The statistical analyses used in research studies are vital to any investigator if the work is to be interpreted correctly and repeated, yet the description of the methods used is often brief and lacking in detail. Furthermore, inappropriate statistical evaluations can result in inaccurate conclusions being drawn and may result in a waste of money and time, which could be viewed as unethical. Sangseok Lee has published a paper outlining the common statistical errors picked up during peer review and explains how statistics should be used during the planning, execution and reporting of research.
Lee highlights a number of statistical errors frequently made by researchers in the design of studies, in data analysis and in the reporting of the data, including:
- Inappropriate sample size
- Inadequate descriptions of methods
- Unsuitable use of t-tests
- Insufficient explanations of statistical power of tests
- Incorrect deductions from statistical conclusions
Lee then provides some suggestions for how to avoid such mistakes. The involvement of a statistician during the planning stage of the study is critical and can ensure that endpoints, sample size and statistical methods used are all applicable for the work to be carried out, resulting in reliable and reproducible research. In addition, a full description of all the methods employed should be included in any report of the data. Finally, Lee describes the problems associated with misinterpretation of P-values and highlights the six principles for using P-values that has recently been published by the American Statistical Association.