Function getblock creates block factors for designs with replications, repeated measurements or split plot designs. Function rerandomize.design rerandomizes an experimental design.
getblock(design, combine=FALSE, ...)
rerandomize.design(design, seed=NULL, block=NULL, ...)Function getblock returns
a single factor with block information (for split plot designs without replication
or replicated designs without randomization restrictions)
or a data frame with several blocking factors (for designs with randomization restrictions and replication).
Function rerandomize.design returns a class design object;
note that it will not be possible to add center points after re-randomization,
i.e. if required, center points have to be added before using the function.
an object of class design, which is a design with replications
or repeated measurements or a split plot design
logical with default FALSE. It has an effect for replicated blocked
and splitplot designs only: If TRUE, all blocking information
is combined into a single factor. Otherwise, a
data frame with separate identifiers is returned.
integer number for initialization of the random number generator
(needed for repeatable rerandomization)
In R version 3.6.0 and later, the default behavior of function sample
has changed. If you work in a new (i.e., >= 3.6.-0) R version and want to reproduce
a rerandomization from an earlier R version (before 3.6.0),
you have to change the RNGkind setting by
RNGkind(sample.kind="Rounding")
before running function rerandomize.design.
It is recommended to change the setting back to the new recommended way afterwards:
RNGkind(sample.kind="default")
For an example, see the documentation of the example data set VSGFS.
character string giving the name of a block factor (only for unreplicated
designs that do not have any prior blocking or split plot structure;
meant for block randomization of designs created with function oa.design)
currently not used
Since R version 3.6.0, the behavior of function sample has changed
(correction of a biased previous behavior that should not be relevant for the randomization of designs).
For reproducing a re-randomization that was produced with an earlier R version,
please follow the steps described with the argument seed.
Ulrike Groemping
The purpose of function getblock is to support users in doing their own analyses
accomodating randomization restrictions like blocking and split plotting
with R modeling functions.
The reason for including designs with proper replications is that these are randomized in blocks by packages DoE.base and FrF2 and partly by DoE.wrapper. While the package author does not consider it generally necessary to analyze these with a block factor, function getblock makes it easy for users with a different opinion (or for situations for which time turns out to be important in spite of not having explicitly blocked for time) to run an analysis with a block factor for the replication.
For unreplicated split plot designs, a whole plot identifier is returned; the design itself contains the plot information via the settings of the whole plot factors only. Thus, it may be useful to be able to create the plot identifier.
For replicated block or split plot designs, there is a randomization hierarchy that will depend on how the experiment was actually conducted. Therefore, a dataframe is generated the columns of which can be used in the appropriate way by a statistically literate user.
Function rerandomize.design rerandomizes a design. This can be useful if
the user wants to obtain unblocked replications (packages DoE.base and FrF2
usually randomize in blocks on time) or wants to freely randomize the center point
position over the whole range of the experiment (or a block, respectively),
or if the user wants to also randomize the
blocks (rather than randomizing the block units to the experimental blocks
outside of the design),
or if the user wants to do block randomization on a block factor specified with
the block option
for a design created with function oa.design or pb (which do not offer
explicit specification of blocking).
It can also be useful for ensuring a randomization that has little correlation
between run order and model matrix columns; this correlation can e.g. be checked with
the help of function corrPlot, using the option run.order=TRUE.
## a blocked full factorial design
ff <- fac.design(nlevels=c(2,2,2,3,3,3), blocks=6, bbrep=2, wbrep=2, repeat.only=FALSE)
getblock(ff)
getblock(ff, combine=TRUE)
rerandomize.design(ff)
ff <- fac.design(nlevels=c(2,2,2,3,3,3), replications=2, repeat.only=FALSE)
getblock(ff)
ff <- fac.design(nlevels=c(2,2,2,3,3,3), replications=2, repeat.only=FALSE)
try(getblock(ff))
## a design created with oa.design
small <- oa.design(nlevels=c(2,2,2,2,2,2,2,2,8))
rerandomize.design(small, block="J")
Run the code above in your browser using DataLab