In swimming, as in other physical activities, athletes, coaches and sport scientists are interested in monitoring, regulating, and prescribing the intensity of exercise. The most common methods used for this purpose are physiological measures such as heart rate (HR), blood lactate concentration (La), and oxygen uptake. However, the use of such physiological measures in training sessions on a daily basis is often limited by the lack of appropriate equipment and the fact that training has to be interrupted to obtain these measurements. An alternative method for monitoring, regulating, and prescribing exercise intensity is the athlete's perception of physical exertion. The perception of exertion can be defined as the subjective intensity of effort, strain, discomfort, and fatigue that is experienced during physical exercise (15). Several psychophysical category scales that quantify exertional perceptions have been developed. The 15-grade scale (6-20) for ratings of perceived exertion (RPE), developed by Borg (1), has been shown to be one of the best indicators of exercise intensity (2). Given that Borg's RPE scale is simple to use and interpret and that it is not restricted from the practical limitations of physiological measures, it has become a popular and useful tool for researchers and coaches and is widely used in laboratory and training settings.
Borg's RPE scale has been used in many swimming studies for the purposes of monitoring and prescribing swimming exercise (5-7,17-19). In some of these studies, the validity of RPE was tested with the use of the aforementioned physiological measures as criteria. None of these physiological measures appeared to be consistently superior to the others with respect to their association with RPE. In general, researchers agreed that RPE is effective as a measure of exercise intensity and can be used for monitoring and prescribing swimming exercise. Nevertheless, Green et al. (6) suggested that RPE adjustments might be required for some levels of exercise intensity, as they found RPE valid for prescribing intensity at RPE level 16 but not at RPE level 12.
Despite the useful findings reported in swimming studies, further research is required to investigate some new areas and to overcome limitations that might restrict the generalizability of existing data. First, there are no data for elite swimmers as recreational or university swimmers were used in all studies. However, data from the latter groups might not be applicable to swimmers of a higher standard because it has been shown that participants of different levels might have significantly different values in variables such as HR, RPE, and La (e.g. 3,9). Second, given that maximum HR (HRmax) depends on age, the fact that HR has been used as a criterion measure to assess validity of RPE in all swimming studies might have caused errors in the findings, especially when the age span of the participants was large, and does not facilitate comparisons between swimmers or groups. A more appropriate measure for this purpose would be the expression of the HR values as a percentage of HRmax (%HRmax) (12). Third, the findings of studies conducted in swimming flumes might have limited generalizability, as it has been reported that swimming technique in the pool and a swimming flume might differ (8,22). Furthermore, many studies have restricted their analysis to 1 testing session, and test-retest reliability of RPE responses has been rarely reported. Given that RPE is used by coaches throughout the training season, it would be of interest to examine the reliability of RPE responses through a longitudinal analysis, as this would reveal whether the RPE responses of individual swimmers remain consistent with regard to the respective levels of the physiological criterion measures.
Given the widespread use of RPE in swimming training, coaches would be also interested in indicative/recommended %HRmax values that should correspond to each RPE grade according to the level of the swimmers. McArdle et al. (10) stated that HRmax values in swimming should be adjusted because HRmax during swimming or upper-arm exercises is approximately 13 beats·min−1 lower than in exercises such as running. These authors calculated the modified %HRmax (%MHRmax) according to this adjustment and provided the recommended values for points 9 to 16 on the RPE scale. However, McArdle et al. (10) based their recommendation for the HRmax adjustment on studies of recreational swimmers or on upper-arm exercises only (e.g., 11). As mentioned above, recreational swimmers' data might not be applicable to swimmers of other levels, and, furthermore, the generalizability of findings from upper-arm exercises to swimming might not be appropriate. Thus, it would be of interest to obtain %HRmax values for elite swimmers during swimming exercise and for the whole range of the RPE scale, in order to examine whether an adjustment in HRmax is required for elite swimmers.
The purpose of the present study was to examine longitudinally the reliability and validity of RPE for monitoring exercise intensity in elite swimmers, with the use of %HRmax and La as the criterion measures. Moreover, the purpose is to calculate the %HRmax that corresponds to each RPE level and to examine whether an adjustment of HRmax is required for subsequent calculations of %HRmax for elite swimmers.
Experimental Approach to the Problem
To examine a range of very low to maximum intensities, an incremental 7 × 200 m test was selected. This swimming test is widely used for training and research purposes and has been a part of the normal training routine of many national teams and swimming clubs (13,14). All swimmers were tested in their specialist strokes (6 freestyle, 4 backstroke, 3 butterfly, 3 breaststroke, and 1 individual medley swimmer), and the 200 m was one of their specialist events.
