Communitarian Statistics
We use a highly applied and intuitive approach to statistics based on the communitarian principle that statistics are not a domain of the elite that should be used to intimidate readers or laypeople; rather, statistical thinking is essential to competent evaluation of many questions in everyday life. Statistical thinking can be taught, and even people who have never had a statistics course can learn basic principles that will help them with everything from evaluating media reports on health products to making decisions on love and marriage.
For example, we all can intuitively understand that the answer to the question, “Are men taller than women?” is one that must simultaneously incorporate the general response (“Yes”) with the nuance that there are many exceptions. Nonetheless, when confronted with questions of men’s and women’s roles in society, we tend not to apply this nuanced understanding, opting instead for black-or-white answers such as “women are built for childrearing and should stay home and take care of the kids” on the one hand, or “any differences in men’s and women’s roles in childrearing are purely cultural” on the other. A statistical understanding of such questions does not tell us what to do as a society, but it helps us avoid making simplifications that could lead to bad decisions.
For this reason, a major goal of the lab is to foster statistical literacy (numeracy?) as a skill that can be acquired based largely on logic, common sense, familiarity with relevant examples, and practice – without any need for math. We are thus great fans of Randall Monroe, who writes the webcomic XKCD. His particular genius is to apply common errors in statistical reasoning to everyday examples in which their absurdity becomes evident, leaving it to the reader to extrapolate back how equally absurd such errors are in the scientific contexts where they often go unnoticed. Some favourites:
On the absurdity of assuming that correlation tells us nothing about causality
On cognitive biases that prevent us from understanding the role of chance in the world (see further: Daniel Kahneman)
On the absurdity of linear extrapolations, such as that less LDL is always better
On the difference between relative and absolute risk
On forgetting to account for multiple testing when interpreting results (and on publication bias and sensational media coverage of science)
On the absurdity of ignoring prior knowledge when interpreting results of statistical tests
