Recapping parametric assumptions

Let’s summarize the four assumptions and generally how to test for them:

1. Interval or ratio (i.e., continuous) dependent variable
   1. Know your levels of measurement and make sure the dependent variable is continuous.
2. Independent scores on the dependent variable
   1. Know your research design and make sure you have a between-subjects design and that scores from one person do not affect scores from another person (e.g., nested data). There are special considerations for within-subjects designs that we’ll learn about later.
3. Normal distribution of the dependent variable

   1. Visualize the distribution with a histogram and/or density plot of the dependent variable

   2. Test the skew and kurtosis by calculating the z-scores

   3. Test using the Shapiro-Wilk test

   4. Examine the Q-Q plot for deviations from the diagonal line

4. Homogeneity of variances

   1. Visualize the distribution by a boxplot, violin plot, and/or data (jittered recommended) with the dependent variable split by the independent variable

   2. Examine the variances of the dependent variable split by the independent variable

   3. Test using Levene’s test