Use of statistics in Physical Education: analysis of national publications between 2009 and 2011

Abstract

The purpose of the present study was to verify the pattern of use of statistics in scientific articles published in national journals of the Physical Education area. Thus, all articles published in the 2009-2011triennium of the Physical Education journals stratified at B2 or higher in the current QUALIS CAPES (Field 21) were analyzed. The bibliographic search found 872 articles in the five journals selected, which were classified into no statistics, descriptive statistics and analytical statistics. For the analysis, descriptive statistics was performed and the 95% confidence interval to compare the difference between the proportions and, when necessary, the chi-square test and Logistic Regression. There was a lower proportion of articles with descriptive analysis (7.1%, 95%CI=5.4%-8.8%) compared with those with no statistics (46.3%, 95%CI=42.9%-49.6%) and analytical statistics (46.6%, 95%CI=43.2%-49.9%). The pattern of use of statistical procedures ranged among the five scientific journals and across fields of concentration (Health, Sport, Leisure, Education and Others). The proportion of articles which did not meet the basic assumptions for the use of parametric test was 43.3%. This proportion was not significantly different over the three years analyzed, first and last authors' region of affiliation of or for the first and last authors' degree. The present study points out a worrying scenario regarding the use of statistics in the area of Physical Education, because besides the high amount of work which do not follow the basic assumptions for statistical use , this situation seems to be common regardless of authors' degree, region of affiliation or journal of publication.