Each swimmer was tested 4 times during a 6-month period, with a minimum of 4 weeks between subsequent tests. The tests were conducted in a 50-m indoor swimming pool during afternoon sessions (between 5 pm and 8 pm). The participants completed each 7 × 200 m test on either a 5-minute cycle for freestyle and backstroke or a 6-minute cycle for breaststroke, butterfly, and medley. Swimmers were instructed to swim the last 200 m with maximal effort. Ten seconds were added to each swimmer's personal best 200-m performance of the year for the sixth swim, and then, working in reverse order, 5 seconds were added for each subsequent interval to establish the full test protocol. All swimmers were free from injuries and illness, and standardized procedures and guidelines were followed regarding participation in the test, nutrition, training before each test, and warming up.
The main aim of the study was to monitor longitudinally the perception of exertion of elite swimmers and to assess its validity and reliability. Borg's 15-grade RPE scale (1) was used to quantify exertional perception because it has been shown to be one of the best indicators for swimming intensity (2), it has been recommended for use during incremental intensity exercise (15), and it is frequently used by researchers and coaches (5-7,17-19). On the basis of the existing literature, La and %HRmax were selected as the physiological criterion measures for the assessment of RPE validity and reliability. Finally, the %MHRmax was calculated for each RPE stage for the purpose of testing whether previous recommendations for the adjustment of HRmax for recreational swimmers are applicable to elite swimmers.
Seventeen swimmers who compete at an international level participated in this study. The group consisted of 9 male (age: 23.4 ± 2.1 yr; height: 186.6 ± 6.0 cm; body mass: 83.7 ± 6.6 kg) and 8 female swimmers (age: 20.5 ± 1.9 yr; height: 170.1 ± 3.8 cm; body mass: 60.9 ± 5.6 kg). The swimmers' best times were 3.9 ± 1.3% slower than the world records in their events. All the test procedures were approved by the human subjects committee of the institutional ethics review board. The full study protocol was explained, and written informed consent forms were obtained from all swimmers before their participation in the test.
Borg's 16-grade RPE scale (Table 1) was shown to each swimmer before each test, and it was explained that the scale is used to describe the overall feelings of physical exertion and fatigue as perceived by the individual during or after exercise. The verbal descriptors ranged from “Very, very light” to “Very, very hard,” and, in line with other studies (6), it was indicated that grade 6 corresponds to no exertion at all, such as seated rest, and grade 20 corresponds to maximal physical exertion. The RPE scale was shown to the swimmers after each 200-m swim, and they were asked to indicate their perceived exertion on the scale. Given that all swimmers had performed the incremental test at least once before first tested and that they were familiar with RPE through frequent use in training, there were no concerns over learning effects during the period of the longitudinal analysis. The HR (beats·min−1) was recorded immediately after each swim with the use of Polar S810 HR monitors (Polar, Kempele, Finland). One minute after the end of each 200-m swim, 10-mL blood samples were taken from a hyperemic earlobe and analyzed immediately with a Lactate Pro lactate analyser (Arkay Factory, Inc., Shiga, Japan). Because the La measures were not obtained in some tests for 3 of the swimmers in this study, for the analyses involving La values, the data of these 3 swimmers were excluded.
The HRmax for each swimmer was estimated by subtracting the swimmer's age (yr) from 220. Then, the HR values were expressed as %HRmax. On the basis of the recommendations of McArdle et al. (10), %MHRmax was also calculated following the same steps as above but subtracting 13 beats·min−1 from the HRmax used for the calculations.
For the purpose of the subsequent analysis, the 15-grade RPE scale was divided to 7 perceived exertion stages, in line with other investigators (e.g., 10) and based on Borg's verbal descriptors (1). These stages are shown in Table 2.
Assessment of Validity
For each test of each swimmer, the average %HRmax, %MHRmax, and La (mmol·L−1) values were calculated for each RPE stage and used for subsequent analysis. In line with other researchers (e.g. 2,7,12), the Pearson's product moment correlation coefficient (r) was calculated between RPE and %HRmax, and RPE and La. These correlation coefficients reflect the association between RPE and each of the 2 criterion measures and can be used as indicators of RPE validity. Furthermore, given that a good association between 2 methods does not necessarily denote an agreement between their scores, it was also of interest to compare RPE scores with those of each criterion measure. As the scores for each method are measured in different units, the raw scores for RPE, %HRmax, and La were standardized by being transformed into z scores [z = (raw score - mean score)/SD], as suggested by Green et al. (6) and Field (4). For these transformed data, the mean values for the 4 tests were calculated for each swimmer, and each stage of the test and the RPE scores were then compared with the %HRmax and La scores using repeated measures analysis of variance (ANOVA) (4).
