Producing Acceptable Results from Statistical Collaboration (303266)

Keywords: Collaboration, Tanzania, LISA, audience, target audience, Acceptability

As practicing statisticians, and collaborators, there is rightfully a heavy focus on correct statistical practices both in terms of use of statistics, and results that answer research questions. This overlooks a frequent problem that arises from working with clients who are extremely unfamiliar with statistics: any method used must be both understood, and more importantly, accepted by other practitioners in their field. While working in a statistics laboratory in Tanzania, I encountered the most extreme case of this issue in advisors who neither understood statistics, nor accepted results contrary to their suggestions. A unique feature of my collaborations in Tanzania is the necessity to correct clear statistical errors, while maintaining results that will pass review by a student’s committee. This poster will show that sometimes it is not enough to provide the correct answer to a client’s research question by using understandable statistics by employing worst case scenario examples that refocus a statisticians mind on acceptability of the research to the intended audience.