Parametric Methods for Comparing Discrete Numerical Variables
Author Information
Author(s): Morten W. Fagerland, Leiv Sandvik, Petter Mowinckel
Primary Institution: Oslo University Hospital, Norway
Hypothesis
How do parametric methods compare to non-parametric methods in analyzing discrete numerical variables?
Conclusion
Parametric methods, particularly the Welch U test, are recommended for analyzing discrete numerical variables as they outperform non-parametric methods.
Supporting Evidence
- The Welch U test performed well in almost all situations considered.
- The Brunner-Munzel test also performed well, except for small sample sizes.
- Non-parametric methods did not perform satisfactorily compared to parametric methods.
Takeaway
This study shows that using certain math methods can help us better understand data that counts things, like how many times something happens.
Methodology
Two simulation studies were conducted to compare hypothesis tests and confidence intervals for discrete numerical variables.
Limitations
The study is limited to specific outcome scales and may not generalize to all discrete numerical variables.
Statistical Information
P-Value
0.0007
Confidence Interval
0.36 to 1.30
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website