ci.asymptotic: Confidence intervals using Normal approximation

In mnel/dsrci: Confidence Intervals on Directly Standardized Rates for Several Parameterizations

Description

Confidence intervals for directly standardized rate estimate. A common approach for constructing confidence intervals around an MLE is to use a normal approximation of the MLE or transformed MLE.

Usage

ci.asymptotic(x, w, level, trans = c("none", "log", "cube.root", "skew"))

Arguments

x	a vector of counts
w	a vector of weights
level	the level of confidence
trans	transformation to apply

Details

Four different transformations are implented - following Ng et al (2008).

• none (no transformation)

• log

• cubic root

• Edgeworth correction for skewness

Value

a vector with the lower and upper bound of the confidence interval.The estimate of the directly standardised rate and the of confidence are returned as attributes to this vector

References

Ng, Filardo, & Zheng (2008). 'Confidence interval estimating procedures for standardized incidence rates.' Computational Statistics and Data Analysis 52 3501–3516. doi: 10.1016/j.csda.2007.11.004

mnel/dsrci documentation built on May 23, 2019, 5:06 a.m.