[Image via wallpapers.ae]
My friend and colleague Merle King from Kennesaw State University sent me the attached yesterday – and it was so good I had to share. Recounts are in the news again – and Merle has a fantastic take that will clarify the issues involved, helping you separate signal from noise in the next few weeks:
Sportsmen across the upper Midwest know the difference between a hunting license and a fishing license. Let’s hope election activists and campaigns do, too.
A hunting license gives you permission to hunt for a specific game animal. A deer license not only permits the hunter to hunt deer, but restricts the hunter to only hunting deer. A fishing license lets you bring up anything in the lake, study it in the bottom of the boat, and keep anything that is not on the endangered list or too small to take home. Fishing trips often lead to “fish” stories.
A recount is a hunting license – not a fishing license. As Wisconsin election officials begin the onerous task of recounting the ballots from the November 8 election, they must follow Wisconsin statute and rule regarding recounts. They will be re-applying tabulation and canvassing procedures to ballots that have already been tabulated and counted to ensure a) accuracy in interpreting voter intent; b) completeness of processing; and c) conformance with Wisconsin procedures for tabulating and canvassing ballots. They cannot introduce new procedures for processing the ballots or new interpretation standards for voter marks. They cannot (and should not) look beyond the legally defined scope and procedures for recounting ballots.
Although recounts are not in and of themselves unusual in elections, their goal is not always clear. Unlike other transaction processing systems in which an audit seeks to produce an accurate and precise result, election recounts tend to produce a reasonable approximation of the initial count. If the recount produces the exact same totals as the election canvass it may arouse suspicion that the recount was mismanaged. If it produces results that are different than the initial canvass, then it can raise suspicions that the election and/or the recount were not conducted properly.
Recounts should validate the uniform and complete application of vetted election procedures to the ballots by producing a total that is “reasonably” close to the original canvass – and reasonableness will be in the eye of the beholder.
Because every election system uses combinations of paper-based and electronic vote capture, there will be the inevitable comparisons between the accuracy of the recount on the two types of ballots. To the frequent surprise of those who do not work in elections, the discrepancies between the canvass and the recount will most likely be concentrated in the paper ballots, not the electronic ballots. Election officials know that paper ballots introduce anomalies in processing – stains on the ballots, folds across timing markets, skewed feeds through scanners, miscalibrated scanners, etc. You can count a million paper ballots, a million times, and potentially get a million different results. The 2004 governor’s race in Washington State illustrated the challenges of producing accurate and consistent recount outcomes. After winning in the canvass and once in the recount, candidate Rossi conceded on June 6, 2005, after the second recount showed him losing by 129 votes. Many observers agree that if there had been another recount, the numbers could have flipped again, ad infinitum.
Like most things in elections, there may be unintended consequences to a recount. Recounts can illuminate vulnerabilities in voting technologies (see lever machines and punch cards in the museum of no-longer-used-voting-technologies), weaknesses in poll worker training, sloppiness in record keeping, inconsistencies in the interpretation of voter intent, lack of adequate accessibility accommodation (for vision- and mobility-impaired voters), ballot design deficiencies, deficiencies in voter instructions and partisanship, internal and external to the elections operations. Recounts are serious business. The outcome of a recount may cost appointed and elected election officials their jobs, and make it harder to recruit election workers for future elections. And it may serve as a fund-raising vehicle or rallying cry for political parties, campaigns or candidates.
Let’s hope that election officials in Wisconsin, Michigan and Pennsylvania approach this recount as a hunt, and not a fishing trip. Focus on the specific goal of accurately and completely recounting the cast ballots and do not unnecessarily expand the scope of the recount beyond this important task. If future audits are needed to improve election procedures, then let’s go about the deliberate business of designing appropriate and effective data collection and audit procedures to evaluate and improve election processes, with due regard to federal and state law, state rules, and generally accepted auditing practices. Auditability of election systems must be designed into these systems and not retro-fitted after-the-fact.
Let’s recount the votes and not rehash the election.
It’s clear that wecount – er, recount – season is upon us once again. Here’s hoping that the various ballot hunters out there remember what exactly they’re licensed to capture.
Thanks to Merle for this piece … be vewwy quiet and stay tuned!