An analysis of the statistics used in psychological papers

Abstract

This article analyses the statistical procedures used in the proceedings of the one of the psychology conferences regularly organized in Romania. 111 papers in six branches of applied psychology were scrutinized. The analysis was structured on descriptive statistics (numerical and graphical), and inferential statistics (parametric and nonparametric). Was also pursued the extent to which papers complied with APA recommendations on the reporting of confidence limits and size effect. Results: 64.9% of the papers reported statistical results. The sections with more statistical based papers are military psychology (85.7%) and industrial-organizational psychology (84%). On the other hand, only 52.4% papers on educational psychology reported statistics procedures. The average of the reported samples was 190 subjects (minimum 10, maximum 1519). A percentage of 26.4% of works includes at least one statistical descriptive indicator, while 31.9% of works containing statistics, have no descriptive indicators. The graphics used are mostly bar and pie. Regarding inferential statistics, 47.2% of work contains at least one statistical test, and 36.1% have none statistical test. The most frequent utilized tests are Pearson r and t tests. It was recorded no single case of reporting effect sizes and confidence limits. Conclusions: The results allow us to appreciate that the papers presented at scientific events, most of them by young psychologists, contain a relative insufficiently matured statistical analysis, and often insufficient connected to the problem studied. We must appreciate, however, as positive the interest in empirical research and statistical analysis of results.