Creating Contingency Tables with StatCrunch

Data Set Overview

• Data Set Name: Two categorical variables
• Columns:
  • var1: 10 values
    • Value 'b': 4 rows
    • Value 'a': 6 rows
  • var2: 10 values
    • Value 'c': 6 rows
    • Value 'd': 4 rows

Steps to Create a Contingency Table

1. Access StatCrunch
   • Navigate to the Stat menu.
   • Select Tables, then Contingency, and With Data.
2. Set Variables
   • Choose var1 as the row variable.
   • Choose var2 as the column variable.
3. Compute the Table
   • Click Compute to generate the contingency table.

Contingency Table Details

• Outputs:
  • Headers: Unique values from each column appear in the first row and column.
  • Cells: Display frequency of each variable pairing.
  • Totals: Shows row totals, column totals, and total number of pairs.
• Chi-Square Test: Automatically computed but not ideal for small cell counts.
  • A warning will be displayed due to low cell counts.

Customizing the Contingency Table

• Edit Options:
  • Go to Options > Edit to customize the table.
  • Display Options:
    • Example: Select Row Percent.
  • Click Compute after selecting options.
• Output Example:
  • 2/3 (66.6%) of 'a' values pair with 'c'
  • 1/3 (33.3%) of 'a' values pair with 'd'
  • 'b' values have a 50/50 split between 'c' and 'd'

Alternative Hypothesis Tests

• Choosing Different Tests:
  • Navigate to Options > Hypothesis Test.
  • Deselect Chi-Square due to small sample size.
  • Select Fisher's Exact Test for Independence.
  • Note: Some tests restricted to 2x2 tables.
• Additional Options:
  • StatCrunch can also compute confidence intervals (not selected in the example).
• Compute New Results
  • Click Compute to see results for Fisher's Exact Test.

Conclusion

• Fisher's Exact Test is more suitable for small samples.
• StatCrunch offers customizable options and multiple hypothesis tests for contingency tables.