Statistical audit of original research articles in International Psychogeriatrics for the year 2003

Abstract

Background: At the request of the Editor of International Psychogeriatrics, a statistical audit of all papers published in the journal during 2003 was undertaken by the statistical advisor to International Psychogeriatrics.

Method: Only research papers using inferential statistical techniques were assessed and only the statistical elements of these papers were evaluated. The following issues were addressed: did the authors report a power calculation or address power issues? Did the authors report an appropriate effect size indicator? When multiple univariate statistical tests were used was a correction for type 1 error employed? Did authors demonstrate the adequacy of the data analyzed for the statistical tests employed? Were sufficient details reported to enable an evaluation of the statistical analyses and reported results?

Results: Twenty papers published during 2003 were suitable for analysis. None addressed power issues. About half reported an effect size indicator and about half adjusted the statistical analysis for the effects of multiple univariate statistical comparisons. Few demonstrated the adequacy of the data being analyzed and few provided sufficient detail to evaluate the statistical analyses and reported results. Most papers used the right statistic in the right way.

Conclusion: The statistical quality of articles published in International Psychogeriatrics could be improved by attention to a few relatively fundamental issues.