Why are some variable types not used with different graph types?
The Problem: Only Multiple variable graphs can use Categorical and Label variables
Beginning with data in a Multiple variables data table, it is possible to create a graph from any of the families that Prism offers (column, grouped, etc.). While this may sometimes be convenient because you do not have to transfer the existing data to a new type of data table, there are some limitations in this functionality. Specifically, only the Multiple variables family of graphs can use Categorical and Label variables from Multiple variables data tables. When trying to create a graph from one of the other families from a Multiple variables data table, only Continuous variables will be used to generate the new graph. In other words, any Categorical or Label variables will simply be omitted from this new graph.
Short-answer solution: only use continuous variables
The only variable type that non-Multiple variable graphs can work with are continuous variables. This sort of data (see below) is the same sort of data that Prism tables have always expected, and so if making a non-Multiple variables graph directly from a Multiple variables data table, be sure that all of the data you want included are continuous variables.
Slightly longer-answer solution: extract the data to a different data table
A more "appropriate" solution in this case is likely to use the Extract and Rearrange analysis from the Multiple variables data table to generate a new (result) table in the format of your choice. For example, if you have a single continuous response variable (height) and a single categorical grouping variable (gender), you may want to create a column scatter graph displaying the heights of each individual, grouped into columns by the individual's gender. Using the Extract and rearrange analysis to create a new column data table will make creating this visualization simple!
What's the difference between variable types and why is it important?
Beginning in Prism 9.0, variable types could be assigned to variables in Multiple variables data tables. These variable types include:
- Continuous
- Categorical
- Label
Continuous variables are numeric variables, and can take on any value. Common examples of continuous variables are height and weight. We often round these values to the nearest inch or centimeter for height, or the nearest pound or kilogram for weight. However, all of the values will be numeric, and there are an infinite number of values that could possibly exist for these variables. In contrast, categorical variables are variables that can only take on a finite number of values (or levels). As an example, the nationality of a large group of individuals would be a categorical variable: there may be individuals from many different nationalities in the group, but there are only a limited number of responses that you would get. Finally, label variables in Prism are generally not used for any sort of analysis, and exist primarily as a means to identify rows of data. For example, someone's name, their case ID, or some other unique identifier could be entered in a label variable so that you can know which values belong to each individual.
By default, Prism will automatically assign each variable a type based on the type of data that is entered into a Multiple variables data table. Variables that contain text data cannot be continuous, and so must be either categorical or label. Variables that contain only numeric data can be continuous (height), categorical (group number), or label (unique ID). However, Prism will generally default to assigning these variables as continuous (if Prism makes the wrong assignment, it's easy to change the assignment manually).
The problem is that all of the graphs that Prism generates - except for Multiple variables graphs - only understand how to work with continuous values. Thus, if a non-Multiple variables graph is created directly from a Multiple variables data table, Prism will simply omit any non-continuous variables when generating this graph.
Bottom line: non-Multiple variable graphs can only use continuous variables. We recommend ensuring that you only use continuous variables when creating non-Multiple variable graphs, or (better) using the Extract and rearrange analysis to restructure your data first.
Keywords: multiple variables data non-multiple variables graphs