Assessment of Reliability
Because the present study consisted of 4 repeated tests of the same group and 7 RPE stages of interest, a factorial ANOVA was used as described by Field (4) and Vincent (20). For the factorial ANOVA, %HRmax and La were the dependent variables and test (4 levels) and stage (7 levels) were the fixed factors. Vincent (20) stated that for repeated measures designs, such as the one in the present study, the above analysis can be used to assess reliability with sensitivity to both order and magnitude. Thus, reliability of the RPE scale was indicated by the between tests significance values for the dependent variables. The between stages comparisons were also calculated for the purpose of assessing the changes in the %HRmax and La values between the different RPE stages. For the factorial ANOVA, any missing values in the data set (caused by a swimmer not indicating exertion in 1 of the RPE stages during a test) were replaced through a missing value analysis by the mean of nearby points (4,16). Levene's test was used to examine the homogeneity of the data (4). To identify any significant differences between pairs of tests and pairs of stages for the dependent variables, post hoc pairwise comparisons were conducted for all such pairs. To eliminate the possibility of type I errors in these post hoc tests, a Bonferroni adjustment to reduce the alpha level was applied as described by Vincent (20).
Although the assumption of homogeneity was not violated for the test data, Levene's test revealed heteroscedasticity for the stage data. Because this problem could not be resolved with any data transformation methods, Friedman's ANOVA was used to examine any differences between the stages of the test, as described by Field (4). The post hoc pairwise comparisons in this case were performed with the use of the Wilcoxon signed-ranked test (4). For all statistical calculations in this study, significance was accepted at p < 0.05. For the pairwise comparisons between stages and tests, after the Bonferroni adjustment, the critical p value was 0.002. All statistical analyses were conducted with the use of the Statistical Package for Social Sciences 14.0 software (SPSS, Inc., Chicago, IL, USA).
Estimation of Maximum Heart Rate
Table 3 shows the La, %HRmax, and %MHRmax values for the swimmers in the present study, as well as the %MHRmax values recommended by McArdle et al. (10) for recreational swimmers. The mean values and the range of values for both %HRmax and %MHRmax of the swimmers in the present study differed considerably from previous %MHRmax recommendations that were based on recreational/university swimmers data. Although %HRmax values slightly over 100% could be expected in some cases (as in stages 6 and 7) given that each swimmer's HRmax was a prediction, the %MHRmax produced values higher than 100% as early as in stage 3, which corresponded to a perception of fairly light exercise. These results suggested that MHRmax appears to underestimate HRmax for elite swimmers. Thus, for swimmers of this level, the normal method for HRmax estimation appears to be more appropriate, and no further adjustments are required. On the basis of these findings, it was decided to use the %HRmax values for subsequent analyses.
There were significant correlations between RPE and %HRmax (r = 0.85, p < 0.001) and RPE and La (r = 0.82, p < 0.001), indicating that RPE is a valid method for monitoring exercise intensity in elite swimmers. There was a good agreement between the RPE and %HRmax values, as no significant differences were found between the z scores in any stages, with the exception of stage 7, at which the RPE scores were significantly higher than the %HRmax scores (p = 0.002). There was a weaker correspondence between RPE and La, with the z scores for RPE being significantly lower than La in stages 1 and 2 (p ≤ 0.001) and significantly higher than La in stages 4 and 5 (p ≤ 0.003).
RPE Reliability and Interstage Comparisons
The longitudinal intertest reliability was high, as no significant differences were found in the values of %HRmax or La between the 4 tests. As expected, the comparison of the 7 RPE stages indicated intratest differences for both %HRmax and La. For %HRmax, significant differences (p < 0.002) were found for all pairs of stages. For La, there were no significant differences for any pairwise comparisons between the first 4 stages. Contrary, all La pairwise comparisons involving stage 5, 6, or 7 were significantly different (p < 0.002).
