How to test for Normality with Prism

Analyzing stacks of data

1. Create a Column data table. In older versions of Prism, this was called "one grouping variable".  
2. Enter each data set in a single Y column. So all values from each group are stacked into a column. 
3. Click "Analyze... Statistical analyses... Column statistics"
4. Prism offers three options for testing for normality. Check one, or more than one, of these options.

Analyzing residuals from nonlinear regression

A residual is the distance of a point from the best-fit curve. One of the assumptions of linear and nonlinear regression is that the residuals follow a Gaussian distribution. You can test this with Prism. When setting up the nonlinear regression, go to the Diagnostics tab, and choose one (or more than one) of the normality tests. 

Analyzing residuals from linear regression

Prism's linear regression analysis does not offer the choice of testing the residuals for normality. But this limitation is easy to work around. Run nonlinear regression, choose a straight line model, and you'll get the same results as linear regression with the opportunity to choose normality testing. This is just one of many reasons to fit straight lines using the nonlinear regression analysis.