Editorial Comment

Abstract

© 2014 The Authors Teaching Statistics © 2014 Te Teaching Statistics hats off to the G Oswald George prize winners 2013 for focussing attention on problems with excessive use of the term ‘population’ in statistics education and discussion of possible alternative approaches. In their article, Lu and Henning (2013) quote an oftenused classical definition of a population as ‘the set of all entities of interest in a particular statistical study’, discuss what is wrong with the population concept, include suggestions of terms such as ‘data generating process’ and the ‘statistical population’ of Kass (2011), and emphasize the importance of understanding and questioning the assumptions of a statistical model, and how a statistical model is a ‘lens’ (Wild, 2006). There is much discussion that can be opened here. Is the prolonged use of the word ‘population’ just habit, or is it because of avoiding words such as ‘theoretical’, ‘model’, ‘distribution’ or ‘parameter’, and lack of energy to find alternative expressions suitable for various student levels? Lu and Henning point to the problem of understanding randomness in experiments; the challenges of observational studies require even more discussion with students. In coming to understand concepts of random representativeness of data with respect to the issues of interest, the words ‘general situation or population’ can be very useful. For example, ‘if the data can be considered randomly representative of a general situation or population with respect to the issues of interest, then we can use the data to comment on the general situation or population with respect to these issues’. Clearly the word ‘population’ is appropriate for some statistical situations, particularly those involving sampling from real populations of entities whether they be marbles or people. Such situations tend to be how young students are introduced to both chance and data. Then the term ‘population proportion’, which does make sense, tends to appear and the challenges of sampling to estimate a population proportion provide rich scenarios for student thinking and learning, with focus on ‘worry’ questions rather than delusions created by emphasis on sampling schemes available to organisations with vast resources. Categorical data do indeed provide valuable doors to statistical concepts and thinking at a