SimComp-package {SimComp} | R Documentation |
This package provides simultaneous tests and confidence intervals for one-way experimental designs with one or many normally distributed, primary response variables (endpoints). Means of several groups or dose levels can be compared
If many endpoints are to be analyzed, the covariance matrices (containing the covariances between the endpoints) may be assumed to be equal or unequal for the different groups. Exact or approximate multivariate t-distributions, respectively, are used for quantiles or p-values.
For example, the well-known conventional all-pair comparison of Tukey (1953) can be performed by specifying only a single endpoint and homogeneous (co-) variances. On the other hand, it's also possible to the same, but for many endpoints simultaneously, with heterogeneous covariance matrices and in terms of ratios.
For multiple comparisons of means of heteroscedastic data, see Hasler and Hothorn (2008). The test procedure for multiple endpoints is described by Hasler (2009).
Package: | SimComp |
Type: | Package |
Version: | 1.4.2 |
Date: | 2009-10-22 |
License: | GPL |
LazyLoad: | yes |
Mario Hasler
Maintainer: Mario Hasler <hasler@email.uni-kiel.de>
Hasler, M. (2009): Extensions of Multiple Contrast Tests. PhD Thesis, Gottfried-Wilhelm-Leibniz-Universitaet Hannover.
Hasler, M. and Hothorn, L.A. (submitted): A Dunnett-type Procedure for Multiple Endpoints
Hasler, M. and Hothorn, L.A. (2008): Multiple contrast tests in the presence of heteroscedasticity. Biometrical Journal 50, 793-800.
data(iris) comp <- SimTestRat(data=iris, grp="Species", alternative="greater", Margin=0.8) summary(comp)