Because of the highly stochastic nature of the training activities in team sports, it is difficult to accurately quantify the training loads experienced by athletes. Moreover, many team sports such as Australian Rules Football (AF), Rugby Union, and Rugby League require collisions and eccentric muscle actions that contribute to the overall load endured by these athletes. In the field setting, there are presently a number of methods that are used quantify the training loads undertaken by these athletes (5). These methods can be described as a measure of either the external training load (a measure of training load independent of individual internal characteristics, e.g., distance covered, training duration), or a measure of internal training load (the physiological stress imposed on the athletes by the external load) (14).
Theory suggests that internal training loads may be most appropriate for monitoring training (i.e., the load endured by athletes), whereas the external load is generally considered to be important for the prescription and planning of training. Indeed, in a review of aerobic training within soccer players (13), it was suggest that although the individual's physiological response to a training stimulus (internal load) may be a more acute marker of training load, it is the combination of both the external load (quality, quantity, and organization of the training stimulus) and the individual characteristics that make up the complete training process (14). Therefore, it might be that these 2 constructs of training provide different information to coaches that can be used to influence decisions about the training process (i.e., external training loads confirm if planned training outcomes are achieved, whereas internal training load measures can be used to determine how players are responding/coping with training). Currently, the relationships between internal and external training loads have not been appropriately examined. It is important to understand these relationships so that coaches and sport scientists can make informed decisions about choosing appropriate methods for quantifying training.
At present, the common methods for quantifying training load in team sports are heart rate (HR) and microtechnology, including global positioning system (GPS) and accelerometers (5). The GPS devices are often used to provide feedback on distance traveled, running speeds, and repeated-sprint efforts of players, whereas accelerometers provide further information on the impacts endured by the athletes, giving feedback on the overall body load these impacts generate. It has recently been demonstrated that accelerometers have acceptable level of technical reliability both within and between devices for measuring physical activity in AF, providing increased practical application within team sports (6). Unfortunately, however, all these methods have a strong reliance on technology and technical expertise. As such, these methods can be quite expensive and difficult to implement with large groups of athletes. Moreover, the risk of data loss is high with these technologies (1,7,13). For example, Wallace et al. (22) reported that approximately 36% of HR measures taken during swimming sessions were lost when monitoring training intensity over a season. Therefore, alternative methods that are low cost and easy to implement with large groups may be more effective and appropriate for team sports.
The session-rating of perceived exertion (sRPE) method has been proposed as an alternative, practical and noninvasive method for evaluating internal training load in athletes (11). Based on psychophysical principles, this simple method calculates internal training load by multiplying the individual's RPE (using Borg's category ratio 10-point [CR10] scale) by the duration of the session (in minutes). The sRPE method has been previously shown to be valid across numerous exercise intensities and activities, comparing favorably with more complicated methods of quantifying training load. Indeed, previous studies have shown strong relationships between sRPE and other measurements of internal training load among team sport (1,13), endurance (10), and resistance trained athletes (8).
Recently, Borg proposed a new category ratio scale that attempts to delimit the sensitivity limitations detected in the CR10 scale (3). The CR100 scale is a refined version of the Borg CR10 scale and contains a 1–100 range, giving the individual a greater numerical range to express their perceived exertion. Although it has been shown previously that the CR100 is a valid and reliable method in the perception of effort (3), these results were found using a 5-stage bicycle ergometer protocol conducted in laboratory testing and therefore may not apply to prolonged, high-intensity exercise such as team sports. Moreover, it has yet to be determined if the CR100 provides a more sensitive measure of training load than the CR10 scale. Furthermore, although previous research has suggested the sRPE method to be valid across numerous exercise modes (1,8,10), no study has examined the reliability levels of the recent CR100 or previous CR10 RPE scales.
At present, no study has examined the validity of the sRPE method in AF, particularly using the CR100 as a measure of intensity. Moreover, the relationships between internal and external load with the CR10 and CR100 measures of intensity are not known. Therefore, the first purpose of this study was to examine the validity of the sRPE method within AF using both the CR10 and CR100 scales as markers of intensity. Additionally, this study will investigate the relationships between various measurements of both intensity and training load. It is hypothesized that both the CR10 and CR100 scale derived sRPE methods would be valid measures of TL within AF. Furthermore, it is hypothesized that strong relationships between sRPE and measures of internal and external TL will be shown.
