2022 Introduction to Statistics in Research Mitchell 2nd ed

I N T R O T O R E S E A R C H : D A T A V I S U A L I Z A T I O N & C O M M O N S T A T T E S T S

Table 139: ANOVA Effect Sizes from Sample Dataset

Tomczak & Tomczak suggest bias for eta-squared and partial eta-squared because the effect is for the sample only . A potential solution is omega-squared because it is relatively unbiased. In the figure above, the random effect is lower (.309) compared to the omega-squared fixed-effect (.573). The most important difference is that random effects uses partial pooling and fixed effect does not. The choice of effect size should match your data set. Remember, this is a very small dataset. Let’s assume for this dataset that you wish to use partial eta-squared. There is a way to do this in SPSS, but you can also calculate it by hand.

To arrive at partial eta squared in SPSS (version 27), do the following:

Analyze > General linear model > Univariate > Dependents = Complaints, Factor = Region > Options > Estimate of effect size.

Table 140: Eta Squared in SPSS

Notice that the partial eta-squared is much lower than eta squared. Essentially, you are saying that 65% of the variability of complaints about COVID-19 policies is explained by the region in which the stores reside.

152