Concurrent Validity of Power From Three On-Water Rowing Instrumentation Systems and a Concept2 Ergometer

Purpose: Instrumentation systems are increasingly used in rowing to measure training intensity and performance but have not been validated for measures of power. In this study, the concurrent validity of Peach PowerLine (six units), Nielsen-Kellerman EmPower (five units), Weba OarPowerMeter (three units), Concept2 model D ergometer (one unit), and a custom-built reference instrumentation system (Reference System; one unit) were investigated. Methods: Eight female and seven male rowers [age, 21 ± 2.5 years; rowing experience, 7.1 ± 2.6 years, mean ± standard deviation (SD)] performed a 30-s maximal test and a 7 × 4-min incremental test once per week for 5 weeks. Power per stroke was extracted concurrently from the Reference System (via chain force and velocity), the Concept2 itself, Weba (oar shaft-based), and either Peach or EmPower (oarlock-based). Differences from the Reference System in the mean (representing potential error) and the stroke-to-stroke variability (represented by its SD) of power per stroke for each stage and device, and between-unit differences, were estimated using general linear mixed modeling and interpreted using rejection of non-substantial and substantial hypotheses. Results: Potential error in mean power was decisively substantial for all devices (Concept2, –11 to –15%; Peach, −7.9 to −17%; EmPower, −32 to −48%; and Weba, −7.9 to −16%). Between-unit differences (as SD) in mean power lacked statistical precision but were substantial and consistent across stages (Peach, ∼5%; EmPower, ∼7%; and Weba, ∼2%). Most differences from the Reference System in stroke-to-stroke variability of power were possibly or likely trivial or small for Peach (−3.0 to −16%), and likely or decisively substantial for EmPower (9.7–57%), and mostly decisively substantial for Weba (61–139%) and the Concept2 (−28 to 177%). Conclusion: Potential negative error in mean power was evident for all devices and units, particularly EmPower. Stroke-to-stroke variation in power showed a lack of measurement sensitivity (apparent smoothing) that was minor for Peach but larger for the Concept2, whereas EmPower and Weba added random error. Peach is therefore recommended for measurement of mean and stroke power.