MaximumBounds: Election Audit Error Bound Functions

Description Usage Arguments Details Value Author(s)

Description

Various bounding functions used to bound the maximum amount of error one could see in a single audit unit.

Usage

1
2

Arguments

Z

The elec.data object.

votes

The data.frame to compute the maximumMarginBounds for. If null, will return all bounds for all precincts in Z.

frac

Fraction of total votes that could be a winner overstatement/loser understatement. So if the worst-case is a 20% flip then enter 0.4

Details

maximumMarginBound return the maximum margin reduction for each precint by computing all margin reductions between pairs of winners & losers and then scaling by that pair's total margin to get a proportion and then taking the max of all such proportions (usually will be the last winner to the closest loser).

fractionOfVotesBound: WPM. The maximum error of the unit is a fixed percentage of the total votes cast in the unit. Typically the 20% WPM is used–meaning a swing of 40% is the largest error possible as 20% of the votes go from the winner to the loser.

Value

Vector (of length of precincts) of maximum possible error for each precinct.

Author(s)

Luke W. Miratrix


elec documentation built on May 30, 2017, 5:51 a.m.