Implementing your data analysis plan

For randomized clinical trials, a statistical analysis plan is obligatory. For other types of research we also recommend making an analysis plan.

Frequently Asked Questions

You should always make an analysis plan prior to analysing your data, but it is preferably to already make the plan before you start collecting data.

Your analysis plan should at least address the following topics:

  • the research question in terms of population, intervention, comparison, and outcomes;
  • a description of the (subgroup of the) population that is to be included in the analyses (list inclusion and exclusion criteria);
  • which data sets are used and if applicable, how data sets are merged;
  • data from which time point (T1, T2, etc.) will be used, if applicable;
  • variables to be used in the analyses and how these variables will be analysed (e.g., continuous or categorical);
  • variables to be investigated as confounders or effect modifiers and how these variables will be analysed;
  • missing value treatment;
  • which analyses are to be carried out in which order.

Note that:

  • You may need to consult a statistician about the choice of statistical techniques.
  • Some of the topics in your data analysis plan may be part of your research protocol and can be referred to as such.
  • The volume of data sets increases rapidly as does the number of locations at which this data reside. It may therefore be worth considering distributed analysis, where data remains at its original location.

Examples of plans can be found on the internet. You should however check with your institute's data governance authority, clinical trial office or statistical support group if there is a preferred template.

Your choice of statistical methods may have an impact on the conclusion that you can draw from your data. Think carefully about the hypothesis and the alternatives before running all kinds of statistical analyses and be open for unexpected outcomes. Do not hesitate to seek expert knowledge.

For ever-larger studies, ever-higher degrees of automation of all procedures are required. You should consider a workflow system rather than running each analysis step by hand. Workflow systems can automatically keep track of the exact processing steps. Workflow systems can also run exactly the same series of steps on a series of input files. For very large studies, it may be required to use a system that can automatically recreate and validate workflows on different computer infrastructures.

Text in preparation

Text in preparation