Related Experiment Videos
Implementation of exact and approximate randomization tests for polynomial growth curves
E D Schneiderman1, S M Willis, C J Kowalski
1Department of Oral and Maxillofacial Surgery and Pharmacology, Baylor College of Dentistry, Dallas, TX 75266-0677.
International Journal of Bio-Medical Computing
|July 1, 1994
Summary
This study introduces two PC programs for comparing growth curves using exact or approximate randomization tests. These tools handle missing data and do not require Gaussian distribution for measurements.
Area of Science:
- Biostatistics
- Statistical Software Development
Background:
- Comparing growth curves is essential in longitudinal studies.
- Existing methods may have limitations with missing data or distributional assumptions.
Purpose of the Study:
- To introduce two user-friendly PC programs for comparing multiple growth curves.
- To provide methods for analyzing growth data with missing measurements.
- To offer both exact and approximate randomization tests for hypothesis testing.
Main Methods:
- Development of two menu-driven, stand-alone PC programs using GAUSS386i.
- Implementation of exact and approximate randomization tests for comparing growth curves.
- Accommodation of incomplete measurement sequences in longitudinal studies.
Main Results:
- The programs facilitate the comparison of mean growth curves across multiple groups.
- Exact P values can be computed using the exact randomization test.
- The approximate test estimates P values and provides confidence intervals, handling missing data effectively.
Conclusions:
- The developed software offers flexible and robust tools for growth curve analysis.
- These programs are valuable for researchers dealing with incomplete longitudinal data.
- The availability of both exact and approximate tests enhances the utility of the software.