December 1, 1994Biometrics
Variance Components Testing in the Longitudinal Mixed Effects Model
Authors
Discussion
Loading...
Member takes
Implication
Theoretical analysis demonstrates large-sample properties of likelihood ratio tests in longitudinal mixed models, highlighting valid inference on parameter boundaries.
Key Points
- To establish the asymptotic behavior of likelihood ratio tests for evaluating nonzero variance components within longitudinal mixed-effects linear models.
- Applied theoretical frameworks for nonstandard hypothesis testing on parameter boundaries (Self and Liang, 1987) to longitudinal mixed-effects linear models (Laird and Ware, 1982).
- Characterized the large-sample asymptotic distribution of the likelihood ratio test statistic when the null value lies on the boundary of the parameter space.
- Established proper reference distributions required to prevent incorrect inference when testing whether random-effects variance components equal zero.
Cite This Study
Stram et al. (1994) studied this question.