ExchFixed: Specify a simple exchangeable prior.

In StatisticsNZ/demest: Bayesian Demographic Estimation and Forecasting

Specify a simple exchangeable prior.

Description

A simpler exchangeable prior than Exch. The units are assumed to be drawn from a normal distribution with no covariates and a known standard deviation.

Usage

ExchFixed(mean = 0, sd = NULL, mult = 1)

Arguments

mean	Mean. Optional.
sd	Standard deviation. Optional.
mult	Multiplier applied to sd, if sd is generated automatically. Defaults to 1.

Details

The prior specified by ExchFixed has the form

parameter[j] ~ N(mean, sd^2).

mean defaults to 0. In most cases, non-zero means are ignored, with a warning, when the model is generated. The exception is when ExchFixed is being used to specify an intercept term, when generating simulated data.

If a value for sd is not supplied by the user, then a value is generated when function estimateModel, estimateCounts, or estimateAccount is called. Let s be the standard deviation of data y, or of log(y) in the case of a Poisson model without exposure, and let m be the mult argument. The default value for sd is then

Model	Default
Poisson with exposure	m
Poisson without exposure	ms
binomial	m
normal	ms.

These value are designed to be weakly informative, in that they favour values for sd that are demographically plausible.

ExchFixed can be useful a dimension with few elements, such as sex dimension, where there is insufficient information to justify more complicated priors.

Value

An object of class SpecExchFixed.

See Also

Exch specifies a more complicated exchangeable prior.

Examples

## default standard deviation
ExchFixed()
## user-specified standard deviation
ExchFixed(sd = 0.5)
## reduce the size of the automatically-generated 'sd'
ExchFixed(mult = 0.5)
## increase the size of the automatically-generated 'sd'
ExchFixed(mult = 2)

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