scaledTMix.psat: Fits a partially saturated model to t-statistics

Description

Fits two-component mixture model to t-statistics, where each t-statistic has a different mixing proportion pi0, but all t-statistics shares a common scale factor.

Usage

scaledTMix.psat(tstat, df, upper0 = 2)

Arguments

tstat

A numeric vector t-statisics

A numeric scalar or vector of degrees of freedom

upper0

A numeric scalar giving the initial upper bound to search for scale factor. It will be automatically increased if the initial bound is not appropriate.

Value

A numeric scalar, being estimated scale factor, with attributes

equiv.sd.ncp

A numeric scalar being the equivalent standard deviation of the noncentrality parameters, given the noncentrality parameter being nonzero.

degrees of freedom

fit

results from optimize

n2ll

Negative 2 times the log likelihood

pi0

A numeric vector of 0 or 1, which are the maximum likelihood estimate of pi0

Details

This function fits a two-component mixture model, with a central t component with probability pi0, and a scaled central t component with scale factor greater than 1. The model assumes a different pi0 for each t-statistic, but a common scale factor for all t-statistics. Maximum likelihood estimates are obtained. upper0 is only used as a hint of the upper bound of the scale factor. If it is too small, it will be automatically increased.

Examples

Run this code

# NOT RUN {
set.seed(99927220)
(tstat=rt(5,1))
scaledTMix.psat(tstat,1)
# }