TEACHING STUDENTS TO USE THE CHI-SQUARE TEST WHEN OBSERVATIONS ARE DEPENDENT

摘要

The purpose of statistical analysis is to extract and assess information contained in data and draw inferences based on the analysis. Students, especially those doing experimental work and not majoring in statistics, need to apply the statistical methods they have learned to their collected data. These data often arise in diverse situations and standard methods cannot be applied directly. The statistical methodology must be developed to address problems associated with such data. Here we use the chi-square test with dependent data to illustrate methodology modification in order to analyse data correctly. The underlying mathematics is summarised, but students are only expected to analyse such data using software and to check modification procedure using simulations. The aim here is to show students that statistical procedures can be modified then applied even though the data does not fit assumptions for a standard procedure. The importance of simulation is evident as an additional benefit for learning.