Empirical evaluation of two-sample statistical tests for differences of stepping phase during insect walking

Area of Science:

• Animal behavior and biomechanics
• Statistical analysis in biology
• Locomotion and movement analysis

Background:

• Analyzing temporal relationships in animal movement requires appropriate statistical methods.
• Previous research has not empirically validated statistical tests for phase data in animal locomotion.
• Cockroach leg muscle activity provides a model system for evaluating statistical test applicability.

Purpose of the Study:

• To empirically evaluate the validity of seven statistical tests for analyzing temporal relationships in animal movement data.
• To identify which statistical tests are reliable for assessing differences in phase data from repetitive movements.
• To provide guidance on selecting appropriate statistical methods for locomotion studies.

Main Methods:

• Seven statistical tests (Student's t, Watson Williams for means and k, variance-ratio F, Wallraff, Mann Whitney, Watson U2) were evaluated.
• Data from cockroach leg muscle activity phases during walking were used.
• Each test was applied to 400 random subset pairs of the data to assess reliability against the null hypothesis.

Main Results:

• Student's t, Mann Whitney, Wallraff, and Watson U2 tests showed potential utility for analyzing phase data.
• The variance-ratio F test and Watson Williams test for concentration parameter k were found unsuitable.
• The Watson Williams test for means demonstrated conditional usefulness.

Conclusions:

• Certain statistical tests, including Student's t and Mann Whitney, are empirically validated for analyzing temporal relationships in animal movement data.
• Specific tests are not appropriate for phase data, highlighting the need for careful test selection in locomotion research.
• Further investigation into the inapplicability of certain tests to phase data is warranted.