ci.one.prop.cluster: Confidence interval for a single proportion estimated from a...

In m-Py/propint: Testing for interactions in proportions

Description

Computes the confidence interval for a proportion arising from clustered data. That is, individual responses are nested within a cluster; for example disease prevalence may be estimated on the basis of responses given from individuals in different communities. The clustering of responses is taken into account in the estimation of the standard error of the proportion using the method described in Bennet et al. (1991).

Usage

ci.one.prop.cluster(ci, successes, N)

Arguments

ci	The confidence level - pass as a natural number (e.g. 95 for the 95% confidence interval)
successes	A vector describing the number of "successes" (i.e., response = 1) in each cluster.
N	A vector describing the total number of responses in each cluster.

Details

When the intraclass correlation is estimated to be lower than 0, the value is set to 0 as intraclass correlations of less than zero are generally implausible (Donner & Klar, 1993)

Value

A list containing the confidence interval boudaries.

p	The estimated proportion of "success" across all clusters
l	The lower bound of the confidence interval of p
u	The upper bound of the confidence interval of p
se.cluster	The estimated standard error of the proportion
design.effect	The design effect
intraclass.cor	The intraclass correlation

Author(s)

Martin Papenberg [e-mail]

References

Bennett, S., Woods, T., Liyanage, W. M., Smith, D. L. (1991). A simplified general method for cluster-sample surveys of health in developing countries. World Health Stat Quarterly, 44(3), 98-106.

Donner, A., & Klar, N. (1993). Confidence interval construction for effect measures arising from cluster randomization trials. Journal of clinical epidemiology, 46(2), 123-131.

Examples

# Example from Bennet et al. (1991)
ci.one.prop.cluster(95, successes = c(2, 5, 3, 3, 1, 0), N = c(2, 7, 4, 6, 4, 3))

m-Py/propint documentation built on July 26, 2019, 8:47 a.m.