Currently, to the author's best knowledge, no study has examined the reliability of either the CR10 or CR100 RPE scales. Therefore, the second purpose of this study was to examine and compare the criterion validity and test-retest reliability of the CR10 and CR100 RPE scales. It was hypothesized that the CR100 would be a more reliable scale because of the increased sensitivity previously shown (3).
It is common practice for professional AF clubs to monitor training loads using sRPE. However, the common CR10 RPE scale has been criticized for its lack of sensitivity to small changes in effort and its validity in AF has not yet been assessed. Therefore, this study was conducted in 2 parts: (a) to assess the construct validity (part A) and (b) to assess the reliability of the CR10 and CR100 RPE scales when used to quantify training load using Fosters's RPE approach.
Experimental Approach to the Problem
Part A of this study was conducted to assess the validity of the sRPE method within AF using both the CR10 and CR100 RPE scales. This study compared the CR10 and CR100 as methods for quantifying exercise intensity and training load during prolonged high-intensity intermittent exercise. Additionally, the relationships between CR10 and CR100 sRPE methods with internal and external training loads were determined. Individual RPE, HR, and GPS measurements were recorded during skill-based training sessions over a 13-week period. Training loads were calculated daily using the session-RPE method (10).
Twenty-one male Australian football (AF) players (age: 19.0 ± 1.8 years, body mass: 83.92 ± 7.88 kg, height: 187.5 ± 6.3 cm, sum of 7 skinfolds: 48.8 ± 6.1 mm) representing the Greater Western Sydney Football Club volunteered to participate in this study. The club competed in the Northern Eastern AF League. All the players were given a verbal and written description of the procedures, possible risks, and potential benefits before participation, and written consent was obtained from each subject. In the week preceding the study period, each player was familiarized with the exact protocols employed in the study. The players were familiarized with both CR10 and CR100 RPE scales and were clearly explained the use and protocol of these scales (4). Test procedures were approved by the University Human Research Ethics Committee.
Criterion Measures of Internal Load
Session-RPE: The training load for each session was calculated using the sRPE method (10) for each player during the study period. This method involved multiplying the training duration in minutes by the mean training intensity. The training intensity was measured using the CR10 (4) and CR100 (3) RPE scales. The players were asked ‘How intense was your session?’ and were requested to make certain that their RPE referred to the intensity of the whole session rather than the most recent exercise intensity. The players were randomly assigned which RPE scale they were first to answer (i.e., CR10 or CR100) and then responded to the alternative scale 5 minutes later.
Heart Rate: The HR was recorded over the course of each training session at 1-second intervals using Polar Team2 System (Polar Electro Oy, Finland) HR monitors. The HR was downloaded from transmitters using Polar Team2 software (Polar Electro Oy, Finland) and was analyzed at the completion of each training session.
Two HR-based methods for quantifying internal training load were used as criterion measures of training load. The TRIMP method proposed by Banister et al. (2) was used as the first criterion measure of internal training load and was determined using the following formula:
where D is the duration of training session and b is 1.67 for female participants and 1.92 for male participants (19).
The HR-based method proposed by Edwards (9) was also used as a criterion measure of internal training load in this study. This method involves integrating the total volume with the total intensity of each physical training session relative to 5 intensity phases. A score for an exercise bout was calculated by multiplying the accumulated duration (minutes) of time spent in these HR phases (50–60, 60–70, 70–80, 80–90, and 90–100%) by the weighing factor allocated to each zone (1 = 50–60%, 2 = 60–70%, 3 = 70–80%, 4 = 80–90%, and 5 = 90–100%) and then summating the results.
Criterion Measures of External Load
Global Positioning Systems: During each training session, 10 players wore a nondifferential, commercially available GPS receiver with a sampling rate of 10 Hz (MinimaxX, Team 2.5, Catapult Innovations, Scoresby, Australia). The GPS unit was placed in a backpack harness and positioned between the subject's shoulder blades in accordance with manufacturer's specifications. For the purpose of validity and reliability, GPS participants were fitted with the same GPS unit each session (15). Distance, meters per minute, and individual player load (the vector magnitude formulated from accumulated data of all 3 axis measured by the MinimaxX accelerometer) were recorded by GPS devices and downloaded after the completion of each training session using Logan Plus 4.7.0 software (Catapult Innovations, Scoresby, Australia) before being exported for analysis. All the training was completed on a grassed AF field, clear of large building to allow clear satellite reception. The mean (±SD) number of satellites during data collection was 12.5 ± 0.6.
