One of the most important lessons investigators learn in research is the value of collaborating with a statistician. Nothing is worse than trying to analyze data from a study only to discover that incorrect assumptions about power and approach severely limit the ability to draw meaningful conclusions from the completed research. Consulting with a statistician having expertise in a specific area (epidemiology, genetics, etc.) before study initiation and throughout data analysis will maximize the impact of the research and its ability to contribute to science.

1. Before data collection. One could argue that seeking feedback from a statistician during the planning phases of a project is even more important than after the study is completed. Although the investigator has expertise regarding the clinical relevance of an outcome measure, the statistician can weigh in on the most appropriate sampling approach to answer the research question. This evaluation will consider the sample size required, the type of and frequency of measurements, the potential threats to study validity, and the most efficient use of resources.
2. During data collection. The unpredictable nature of research can result in the need to change the study design, perform interim analyses, or otherwise adapt study operations. A statistician can help update sample size estimates and continue to offer feedback on challenging issues, such as how to handle missing data.
3. After data collection. Although it is ideal to engage a statistician before initiating a project, statisticians are often consulted after data have already been collected. At this stage, working with a statistician helps ensure appropriate analysis methods for the types of data collected. Often, the statistician incorporates novel techniques that help clarify study findings. The statistician also advises on potential confounders to consider and the interpretation of results from outliers. Whether or not investigators perform the analyses themselves, it is always prudent to ask the project statistician to check the accuracy of the code and statistical output. Finally, as coauthors on abstracts and manuscripts, statisticians help hone the research message and ensure that conclusions are supported by the data presented.

When should an investigator collaborate with a statistician? Answer: early and often.