Why do you need to know about statistics?

Lars Valter works as a statistician at Region Östergötland. In the interview below, he tells us why it might be a good idea to make contact with a statistician when planning to conduct a clinical study.

Lars Valter

Why is it important to think about statistical methods already when planning a clinical study?

You often need to decide how large a sample is necessary to achieve a sufficiently large statistical power in a study. This calculation is necessary to make before the study starts, and the discussion on statistical power is often quite comprehensive.

It is a good idea to discuss your study with a statistician before it is started, to make sure the design you choose actually is able to answer the question you have. The discussion with a statistician is also important for knowing what data it is necessary to collect. When it comes to statistical analysis, there are often different methods that can be used. Should you use a factorial design, or a longitudinal design with repeat measurements? Perhaps it is necessary to adjust the analysis in relation to one or several disrupting factors, and then it is of course important to have collected these data. Should the analysis perhaps be stratified? All these usually necessitate discussion of analysis methods. Discussions of design, data collection and analysis methods are best held simultaneously.

What is the most common error that people make?

As most people only consult a statistician once they have collected their data and are ready to do the statistical analysis, we often end up with a discussion of what the data file should look like. It is common for researchers to have to go back to base and spend quite a lot of hours correcting the data file. This could have been avoided if we had had a meeting before the data was collected.

Who can you turn to if you need the help of a statistician?

You would, for example, contact your regional node and ask about statistics advice.