What sort of improvement we need in our research?

This is not a new study but I came across to this study by Brophy et al. titled “Update on the Methodological Quality of Research Published in The American Journal of Sports Medicine“.

They concluded:

Despite a dramatic increase in the number of published articles, the research published in AJSM shifted toward more prospective, randomized, controlled, and blinded designs during 2011-2013 compared with 2001-2003 and 1991-1993, demonstrating a continued improvement in methodological quality.

Some of the aspects in the research methodology they assessed was controlling, randomization, blinding of patients, blinding of people assessing the outcome, funding disclosure and prospective data collection. These are of course important things to consider in “research methodology”.

These improvements are, however, of very little benefit if we don´t improve other aspects. These are of course appropriate use of and understanding in statistical methods. Misconceptions in p-values and confidence intervals, claims of “no difference”, dichotomization of research findings, arbitrary categorization of continuous variables, small sample sizes, negligible understanding of causal inference and mechanistic use of analyses is rampant in our field. I think these are far more important than those usually reported in “improvement reviews”. While blinding and controlling etc. are of course important, they will be no good if there is no knowledge of applied statistical methodology.

Leave a Reply

Your email address will not be published. Required fields are marked *