Thirty-year trends of study design and statistics in applied sports and exercise biomechanics research
Vagenas, G., Palaiothodorou, D., & Knudson, D. (In Press). Thirty-year trends of study design and statistics in applied sports and exercise biomechanics research. International Journal of Exercise Science.
This study documented the change in study design and statistics employed in applied sports and exercise biomechanics research from 1985 to 2014. The sample comprised 676 data based original research reports published in the Journal of Applied Biomechanics (JAB) from 1985 to 2014. Eight design and 10 statistical criteria were extracted from each study. Descriptive statistics were calculated and change in study criteria over time were documented. Design criteria that did not change over time, remaining at relatively low levels of rigor, were widespread (71%) use of small (2-20) sample sizes and examination of numerous dependent variables (26.6% with >13). The number of experimental groups and independent variables also did not change with typically 1 to 2 reported. There was a significant 61% linear increase in randomization of participants into groups, however by 2014 still a minority (39%) of studies were not reporting randomized assignment. Types of statistical analysis showed positive changes over time with a 48% quadratic decrease in descriptive analyses, a 3% linear increase in nonparametric statistics, and a 45% linear increase in reporting parametric statistical analysis. Changes in specific statistical methods included a 9% linear decrease in bivariate correlation and a 73% linear increase in ANOVA. Reporting of assumptions had a 35% linear increase, yet in 2014 sixty-five percent still did not report on meeting statistical assumptions. Changes in test statistics included a linear 56% increase of reporting observed P values and a quadratic 29% increase in reporting effect sizes beginning in the late 1990s. It was concluded there was evidence of small improvements in research design and statistics in JAB over the last 30 years; however, there is still room for improvement to meet higher levels of research rigor and current recommendations on statistical analysis and reporting.