Use of numerically blinded ratings of perceived exertion in soccer : assessing concurrent and construct validity

Purpose: To examine the concurrent and construct validity of numerically blinded ratings of perceived exertion (RPEs). Methods: A total of 30 elite male youth soccer players (age 16.7 [0.5] y) were monitored during training and matches over a 17-wk in-season period. The players’ external loads were determined via raw 10-Hz global positioning system. Heart rate (HR) was collected continuously and expressed as Bannister and Edwards training impulses, and minutes >80% of the players predetermined the maximum HR by the Yo-Yo Intermittent Recovery Test Level 1. RPE was collected confidentially 10 to 15 min after training/matches using 2 methods: (1) a traditional verbal response to the 0 to 100 category-ratio “centiMax” scale (RPE) and (2) numerically blinded RPE centiMax scale (RPEblind) with the response selected manually via a 5 × 7-in tablet “slider.” The RPE and RPEblind were divided by 10 and multiplied by the duration to derive the sessional RPE. Linear mixed models compared ratings, and within-subject repeated-measures correlations assessed the sessional RPE versus HR and external load associations. Results: There were no differences between the RPE and RPEblind (0.19; 95% confidence intervals, −0.59 to 0.20 au, P = .326) or their session values (13.5; 95% confidence intervals, −17.0 to 44.0 au, P = .386), and the ratings were nearly perfectly correlated (r = .96). The associations between the sessional RPE versus HR and external load metrics were large to very large (r = .65–.81), with no differences between the RPE methods (P ≥ .50). The RPEblind also reduced verbal anchor clustering and integer bias by 11% and 50%, respectively. Conclusions: RPEblind demonstrated concurrent and construct validity versus the traditional method, and may be used in situations where practitioners have concerns regarding the authenticity of athlete ratings.