Researchers have to communicate the aims, hypotheses, and data of their projects to their statistician colleagues. Often, that’s done in an email with the text, “Thank you in advance for your help. Email me with any questions,” along with an attached scientific proposal PDF and an Excel file of data.
That approach works on some level. It gets the job done. But applied statistics is a broad, nuanced field, more art than mathematical fact. Statistics is like cooking: food tastes better when cooked with love, and statistics are better when the statistician cares about the project.
It’s hard to care about a project when it’s poorly understood. Your statistician became a statistician because he or she likes statistics, not medicine, biology, geology, and so on. Asking your statistician to read through a proposal attached to an email and glean information about biology would be like asking most biologists to read a statistics theory paper and then apply its methods.
Caring about projects is important for statisticians because most of them want to be part of important projects, and because statistics really is like cooking. A chef can throw together an acceptable, healthy meal, but if a chef really cares about the meal, he or she will dig deep and really bring out the flavors. A good statistician can do exactly the same thing: put together a professional, satisfactory analysis, or dig deeper. That may mean a more sophisticated analysis, developing new statistical methods, or it may mean a better statistical report with cleaner language and prettier graphs.
Researchers can help statisticians by explaining the project during the design phase in terms that a family member might understand. For example, tell your statistician, “Disease A has a 20% mortality rate, and we think we can reduce that with a novel drug combination. We plan a clinical trial using hundreds of subjects, whatever you suggest. If you can come over to the lab, you are welcome to sit in on our study design meetings. If this drug combination works, we really can make a difference for our patients.”
Of course, tune your language to the specific statistician. Statisticians come from many backgrounds and experiences. Some have theoretical statistical backgrounds, others have advanced scientific degrees, even medical or veterinary degrees, and they can understand sophisticated scientific language.
Researchers should also understand what kind of statistician they are working with to assess the statistician’s experience with certain analyses. For example, some statisticians work primarily in cancer clinical trials, so they may need more help understanding the subtleties of human motion analysis than a statistician with a background in orthopedics.
Whatever their backgrounds, understanding key features of the project in everyday language helps statisticians identify features of the project that affect statistical analyses, such as correlated data or stopping rules. These kinds of data features often need more than careful analyses, they need caring analyses.
Richard Evans, PhD, PSTAT, is a biostatistician and adjunct professor at the University of Missouri, USA. He was previously an associate professor at Iowa State University, where he worked with researchers from many fields. Richard is also a trained sports journalist, covering sailing and climbing.