randomizeR packageThis list of parameters yields a comprehensive overview of the parameters
used in the randomizeR package.
integer representing the number of balls that are added to the urn in each step.
the level of the t.test in each simulation.
vector which contains the lengths k_1,...,k_l of each block.
This means that the vector bc will have one entry for each block.
numeric vector of length at most 2 specifying the weight(s) for the punishment of
integer representing the initial urn composition.
factor of compression for the sigmoid-time trend.
effect size.
first noncentrality parameter of the doubly noncentral t-distribution.
degrees of freedom (i.a. N-2).
numeric specifying the magnitude of selection bias.
A connection, or a character string naming the file to write to.
logical whether the last block should be filled or not.
final tolerated imbalance. This is the difference in number of patients of groups A and B that is permitted at the end of a trial. Usually this is set to zero.
character vector of labels for the different treatments.
length of the block to be permuted. k should be divisible by
the number of treatment arms.
number of treatment groups (e.g. K=2 if we compare one experimental against one control treatment).
lower bound for the starting value of the poisson distribution.
(second) noncentrality parameter of the doubly noncentral t-distribution.
method that is used to generate the (random) allocation
sequence. It can take values PBR, RAR, HAD, PWR,
EBC, BSD, CR, TBD, UD, and MP.
maximum tolerated imbalance in patient numbers during the trial.
integer for the total sample size of the trial.
name of a variable.
vector of the expected responses of the treatment groups, should have
length K
(i.e. one entry for each treatment group).
object specifying the randomization procedure, see randPar
or createParam.
any R object.
success probability of the biased coin (e.g. in Efron's Biased Coin Design).
vector with patient responses, i.e. each patients resulting value after the treatment.
"cut-off" value in [0.5,1]. This is the ratio of patients up
from which the experimenter imposes selection bias on the data.
numeric indicating the number of random sequences to be generated at random, or missing.
vector of length K. The total sample number N and
all used block lengths (bc) have to be divisible by sum(ratio).
block lengths of the blocks that can be selected equiprobable at random.
randomization sequence (of one block).
randomization sequence (of all blocks).
matrix for the computation of the probabilities in the maximal procedure.
integer or missing specifying the patient index (i.e. position)
of the step in case of step time trend.
a single value, interpreted as an integer, that specifies the seed for the random number generation.
vector of the standard deviations in each the treatment group,
should have length K (i.e. one entry for each treatment group).
numeric vector of length at most 2 specifying the lower and/or upper specified border.
factor of the time trend for further details see type.
character vector indicating which biasing strategy the
experimenter is using (selection bias) and which other bias is present in the
clinical trial (e.g. time trend). All biases included in the vector are
combined (i.e. added up) to form the total bias. Possible values are
"none" (if no bias occurs), "CS" (resp. "DS") (if the
experimenter uses the convergence (resp. divergence) strategy to invoke
selection bias), LinT for linear time trend, LogT for
log-linear time trend, StepT for step time trend, SigT for
sigmoid time trend, PWR for knowledge of all up to the first
observation in each block, MTI the next observation after reaching the
maximal tolerated imbalance is reached will be known to the physican.
numeric specifying the optimal desired value called the target value.
logical parameter for the t.test: Shall the variances be treated
as equal (TRUE= t.test) or different (FALSE= Welch.test).
upper bound for the last value of the poisson distribution.
a variable x.
nonnegative parameter which my be adjusted according to how strongly it is
desired to balance the experiment. If rho = 1, we have Wei's urn design with
alpha = 0. If rho = 0, we have complete randomization.
nonnegative parameter which my be adjusted according to how strongly it is
desired to balance the experiment. a = 0 gives the complete randomization, while the
assignments become more deterministic as a increases.
nonnegative parameter which controls the degree of randomness:
For decreasing a the allocations become deterministic, while for increasing
a this procedure tends to complete randomization.