Batch-Level Comparison Audits

Topic for all things related to batch-level comparison audits.

@lynn, thanks for the question you added in the Welcome section! I’m moving my response here so we can keep the conversation somewhat categorized.

I’m not sure what you are referencing in your question. Can you remind me where my quote about “two main flavors” comes from? I do provide a definition of batch-level comparison audits in Part 2, page 8 of the RLA Implementation Workbook.

Also, I’ve been meaning to add an estimated sample size for batch-level comparison to my estimated workload chart on page 9 of Part 2. Are you aware of an RLA tool that can conduct a batch-level comparison audit, including determining the initial sample size?

1 Like

I also think batch comparison audits may be a more insightful and reliable way to perform audits when ballot-level comparison audits aren’t feasible. Even if it is hard to hit a satisfying risk level, batch audits provide usable quality information and aren’t unpredictable like ballot-polling audits.

My ElectionAudits: Post-election risk-limiting audit implementation, for batch comparison audits tool is one option.
It is using a very old version of Django, but that probably wouldn’t be too hard to fix.

For lots more software, see: ElectionAuditWareRepo: Repository for Google Group ElectionAuditWare, re post-election audit software and tools.

For example, that ElectionAudits tool performed a batch-level risk-limiting audit in 2010 in Boulder Colorado.


The quote comes from: under the heading Types of RLA.

I think Mark L may have a RLA tool for batch level comparisons.



The first several pilot RLAs in 2008 were batch-level comparison audits, and the original theory (the first several papers, starting in 2007) was developed for batch-level comparison audits. Ballot-level comparison audits were introduced in 2010.

See, e.g.,

Stark, P.B., 2008. Conservative statistical post-election audits, The Annals of Applied Statistics, 2, 550–581. Reprint:

Stark, P.B., 2008. A Sharper Discrepancy Measure for Post-Election Audits, The Annals of Applied Statistics, 2, 2008, 982–985. Reprint:

Hall, J.L., L.W. Miratrix, P.B. Stark, M. Briones, E. Ginnold, F. Oakley, M. Peaden, G. Pellerin, T. Stanionis and T. Webber, 2009. Implementing Risk-Limiting Audits in California, 2009 Electronic Voting Technology Workshop/Workshop on Trustworthy Elections (EVT/WOTE ’09).

Stark, P.B., 2009. CAST: Canvass Audits by Sampling and Testing. IEEE Transactions on Information Forensics and Security: Special Issue on Electronic Voting, 4, 708–717.