Create crosstabs in Protobi by dragging the label of one element onto another:
Crosstab elements individually
You can also drag from the element tree on the left. This is useful when crosstabbing variables that are not on the same tab.
Crosstab elements globally
Drag onto the the "Crosstab" button in the toolbar to define a global banner. Any elements and charts that can take a banner variable will be crosstabbed.
To remove, click on the cross tab button, and select "clear."
Crosstab with two banner variables
Elements can take up to two crosstab banners. Once an element is crosstabbed by one banner, a second "+" button appears, and you can drop another banner.
The columns are sorted by the first level banner, and then the second.
If you double click on either banner, it will switch the order:
Global crosstab with two banner variables
You can apply two crosstab banners globally. Drag the first banner to the crosstab button, then start dragging the second banner over. Before you drop the second banner, hold down the shift key.
This should allow the crosstab button to take both banners:
When you create a global crosstab with two banner variables, an option to "swap" will appear. Below, the project is crosstabbed by specialty (S1) first, then practice start (S3). Press swap to crosstab by S3 first, then specialty.
Crosstab significance tests
Admins can change crosstab significance tests in Project properties. The default mode is Complement testing.
In Complement testing, each column is compared with the average of all other columns (excluding itself). Blue cells indicate the value is significantly higher in the specified column compared to all other columns, and grey indicates significantly lower.
Pairwise testing compares each column with each other individual column. This mode shows detailed superscripts like a traditional crosstab.
Below we see that the percentage of respondents who rated their health as "Excellent" is much higher in column A (Very happy) than columns B, C or D.
Note: Protobi limits significance testing to N>=20 to avoid testing when the sample size is too small.