Example of the Test for Response Homogeneity

This example uses the Car Poll.jmp sample data table, which contains data collected from car polls. The data include demographics about the individuals polled and information about the cars that they own. You want to explore the relationship between marital status and origin of car. You also want to test for the homogeneity of the responses. That is, you want to test to see whether the distribution of the origin of cars is the same for married and single respondents.

There are two versions of this test, the Pearson and the Likelihood Ratio, both provide chi-square statistics and p-values.

Select Help > Sample Data Library and open Car Poll.jmp.

Select Analyze > Consumer Research > Categorical.

Select country and click Responses on the Simple tab.

Select marital status and click X, Grouping Category.

Click OK.

Click the Categorical red triangle and select Test Response Homogeneity.

Figure 3.7 Test Response Homogeneity

The Share Chart indicates that the married group is evenly split between ownership of American and Japanese cars. In the single group, Japanese cars are the most frequently owned. The test for response homogeneity has a significance of about 0.08. Therefore, the difference in response probabilities across marital status is not statistically significant at an alpha level of 0.05.