This study presented the first set of data on %HRmax that correspond to different RPE levels for elite swimmers. Such data can be useful to swimmers and coaches as a general indicator for prescription and regulation of exercise intensity in swimming training. However, given that the data in the present study were collected through the monitoring of exercise intensity, prescriptive and regulative protocols should be used in future studies to confirm these findings. Contrary to previous suggestions based on recreational/university swimmers' data, the HRmax values used for calculations of %HRmax during swimming should not be adjusted for elite swimmers because the modified values underestimate HRmax. A possible explanation of the differences between elite and recreational swimmers in the HRmax observed during swimming might lie in differences in the number and activation of muscle groups used. Robertson and Noble (15) stated that the volume of activated muscle mass can influence perceptual responses during exercise evaluation. Although skilled performers generally tend to take advantage of the large muscle groups and use many parts of the body to produce movement, nonskilled athletes normally use fewer body segments and are unable to use them in an effective and efficient way. For example, McArdle et al. (10) implied that recreational swimmers use mainly upper-body muscles when swimming. These authors also stated that one of the factors that might relate to lower HRmax is less feedback stimulation from the smaller, active upper-body muscle mass. Thus, it would not be unreasonable to assume that the higher HRmax values observed in elite compared to recreational swimmers are related to the ability of the former to activate more muscle groups, such as the large muscles of the legs and the trunk, and use them effectively in swimming.
Interindividual variability was observed in the %HRmax values for each RPE stage, implying that generalization of these values for exercise prescription purposes should be done with caution. Interindividual variability is not surprising and has also been reported in other studies (e.g., 21). A factor that is possibly related to the variability in the present study is that the points in the Borg scale were grouped according to the verbal descriptors, and, therefore, 2 to 3 scale points were used for the analysis of each RPE stage. Another explanation for these differences is that the %HRmax values are subject to errors caused by the fact that HRmax is predicted and not measured. In the present study, for example, a swimmer who had a rather high %HRmax value of 89% in stage 1 also had a 107% value in stage 7, which indicates that HRmax was underestimated for this swimmer. Although the values reported in this study could be considered as indicative of elite swimmers, for individualized training and research programs, it is recommended that when the observed HRmax values exceed the predicted ones, the former values should be used for calculations of %HRmax and subsequent analyses. Moreover, it should be noted that the tests in the present study were performed in afternoon sessions. However, competitive swimmers also have frequent morning training sessions throughout the year. Given the possibility of HR variations due to circadian rhythms during the day, it is suggested that possible differences in HRmax between morning and afternoon sessions are assessed and taken into account when designing personalized training programs.
The high correlation coefficients with both La and %HRmax showed that RPE is a valid method for monitoring exercise intensity in elite swimmers. This is in line with Chen et al. (2), who, after conducting a meta-analysis on RPE studies in a range of sports, concluded that validity of RPE was quite high for swimming exercise (0.78 ≤ r ≤ 0.84). As mentioned before, however, the interpretation and generalization of findings of previous studies should be made with consideration to the use of raw HR data as the criterion measure and of recreational/university level participants.
Borg (1) developed the RPE scale aiming for its ratings to correspond to exercise intensities from 60 to 200 beats·min−1. The statistical analysis of the z-scores in the present study indicated good agreement between the RPE and %HRmax measures of exercise intensity (with the exception of the last RPE stage), suggesting that RPE reflects accurately the HR levels of elite swimmers. However, the RPE and La z-scores were in agreement only for 3 of the 7 test stages. Green et al. (6) reported similar patterns when examining the association between RPE, HR, and La during interval cycling. These investigators found no differences between RPE and HR, but RPE scores were not always in agreement with the corresponding La scores, with RPE being either significantly lower or significantly greater than La during some high-intensity and some recovery periods.
It must be pointed out that the delay between lactate production and appearance in blood creates difficulties in interpreting La measures (and their association with RPE measures) during incremental exercise. For example, if more samples were taken during each exercise interval in the present study, it is possible that higher La values would have been recorded in some cases, meaning that the recorded La values might not accurately represent the peak La values after each 200-m swim. Green et al. (6) stated that, although from a practical perspective it is usually impossible to obtain more La measures during exercise intervals (as this would require modification of the incremental set and compromise exercise quality), such La values should be used as indicators rather than as an accurate reflection of intramuscular metabolism.
