EXAMINATION OF THE VALIDITY OF MEASUREMENT TOOLS FOR CONTROLLING THE PHYSICAL TRAINING PROCESS IN SOCCER

Training can be described as its outcomes (tests) and process (training load control). The training process involves the repetitions of exercises in order to improve technical skills, tactical situations and develop the ability to cope with physical requests of the competitions. Coaches, physical trainers and sport scientists generally use physiological tests to evaluate the fitness of players and to assess training outcome. The daily monitoring of training load is important to control all the process and to allow the achievement of optimal physical condition. This model requires the quantification of both training loads and outcomes. The aims of this thesis were to fully validate one of most used soccer test (the Yo-Yo Intermittent Recovery test, YYIRT) and to examine some measurement properties and methodological issues of the rating of perceived exertion (RPE) method to measure the training load (TL). As suggested by Impellizzeri and Marcora the tests used in soccer are probably not appropriately validated, at least not to the same extend like in other field such as medicine, psychology and social sciences. Indeed, the tests should be developed and validated using a rigorous approach and methods such as those derived from psychometrics or clinimetrics (an area focusing on the quality of clinical measurements). The Scientific Advisory Committee of the Medical Outcomes Trust for Health Status and Quality of Life instruments have proposed eight attributes by which instruments would be reviewed. In sport science at least five should be verified: conceptual and measurement model; validity; reliability; responsiveness; interpretability. The Session-RPE is frequently used to monitor and control the training process, as a valid measure of internal training load in soccer. Nevertheless some important aspects that could influence the session-RPE are not studied yet: a) the validity of the new Borg CR100 Scale and its interchangeability with the most used CR10, b) the effect of collection, c) the presence of response shift and recall bias in RPE. The Yo-Yo Intermittent Recovery Test (YYIRT) is certainly the most valid soccer-specific endurance test; it is correlated to the high-intensity activity performed during a match, differentiates between competitive levels and playing positions, and is responsive to training. In addition, it is the only test for which some evidence of external validity has been provided. There are two versions of this test suggested to assess different physiological characteristics: the level 1 (YYIRTL1, more aerobic) and the level 2 (YYIRTL2, more anaerobic). In addition there is also a sub-maximal version of the YYIRTL1, that may be useful during rehabilitation process in injured players or when it is not possible to evaluate athletes with maximal tests such as during congested competitive period. The aim of the first part of the research project is to conduct two studies to examine the redundancy of these YYIRT versions by comparing their reliability, responsiveness and interchangeability (convergent validity) (first study) and to assess the longitudinal validity or external responsiveness (second study). The absolute and relative reliability of the two tests (YYIRTL1 and YYIRTL2) was found similar, but the improvement induced by training was higher for YYIRTL1 compared to YYIRTL2, therefore the signal-to-noise ratio of YYIRTL1 is better than YYIRTL2 for detecting training changes. However, the minimal detectable change values in both tests suggested a limited ability to detect substantial changes at individual level. The two tests are correlated but they measure different physical characteristics (low convergent validity) and therefore they are not interchangeable. Therefore, the two tests could be both used, although the YYIRTL1 appears to be superior in terms of measurement properties. In the second study the YYIRTL1 showed longitudinal validity (external responsiveness) and therefore it is able to detect changes in the potential to run at high intensity in a match after training. In addition the construct validity of the YYIRTL1 was confirmed. The sub-maximal version of the YYIRTL1 seems to be less responsive to training effect compared to YYIRTL1. Unfortunately the YYIRTL1-sub did not show adequate longitudinal and construct validity. Therefore, its use cannot be recommended and other studies are necessary before its use in the routine assessment. The second part of this project focused on the control of the training load and some studies were conducted to examine some methodological aspects of session-RPE assessed with the new Borg CR100 scale®. The rating of perceived exertion has been showed to be a valid indicator of intensity and it is used to assess training loads in sport by multiplying its value for the duration of the session (session-RPE). The most used scale to assess RPE in research and routine practice is the Borg CR10 scale but a new scale has been recently developed. The new Borg CR100 has been suggested to be more fine-graded and to provide ratings less clustered around the verbal anchors. However, the CR100 has not been validated in soccer players and therefore before its validity should be verified. The CR100 (third study) was found to be valid and can thus be used for calculating the S-RPE in top-level soccer players. The two scales are interchangeable and, importantly from a practical point of view, the data collected with the previous scale can be appropriately converted on a CR100 score. In addition, the scores given using the CR100 tended to cluster less on the verbal anchors suggesting this scale may allow more accurate training load determination given it is more fine-graded than the CR10. As a consequence the new CR100 scale can improve the quality of monitoring the training process. The session-RPE was suggested to be assessed 30 min after the end of the session to avoid the influence of the exercise intensity of the activities performed. However the effect of different intensity distributions and time delay on session-RPE has not been examined. The results of the fourth study showed that session-RPE was not influenced neither by exercise intensity distribution nor by the time delay (the rating given immediately after is the same as the one given 30 min after the end of the exercise irrespectively from the intensity distribution of the session). Another methodological aspect that is important in the routine use of session-RPE is the possibility to collect RPE after 48 hours instead of immediately after the training or match, particularly when players are not too collaborative as can happen after a match. To answer this question, it is needed to investigate two attributes: response shift and recall bias (fifth study). Response shift is a complex factor concerning the change in perception due to reconceptualization, reprioritization, scale recalibration after a period of time. Recall bias concern the ability to remember the RPE given and can influenced the retrospective assessment of RPE. The session-RPE provided after 48 hours was not influenced by response shift and recall bias. Therefore it is possible to collect RPE also after 48 hours in both matches and training sessions.