logo
The Danger of Blindly Repeating Statistical Tests on Clinical Data
8th December 2022 By MARS Research Hub
blog

 

Clinical research is a demanding and tedious process. It's very easy to run into a dead-end, and it can be difficult to trace back the cause of the problem.
it is very common for us at MARS to receive the after rejection statistical service requests
Often, it's because the researcher blindly replicated the statistical test on their data from a similar publication. they believe that it is the right approach, but it is not.


and I can assure you that the process is very similar to diagnose a patient, the patients come to you with some symptoms that could be anything, and you will carry on further investigations, labs, radiology and so on to properly diagnose the patients then treat them
just repeating the same statistical tests from a similar published study means you are dealing with the symptoms only!
you'll need to identify signals, objectives, of your data, even you may head back to the drawing board.
so before just repeating let’s start:
First things first, properly calculate your sample size , define the stat power and the level of significance
then identify exactly your dependent, and independent variables and also the confounding ones
then tailor your research objectives in statistical terms, depending on the variable’s distributions and nature
always define why, where, when to use a stat test or model, your published clinical research is affecting millions of lives