68.4 (47.0) 0.49 0.002 Volume/week (km) 30.0 (18.8) 22.2 (11.8) * 46.25 (32.0) <0.001 0.054 42.8 (34.6) 28.1 (13.8) *, 93.8 (58.4) <0.001 0.092 Running pace (s) 354.7(56.1) 405.3 (88.1) * 308.25 (94.4) <0.001 0.102 300.7 (53.8) 343.4 (67.4) *, 226.3 (50.8) <0.001 0.112 MANOVA Test [(Pillai's trace = 0.144); F (8,840) = 612.21, p< 0.001; n 2 = 0.072] [(Pillai's trace = 0.182); F (8,1310) = 16.37, p< 0.001; n 2 = 0.091] Note: * statistically different from amateur runners; statistically different from professional athletes; n 2 partial eta squared. The discriminant analysis indicated that only the rst function explained the group variance in both sexes. So, for women and men, this variance explanation is 95.4% and 96.1%, respectively. For both sexes, running pace was the variable that better discriminated the groups, followed by BMI and volume/week (Table). The practice time was not a good indicator to differentiate the types of runners. Table 3.Function discriminant results, split by sex. Women Men Wilks'USV Symbol Macro(s) Description 0241 Ɂ \textglotstopvari LATIN CAPITAL LETTER GLOTTAL STOP 0242
is 95.4% and 96.1%, respectively. For both sexes, running pace was the variable that better discriminated the groups, followed by BMI and volume/week (Table). The practice time was not a good indicator to differentiate the types of runners. Table 3.Function discriminant results, split by sex. Women Men Wilks'USV Symbol Macro(s) Description 0241 Ɂ \textglotstopvari LATIN CAPITAL LETTER GLOTTAL STOP 0242 ɂ \textglotstopvarii LATIN SMALL LETTER GLOTTAL STOP 0243 Ƀ \textbarcapitalb LATIN CAPITAL LETTER B WITH STROKE 0244 Ʉ \textbarcapitalu LATIN CAPITAL LETTER U BAR 0245 Ʌ \textturnedcapitalv LATIN CAPITAL LETTER TURNED V 0246 Ɇ \textstrokecapitale LATIN CAPITAL LETTER E WITH STROKE 0247 ɇ \textstrokee LATIN SMALL LETTER E WITH STROKE 0248 Ɉ \textbarcapitalj LATIN CAPITAL LETTER J WITH STROKE 0249 ɉ \textbarj LATIN SMALL LETTER J WITH STROKE 024A Ɋ \texthtcapitalq LATIN CAPITAL LETTER SMALL Q WITH HOOK TAIL 024B ɋ \texthtq LATIN SMALL LETTER Q WITH HOOK TAIL 024C Ɍ \textbarcapitalr LATIN CAPITAL LETTER R WITH STROKE 024D ɍ \textbarr LATIN SMALL LETTER R WITH STROKE 024E Ɏ \textbarcapitaly LATIN CAPITAL LETTER Y WITH STROKE 024F ɏ \textbary LATIN SMALL LETTER Y WITH STROKE 0250 ɐ \textturna LATIN SMALL LETTER TURNED A 0251 ɑ \textscripta LATIN SMALL LETTER ALPHA 0252 ɒ \textturnscripta LATIN SMALL LETTER TURNED ALPHA 0253 ɓ \m{b} \m{b} \texthtb \textbhook LATIN SMALL LETTER B WITH HOOK 0254 ɔ \m{o} \textopeno \textoopen LATIN SMALL LETTER OPEN O 0255 ɕ \textctc LATIN SMALL LETTER C WITH CURL 0256 ɖ \M{d} \textrtaild \textdtail LATIN SMALL LETTER D WITH TAIL 0257 ɗ \m{d} \texthtd \textdhook LATIN SMALL LETTER D WITH HOOK 0258 ɘ \textreve LATIN SMALL LETTER REVERSED E 0259 ə \schwa \textschwa LATIN SMALL LETTER SCHWA 025A ɚ \m{\schwa} \texthookabove{\schwa} \textrhookschwa LATIN SMALL LETTER SCHWA WITH HOOK 025B ɛ \m{e} \textepsilon \texteopen \textniepsilon LATIN SMALL LETTER OPEN E 025C ɜ \textrevepsilon LATIN SMALL LETTER REVERSED OPEN E 025D ɝ \texthookabove{\textrevepsilon} \textrhookrevepsilon LATIN SMALL LETTER REVERSED OPEN E WITH HOOK 025E ɞ \textcloserevepsilon LATIN SMALL LETTER CLOSED REVERSED OPEN E 025F ɟ \B{j} \textbardotlessj \textObardotlessj LATIN SMALL LETTER DOTLESS J WITH STROKE 0260 ɠ \texthookabove{g} \texthtg LATIN SMALL LETTER G
Description
This study analyzes anthropometric and training variables among Brazilian runners.