elec.strat (version 0.1.1)

first.r: Obtain a Vector of Sample Sizes with Total Number of Samples Fixed

Description

first.r, next.r, and propSizes obtain sample sizes so that the total number of samples is fixed. first.r uses the first.r algorithm, next.r uses the next.r algorithm, and propSizes finds a vector of sample sizes that is proportional to stratum sizes. See Higgins, Rivest, Stark for details about the first.r and the next.r algorithms.

Usage

first.r(Z, n, t = 0, asTaint = FALSE, asNumber = FALSE, M = NULL, initSamp = NULL)
	next.r(Z, n, t = 0, asTaint = FALSE, asNumber = FALSE, M = NULL, initSamp = NULL)
	propSizes(Z, n)

Arguments

Z

A strat.elec.data object.

n

The fixed number of samples. When initSamp is provided, first.r and next.r will run for n iterations, adding samples iteratively to initSamp; first.r and next.r will produce a vector of sample sizes with a total of sum(initSamp) + n) samples.

t

Value of the observed maximum, either as the MRO, as taint, or as the overstatement of the margin in votes.

asTaint

Set asTaint = TRUE if t is the maximum observed taint.

asNumber

Set asNumber if t is the maximum observed overstatement of the margin in votes.

M

A priori margin. If NULL, M defaults to 1.

initSamp

An initial choice of sample sizes. Used in call of get.first.r.samp and get.next.r.samp to reduce computational time.

Details

The arguments t, asTaint, asNumber, M are used in first.r and next.r in the call of getEbsMargin. The getQ function is bypassed to increase efficiency. propStrat obtains a vector of sample sizes that has exactly n samples. It obtains such a sample by sorting values of k*sum(Z$strat$n)/Z$strat$n, where k = 0, 1, ..., in increasing order and allocating a sample to the strata corresponding to the first n values. Ties are broken by choosing the strata with the largest number of batches. See Higgins, Rivest, Stark for details.

References

M. Higgins, R. L. Rivest, P. B. Stark. Sharper p-Values for Stratified Election Audits

See Also

See get.first.r.samp, get.next.r.samp, and get.prop.samp for finding sample sizes given constraints on the p-value and the largest observed overstatement.

Examples

	data(MN_Senate_2006)
	MN_Senate_2006.strat$strat$audit <- first.r(MN_Senate_2006.strat, n = 150)
	BaB(MN_Senate_2006.strat)
	MN_Senate_2006.strat$strat$audit  <- next.r(MN_Senate_2006.strat, n = 150)
	BaB(MN_Senate_2006.strat)
	MN_Senate_2006.strat$strat$audit  <- propSizes(MN_Senate_2006.strat, n = 150)
	BaB(MN_Senate_2006.strat)