Resources

Academic paper examining the use of audits following elections to improve voter confidence.

Using monthly data from the Colorado Department of Motor Vehicles from 2017 to 2021, the research studies a series of reforms to the voter registration process conducted by the DMV between 2018 and 2020. Prior to the reforms, a large majority of unregistered DMV patrons declined the opportunity to register when conducting a transaction. When voter registration became the clear default option for certain unregistered Colorado DMV patrons in 2020, very few of them subsequently opted out, which resulted in a sudden, large increase in the rate at which DMV patrons registered to vote.

In this paper, authors present ALPHA, a flexible risk limiting audit method that can handle sampling without replacement and stratification while learning from audited ballots.

The results in this article suggest that while convenience measures are designed to improve registration and voting rates, they may not result in across-the-board increases that some policymakers and advocates hope for. Nevertheless, there may be some differential impacts of these innovations on registration and turnout, particularly for youth and minority voters.

This report provides practical guidance for conducting tabulation audits, with discussion of audit methods and links to risk limiting audit resources.

This report outlines Virginia's 2022 risk limiting audit process, results, observations, and recommendations for future audits.

This report describes the Carter Center's observation of the 2022 risk limiting audit conducted in Georgia.

This research focuses on how the timing of voter file snapshots affects the most commonly cited advantage of voter file data: accurate measures of who votes.

This research uses difference-in-differences estimates that suggest that same day voter registration disproportionately increases turnout among individuals aged 18–24 (an effect between 3.1 and 7.3 percentage points).

Explains risk limiting audits (RLA) for observers, including steps, what observers should monitor, and how RLAs compare with other post-election checks.

In researching how to ensure a statewide voter registration database’s accuracy and integrity, this analysis develops a Bayesian multivariate multilevel model to account for correlated patterns of change over time in multiple response variables, and label statewide anomalies using deviations from model predictions.

In this paper, authors use snapshots of voter registration files (VRF) over time and machine learning models to test the effectiveness of unsupervised anomaly detection methods in detecting VRF modifications. They find that statistical models comparing administrative districts within a short time span and non-negative matrix factorization are most effective for surfacing anomalous events for review.