Survey of the use of statistical methods in the international journal of prosthodontics.

PURPOSE

This survey aimed to review how scientific articles were reported and to describe the types of statistical tests that had been recently and commonly used in The International Journal of Prosthodontics.

MATERIALS AND METHODS

All 174 articles published in 2012 and 2013 were hand-searched to identify scientific articles (n = 151) and those using at least one statistical test to explore results (n = 111). Editorials, letters, comments, erratum, and award proceedings were excluded. The number and type of statistical tests used within articles were collated, and the 10 commonly used methods were identified and described.

RESULTS

Of the 151 scientific articles, 76% (n = 111) used at least one statistical test and 24% (n = 40) used qualitative methods. Up to 10 tests were used per article, with 237 in total, of which 36 were unique. The 10 most commonly used tests were analysis of variance (ANOVA; n = 34), survival analyses (n = 29), Student t test (n = 19), chi-square (n = 19), Mann-Whitney U (n = 14), logistic regression (n = 13), Wilcoxon signed rank (n = 12), Fisher exact (n = 11), log-rank (n = 10), and Cox proportional hazards (n = 8), and they accounted for 71% (n = 169) of all tests used.

CONCLUSIONS

The vast majority of articles published in recent years in The International Journal of Prosthodontics employed statistical analyses. Across 2 years, nearly 250 tests were completed, including 36 unique tests. Statistical test use was common but diverse.