micd (version 1.1.1)

gaussMItest: Test Conditional Independence of Gaussians via Fisher's Z Using Multiple Imputations

Description

A modified version of pcalg::gaussCItest, to be used within pcalg::skeleton, pcalg::pc or pcalg::fci when multiply imputated data sets are available.

Usage

gaussMItest(x, y, S, suffStat)
gaussCItestMI(x, y, S = NULL, data)

Value

A p-value.

Arguments

x, y, S

(Integer) position of variable X, Y and set of variables S, respectively, in the adjacency matrix. It is tested, whether X and Y are conditionally independent given the subset S of the remaining nodes.

suffStat

A list of length m+1, where m is the number of imputations; the first m elements are the covariance matrices of the m imputed data sets, the m-th element is the sample size. Can be obtained from a mids object by getSuff(mids, test="gaussMItest")

data

An object of type mids, which stands for 'multiply imputed data set', typically created by a call to function mice()

Details

gaussMItest is faster, as it uses pre-calculated covariance matrices.

Examples

## load data (numeric variables)
dat <- as.matrix(windspeed)

## delete some observations
set.seed(123)
dat[sample(1:length(dat), 260)] <- NA

## Impute missing values under normal model
imp <- mice(dat, method = "norm", printFlag = FALSE)

## analyse data
# complete data:
suffcomplete <- getSuff(windspeed, test = "gaussCItest")
gaussCItest(1, 2, c(4,5), suffStat = suffcomplete)
# multiple imputation:
suffMI <- getSuff(imp, test = "gaussMItest")
gaussMItest(1, 2, c(4,5), suffStat = suffMI)
gaussCItestMI(1, 2, c(4,5), data = imp)
# test-wise deletion:
gaussCItwd(1, 2, c(4,5), suffStat = dat)
# list-wise deletion:
dat2 <- dat[complete.cases(dat), ]
sufflwd <- getSuff(dat2, test = "gaussCItest")
gaussCItest(1, 2, c(4,5), suffStat = sufflwd)

## use gaussMItest or gaussCItestMI within pcalg::pc
(pc.fit <- pc(suffStat = suffMI, indepTest = gaussMItest, alpha = 0.01, p = 6))
(pc.fit <- pc(suffStat = imp, indepTest = gaussCItestMI, alpha = 0.01, p = 6))