Carter Center Report on an Analysis of the Representativeness of the Second Audit Sample, and the Correlation between Petition Signers and the Yes Vote in the Aug. 15, 2004 Presidential Recall Referendum in Venezuela
This study was conducted by The Carter Center and confirmed by the OAS in response to a written request from Sumate presented to The Carter Center Sept. 7, 2004. Sumate asked that The Carter Center evaluate a study performed by Professors Ricardo Hausmann and Roberto Rigobon.
The Hausmann/Rigobon study states the second audit conducted Aug. 18-20 and observed by The Carter Center and the OAS was based on a sample that was not random and representative of the universe of all voting centers using voting machines in the Aug. 15, 2004, recall referendum. 1 The study further indicates that the correlation coefficient (elasticity) for the correlation between the signers and the YES votes for the sample was 10 percent higher than that for the universe. The Hausmann/Rigobon study came to these conclusions through an analysis of the exit poll data, petition signers data, and electoral results data provided by Sumate.
1 Objectives of the Carter Center Study
1. Determine the correlation between the number of signers of the presidential recall petition and the electoral results of the Aug. 15 recall referendum.
2. Compare the characteristics of the universe of voting machine results with those of the sample for the 2nd audit performed Aug. 18.
3. Determine the universe from which the sample generation program used Aug. 18 was drawn.
The sample drawing program used Aug. 18 to generate the 2nd audit sample generated a random sample from the universe of all mesas (voting stations) with automated voting machines. The sample was not drawn from a group of pre-selected mesas. This sample accurately represents different properties of the universe, including the accuracy of the machines, the total YES and NO votes and the correlation between the YES votes and signer turnout.
There is a high correlation between the number of YES votes per voting center and the number of signers of the presidential recall request per voting center; the places where more signatures were collected also are the places where more YES votes were cast. There is no anomaly in the characteristics of the YES votes when compared to the presumed intention of the signers to recall the president.
The second audit showed a high accuracy of the voting machines with discrepancies of less than 0.1 percent. The sample was analyzed, and it does not have different properties than the universe. The sample generation program was analyzed as part of the 2nd audit process and again in this study. Both studies showed that the sample does not operate on a subset of the universe, thus hiding or masquerading some of the properties of the universe. Consequently the results of the 2nd audit accurately confirm the electoral results of Aug. 15.