The final 10 s of
\(\dot{V}{\text{O}}_{2}\) data, from each stage of the GXT, were averaged and used in subsequent analyses. A paired samples
t test was used to compare the difference in the terminal RPE between the CERT and the EP Scale. Similarly, a paired samples
t test was used to compare the submaximal
\(\dot{V}{\text{O}}_{2}\) recorded at CERT 5 against that recorded at EP 5, as well as between CERT 7 and EP 7. Simple linear regression analysis was employed on the submaximal RPE and
\(\dot{V}{\text{O}}_{2}\) values up to- and including both an RPE of 5 and 7 on CERT, extrapolated to both the theoretical maximal- (CERT 10) and peak terminal RPE (CERT 9), for each participant. For the EP Scale, due to the curvilinear nature of the RPE response (Eston et al.
2009; Lambrick et al.
2011), the corresponding submaximal RPE and
\(\dot{V}{\text{O}}_{2}\) values (≤RPE of 5 and 7) for each participant were log-transformed and then linearly regressed to obtain an appropriate
b coefficient and constant. Thereafter, logged variables were transformed into their inverse to obtain a prediction of
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\), when extrapolated to both EP 10 and EP 9. RPE 5 and RPE 7 were chosen as points for extrapolation as they correspond to moderate- and high exercise intensities, respectively, in children. In adults, accurate predictions of
\(\dot{V}{\text{O}}_{{2{ \text{max} }}}\) have been elicited when extrapolating from both a moderate- (e.g., RPE 13, when employing the Borg 6–20 Scale) and high- (e.g., RPE 15, when employing the Borg 6–20 Scale) exercise intensity (Faulkner and Eston
2007; Lambrick et al.
2009). A one-way analysis of variance (ANOVA) was used to compare the predicted
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\), which had been extrapolated from an RPE of 5 to both RPE 10 and RPE 9, using both CERT and the EP Scale, to measured
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\). An identical analysis was performed when extrapolating from RPE 7. Pearson’s correlations (
r) were used to assess the strength of the relationship between the measured- and predicted
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\) (from RPE 5 and 7, to RPE 9 and 10, for both CERT and the EP Scale). In general,
r values above 0.75 are considered to indicate excellent agreement, and values between 0.4 and 0.74 are considered a fair to good agreement. The uniformity of error was assessed by visual analysis of regression plots, and the standard error of estimate (SEE) was derived from the regression analysis to provide an estimation of random error (Hopkins
2000). In addition, the SEE was divided by the SD of the predicted
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\) to provide a standardised indicator of error, whereby <0.20 is considered a trivial difference, 0.2-0.6 small, 0.6–1.2 moderate, 1.2–2.0 large, and >2.0 very large difference. The relative standard error (RSE) was also calculated by expressing SEE relative to the mean of the criterion (measured
\(\dot{V}{\text{O}}_{{2{\text{peak}}}}\)). All data were analysed using the statistical package SPSS for Windows (PC software, Version 21.0).