AbstractAlthough the number of participants required for a study can be readily determined via power analyses, the number of trials completed in any given condition is typically chosen out of convenience. One of the presumptions of the central-limit theorem is that if sufficient trials have been collected, then the sample mean should approximate the population mean. The purpose of the current study was to evaluate the number of trials required for stable estimates of of basic kinematic variables commonly used for pointing tasks across basic, full-vision and no-vision conditions. Ten participants completed twenty pointing movements to each of three visual targets (27, 30, and 33 cm amplitude) in both full-vision, and no-vision viewing conditions. Running, cumulative means were computed on a trial-by-trial basis for the following kinematic measures: reaction time, time-to-peak-limb velocity, time-after-peak-limb velocity, peak-limb velocity, movement time, and constant error. These running estimates were considered stable when they entered and remained within a +/- 5 % Z-score bandwidth around the final, cumulative estimate. Across all measures and conditions, between 14.5 (+/- 1.3) and 15.4 (+/- 1.3) trials were required to achieve cumulative mean stability. Therefore, if 18 or more are to be collected for a given vision condition, one can be reasonably confident that their mean data are representative of the true condition parameters. That is, under the current experimental conditions, twenty trials is plenty.
Acknowledgments: Acknowledgements: Natural Sciences and Engineering Research Council of Canada (NSERC)