Data Matters with Fathom! Dynamic Statistics™ software

Activity 4.2

Here are the steps this project requires.

1. Pick two of the categorical attributes in the Rep US Sample and create a cross-tabulation.
2. Find the marginal totals and proportions. Run a chi-square test to get a chi-square value.
3. Use Fathom to find the expected counts.

Here’s how to do each step.

Step 1: Pick two of the categorical attributes in the Rep US Sample and create a cross-tabulation.

On the toolbar at the top of the screen, there is a New Summary Table next to the New Graph icon. Pull a summary table onto the workspace. The summary table has two gray arrows. The down-arrow determines the rows. The right-arrow determines the columns. Select the attribute you want in the rows, drag it over the down-arrow, and let go. The marginal counts for that attribute will appear.

Select the attribute you want in the columns and drag it over the right-arrow. There’s your cross-tab.

Step 2: Find the marginal totals and proportions. Run a chi-square test to get a chi-square value.

Fathom shows the marginal totals right away. Right-click on the table, select Add Formula, type proportion() , and click OK. That adds the proportions, including the marginal proportions.

To run a chi-square test, select Analyze, Test Hypothesis. That adds a new table to the workspace. Click on Empty Test and select Test for Independence. A table appears. There’s something about that table that might fake you out a little. Drag your row variable over Second attribute (categorical): <unassigned>. (I know the diagram that appears in the table indicates that it will go into the columns, but it won’t.) Drag your column variable over First attribute (categorical): <unassigned>.

Find the chi-square value in the analysis that appears.

Step 3: Use Fathom to find the expected counts.

The expected counts appear in parentheses in each cell.

Where are the big differences between expected and observed? Can you guess why those differences appear in the data?