Within-individual correlations between CR10, CR100, and measures of internal (HR) and external (GPS-based measures) training load and intensity were analyzed using Pearson's product moment correlation. Ratio measures for 95% limits of agreement were calculated. SPSS statistical software package version 19 (SPSS Inc., Chicago, IL, USA) was used for all statistical calculations. The level of statistical significance was set at p ≤ 0.05. All data are mean ± SD unless otherwise stated.
Experimental Approach to the Problem
To date, no studies have examined the reliability of either the CR10 or CR100 RPE scales. Reliability is important to coaches and sports scientists because it indicates the reproducibility of a measurement. Indeed, in this case, poor reliability reduces your ability to track changes in measurements of an individual's training load. Therefore, the purpose of part B of the present investigation was to examine and compare the criterion validity and test-retest reliability of the CR10 and CR100 RPE scales. To achieve this, RPE measures were taken after 8 minutes of controlled intermittent running of 3 different intensities. The players completed each running speed twice so that the test-retest reliability could be calculated. This study was conducted using a randomized group design to ensure speeds were kept anonymous from participants.
The study was completed over a 3-week period; during which each athlete was required to complete the testing protocol twice a week. Testing was conducted at the same time on the 2 same weekdays over the 3 weeks to determine the test-retest reliability. Athletes were familiarized with both CR10 and CR100 RPE scales and were clearly explained the use and protocol of these scales (4), and protocols used throughout the study. Individual training programs and periodization strategies were kept identical during the study period. In the 24 hours before testing, each athlete avoided any moderate to heavy training and to standardize their food and fluid intake to increase the reliability of perception of effort responses.
Ten young male team sport athletes (age: 16.1 ± 0.5 years, body mass: 83.1 ± 8.4 kg, height: 183.2 ± 7.6 cm) volunteered to participate in part B of the study. All the athletes were given a verbal and written description of the procedures, possible risks, and potential benefits before participation. Written consent was obtained by each athlete and their guardian, and ethical approval was granted by the University Human Research Ethics Committee for all experimental procedures.
Standardized Intermittent Running Test
This study used 3 different speeds of the Yo-YoIR1 (Yo-Yosubmax) speed protocols to control the external training load of each athlete and assess the sensitivity of the CR10 and CR100 RPE scales. These Yo-Yosubmax protocols were undertaken for 8 minutes at speeds of 10 km·h−1 (level 1), 11.5 km·h−1 (level 2), and 13 km·h−1 (level 3). The athletes were asked to run intermittently back and forth a 20-m marked course to standardized audio signals following similar methods previously described for the Yo-YoIR1 (16). All the trials were completed outdoors on a grass playing field. The 3 testing protocols were allocated to both groups in a randomized order while the speed profiles of each test were kept anonymous from athletes.
The HR and RPE-based measures were recorded with the equipment and protocol methods similar to that done in part A.
Typical error as a coefficient of variation (CV) and interclass correlation coefficients (ICC) were calculated to establish the reliability of the CR10 and CR100 scales and Edwards TRIMP and %HRpeak. Ratio measures for 95% limits of agreement were calculated. SPSS statistical software package version 19 (SPSS Inc., Chicago, IL, USA) was used for all statistical calculations. The level of statistical significance was set at p ≤ 0.05. All data are mean ± SD unless otherwise stated.
A total of 38 training sessions with a mean duration of 63 ± 23 minutes were observed during a 13-week period. Twenty-one subjects completed an average of 23 ± 4 sessions. Session-RPE and HR-based measures were collected from all the players for each session completed. Because of limitations with the number of GPS devices available, data from 10 subjects over 18 ± 3 training sessions were recorded. Training load measures were recorded for sessions completed by all the players (n = 21) (CR100-training load: 3,740 ± 2,470 AU, CR10-training load: 382 ± 246 AU, Edwards' TRIMP: 157 ± 79, Banister's TRIMP: 129 ± 67).
The within-individual correlation (r) of CR100-training load and CR10-training load, 2 HR-based training load methods and various measurements of external training load are shown in Figure 1. All correlations were statistically significant (p ≤ 0.05). There were also significant correlations between the sRPE and all external training load measures (distance, high speed running, and player load).
