Statistical Analysis of the Post-Election Audit Data 2014 August Primary Elections

Posted: December 30th, 2014 | Author: | Filed under: Statistical Analysis | Tags: , , , , , ,

The Center for Voting Technology Research (VoTeR Center) at the School of Engineering of the University of Connecticut received the data gathered in the post-election audit performed in the State of Connecticut following the August 12, 2014 election. The audit involved the randomly selected 10% of the districts and the audit returns were conveyed by the Office of the Secretary of the State (SOTS) to the VoTeR Center on December 4, 2014. The audit data received by the Center contains 305 records, where each record represents information about a given candidate: date, district, machine seal number, office, candidate, machine counted total, hand counted total of the votes considered unquestionable by the auditors, hand counted total of the votes considered questionable by the auditors, and the hand counted total, that is, the sum of undisputed and questionable ballots. This report contains several statistical analyses of the audit returns and recommendations.
Read the rest of this entry »


A Systematic Approach to Analyzing Voting Terminal Event Logs

Posted: December 5th, 2014 | Author: | Filed under: Other Publications | Tags: , , , ,

A Systematic Approach to Analyzing Voting Terminal Event Logs
Laurent D. Michel, Alexander A. Shvartsman and Nikolaj Volgushev
2014 Electronic Voting Technology Workshop/Workshop on Trustworthy Elections (EVT/WOTE’14)
USENIX Journal of Election Technology and Systems (JETS), Volume 2, Number 2, April 2014 www.usenix.org/jets
August 18-19, 2014, San Diego, CA, USA www.usenix.org

Abstract
This paper presents a systematic approach to automating the analysis of event logs recorded by the electronic voting tabulators in the course of an election. An attribute context-free grammar is used to specify the language of the event logs, and to dis- tinguish compliant event logs (those that adhere to the defined proper conduct of an election) and non-compliant logs (those that deviate from the expected sequence of events). The attributes provide additional means for semantic analysis of the event logs by enforcing constraints on the timing of events and repetitions of events. The system is implemented with the help of commodity tools for lexical analysis and pars- ing of the logs. The system was rigorously tested against several thousand event logs collected in real elections in the State of Connecticut. The approach based on an at- tribute grammar proved to be superior to a previous approach that used state machine specifications. The new system is substantially easier to refine and maintain due to the very intuitive top-down specification. An unexpected benefit is the discovery of revealing and previously unknown deficiencies and defects in the event log recording systems of a widely used optical scan tabulator.

Download full paper:: evt14.pdf


Statistical Analysis of the Post-Election Audit Data 2011 November Election

Posted: June 7th, 2012 | Author: | Filed under: Statistical Analysis | Tags: , , , , , ,

The Center for Voting Technology Research (VoTeR Center) at the School of Engineering of the University of Connecticut received the data gathered in the post-election audit performed in the State of Connecticut following the November 2011 election. The audits involved the randomly selected 10% of the districts and the audit returns were conveyed by the Office of the Secretary of the State (SOTS) to the VoTeR Center on December 22nd of 2011. The audit data received by the Center contains 887 records, where each record represents information about a given candidate: date, district, machine seal number, office, candidate, machine counted total, hand counted total of the votes considered unquestionable by the auditors, hand counted total of the votes considered questionable by the auditors, and the hand counted total, that is, the sum of undisputed and questionable ballots. This report contains several statistical analyses of the audit returns and recommendations.

The VoTeR Center’s initial review of audit reports prepared by the towns identified a number of returns with unexplained differences between hand and machine counts. Audit returns included 45 records with discrepancies higher than 5, with the highest reported discrepancy of 40. It is worth noting that 75% (30 out of 45) of the records that were subject to the follow up investigation already contained information indicating that the discrepancies were due to the human error. Following this initial review the SOTS Office performed additional information gathering and investigation of those 45 records. The final information was conveyed to the Center on May 18th of 2012. The rest of the records (842 out of 887) discussed in this audit report are the original records reported by the towns.
Read the rest of this entry »


Statistical Analysis of the Post-Election Audit Data 2011 September Primary Election

Posted: November 30th, 2011 | Author: | Filed under: Statistical Analysis | Tags: , , , , , ,

The Center for Voting Technology Research (VoTeR Center) at the School of Engineering of the University of Connecticut received the data gathered in the post-election audit performed in the State of Connecticut following the September 2011 election. The audits involved the randomly selected 10% of the districts and the audit returns were conveyed by the Office of the Secretary of the State (SOTS) to the VoTeR Center on November 18th of 2011. The audit data received by the VoTeR Center contains 83 records, where each record represents information about a given candidate: date, district, machine seal number, office, candidate, machine counted total, hand counted total of the votes considered unquestionable by the auditors, hand counted total of the votes considered questionable by the auditors, and the hand counted total, that is, the sum of undisputed and questionable ballots. This report contains several statistical analyses of the audit returns.

This report presents the analysis of 83 records. The data presented in this analysis show that the absolute average reported discrepancy is lower than the number of questionable ballots (0.04 versus 0.93).
Read the rest of this entry »