Regardless of the limitations in obtaining the true peak La value during incremental exercise, Green et al. (6) suggested that the weak correspondence between RPE and La could be attributed to dissimilar response times or acute sensitivity of the different measures to changing intensities. The authors stated that because adenosine triphosphate turnover is not dependent on lactate supply, lactate removal from blood for energy purposes is not critical, owing to the relatively slow response to intensity changes for La compared with %HRmax. The HR, on the other hand, should be quicker to respond because metabolic demand changes rapidly and is partially dependent on cardiac output and, therefore, indirectly reliant on HR. The latter could also partially explain the good agreement observed between RPE and %HRmax.
The lack of agreement between RPE and La during the first stages is also linked to the fact that La values did not change significantly between stages 1 and 4 (as discussed in the next section). Significant increases in La during the last stages resulted in stronger correspondence with the RPE values. In agreement with Green et al. (6), although La might not be the primary RPE mediator during incremental exercise, it does appear to be a mediator of RPE, especially for high-intensity exercise. It is possible that the fatigue accumulated during the incremental test (and, therefore, the increased La) was reflected in the RPE values. Nevertheless, as suggested by Robertson and Noble (15), it should also be emphasized that changes in RPE may be attributed to several factors, with no single factor being exclusively dominant.
There was high consistency in the %HRmax and La values that corresponded to each RPE stage throughout the 4 tests. This longitudinal RPE analysis indicated that RPE is a reliable tool for monitoring swimming intensity in elite swimmers and can be used with confidence by researchers, coaches, and elite swimmers throughout the training season. It must be noted that all swimmers were familiar with the test and RPE measurements before participating in the study. For subjects with no previous experience in a test or RPE measurements, a familiarization period would be necessary to maximize RPE reliability and minimize any variability that could be caused by learning effects.
The interstage comparisons revealed some noteworthy patterns. Although the %HRmax values significantly increased with RPE stage, such a pattern was not observed for the La values during the first RPE stages. The La values recorded for perceived exertion levels of “Very very light” up to “Somewhat hard” (RPE stages 1-4) were not significantly different, with the La levels remaining below a potential lactate threshold point of 4 mmol·L−1 (with few exceptions in stage 4). It appears that for perceived exertion of up to “Somewhat hard” during incremental swimming exercise, most elite swimmers were able to work on HRs up to 83.5 ± 6.3% of HRmax while remaining below lactate threshold levels.
It must also be noted that, for some swimmers, La values only exceeded 4 mmol·L−1 in stages 6 and 7, meaning that the perception of exercise intensity at lactate threshold level varied among swimmers from “Somewhat hard” to “Very hard.” This is in line with other studies that also reported that RPE values at the lactate threshold might fall in the above range (3,15). However, as suggested in the previous section, La data should always be interpreted with consideration to the possible limitations in obtaining the true peak La values during the different stages of a test protocol. Despite the potential difficulties that might arise with different protocols, the identification of the RPE points that correspond to the lactate threshold could be of particular interest to swimmers and coaches. Robertson and Noble (15) suggested that the regulation of exercise intensity by producing an RPE that corresponds to a lactate threshold point provides an appropriate overload stimulus to improve functional aerobic power and can be considered a physiologically valid prescriptive procedure. For the purposes of designing individual training programs, researchers and coaches should assess the RPE points that correspond to potential lactate thresholds with such methods that would allow the identification of peak La values after each exercise intensity of interest.
The present study was conducted over a period of 6 months and that the group consisted of 17 male and female swimmers specialized in all 4 strokes. The variation in their training programs and competition goals suggested that attempting to control or evaluate the influence of their training programs on the variables studied would be both impractical and beyond the scope of this research. In light of the findings of the present study, which suggested that RPE is a valid and reliable tool, it would be of interest in future studies to examine the use of RPE in monitoring, regulating, and prescribing swimming intensity during different training periods.
The reliability and validity of RPE in elite swimmers was examined through a longitudinal analysis of incremental swimming, with the use of %HRmax and La as the criterion measures. RPE was found to be a valid and reliable method that swimmers and coaches may use to monitor exercise intensity throughout the year. The %HRmax was found to be the primary RPE mediator during incremental swimming exercise, whereas La appeared also to be an RPE mediator during high swimming intensities. Contrary to previous recommendations for recreational/university swimmers, when coaches prescribe or regulate swimming intensity for elite swimmers, the estimated HRmax value used for subsequent calculations should not be adjusted because such an adjustment appears to cause an underestimation of HRmax.
The author thanks Dr. Geraint Florida-James for his useful comments on this article.
1. Borg, G. Psychophysical bases of perceived exertion. Med Sci Sports Exerc
14: 377-381, 1982.
