compute_glsPower: Compute power via weighted least squares

View source: R/wlsPower.R

compute_glsPowerR Documentation

Compute power via weighted least squares

Description

This function is not intended to be used directly, but rather to be called by glsPower - the main function of this package. It expects the design matrix as an input argument DesMat and construct the covariance matrix (if not given as well). These matrices are used to calculate the variance of the treatment effect estimator which is then used to calculate the power to detect the assumed treatment effect.

Usage

compute_glsPower(
  DesMat,
  EffSize,
  sigma,
  tau = 0,
  eta = NULL,
  AR = NULL,
  rho = NULL,
  gamma = NULL,
  psi = NULL,
  CovMat = NULL,
  dfAdjust = "none",
  sig.level = 0.05,
  INDIV_LVL = FALSE,
  INFO_CONTENT = FALSE,
  verbose = 1
)

Arguments

DesMat

object of class DesMat.

EffSize

raw effect, i.e. difference between mean under control and mean under intervention

sigma

numeric, residual error of cluster means if no N given.

tau

numeric, standard deviation of random intercepts

eta

numeric (scalar or matrix), standard deviation of random slopes. If eta is given as scalar, trtMat is needed as well.

AR

numeric, vector containing up to three values, each between 0 and 1. Defaults to NULL. It defines the AR(1)-correlation of random effects. The first element corresponds to the cluster intercept, the second to the treatment effect and the third to subject specific intercept. If only one element is provided, autocorrelation of all random effects is assumed to be the same. Currently not compatible with rho!=0 !

rho

numeric (scalar), correlation of tau and eta. The default is no correlation.

gamma

numeric (scalar), random time effect

psi

numeric (scalar), random subject specific intercept. Leads to a closed cohort setting

CovMat

numeric, a positive-semidefinite matrix with (#Clusters \cdot timepoints) rows and columns. If CovMat is given, sigma, tau, eta, rho, gamma and psi as well as alpha_0_1_2 must be NULL.

dfAdjust

character, one of the following: "none","between-within", "containment", "residual".

sig.level

numeric (scalar), significance level, defaults to 0.05

INDIV_LVL

logical, should the computation be conducted on an individual level? This leads to longer run time and is mainly for diagnostic purposes.

INFO_CONTENT

logical, should the information content of cluster cells be computed? The default is TRUE for designs with less or equal than 2500 cluster cells, otherwise FALSE. Ignored if verbose=0.

verbose

integer, how much information should the function return? See also under Value.

Value

The return depends on the verbose parameter. If verbose=0, only the power is returned If verbose=1 (the default), a list containing power and the parameters of the specific setting is returned. If requested (by verbose=2) this list also contains relevant matrices.


PMildenb/SteppedPower documentation built on Nov. 24, 2024, 3:06 a.m.