SimLongiMMM: Simultaneous Inference for Longitudinal Data Using Multiple...
In PhilipPallmann/SimLongi: Simultaneous Inference in Longitudinal Data Settings

Description Usage Arguments Details Value Note Author(s) References See Also Examples

Simultaneous inference for a set of contrasts (linear combinations) of means in longitudinal scenarios. Computes multiplicity-adjusted p-values and simultaneous confidence intervals for comparing groups at multiple time points by combining time-point-specific marginal models as described by Pipper et al. (2012).

1
2
3

SimLongiMMM(data, response, group, time, id, covariates=NULL,
            contrasts=NULL, type="Dunnett", base=1,
            alternative="two.sided", level=0.95, refdist="normal")

`data`	A data frame.
`response`	A character string giving the name of the response variable in `data`.
`group`	A character string giving the name of the (treatment) group variable in `data`.
`time`	A character string giving the name of the time variable in `data`.
`id`	A character string giving the name of the subject variable in `data`.
`covariates`	ccc
`contrasts`	. Default to `NULL`. See examples.
`type`	A character string defining the type of contrast matrix (i.e., the set of comparisons); ignored unless `contrasts=NULL`. Two widespread and often useful choices are `Dunnett` (many-to-one i.e., compare each group against a common reference) and `Tukey` (all-pairs i.e., pairwise comparisons of all groups). Further options are `Sequen`, `AVE`, `Changepoint`, `Williams`, `Marcus`, `McDermott`, `UmbrellaWilliams`, and `GrandMean`.
`base`	An integer specifying the reference group with many-to-one comparisons; ignored unless `contrasts=NULL` and `type="Dunnett"`.
`alternative`	The direction of the alternative to be tested against. Default is `two.sided`. Options for one-sided testing are `greater` or `less`.
`level`	A numeric value giving the simultaneous confidence level (1 - alpha).
`refdist`	.

xxx

A list of class silo with elements

`Results`	A table listing comparisonwise the estimated difference with standard error, lower and upper simultaneous confidence bounds, value of the test statistic, and multiplicity-adjusted p-value.
`CovStat`	The covariance matrix of test statistics.
`CritValue`	The critical value (equicoordinate quantile from a multivariate t-distribution).
`Alternative`	The direction of the alternative.
`ConfLevel`	The confidence level as specified via `level`.
`RefDist`	The reference distribution (multivariate normal or t).
`DF`	The degrees of freedom used for the multivariate t distribution (zero if multivariate normal).
`Corr`	The estimated correlation matrix of time points and test statistics.
`ContMat`	The contrast matrix applied to each point in time.

bla

Philip Pallmann pallmann@biostat.uni-hannover.de

Hothorn, T., Bretz, F., Westfall, P. (2008) Simultaneous inference in general parametric models. Biometrical Journal, 50(3), 346–363.

Pipper, C. B., Ritz, C., Bisgaard, H. (2012) A versatile method for confirmatory evaluation of the effects of a covariate in multiple models. Journal of the Royal Statistical Society, Series C: Applied Statistics, 61(2), 315–326.

SimLongi, SimLongiMix

data(heart)

# Many-to-one comparisons of groups per time point
# taking the third group ("control") as reference

SimLongiMMM(data=heart, response="heartrate", group="drug",
            time="time", id="person", type="Dunnett",
            base=3)$Results

# For comparison:

SimLongi(data=heart, response="heartrate", group="drug",
         time="time", id="person", direction="gpt",
         type="Dunnett", base=3, df="normal")$Results

# With multivariate t as reference distribution:

SimLongiMMM(data=heart, response="heartrate", group="drug",
            time="time", id="person", type="Dunnett",
            base=3, refdist="t")$Results