INTRODUCTION: Evidence for the effects of changes in pacing profile on endurance performance is sparse. Here we report a method for characterizing pacing in 800-m freestyle swimming and for relating changes in pacing to changes in performance time. METHODS: The swimrankings.net website provided 50-m split and final race times for 192 swims of 20 elite female swimmers in 20 national and international competitions between 2006 and 2012. A plot of the log of lap time vs lap number for each swim indicated that the pacing profiles could be characterized parsimoniously by five parameters derived from a linear model: linear and quadratic coefficients for the effect of lap number, reductions in time for the first and last laps, and the residual standard error of the estimate (summarizing random and systematic deviations from the model). Each parameter was then included separately in a mixed model as a simple linear predictor of the log of final race time to determine the relationship between within-swimmer change in the parameter and change in final race time. The magnitude of the relationship was expressed as the effect on race time of two within-swimmer standard deviations of the parameter (a typically low to a typically high value) and evaluated by magnitude-based inference with reference to a smallest effect on swimming performance of 0.3%. RESULTS: The average quadratic profile represented a reduction in pace of 2.1% between the first lap and the nadir in the eleventh lap, followed by an increase in pace of 0.6% by the last lap. Times for the first and last lap were shorter by an additional 6.6% ± 1.0% and 3.6% ± 1.6% (mean ± SD). The residual error was 0.6% ± 0.2%. The within-swimmer analyses showed that the linear and quadratic parameters had likely trivial effects on final time (0.2% and 0.1% respectively), while the first lap, last lap and residual error had possibly trivial-small effects (-0.3%, 0.3% and 0.2% respectively). Uncertainty in all these effects was ±0.3% (90% confidence limits). DISCUSSION: The surprising finding was the curvilinear shape of the pacing profile and the lack of any substantial effect of the slope or curvature of the profile on final time. At this elite level there is apparently nothing to be gained from changing the profile. The possibility of a small negative effect of the first lap presumably reflects the antagonism of strength and endurance fitness, while the possible positive effect of the last lap is of the expected magnitude and reflects the effect of changes in fatigue between races. Our study shows that a fiveparameter mixed-modeling method can be used to assess pacing in distance swimming events.
© Copyright 2014 19th Annual Congress of the European College of Sport Science (ECSS), Amsterdam, 2. - 5. July 2014. Published by VU University Amsterdam. All rights reserved.