Error: Specify the error term in a prior for a main effect or...

In StatisticsNZ/demest: Bayesian Demographic Estimation and Forecasting

Specify the error term in a prior for a main effect or interaction.

Description

Priors for main effects or interactions, such as age effects or age-sex interactions, typically include an error term. The error terms can be specified using function Error.

Usage

Error(robust = FALSE, df = 4, scale = HalfT())

Arguments

robust	Whether to use a t distribution. Defaults to FALSE.
df	The degrees of freedom for the t distribution. Defaults to 4.
scale	Prior for the scale parameter. An object of class HalfT.

Details

If robust = FALSE (the default), then the error term has a normal distribution,

error[j] ~ N(0, scale^2).

The scale parameter has a half-t prior, which is described in the documentation for function HalfT.

If robust = TRUE, then, by default, the error term has a t distribution with 4 degrees of freedom,

error[j] ~ t(4, 0, scale^2).

Users can supply alternative values for the degrees of freedom. The scale parameter has a half-t prior, described in the documentation for HalfT.

Value

An object of class Error

Examples

Error()
Error(robust = TRUE)
Error(robust = TRUE, df = 10)
Error(scale = HalfT(scale = 0.5, max = 1))
Error(scale = HalfT(mult = 0.5))

StatisticsNZ/demest documentation built on Nov. 2, 2023, 7:56 p.m.