Related Experiment Videos
A regression approach to the analysis of data arising from cluster randomization
International Journal of Epidemiology
|June 1, 1985
Summary
A new generalized least squares regression method analyzes clustered data from randomized and non-experimental studies. This approach offers greater flexibility and adjusts significance levels for within-cluster correlations.
Area of Science:
- Statistics
- Biostatistics
- Epidemiology
Background:
- Cluster randomization is common in experimental studies.
- Non-experimental studies often have cluster-level treatment factors.
- Existing methods like ANOVA may lack flexibility for clustered data.
Purpose of the Study:
- To propose a generalized least squares (GLS) regression approach.
- To analyze data from cluster-randomized and non-experimental studies with cluster-level factors.
- To provide a more flexible analytical method than traditional approaches.
Main Methods:
- Generalized least squares (GLS) regression.
- Accounting for within-cluster correlation.
- Application to experimental and non-experimental study designs.
Main Results:
- The proposed GLS approach is more flexible than analysis of variance (ANOVA).
- GLS provides significance levels adjusted for intra-cluster correlation.
- Demonstrated applicability through two distinct examples.
Conclusions:
- GLS regression is a suitable method for analyzing clustered data.
- The approach enhances statistical rigor in studies with cluster-level effects.
- Offers improved analytical capabilities over standard methods for correlated data.