This study compares, using simulated data, two methods for estimating variance components in generalizability (G) studies. Traditionally variance components are estimated from an analysis of variance of sample data. The alternative method for estimating variance components is restricted maximum likelihood (REML). The results indicate that REML provides estimates for the components in the various designs that are closer to the true parameters than the estimates from analysis of variance.
Get full access to this article
View all access options for this article.
References
1.
CornfieldJ.TukeyJ. W. (1956) Average values of mean squares in factorials. Annals of Mathematical Statistics, 27, 907–949.
2.
CronbachL. J.GleserG. C.NandaH.RajaratnamN. (1972) The dependability of behavioral measurements. New YorkWiley.
3.
FleishmanE. I. (1978) A method for simulating nonnormal distributions. Psychometrika, 43, 521–532.
4.
HarvilleD. A. (1977) Maximum likelihood approaches to variance component estimation and to related problems. Journal of the American Statistical Association, 77, 320–340.
5.
LinC. Y.McAllisterA. J. (1984) Monte Carlo comparison of four methods of estimation of generic parameters in the univariate case. Journal of Diary Science, 67, 2389–2398.
6.
MarcoulidesG. A. (1986) GIBBY: a multiple group multivariate random number generator. Computer Program, University of California, Los Angeles.
7.
MarcoulidesG. A. (1987) An alternative method for variance component estimation: application to generalizability theory. Unpublished doctoral dissertation, University of California, Los Angeles.
8.
MarcoulidesG. A. (1988a) A comparative study of variance component estimation methods. In HouckE. (Ed.), Proceedings of the Decision Science Institute. Vol. 1. Pp. 163–164.
9.
MarcoulidesG. A. (1988b) GIBBY: a computer simulator for statistics. Collegiate Microcomputer, 7, 67–72.
10.
MuthenL. (1983) The estimation of variance components for dichotomous dependent variables: applications to test theory. Unpublished doctoral dissertation, University of California, Los Angeles.
11.
OlsenA.SeelyJ.BirkesD. (1976) Invariant quadratic unbiased estimation for two variance components. Annals of Statistics, 4, 878–890.
12.
RobinsonD. L. (1987) Estimation and use of variance components. The Statistician, 36, 3–14.
13.
SAS Institute. (1979, 1982, 1986) SAS user's guide. Raleigh, NC: SAS Institute.
14.
SearleS. R. (1971) Topics in variance component estimation. Biometrics, 27, 1–76.
15.
ShavelsonR. J.WebbN. M. (1981) Generalizability theory: 1973–1980British Journal of Mathematical and Statistical Psychology, 34, 133–166.
16.
SmithP. (1982) A confidence interval approach for variance component estimates in the context of generalizability theory. Educational and Psychological Measurement, 42, 459–465.
17.
StroupW. W.EvansJ. W.AndersonR. L. (1980) Maximum likelihood estimation of variance components in a completely random BIB designCommunications of Statistics, 9, 725–756.
18.
SwallowW. H.MonahanJ. F. (1984) Monte Carlo comparison of estimators of variance components. Technometrics, 26, 46–57.
