Comelec's 16-Stage Senate Canvas

The May 2013 Philippine Midterm Elections are now a week-old. But the age-old questions about all Philippine elections remain as may be gleaned from this article by Raissa Robles, Did the Comelec Just Go Rogue?

To help in answering these questions and promote continuous improvement for future elections, I am writing this series of blog posts in which  I am analyzing the Official tally of votes for the 2013 senatorial race as recorded by the superb new-media enterprise, Rapplerdotcom, led by veteran journalist Ms. Maria Ressa.

As in the 2010 National Elections, the Smartmatic Precinct Count Optical Scanners, or PCOS machines were used to record the votes of 39,898,232 voters out of some 52 million registered, a 75% turnout.  A major source of discomfort and skepticism is the manner by which Comelec CANVASSED 304 Certificates of Canvas (COCs) from Municipal and Provincial Boards of Canvas.   It did so in 16 Canvassing Sessions of the National Board of Canvassers. Based on the published Rappler Tally, here is a snapshot of the Comelec Canvassing process:

By my count, 294,683,224 votes were cast by 39 million voters for Senator.

The x-axis represents the 16 Canvas Reports published by Comelec and used to proclaim 12 winning Senators.

The y-axis represents the Incremental Number of Votes tallied or added to previous canvas sessions at each of the 16 stages of canvas, shown below in tabular form as I computed them:

1:	13278521
2:	10185590
3:	28301098
4:	8142941
5:	10295316
6:	26488085
7:	6849981
8:	43750447
9:	43908974
10:	46207109
11:	24012234
12:	11473986
13:	10926269
14:	6667463
15:	2618914
16:	1576296

The National Canvassing process was marred by delays in transmission from the local boards of canvass as detailed by the article of Raissa Robles, resulting in the wide variance of votes tallied at each stage. Notice for example the large number of votes in Canvas stages 8 to 13.

Many bloggers and other analytical types (including me!) have been poring over the Official Data since the 12 winning senators were proclaimed (in 3 batches of 6+3+3) by last Friday, May 18, 2013 just to see if there is anything in the data to verify or falsify accusations and suspicions that the elections may have been rigged or programmed to achieve the final result.

I think it is important for these analyses to take into account the NON-UNIFORM vote increments at each stage of the canvas. 

In the next blog post, I shall consider the time series data itself to shed further light on its provenance.