The within-individual correlations of CR100 and CR10, %HRpeak, and various external measures of intensity are outlined in Figure 2. Moderate correlations were observed between both RPE scales and %HRpeak (p ≤ 0.05). Poor to moderate correlations were observed between RPE and external measures of intensity (p ≤ 0.05).
The %CV and %ICC from both trials, each of the RPE scales showed a poor level of reliability over all speed categories (CR10 [31.9% CV, 0.66 ICC] and CR100 [38.6% CV, 0.70 ICC]). A poor level of reliability was shown for both scales at the slowest speed protocol (CR10-level 1 [34.8% CV, 0.55 ICC] and CR100-level 1 [52.4% CV, 0.55 ICC]). These reliability levels were improved for faster Yo-Yosubmax protocols (CR10-level 3 [21.2% CV, 0.66 ICC] and CR100-level 3 [25.5% CV, 0.79 ICC]).
The first purpose of this study was to examine the validity of the CR10 and CR100 sRPE methods in quantifying training load in AF. Additionally, this study compared the efficacy of the CR10 and CR100 as methods for assessing exercise intensity and training load in intermittent running. The main findings of this study support previous research (1,13), indicating that the CR10 and CR100 sRPE methods are valid for monitoring training load in team sports. However, both the CR10 and CR100 sRPE measures show poor levels of reliability for assessing internal training load.
The present results are the first to show that sRPE is a valid method for assessing the training load in AF. These findings are in agreement with those of previous studies in team sports such as soccer (1,13) and basketball (10) that have shown strong relationships between HR-based and RPE-based methods for quantifying training load. For example, Impellizzeri et al. (13) assessed 19 young soccer players and reported correlation coefficients between Banister's TRIMP (range: 0.50–0.77) and Edwards' TRIMP (range: 0.54–0.78) and sRPE training load. A possible explanation for the higher correlation coefficients in this study compared with this previous study could be because of the nature of the AF training sessions monitored in this study. For example, although the training drills used in this study typically required the players to complete high-intensity, intermittent running, the time periods between the drills was kept to a minimum so that there was relatively little rest time. It has previously been reported that there is a poorer relationship between HR and RPE during interval type training sessions that required long periods of rest between exercise efforts in both soccer training (1) and resistance training (8,21). Specifically, Alexiou and Coutts (1) showed higher correlation coefficients between sRPE training load and Banisters TRIMP and Edwards' TRIMP during conditioning sessions compared with matches. It was suggested, that these sessions require a greater proportion of time active with minimal rest periods. It is likely that both the differences in the nature of the sport-specific requirements and differences in the ratio of work to rest, between the present and previous studies, also explain the differences in the measured correlations. Collectively, the present results show that sRPE relates well to HR-based measures of training load in AF, especially in skills-based training with a relatively low proportion of time spent resting. However, further work is still required to determine the most appropriate method for determining ‘session duration’ when using the sRPE method for quantifying training load.
A new finding of this study was the strong correlation between measures of external training load (distance, higher speed running, and player load) and sRPE. These observations confirm the training theory (14) that suggests internal load is a product of an individual's external load and provides further evidence for the construct validity of sRPE as a tool for assessing training load. Additionally, these findings provide evidence for accelerometer-derived player load as a valid method for assessing training load in team sports, particularly in AF. Notably, however, these results are in contrast with those of previous research (12), which reported no relationship between sRPE and accelerometer-derived player load during soccer small-sided games in professional players. Although it is difficult to accurately compare these studies because of the different nature of the sports and the different RPE scales (6–20 vs. CR10 and CR100 scales) used in the experimental protocols, it is possible that the lack of differences in the previous study may be attributed to the use of pooled correlations, rather than within-individual correlations used in this study. Further differences may be because of both the different accelerometer devices used to assess player load and the software algorithms for calculation. These findings suggest further research is required to examine the validity of player load data in team sports that require high-intensity intermittent exercise, frequent collisions and sports specific skills.
In this study, we observed stronger relationships between HR-based methods and sRPE when compared with external load measures (i.e., distance, player load) and sRPE for both exercise intensity and training load. This observation is not surprising because HR is a measure of internal training load (similar to sRPE), whereas other measures report the external load, which examines a different construct of training load. Taken collectively, these results suggest that the external load is only one, albeit important, contributor to the overall internal load experienced by the individual. Other factors such as training status, fatigue state, previous training, and genetics may also influence an athlete's response to training (14). These results also highlight that different information provided by external and internal training load measures. Accordingly, we suggest that external training load measures be used by coaches and sport scientists to prescribe training stimulus and evaluate whether the load being undertaken by players is intended, whereas internal training load be used to monitor how players are responding to with the applied training load. It is also possible that the relationship between internal and external training loads may be a valuable marker of how athletes are coping with training, especially during standard training sessions.
