The Role of Statistics in Fighting the Coronavirus

Source: geralt / Pixabay

Biostatisticians and epidemiologists will play a role in solving the coronavirus crisis and helping to normalize our world. These professionals, who use standard statistical processes to address issues of human health, have specialized skills that, in fact, will be essential in the coming months as we fight this pandemic on behalf of our shared humanity.

As someone who has taught statistics courses for more than two decades, I regularly try to get my students to see the practical utility of stats in solving real-world problems. Because once you can see and appreciate something as having high practical utility, you can begin to appreciate its value.

Soon after my wife, Kathy, and I first moved to the Hudson Valley in 2000, she got a job as a biostatistician working for the Dutchess County Health Department. She’d never had formal training as a biostatistician, but armed with a Ph.D. in experimental psychology from the University of New Hampshire, she had powerful statistical skills that were considered valuable by the people who hired her. And for years in this position, she did great work that helped advance our understanding of the factors that predict chronic Lyme disease.

In my efforts to try to get students to appreciate the utility and value of statistical skills, I regularly teach and write about the powerful and useful nature of truly understanding statistical processes. (See my and Sara Hall’s Textbook: Straightforward Statistics and my book Own Your Psychology Major!)

Following are five specific ways that people who understand statistics will play a role in helping to put the coronavirus to rest.

1. A basic understanding of inferential statistics. Without question, people who understand inferential stats will play a role in solving the coronavirus crisis. Inferential statistics are a family of statistics that focus on the ability to draw inferences found in samples to broader populations of interest. If you’re working for a pharmaceutical company in an effort to try to develop an effective vaccine, you better have someone on your team who gets inferential stats. This person can look at findings found in your sample (e.g., the efficacy of the virus in a sample of, say, 20 young American adults) and determine whether the findings likely generalize to people in general. This is a critical part of the entire process, in fact.

2. A basic understanding of the concept of effect size. Effect size is what it sounds like. It includes a broad group of statistics that speak to how big or small your effect is in your sample data. If you're part of a team working on a vaccine and you have evidence that the effects of the trial vaccine are, in fact, quite small, and barely more effective than placebos, you have important information to drive the potential future productivity of the drug.

3. Skills related to SPSS or R or another cutting-edge statistical software package. In your statistics education, you’ll likely be exposed and trained in either SPSS or R, the most top-tier, cutting-edge statistical software packages. Not everyone can use these packages and I promise you that if you get yourself halfway literate in one or both, you’ll have mad skills that others will rely on. You’ll be able to quickly provide descriptive reports on variables, statistics that examine relationships among different variables, statistics that speak to whether some findings likely generalized to broader populations, and more.

4. An understanding of statistical power. Statistical power is essentially the ability to find a significant statistical effect given a particular methodological design. If you understand how to ballpark statistical power, you can help shape this kind of research from the ground up, including how to manipulate the main independent variables, how many participants to include in the study from the outset, etc. Statistical power is, well, powerful. And understanding it will help give you a huge leg up on designing good research that has the capacity to answer important questions.

5. An understanding of potential errors in hypothesis testing. A core part of statistical procedures these days is found in hypothesis testing—testing if some finding in a sample likely generalizes to a broader population of interest. A good statistician understands that there are potential errors embedded in the process, including the error that sees a finding as “significant” even though you are actually wrong, or a finding that does not come across as significant even though it should have been found as such. The understanding of these concepts is a hallmark of good science and absolutely essential in work on developing effective medical procedures.

Bottom Line: The next time that you hear a college kid complain about having to take a statistics class, just smile, and then explain these five ways that expertise in statistics will play a role in solving the crisis that we face today as a world community. Statistics matters. Hopefully this post makes clear why that is.