Tuesday, June 14, 2016

How To... Perform a Chi-Square Test for Independence in Excel @YouTube #ILoveStats #201

Adding to my recent set of new videos, my latest is using the Chi-Square Test for Independence. This is a non-parametric test that is a bit like a two-way ANOVA. It is applied when you have two categorical variables from a single population. It is used to determine whether there is a significant association between the two factors. 

In this video, the example used is based on an opinion poll. Voters might be classified by gender (male or female) and voting preference (Left Wing, Centre, or Right Wing). We could use a chi-square test for independence to determine whether gender is related to voting preference. Note that it is just a relationship that is established (or not) - we do not establish causation.


I will be adding this test to the Advanced Business Data Analysis module next year. The module currently does two-way ANOVA, which assumes that the data are normally distributed. If the data are not normally distributed, the ANOVA test is not appropriate - use Chi-square instead.

No comments:

Post a Comment