Lecture Notes: Using the McNemar Test with SPSS

Introduction

• Presenter: Dr. Grande
• Topic: Using the McNemar test in SPSS, particularly for pre-test/post-test research designs.
• Test Type: Non-parametric.

Requirements for McNemar Test

• Data Characteristics:
  • Variables must be paired or matched.
  • Variables must be dichotomous.
  • Minimum frequency for each category in variables should be at least 5 (although the test can be conservative if less than 5).
• Sampling: Participants must be randomly sampled.
• Groups: Must be mutually exclusive.

Study Example

• Participants: 90 participants, categorized by substance use.
  • Pre-test: 70 using substances, 20 not using.
  • Post-test: Changes observed in substance use status.

McNemar Test Steps in SPSS

Method 1: Using Descriptive Statistics

1. Path: Analyze > Descriptive Statistics > Cross tabs.
2. Setup:
   • Pre-test in row, Post-test in column (interchangeable without affecting results).
   • Check off McNemar under statistics.
   • Optionally check risk for magnitude and odds ratio.
   • Add percentages for row and column under cells options.
3. Results:
   • Observed changes in substance use (15 switched to no use, 2 switched to use).
   • Significant McNemar test result: .002 < .005 (statistically significant).
   • Risk estimate: Odds ratio for pre-test is 33.

Method 2: Using Non-Parametric Tests

1. Path: Analyze > Non-Parametric Tests > Legacy Dialogs > Two Related Samples.
2. Setup:
   • Pre-test to variable 1, Post-test to variable 2.
   • Check McNemar, uncheck others like Wilcoxon.
3. Results:
   • Similar outcomes with fewer outputs.
   • Test statistics confirm significant difference (value .002).

Conclusion

• Findings: Statistically significant difference between pre-test and post-test results.
• Outcome: Null hypothesis rejected.
• Assistance: Dr. Grande offers help for any questions or concerns regarding the use of McNemar test in SPSS.