Application of Model Averaging for Measurement in the Presence of Unknown Familiarization Phase or Fatigue Phase

ABSTRACT

A number of familiarization trials are needed for reliable measurement, particularly for inexperienced subjects. Researchers have studied and developed familiarization protocols that vary by exercise and study population. The pace of familiarization and fatigue may be an individual-level characteristic, so a population-level protocol may not fit all subjects. In this article, the authors view this practical challenge as a statistical problem. We apply piecewise linear regression and model averaging for estimating the true performance level of a subject after familiarization and before fatigue. This statistical method does not require an experimenter to determine when a study participant is familiarized and fatigued. Simulation studies demonstrate that this statistical approach provides a more reliable measurement than the best-fit model. An online interactive applet is provided for those who are not familiar with statistical programming. Detailed instructions for the applet and case study are provided for demonstration.

KEYWORDS:

• Familiarization
• fatigue
• model averaging
• piecewise linear regression

Acknowledgments

This research is supported by the National Research Experience for Undergraduates Program of the Mathematical Association of America funded by the NSF Grant #DMS-1950644. The case study described within was supported by the US Department of Education Hispanic-Serving Institution STEM Grant #P031C160221.

Disclosure statement

No potential conflict of interest was reported by the author(s).