Please enable JavaScript to view this site.

As part of the results output for simple logistic regression, Prism will test if the slope of the simple logistic model (β1) is significantly different from 0, which is equivalent to whether the odds ratio is 1.0. Prism tests this null hypothesis in two ways.

Wald test. This result is shown under the heading “Is slope significantly non-zero”.

Likelihood ratio test. This tests whether  an intercept-only model (a model with only β0) fits the data better than the model containing both β0 and β1.

We include the Wald test so you can compare with other program, but recommend that you rely on the results of the likelihood test. The Wald test is an approximation of the likelihood test, and approximation that is quite good with large sample size.

© 1995-2019 GraphPad Software, LLC. All rights reserved.