Statistics: From arbitrary to accurate reporting is the need of the hour: Results of a preliminary study

Abstract

Statistics has to be used appropriately in order to draw credible result. Statistical flaws have been detected in many biomedical journals but have not been evaluated for and published in dental journals. Hence, the present study was envisaged. Original Research Articles published in Contemporary Clinical Dentistry from January to December 2012 were downloaded from the journal's website and were assessed in the present study. A structured, validated questionnaire was developed to assess the quality of reporting statistics. Descriptive statistics were used to represent the data. A total of 62 articles were retrieved for the assessment. Only 1 out of 55 articles mentioned a prior sample size calculation. One-sixth of randomized clinical studies have mentioned the method used for generating random sequence, and none explained the process of allocation concealment. Only one-tenth of the articles with nonrandomized study design compared baseline characteristics. The majority of the studies (97.3%) employing parametric tests have not mentioned whether the assumptions have been checked. Except for one, no other articles have reported confidence interval at least for the primary outcome. None, except two in our study, albeit having mentioned multiple P values, made an attempt to adjust the same using any of the tests. We found poor reporting of statistics and inferred that more attention is needed from both the researcher and journal editor.