rsbWt: Rao-Scott weighting.

In ABS-dev/PF: Prevented fraction

Rao-Scott weighting.

Description

Rao-Scott weighting of clustered binomial observations.

Usage

rsbWt(fit = NULL, subset.factor = NULL, fit.only = TRUE)

Arguments

fit	A stats::glm object.
subset.factor	Factor for estimating phi by subset. Will be converted to a factor if it is not a factor.
fit.only	Return only the new fit? If FALSE, also returns the weights and phi estimates.

Details

Estimates the cluster design effect {d}_{i} as the variance inflation due to clustering by the method of Rao and Scott. Observations are then weighted by the inverse of the {d}_{i}.

Value

A list with the following elements.

• fit: the new model fit, updated by the estimated weights

• weights: vector of weights

• d: vector of {d}_{i} estimates

Author(s)

PF-package

References

Rao JNK, Scott AJ, 1992. A simple method for the analysis of clustered binary data. Biometrics 48:577-585.

See Also

RRor, rsb.

Examples

birdm.fit <- glm(cbind(y, n - y) ~ tx-1, binomial, birdm)
RRor(rsbWt(birdm.fit))
#
# 95% t intervals on 4 df
#
# PF
#     PF     LL     UL
#  0.479 -1.061  0.868
#
#       mu.hat    LL     UL
# txcon  0.768 0.968 0.2659
# txvac  0.400 0.848 0.0737
#

ABS-dev/PF documentation built on Sept. 19, 2024, 10:31 a.m.