Discouraged Voters Affected VP and Senator Elections 209.0

Those with interest in highlighting the success of automated voting like the Comelec, the PPCRV and the Makati Business Club are already extolling the elections as a total success.

I think we should celebrate the success of automated voting soberly and with the thought that it may not have delivered the true will of the people, the key purpose of elections, though in a manner that is very difficult to prove or disprove.

Note: The columns Pres and VP are for estimates using total Pres and VP voters

My preliminary analysis on the latest PPCRV data with 88.5% of the precincts reporting and various claims by Comelec Chairman Jose Melo and historical voter turnout percent indicate that between 1.91 million to 7.92 million voters may have decided not to turn out and vote.

A summary. A more technical review of the data shows that turnout may have been reduced enough because of congestion from clustering and other problems to have affected, given the existing vote count differences, the elections for the Vice President and the last three senator positions.

The question that seems the better one to ask is:

How much would the turn out have been if there were no glitches and long lines on election day?

Common demographic. Furthermore, I am inclined to believe that those discouraged voters came from the demographic of the elderly and the disabled mainly in the urban areas and so may have voter preferences skewed to certain candidates.

Note: You may scroll to the section below on My approach to counting discouraged voters for the computation.

Ought I also move on? I debated hard with myself and consulted friends about this post.

As typical when a passable result from a muddled process happens in the Philippines, the mood on the ground and as abetted by mainstream media seems to be just to move on. Life is too hard so short cuts are ok.

Yet, to me as a teacher, to move on means not to learn from the problems that hounded this latest election exercise.

As I have constantly preached in this blog from the experience of developed nations, the key to development is learning. And learning is obtained by making tacit knowledge explicit for everybody to learn from. So here goes.

Separate problems. On the main, the problem of long lines is different (though with some overlap) from the issue of automation. While there were some problems with the PCOS machine, the line bottlenecks came because people were asked to come early even as most BEIs were not rehearsed enough to be up to speed in the first hours of the morning.

I predicted the morning congestion problem on a March 30 post here and its disappearance in the afternoon (the queue problem being heuristic with total voters bounded).

The Comelec planned for the potential queue problem as early as December 22, 2009 but, I think by hard choice, did nothing to avoid it.

As Ambassador Tita de Villa, who is a member of the Comelec steering committee said in a May 11 interview at ANC, voters always line up in elections. It was ok to line up.

Yet, the Comelec is a service institution and voters its clients. I think a one-hour wait is acceptable to most voters but a 2 – 5 hour waiting time on line and in the heat, when this wait was known and is avoidable with good planning, is intolerable.

Ambassador de Villa ought to smell the flowers, just talk even with PPCRV volunteers at the precincts, and be more empathetic.

On one hand, most members of the Committee probably agree with her with respect to the congestion problem. On the other hand, Comelec may also have been overwhelmed by the continuous bombardment of brickbats coming their way on the automation side.

The voter verification process most BEIs implemented were for the manual system and slow (the waiting room just mitigated the problem – its main value may have been to stop flying voters). It was not adapted to the requirements of a clustered high-volume operation.

The problem could not be avoided because most responsible people were not able to distinguish that the issue of clustering (from lack of budget) was separate from the issue of automation (tech usability and adoption).

My approach to counting discouraged voters. The basic method I used combines basic algebra, an estimate of actual voters, published claims of turnout by knowledgeable people, and historical turnouts on preliminary reports of votes cast.

Estimate of voters who turned out. Based on the latest report of the PPCRV with 88.5% of precincts voting, a linear estimate of final voters who participated in the May 10 elections is between 35.2 million to 36.2 million as below:

Estimate of Expected Turnout. An estimate of expected turnout can be derived from various published pronouncements of knowledgeable officials and accepted historical data as follows:

The high-low estimate of expected voter turnout ranges from 38 million to 43 million from the combination of data sources.

Estimate of reduction in voter turnout. Comparing the data from the two tables above, the table below shows the range of possible reductions based on the diffrent turnuots and the estimates of actual voters from the Pres and VP data sets from PPCRV.

The reduction in voter turnout ranges from a low 1.9 million to a high of 7.9 million:

Note: The columns Pres and VP are for estimates using total Pres and VP voters

The main possible reasons for turnout reduction that can be identified as unique conditions in this 2010 automated elections are related to congestion from clustering, PCOS machine problems, warm temperature, and computer anxiety (tech usability and adaptation on the part of voters and election workers).

Other possible reasons. Other usual reasons can also be listed like missing names from the CVL, voter harassment, vote buying that are typical reasons affecting turnout. A residual value can be allocated to these causes.

Need for confirmation. The algebra used in this estimates is pretty straightforward and the ranges used in the assumptions can be further narrowed down.

Further confirmation can be obtained in a few weeks when actual data on voter participation is obtained and actual voter turnout % can be determined.

By that time , surveys on final national data, and better still regional, provincial, or better still precinct-to-precinct comparison, can be conducted to identify the extent and the reasons for voters deciding not to vote.

Note: If actual turnout comes out to be 71%, for example, a oido approach that compares against historical turnout of 80% or a reduction of 9% multiplied with registered voters of 50 million indicates a notional reduction of 4,500,000 who did not turn out this time. Surely, that is a significant number.

I highlight the issue here mainly to prevent it from being swept under the rug.

Only when we do this type of self-criticism and evaluation are we able to learn and improve ourselves as a people.