samplingR (version 1.0.1)

srs.samplesize: Simple Random Sample size.

Description

Calculates the required sample size in order to achieve a relative or absolute sampling error less or equal to the specified for an specific estimator and an optional confidence interval in simple random sampling.

Usage

srs.samplesize(
  N,
  var,
  error,
  alpha,
  estimator = c("total", "mean", "proportion", "class total"),
  p,
  mean,
  replace = FALSE,
  relative = FALSE
)

Value

Number of instances of the sample to be taken.

Arguments

N

Number of instances of the data set.

var

Estimated quasivariance.

error

Sampling error

alpha

Significance level to obtain confidence intervals.

estimator

One of "total", "proportion", "mean", "class total". Default is "total"

p

Estimated proportion. If estimator is not "proportion" or "class total" it will be ignored.

mean

Estimated mean. If relative=FALSE it will be ignored.

replace

Whether the sample to be taken can have repeated instances or not.

relative

Whether the specified error is relative or not.

Details

If the sample size result is not a whole number the number returned is the next whole number so srs.samplesize>=n is satisfied.
To estimate sample size of estimators "total" and "mean" estimated quasivariance must be provided. If the error is relative then estimated mean must also be provided.
To estimate sample size of estimator "proportion" and "class total" estimated proportion must be provided. If p is not specified sample size will be estimated based on worst-case scenario of p=0.5.
N must be always be provided for calculations.

Examples

data<-rnorm(200, 100, 20)
n<-srs.samplesize(200, var(data), estimator="total", error=400, alpha=0.05);n
sample<-data[srs.sample(200, n)]
srs.estimator(200, sample, "total", alpha=0.05)$sampling.error