Asking Great Questions: Part of a Theory of Communication in Interdisciplinary Collaborations

Statisticians and data scientists rarely collect the data they analyze, nor do they typically originate the research, business, and policy questions they help answer. Therefore, applied statisticians and data scientists must routinely collaborate with domain experts who have collected or will collect data to help them answer questions from their domain. Complicating efforts to solve interdisciplinary problems is that communication between statisticians and scientists is a serious problem. Statisticians do not fully understand the problems they are trying to solve, and industry, government, and academia increasingly demand that the statisticians they hire can effectively interact with non-statisticians. For many statisticians, inadequacy in communication skills inhibits their ability to make a positive impact on society. To help overcome the challenges of communication, the literature on statistical consulting and collaboration is filled with advice for statisticians to ask good questions. Kimball (1957) states that asking good questions can help prevent the commission of Type III errors, providing the right answer to the wrong question. Lurie (1958) writes that a statistician has the responsibility to ask scientists three "impertinent" questions about their specific hypotheses, how broad or narrow the model and scope of inference will be, and the level of statistical significance the scientist cares about. Hand (1994) discusses methods for mapping the domain expert's research questions to statistical techniques, a task which he considers more difficult than establishing the mathematical properties of statistical techniques. He calls for more research into the topic of "asking good questions" and writes that establishing a valid mapping is more important than applying rigorous mathematics to the problem formulation. In Derr's chapter on "Asking Good Questions" (2000, chap. 5), she asserts that statistical problem-solving begins with questions. The statistician needs to know what questions to ask and how to ask them to get accurate and complete information about the domain expert's problem and the statistical issues in their field. With that information, the statistician can identify the domain problem to be solved and then translate that into an appropriate statistical model. A great question elicits information necessary to successfully accomplish the tasks of the project toward making a deep contribution or strengthens the relationship between the statistician and domain expert. A great question does both. It is characterized by three parts: the question, the answer, and the paraphrasing of the answer to create shared understanding. Statisticians can improve their statistical collaboration skills and thereby increase their potential to help address societal challenges by asking great questions. This article provides a framework for asking great questions and how to implement it in practice. The framework consists of three strategies for asking great questions: preface questions with statements about the intent behind asking the question, follow the question with behaviors and actions consistent with the prefaced words, and model a collaborative relationship via the asking of a great question. The framework for asking great questions can improve the practice of statistics and data science by enabling statisticians to ask domain experts questions that will probe domain experts to improve their understanding of their own research questions, advance their understanding of the statistical analyses that will answer these questions, gather information to improve the statistician's understanding of the domain problem, and improve the statistician's understanding of the statistical issues to enable the statistician to develop appropriate analyses that will answer the domain questions. By asking great questions, statisticians can improve their statistical collaboration skills and thereby increase their potential to help address societal challenges. This framework provides a practical guide for statisticians to improve their collaboration skills and make a positive impact on society.