Reliability of W´ using time-trials under laboratory conditions

Christoph Triska



Critical power (CP) and the maximum work above CP () serve as important parameters to characterize high intensity exercise (Jones et al. 2010). However, recent research reported poor reproducibility of (Galbraith et al. 2011; Karsten et al. 2015; Triska et al. 2015) and it was suggested that differences in the environmental conditions (e.g. terrain, seating position) or exhaustive durations might have influenced between tests.

Therefore, the aim of this study was to determine under controlled laboratory conditions using time-trials (TT). We hypothesized non-significant differences and a high reliability for .



Ten well-trained cyclists (MAP: 329±41 W) participated in this study. Reliability was assessed across three tests comprising three exhaustive TT each. These TT were 12, 7, and 3 min in duration and interspersed by 30 min passive rest. Tests were separated by at least 72 h. TTs were performed on a Cyclus2 ergometer (RBM Elektronik GmbH, Leipzig, Germany) where the participants’ personal bikes were mounted on. To replicate real-world TT cycling, participants consequently utilised a self-pacing strategy were gearing was adjusted throughout efforts using the virtual gear changer mounted to the handlebars. was estimated using a linear regression where power is plotted against the inverse of time (1.s-1):

P = .t-1 + CP

where P is the power output (W)  and t is the time (s). The standard error (SE) for was calculated in absolute and relative values for each test.

A repeated measures ANOVA assessed the differences between the trials and significant main effects were followed-up by Bonferroni post-hoc procedures. Reliability was analysed using the intra-class correlation coefficient (ICC) and the coefficient of variation (CoV) (Hopkins 2000). Statistical significance was set at P<.050.



Table 1 illustrates the results of the tests. Non-significant differences between repeated tests were revealed for (F1.513,9.935 = 2.951; P = .115). However, significant differences were found for the absolute and relative SE (F2,18 = 10.865; P = .001; and F2,18 = 5.428; P = .014, respectively). Using Bonferroni post-hoc procedures, absolute SE of Test I was significantly higher compared to Test II and Test III (P = .008-.042).


Discussion and Conclusions

Using TT efforts in trained cyclists, a learning effect for between Test I-Test II was identified. Furthermore, the ICC was low between Test I-Test II, but high for Test II-Test III. This is supported by CoV values for Test I-Test II which were notably above the recommended upper limit of 10% (Atkinson and Nevill 1998), however, improved to acceptable values in the following tests. The relative SE of the first test was slightly above the accepted upper limit of 10% (Ferguson et al. 2013), but well below that in the two following tests. Furthermore, absolute SE was significantly lower in Test II and Test III. To accurately determine our findings suggest a familiarisation trial even when testing trained cyclists.

Full Text: PDF


  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.


Journal of Science and Cycling (JSC). eISSN: 2254-7053. Cycling Research Center, Inc. All Rights Reserved. Carretera Jaén s/n Km. 426.5. Pulianas-Granada CP 18197 (Pol. ASORIA).