How to Test Against the Previous Period Using a Nominal or Ordinal Variable

Typically, in order to compare one period to a previous period (or one column to the previous column), you would need to use a variable with a Structure of Date/Time in the columns of your table. This is so that the periods are based on a timeline, and the previous period can be identified for testing. However, Displayr allows you to use a nominal or ordinal variable in place of a Date/Time variable to perform statistical tests against the previous period.

This article describes how to go from using the default exception (complement) testing on a table with a nominal or ordinal wave variable:

To a state where a nominal or ordinal wave variable is treated like a date/time variable to test against the previous period:

Requirements

A nominal or ordinal variable representing a wave, month, quarter, etc. If you need to construct one from an existing date/time variable, see How to Create a Categorical Variable from a Date/Time Variable.
- This variable will need to be used in the columns of a crosstab.
Ensure that your Advanced Statistical Testing Assumptions > Test Type > Date/Time/Wave is set to Compare to previous period. See How to Conduct Significance Tests by Comparing to Previous Time Periods for more details.

Method - Use the Data Preparation Agent

Ensure that you have the appropriate variable type. See requirements for more details.
Run the Data Preparation Agent, including the step to identify nominal/ordinal variables to treat as waves in statistical tests.

Method - Set manually

In the Data Sources tree, select the nominal or ordinal variable. See requirements for more details.
In the object inspector , go to Data > Properties and tick Treat as wave in statistical tests.