A further aim of this study was to compare the CR10 and CR100 as methods for assessing exercise intensity and training load during AF training. To the author's best knowledge, no study has assessed the validity of the CR100 in team sport activities nor compared the information provided by each of these scales. The CR100 scale is a refined version of the Borg CR10 scale and has been suggested to overcome the sensitivity limitations detected in the CR10 scale increasing the validity and reliability of this method (3). However, in contrast to these suggestions, we observed similar levels of validity for all measures of intensity and training load. Interestingly, although there were no significant differences between the scales, the CR10 had a slightly larger correlation coefficient with every measure of intensity and training load. Overall, however, these methods provide similar information on the athlete's perceptions of training intensity and training load during AF training, and the CR100 does not improve the reported limitations of the CR10 scale when applied in the manner used in this study.
In summary, our findings provide further evidence to show that sRPE is a valuable tool for assessing training load and training intensity and is the first to demonstrate these relationships in AF. Moreover, the strong relationships between sRPE and external training load measures (distance, higher speed running, and player load) suggest that an athlete's internal load is related to the external load applied during training (13). It also seems that the CR10 and CR100 scales provide similar information regarding exercise intensity. Finally, because other methods of monitoring training intensity and load can be expensive, require technical expertise, may result in data loss, and be time consuming, we suggest that sRPE provides a practical and valid alternative for monitoring training in AF.
In part B of this study, the reliability of both the CR10 and CR100 scales for quantifying training intensity in intermittent activity was examined. The results showed relatively poor levels of reliability for each of the RPE methods (CR10 [31.9% CV] and CR100 [38.6% CV]) for quantifying internal training intensity. These results are similar to those previously reported (20) and indicate that the use of RPE may be limited by its reliability in brief bouts of intermittent activity, which is common in team sports. However, it is important to acknowledge that the determination of the CV may be a poor method for determining the reliability of ordinal scales such as the CR10 and CR100 scales. Still, to our knowledge, there is no method available to appropriately measure the reliability of these scales or allow comparisons with ratio scales (e.g., HR, RPE 6–20 scales). Furthermore, previous authors have also suggested an increased difficulty in determining the reliability of the CR10 scale because of the multifactorial nature of the scale, which is mediated by both physiological and psychological factors (4,18). Nonetheless, we suggest sRPE data should be interpreted in the context of its %CV and that future studies should assess the %CV in a more ecologically valid environment, using longer exercise bouts that the 8-minute efforts used in this study and also a wider range of exercise intensities.
Interestingly, both perception of effort scales showed an increased reliability with higher exercise intensity. Therefore, it would appear that perception of effort becomes more reliable as a marker of exercise intensity as it increases. These observations are supported by the findings of Lamberts et al. (17), who showed that the HR has improved lower levels of reliability during increase speeds of interval shuttle running. Collectively, this showed that the CR10 and CR100 both had similarly poor levels of reliability. This may be related to the short exercise bout (8 minutes) efforts used in this study. Indeed, it is suggested that future studies examine the reliability of sRPE over a wider range of intensities and for longer exercise durations. It is therefore important that these levels of reliability should be taken into account when interpreting data using the CR100 and CR100 as an sRPE measure, while care should be taken when applying these observations to sRPE measures taken during longer training sessions.
There are a variety of methods available to coaches for determining the individual stress placed on an athlete from an exercise bout. The current results show that sRPE is a valid indicator of global internal training load in AF. Indeed, these findings indicate that both the CR10 and CR100 sRPE are valid methods of quantifying training load in team sports. However, it is important that coaches take care when assessing and comparing training loads between athletes and different training load methodologies. We suggest that these measures (internal and external) be used to provide a greater depiction of the overall load undertaken by the athlete and not be used interchangeably to compare exercise intensity or training load.
Findings from this study revealed poor reliability levels for both the CR10 and CR100 RPE scales. These levels of reliability should be taken into account by coaches and sport scientists when interpreting data using the CR10 and CR100 as an sRPE measure during short bouts of intermittent exercise. However, because of the strong validity of this measure across numerous exercise intensities coupled with the simplistic and noninvasive nature of this method, it is still suggested that the sRPE remains a valid method to quantify training loads in high-intensity, intermittent team sport.
