Is Real-Time Poolside Assessment of Upper Limb Errors in Front Crawl Swimming Technique Reliable and Equivalent to Video Analysis?

Context: Swimming technique is widely believed to influence performance, but this relationship has rarely been tested objectively using a real-time poolside assessment. Objective: To determine the (1) test-retest reliability, interrater reliability, and criterion validity (live vs video) of real-time poolside assessment of upper limb (UL) errors in front crawl (FC) swimming technique and (2) the relationship between UL errors and FC swimming performance. Design: Cross-sectional reliability, validity, and correlational study. Setting: Swim team practice at a college natatorium. Participants: Thirty-nine Division III college swimmers (21 women and 18 men, age = 19 [1] y, swimming experience = 11 [3] y). Main Outcome Measures: Seven UL errors in FC swimming technique, many of which involved unnecessary vertical and mediolateral motions, were assessed in real time from outside the pool during swim practice. Test-retest reliability, interrater reliability, and criterion validity were calculated using Cohen kappa (kappa) and weighted kappa (kappa w). Swimming performance was determined by the participants' best FC events relative to the conference records. The correlation between total UL errors and FC swimming performance was assessed with Pearson r. Results: Cohen kappa and kappa w were moderate for the majority of errors, with the following ranges: 0.46 to 0.90 (test-retest), -0.01 to 1.00 (interrater), and 0.36 to 0.66 (criterion validity). There was a significant correlation between total UL errors and FC swimming performance: r(24) = -.59 (P = .001, R2= .35). Conclusions: Reliability and validity were moderate for the majority of errors. The fewer UL errors swimmers made while practicing FC, the faster their best FC race times tended to be relative to the conference record. UL errors in FC swimming technique explained 35% of the variance in performance.