ICOTS8: Scientific Programme: Paper 4D2

This paper is from Session 4D: Innovations in teaching statistics at the tertiary level

Full topic list

which comes under Topic 4: Statistics education at the post secondary (tertiary) level

(Monday 12th, 14:00-16:00)

Visualising inference

Presenter

James Curran (University of Auckland, New Zealand)

Co-author

Chris Wild (University of Auckland, New Zealand)

Abstract

Plots and plot-annotations that facilitate the making of visual inferences have a number of benefits in statistical analysis and statistics education. We explore types of plots for a range of applications from standard intro-stat situations to the effects of factor variables in generalized linear models. The benefit for a data analyst is being able to get an instant, if approximate, sense of what is going on in the data, including being able to see such things as confidence intervals for effect sizes. From an educational viewpoint, the addition of inferential information to raw-data plots provides a promising approach to preventing the misconceptions students have as a result of a disconnect between their data and the resulting inferential summaries. As a small side-benefit, there is a simple least squares problem embedded in our methodology which may provide a useful extension problem for use in theory courses on regression.