mix_re: mix_re

In rms: Regression Modeling Strategies

mix_re

Description

Mixing Function for Dual Random Effect Sigmas

Usage

mix_re(mre)

Arguments

mre

mixing proportion for dual sigmas for random effects with orm(). These are typically a function of time in a Markov model, scaled to be between 0 and 1 inclusive. When mre is zero the multiplier of normal(0,1) random effects is the first sigma, and when mre is 1.0 the multiplier is the second sigma. Otherwise it is a weighted combination of sigmas. The first sigma is constrained to be positive.

Details

Used by orm to specify weights for mixing two scaled normal(0,1) random effects. The main purpose of mre is to get a more realistic correlation structure in the raw data space when fitting Markov ordinal models with lags in the raw Y values. What works well in practice is to set mre to 0.0 for an observation measured at the earliest follow-up time, and mre to 1.0 when the observation is taken at any time after this initial visit time.

Value

mre unchanged

Author(s)

Frank Harrell

Examples

# mix_re(ifelse(t == 1, 0, 1))

rms documentation built on Nov. 2, 2026, 1:10 a.m.