1. Alexiou H, Coutts AJ. The validity of the session-RPE method for monitoring training
load in elite female soccer players. Int J Sports Physiol Perform 3: 320–330, 2008.
2. Banister EW. Modeling elite athletic performance. In: Physiological Testing of Elite Athletes. Green H. J., McDougal J. D., Wenger H. A., eds. Champaign, IL: Human Kinetics, 1991. pp. 403–424.
3. Borg E. On Perceived Exertion and Its Measurement. Stockholm, Sweden: Stockholm University, Department of Psychology, 2007.
4. Borg G, Hassmén P, Lagerstrom M. Perceived exertion related to heart rate and blood lactate during arm and leg exercise. Eur J Appl Physiol Occup Physiol 56: 679–685, 1987.
5. Borresen J, Lambert MI. The quantification of training load, the training response and the effect on performance. Sports Med 39: 779–795, 2009.
6. Boyd LJ, Ball K, Aughey RJ. The reliability of MinimaxX accelerometers for measuring physical activity in Australian football. Int J Sports Physiol Perform 6: 311–321, 2011.
7. Coutts A, Sirotic AC, Catterick C, Knowles H. Monitoring training
loads in professional rugby league. In: Science and Football VI: The Proceedings of the Sixth World Congress on Science and Football. New York, NY: Routledge, 2008. pp. 272–277.
8. Day M, McGuigan MR, Brice G, Foster C. Monitoring exercise intensity during resistance training using the session-RPE scale. J Strength Cond Res 18: 353–358, 2004.
9. Edwards S. High performance training and racing. In: The Heart Rate Monitor Book. Edwards S., ed. Sacramento, CA: Feet Fleet Press, 1993. pp. 113–123.
10. Foster C, Florhaug JA, Franklin J, Gottschall L, Hrovatin LA, Parker S, Doleshal P, Dodge C. A new approach to monitoring exercise training. J Strength Cond Res 15: 109–115, 2001.
11. Foster C, Hector LL, Welsh R, Schrager M, Green MA, Snyder AC. Effects of specific versus cross-training on running performance. Eur J Appl Physiol 70: 367–372, 1995.
12. Gomez-Piriz PT, Jimenez-Reyes P, Ruiz-Ruiz C. Relation between total body load and session rating of perceived exertion in professional soccer players. J Strength Cond Res 25: 2100–2103, 2011.
13. Impellizzeri FM, Rampinini E, Coutts AJ, Sassi A, Marcora SM. Use of RPE-based training load in soccer. Med Sci Sports Exerc 36: 1042–1047, 2004.
14. Impellizzeri FM, Rampinini E, Marcora SM. Physiological assessment of aerobic training in soccer. J Sports Sci 23: 583–592, 2005.
15. Jennings D, Cormack S, Coutts AJ, Boyd L, Aughey RJ. The validity and reliability of GPS units in team sport specific running patterns. Int J Sports Physiol Perform 5: 328–341, 2010.
16. Krustrup P, Mohr M, Amstrup T, Rysgaard T, Johansen J, Steensberg A, Pedersen PK, Bangsbo J. The yo-yo intermittent recovery test: Physiological response, reliability, and validity. Med Sci Sports Exerc 35: 697–705, 2003.
17. Lamberts RP, Lemmink KA, Durandt JJ, Lambert MI. Variation in heart rate during submaximal exercise: Implications for monitoring training
. J Strength Cond Res 18: 641–645, 2004.
18. Morgan WP. Psychological components of effort sense. Med Sci Sports Exerc 26: 1071–1077, 1994.
19. Morton RH, Fitz-Clarke JR, Banister EW. Modeling human performance in running. J Appl Physiol 69: 1171–1177, 1990.
20. Rampinini E, Impellizzeri FM, Castagna C, Abt G, Chamari K, Sassi A, Marcora SM. Factors influencing physiological responses to small-sided soccer games. J Sports Sci 25: 659–666, 2007.
21. Sweet TW, Foster C, McGuigan MR, Brice G. Quantitation of resistance training using the session rating of perceived exertion method. J Strength Cond Res 18: 796–802, 2004.
22. Wallace LK, Slattery KM, Coutts AJ. The ecological validity and application of the session-RPE method for quantifying training loads in swimming. J Strength Cond Res 24: 33–38, 2009.