New ‘Risk Limiting’ Audits Promise More Accurate Vote Counts
New 'Risk Limiting' Audits Promise More Accurate Vote Counts
At Common Cause, we work hard to protect every citizen’s voting rights. Among other things, that means reducing barriers to registration, educating voters, and helping voters get access to the polls.
We also work hard to protect voting rights at the back end of the election — during the tabulation and counting process — to make sure outcomes are what voters intended. Unfortunately, computer errors and human errors can produce vote miscounts. There have been elections in which the losing candidate has been declared the winner because of a simple “software glitch” or programming error.
So we advocate strong post-election audits as a check on the election process. Good post-election audits check machine results by manually inspecting some of the paper ballots. Earlier this month, Arapahoe County, CO., piloted a new type of post-election audit – a risk limiting audit – which is the most robust post-election audit yet developed. I was on site in Arapahoe County to observe the pilot as were several reporters who described the audit in The Denver Post and the Colorado Statesman.
A “risk-limiting” post-election audit provides strong statistical evidence that the reported outcome is accurate or leads to a full hand count, which corrects the outcome if it was inaccurate. This contrasts with most post-election audits in use today. The difference is that risk limiting audits check the outcome, and correct it if it is wrong. Conventional audits simply spot check the machines to see if they are correctly counting – they don’t check the actual outcome.
Election officials in Arapahoe County ultimately audited two contested races from the June 24 balloting: the Republican gubernatorial primary race and the Republican primary race for House District 37. They were able to determine to a 90% confidence level that the reported outcomes of both contests were correct. In the world of post-election audits – confirming that the outcome of a given contest is correct to high level of confidence is unique. But it also is vital to ensuring that our voting rights are protected.
This was the second pilot or the risk limiting audit that Arapahoe County has conducted. They conducted the first in 2013 (receiving partial funding for the project from the Elections Assistance Commission [EAC]) All Colorado counties should be performing these ramped-up post-election audits over the next several years (legislation requires full implementation by 2017, but there’s room for extensions.)
Arapahoe County isn’t the only county to pilot this new type of audit. In California, 13 counties have conducted risk limiting audits as part of a project funded by the federal Election Assistance Commission. (A description of pilots can be viewed here. http://www.sos.ca.gov/voting-systems/oversight/risk-limiting-pilot.htm. )
In December 2012, election officials in Cuyahoga County, Ohio, conducted a risk limiting audit pilot on three major contests in the November General Election — the Presidential contest, the Justice of the Supreme Court contest, and the Judge of the Court of Common Pleas Contest. http://boe.cuyahogacounty.us/pdf_boe/en-US/ElectionAudit/Nov2012/NOV12GENPostElectionAudit.pdf
Matt Crane, the Arapahoe County clerk, and his team of election administrators in Littleton should be highly commended for performing the pilot. They worked long hours ironing out kinks in the process. The county even invested $20,000 in high speed scanners to enable the audit to go forward. The work these election officials did will serve all election jurisdictions in the future and help ensure that voters’ rights are protected from the very beginning of the elections process to the very end.