Educating clinical researchers on how to design a study with mock tables & graphics as well as simulated data

29 July 2013

One of the main problems in educating novice clinical researchers is that they have difficulty seeing what exactly their final paper is going to say once it is completed. Is it too much, too little? Is it feasible given our resources? Does it align with what we said we were going to do or even with what our real goals might be?

While the answers to these questions are relatively easy for somebody who has been doing clinical research for a while, to the novice it's all unclear, making the design a trip into to the unknown. Problem is, when novices get to the end the results might not be necessarily a good surprise.

Good news is that there are at least two approaches that can be used to address this problem.

  1. Draw mock tables and graphics: By drawing mock tables and graphics a researcher can test a number of things. Are the tables and graphics providing an answer to the objective outlined in the Introduction section? Are the results aligned with the available data variables described in the methods? Do we need other analytical methods that might not have been described in the Methods? Are the tables and graphics providing a clear enough message to most readers? Do these tables align with the results that will be explained in the Discussion?
  2. Simulate the data and create real tables and graphics with fake data: This is the next level after creating mock tables and graphics. The advantage here is that the researcher will not only be testing all the points described in the previous item, but will also be forced to create scripts that can later be used with the actual data. Of importance, simulated data is particularly important if the definitive data will take a while to obtain, either because it is being collected or because it hasn't been properly formatted yet.

by Ricardo Pietrobon