2. Chen, MJ, Fan, X, and Moe, ST. Criterion-related validity of the Borg ratings of perceived exertion scale in healthy individuals: a meta-analysis. J Sports Sci
20: 873-899, 2002.
3. DeMello, JJ, Cureton, KJ, Boineau, RE, and Singh, MM. Ratings of perceived exertion at the lactate threshold in trained and untrained men and women. Med Sci Sports Exerc
19: 354-362, 1987.
4. Field, A. Discovering Statistics Using SPSS
(2nd ed.). London: SAGE Publications, 2005.
5. Gergley, TJ, McArdle, WD, DeJesus, P, Toner, MM, Jacobowitz, S, and Spina, RJ. Specificity of arm training on aerobic power during swimming and running. Med Sci Sports Exerc
16: 349-354, 1984.
6. Green, JM, McLester, JR, Crews, TR, Wickwire, PJ, Pritchett, RC, and Lomax, RG. RPE
association with lactate and heart rate during high-intensity interval cycling. Med Sci Sports Exerc
38: 167-172, 2006.
7. Green, JM, Michael, T, and Solomon, AH. The validity of ratings of perceived exertion for cross-modal regulation of swimming intensity. J Sports Med Phys Fitness
39: 207-212, 1999.
8. Hay, JG and Do Carmo, J. Swimming techniques used in a flume differ from those used in a pool. In: XVth Congress of the International Society of Biomechanics
. Hakkinen, K, Keskinen, KL, Komi, PV, and Mero, A, eds. Jyvaskyla: University of Jyvaskyla, 1995. pp. 372-373.
9. Kurokawa, T and Ueda, T. Validity of ratings of perceived exertion as an index of exercise intensity in swimming training. Ann Physiol Anthropol
11: 277-288, 1992.
10. McArdle, WD, Katch, FI, and Katch, VL. Exercise Physiology: Energy, Nutrition and Human Performance
(6th ed.). Philadelphia: Lippincott Williams & Wilkins, 2007.
11. McArdle, WD, Magel, JR, Delio, DJ, Toner, M, and Chase, JM. Specificity of run training on V̇o2
max and heart rate changes during running and swimming. Med Sci Sports Exerc
10: 16-20, 1978.
12. Pfeiffer, KA, Pivarnik, JM, Womack, CJ, Reeves, MJ, and Malina, RM. Reliability and validity of the Borg and OMNI rating of perceived exertion scales in adolescent girls. Med Sci Sports Exerc
34: 2057-2061, 2002.
13. Psycharakis, SG, Cooke, CB, Paradisis, GP, O'Hara, J, and Phillips, G. Analysis of selected kinematic and physiological performance determinants during incremental testing in elite swimmers. J Strength Cond Res
22: 951-957, 2008.
14. Pyne, DB, Lee, H, and Swanwick, KM. Monitoring the lactate threshold in world-ranked swimmers. Med Sci Sports Exerc
33: 291-297, 2001.
15. Robertson, RJ and Noble, BJ. Perception of physical exertion: methods, mediators and applications. Exerc Sport Sci Rev
25: 407-452, 1997.
16. Tabachnick, BG and Fidell, LS. Using Multivariate Statistics
(4th ed.). Boston: Allyn & Bacon, 2007.
17. Ueda, T, Choi, TH, and Kurokawa, T. Ratings of perceived exertion in a group of children while swimming at different temperatures. Ann Physiol Anthropol
13: 23-31, 1994.
18. Ueda, T and Kurokawa, T. Validity of heart rate and ratings of perceived exertion as indices of exercise intensity in a group of children while swimming. Eur J Appl Physiol
63: 200-204, 1991.
19. Ueda, T and Kurokawa, T. Relationships between perceived exertion and physiological variables during swimming. Int J Sports Med
16: 385-389, 1995.
20. Vincent, WJ. Statistics in Kinesiology
(3rd ed.). Leeds: Human Kinetics Europe, Ltd., 2005.
21. Whaley, MH, Brubaker, PH, Kaminsky, LA, and Miller, CR. Validity of rating of perceived exertion during graded exercise testing in apparently healthy adults and cardiac patients. J Cardiopulm Rehabil
17: 261-267, 1997.
22. Wilson, B, Takagi, H, and Pease, D. Technique comparison of pool and flume swimming. In: Biomechanics and Medicine in Swimming VIII
. Keskinen, KL, Komi, PV, and Hollander, AP, eds. Jyvaskyla, Finland: University of Jyvaskyla, 1998. pp. 181-184.