Randomness and inference in medical and public health research

Abstract

Background: The purpose of this study was to provide a basis for describing the types of randomness used and statistical inferences reported in the medical and public health research literature. Methods: A study was conducted to quantify the types of research designs and analyses used and reported in medical and public health research studies. A stratified random sample of 198 articles from three top-tier medical and public health journals was reviewed, and the presence or absence of random assignment, random sampling, p-values, and confidence intervals, as well as type of research design, were quantified. Results: Random sampling was used in 58 (29.3%) and random assignment in 21 (10.6%) articles. Most (n=125; 63.1%) research studies did not report random assignment or random sampling; however, statistical inference was applied in more than 90%. Conclusions: Results revealed a concerning overuse of statistical inference. Incorrectly applying statistical inference when not warranted has potentially damaging medical and public health consequences. Researchers should carefully consider the appropriateness of using statistical inference in medical and public health research.