Details and restrictions of the paired preference test
The paired preference test is not suitable for all tables. It is up to you to make sure that the data in the table is generally suitable for testing, that the sample size is suitable, etc.
The paired preference element is usually specified in a row to compare two rows for each column independently. However, you can specify it in a column to compare two columns for each row independently. The following information assumes specification as a row.
Hierarchical data
This test is unsuitable for running on lower level data when you are working with hierarchical data a hierarchical view of the data. See the topic “Hierarchical Data” for more information.
Rows
The test searches for the nearest two categories preceding the paired preference row (or column) to perform the test on, but it stops searching at a base or another paired preference item, and if it has not found two categories by then it does not perform the test. The test ignores categories at a different net level.
Columns
The test works independently on all columns.
Sample size
This test relies on a large sample, which means that it may not be valid for a small sample--for example, fewer than about 30 cases. UNICOM Intelligence Professional checks for small sample sizes, and does not carry out the test on columns with a base below 30. You can change the minimum sample size if required, using the MinBase property. You can change the minimum sample size if required, by entering a new value in the Minimum Base field in the Preferences tab. You can change the minimum sample size if required, by entering a new value in the Minimum Base field in the Statistics tab.
Multiple response variables
The test is invalid if the two rows being tested can have overlap (that is, one person can belong in both of them). However, there is no way that UNICOM Intelligence Professional can check for this.
Two-tailed test
This is a two-tailed test, which means that it reports all significant differences between the proportions in all of the columns regardless of which row contains the greater proportions.
Notes
The Paired Preference and Net Difference tests will not calculate successfully until the Paired Preference or Net Difference properties are added, as shown below:
Solution 1 (Paired Preference and Net Difference)
Paired Preference
TableDoc.Tables["Table5"].Statistics.Add("PairedPreference")
TableDoc.Tables["Table5"].Statistics.PairedPreference.SigLevel = 5
TableDoc.Tables["Table5"].Statistics.PairedPreference.SigLevelLow = 10
Net Difference
TableDoc.Default.Statistics.Add("NetDifference")
TableDoc.Default.Statistics.NetDifference.SigLevel = 5
Solution 2 (Net Difference)
Select the [Net Difference] option from the following UNICOM Intelligence Reporter menu:
Tables > Properties > Statistics
See also