Executive Summary of NAMFREL's Report on the conduct of its Random Manual Audit Monitoring project for the May 13, 2013 National and Local Elections

from NAMFREL Election Monitor Vol.3, No.2
NAMFREL deployed on May 13, 2013 teams of volunteers (including members of the Philippine Institute of Certified Public Accountants and Jaycees) for Random Manual Audit Monitoring (RMAM), to observe the conduct of the RMA by the Comelec's RMA teams, and to make a report on the results of the manual count vis-a-vis the PCOS count. Of the 234 clustered precincts chosen by the Comelec for the RMA, NAMFREL volunteers were able to submit reports with complete data for only 130, which were the basis for this analysis.
In summarizing and analyzing the RMAM tally form data submitted by the observers, NAMFREL focused on the variances between the Automated Count and the RMA Count considering three metrics contained in official documents and directives coming from the Comelec. NAMFREL’s findings are presented without the benefit of an analysis of the nature or evaluation of the causes of the variances, thus, any variances noted between the PCOS machine count (AES result) and the manual count (RMA Result) are treated as a machine error for the purpose of this report.
Metric I.

The expected accuracy (99.995%) of the
PCOS machines under the Terms of Reference
and Request for Proposal for the AES contract
and as published in the daily papers;
Metric II. Discrepancies of ten votes per candidate per position stated in Section 12 of Comelec Resolution No. 9595;
Metric III. Aggregate difference of ten votes under
Section 13 of the same Resolution (No. 9595,
as amended).
Below is a summary of the findings resulting from the analysis of the RMA as observed by NAMFREL volunteers on the
ground using the three different metrics, discussed in greater detail in the report:
Metric 1 Metric 2 Metric 3

99.995 % accuracy of the PCOS machines
Margin of ten (10) votes per
candidate per position (SECTION
12 of Res. 9595)
Note: 10 or less is allowed
Aggregate variance of 10 votes”
(SECTION 13 of Res. 9595, as
amended by Res. 9647
Note: 10 or less is allowed

1.1) Allowable error of one (1) for every 20,000 marks

Results: Total of 7,210 variances out of 633,297 marks translates to an accuracy rate of 98.86 %.
Results: Met the allowed margin -
118 PCOS or 90.77 % .

Exceeded the allowed margin –
12 PCOS or 9.23 %
3.1) For the 33 Senatorial Candidates per Clustered Precinct or Ballot Box

Results: Met the allowed margin -
43 PCOS or 33.08 %

Exceeded the allowed margin –
87 PCOS or 66.92 %

1.2) Allowable error of one (1) per PCOS –

Results: Met the allowed margin -
15 PCOS or 11.54 %

Exceeded the allowed margin -
115 PCOS or 88.46 %
  3.2) For All Positions:

Results: Met the allowed margin –
32 PCOS or 24.62 %

Exceeded the threshold –
98 PCOS or 75.38 %
Based on the results of the RMA as well as the observation reports submitted by NAMFREL volunteers, NAMFREL recommends to Comelec to, among others, seriously consider the propriety of reusing the same PCOS machines in future elections; design a clearer and simpler audit procedure; and conduct the "root cause determination" for variances at the precinct level in the interest of transparency. NAMFREL also recommends the amendment of the election automation law to define in clearer terms the purpose of random manual audit, taking into consideration its possible role in the resolution of electoral protests.

(NAMFREL's full report on its RMA monitoring project was submitted to the Comelec on July 5, 2013.)