R: SE calculations for mean and variance regression models

seVarReg {VarReg}

R Documentation

SE calculations for mean and variance regression models

Description

seVarReg calculates SE for an object of class VarReg. If the result is not on a boundary, the Fishers Information matrix SE are given. The bootstrapped 95% CI can also be calculated. Designed to be called by the plot function plotVarReg, rather than run by a user.

Usage

seVarReg(
  x,
  boot = FALSE,
  bootreps = 1000,
  vector.mean = x$data[, 2],
  vector.variance = x$data[, 2],
  control = list(...),
  ...
)

Arguments

`x`	Object of class `VarReg` to determin the SE (eg. result from `semiVarReg`).
`boot`	Logical to indicate if bootstrapped CI should be calculated. Default is `FALSE`.
`bootreps`	Number of bootstraps to be performed if `boot=TRUE`. Default is `1000`.
`vector.mean`	Vector of `x` values for which the SE of the mean is to be calculated. Default is the `x` covariate from the model.
`vector.variance`	Vector of `x` values for which the SE of the variance is to be calculated. Default is the actual `x` covariate from the model.
`control`	List of control parameters for the bootstrapped models. See `VarReg.control`.
`...`	arguments to be used to form the default control argument if it is not supplied directly

Value

The result is a list of results. This includes:

mean.est: dataframe of overall results from the mean model, including parameter estimates from the model, SEs from information matrix (if boundary=FALSE) and if specified, the SE from bootstrapping with the bootstrapped 95% CI.
variance.est: dataframe of overall results from the variance model, including parameter estimates from the model, SEs from information matrix (if boundary=FALSE) and if specified, the SE from bootstrapping with the bootstrapped 95% CI.
mean.im: dataframe of the expected information matrices for the mean (as appropriate)
variance.im: dataframe of the expected information matrices for the variance (as appropriate)
mean.outputs: dataframe with complete output for mean graphics. Includes the vector.mean as input, and the mean vector (mean.mean) and the SE vector mean.se.im, and bootstrapping outputs as appropriate.
variance.outputs: dataframe with complete output for variance graphics. Includes the vector.variance as input, and the mean vector (var.mean) and the SE vector var.se.im, and bootstrapping outputs as appropriate.

Examples

data(mcycle)
##Fit model with range as a covariate in the mean and the variance model
semimodel<-semiVarReg(mcycle$accel, mcycle$times, meanmodel="semi", varmodel="linear",
knots.m=4, maxit=10000)
##Calculate SE
se1<-seVarReg(semimodel, boot=FALSE)
##not run: with bootstrapping
##se2<-seVarReg(semimodel, boot=TRUE, bootreps=10)
##not run: calculate mean and SE for a given sequence
##test.seq<-seq(min(mcycle$times), max(mcycle$times),
##by=((max(mcycle$times)-min(mcycle$times))/999))
##se2<-seVarReg(semimodel, boot=TRUE, bootreps=10, vector.mean=test.seq)