The statistical and practical significance and "effect size" coefficients for the evaluation of physical activity

Abstract

BACKGROUND: An adequate statistical analysis of obtained data is a necessary condition for the right interpretation of results followed by correct formulation of conclusions. The use of appropriate statistical procedures is repeatedly a subject of justified criticism and recommendations in our and other international publications. For example, during the interpretation of results we often "blindly" rely on the statistical significance, and the practical significance of the research is ignored. OBJECTIVE: The main aim of this study is to highlight the formulation of the practical significance of the results and its correct interpretation. We used an analysis of the international studies and own findings from physical activity monitoring. Another aim is to introduce an applicable effect size coefficient as a guide for assessing the practical significance. METHODS: Materials for formulating rules to assess the practical significance and introduction of effect size coefficients was comprised of the 29 international and Czech studies about the level of week and short time (PE lesson, training, exercise lesson) physical activity of Czech children and adolescents (1,129 girls and 938 boys), and adults (5,727 females and 5,426 males) with use of accelerometers, pedometers and IPAQ questionnaires during 2000-2010. RESULTS: Rules for assessing the practical significance consider measurement error, data variability and the size of position from the beginning on the measurement scale. The formulation of the practical significance should include - a) a determination of the minimal values of certain measurement unit that will be limiting for assessing the significancy of the difference; b) the determination of a minimal size of mutual relationship between the expected results and findings. The presentation of the "effect size" coefficients (d, r, r2, K2, ω2) comprises of their definition, the conditions for use as well as the calculation and interpretation of the result. CONCLUSIONS: The presented information about formulating the practical significance and introduction of effect size coefficients should contribute to the improvement of statistical evaluation of data and its consequent correct interpretation of the results.

Keywords

• data
• level of significance
• correlation coefficient
• kinanthropology