From Verified Voting:

Rhode Island is making good on its promise to road-test risk-limiting election audits, following 2017 passage of legislation by the Rhode Island General Assembly, requiring them. Beginning with the presidential primary in April 2020, Rhode Island will become the second state to require these audits to verify election results. A “risk limiting” audit checks if the election result is correct. Specifically it checks the counting of the votes. A “risk-limiting” audit limits the risk that the wrong election result will be certified. It can catch errors which change the result and correct a wrong result.

For more background on the legislation, visit here: https://www.commoncause.org/democracy-wire/rhode-island-adopts-risk-limiting-election-audits/ and here: https://webserver.rilin.state.ri.us/BillText/BillText17/HouseText17/H5704.pdf

To prepare for next year’s full implementation, the Rhode Island Board of Elections will conduct three pilot audits on January 16 and 17 at 50 Branch Avenue in Providence, Rhode Island beginning at 9:30 a.m. These pilot audits will be conducted with local election officials from Bristol, Cranston and Portsmouth, Rhode Island.

The purpose is to test three different methods for conducting risk-limiting audits. A variety of tasks will be conducted over two days, including hand tabulation of a sample of ballots. For purposes of planning future audits a time and measurement study will be conducted over the two days.

Rhode Island will demonstrate three types of audits:

• Ballot-level comparison audit for Bristol precincts: This method A ballot-level comparison audit is an audit that is similar to checking an expense report. First the audit checks that the subtotals add up to the reported totals. And then individual ballots are checked against how they are recorded by the machine – similar to checking receipts against numbers in a spreadsheet.
• Batch-level comparison audit for Cranston precincts: This method will check a random sample of ballot “batches” and compare the total vote count of those batches against the voting machine’s count. A batch will consist of between 250-300 ballots.
• Ballot-level polling audit for Portsmouth precincts: This method will check a random sample of ballots with the reported outcome, not against the voting machine’s record of those votes. This is comparable to an exit poll. But instead of using the voters’ responses to questions, it checks marking of the actual ballots. Enough ballots are sampled to give election officials confidence that the outcome is correct.

Read the full Verified Voting article here