Specify Poisson sampling design

Specify a design where units are sampled independently from the population, with known probabilities. This design is often used theoretically, but is rarely used in practice because the sample size is variable. This function calls ppscov to specify a sparse sampling covariance matrix.

poisson_sampling(p)

Arguments

p

Vector of sampling probabilities

Value

Object of class ppsdcheck

See also

ppscov, svydesign

Examples

data(api)
apipop$prob<-with(apipop, 200*api00/sum(api00))
insample<-as.logical(rbinom(nrow(apipop),1,apipop$prob))
apipois<-apipop[insample,]
despois<-svydesign(id=~1, prob=~prob, pps=poisson_sampling(apipois$prob), data=apipois)

svytotal(~api00, despois)
#>         total     SE
#> api00 3911368 279007

## SE formula
sqrt(sum( (apipois$api00*weights(despois))^2*(1-apipois$prob)))
#> [1] 279007.3