This paper is from Session 5F: Bridging the gap between current statistical practice in the workplace and modern statistics
Full topic list
which comes under Topic 5: Statistics education in the disciplines and the workplace

(Monday 14th, 16:15-17:45)

Tradition should not supplant understanding and insight




Technological changes and theoretical advances in the past several decades have created new demands and opportunities for cooperation between the statistical mainstream and application area specialists. Against this, traditions of statistical analysis have become embedded in some places that too often hinder understanding and insight. This paper will discuss: a) approaches that were never a good idea; b) common approaches which are outdated due to advances in modeling and computer technology; c) use of over-simplistic modeling assumptions; d) data pre-processing which removes key information; and e) discipline focus on one specific statistical paradigm, rather than choosing the paradigm that relates best to the research question. How can change best be effected through education and collaboration? What role may applied and mathematical statisticians have in such change?