How to Apply Significance Testing to Grid Tables With Lots of Missing Data

This article describes how to go from a grid table with Column Comparison significance testing that includes a lot of missing data...

...to a table where you modify the Column Comparisons to use an Independent Samples z-test instead:

Requirements

A table of a Pick Any - Grid or Number - Grid.
Note, this rule is only provided for backwards compatibility. In general, it is preferable to instead use the Statistical Assumptions settings to control testing with missing data on grid questions (see Column Comparisons with Missing Data and Grid Questions) or using Significance Testing in Tables - Independent Samples Column Means and Proportions Tests.

Select your table.
In the toolbar go to Automate > Browse Online Library > Significance Testing in Tables > Column Comparisons on Grids with Lots of Missing Data.
OPTIONAL: Change the Z-Statistic thresholds for displaying uppercase and lowercase Column Comparison letters.
Press OK.

Please note the following:

This rule runs an independent samples z-test on columns in a grid, testing be default at the 0.05 level and 0.10 level with no multiple comparison corrections.
You must set Show significance to Compare columns in order for the results of this rule to be visible.
The reason that some people use this test instead of the in-built tests in Q is that, in general, this rule uses equal or higher sample sizes to those used in Q. For example, if 20 people have provided data on one cell and 20 on another, but only 2 people have provided data for both cells, then the sample size of 2 would be used by default in Q's tests but 40 in the tests performed using this rule. Although this may seem a clear benefit, the problem is that, in general, the cells are not independent so the assumptions implicit in the application of this rule are rarely met in practice (the default test in Q is a related sample test). Additionally, this rule does not apply any Multiple Comparison Correction nor any other of the settings in Statistical Assumptions.
Testing is conducted on NET and SUM categories.