Fixing India’s VVPAT-based audit of EVMs

GS Paper II

News Excerpt:

The Election Commission of India (ECI) has been criticized for limiting the Voter Verified Paper Audit Trail (VVPAT) - based audit of Electronic Voting Machines (EVMs) to a token exercise and for a lack of transparency in the matter.

What is a VVPAT machine?

  • The VVPAT machine is attached to the ballot unit of the EVM and provides visual verification for the vote cast by a voter by printing a slip of paper with the voter’s choice on it. 
  • This slip of paper, containing the candidate’s serial number, name, and party symbol, is displayed in the machine behind a glass window, giving the voter seven seconds to verify her vote. Following this, the slip falls into a compartment underneath.
  • No voter can take the VVPAT slip back home, as it is later used to verify votes cast in five randomly selected polling booths. 
  • The idea is that by allowing for a physical verification of the electronically cast vote, both voters and political parties have greater faith in the process — that their vote is being recorded correctly.

VVPAT-based audit of EVMs:

  • The VVPAT-based audit of EVMs is a simple problem of statistical quality control.
  • It is very similar to the “lot acceptance sampling technique” that is widely used in industry and trade.
    • If the number of defectives found in a randomly drawn statistical sample is less than or equal to a specified acceptance number, the lot (or ‘population’) is accepted; otherwise, the lot is rejected.
  • Here, a ‘defective EVM’ is defined as one with a mismatch between the EVM count and the VVPAT’s manual count of voter slips due to EVM malfunction or EVM manipulation. 
  • Unlike industry and trade, where a few defectives in the sample may be tolerated, the acceptance number will have to be ‘zero defective EVM’ in the context of elections. 
    • Even if there is a single instance of mismatch between the EVM count and VVPAT manual count in the randomly drawn sample of EVMs, the ‘population’ of EVMs from which the sample was drawn should be ‘rejected’. 
    • ‘Rejection’ here means non-acceptance of the EVM counts for that ‘population’ and doing manual counting of VVPAT slips for all the remaining EVMs of that ‘population’. 
    • In such a scenario, the election result should be declared only based on the VVPAT count.
  • Thus, the VVPAT-based audit of EVMs involves three essential elements — 
    • It could be all the EVMs deployed in an Assembly constituency, a Parliamentary constituency, a State as a whole, India as a whole, a region (or group of districts) within a State, or any other. 
      • The population size (N) could vary widely depending on how we define the ‘population’.
      • A clear definition of the ‘population’ of EVMs from which the statistical sample would be drawn. 
    • Determination of statistically correct and administratively viable sample size (n) of EVMs whose VVPAT slips will be hand-counted.
    • Application of the ‘decision rule’, viz., in the event of a mismatch between the EVM count and the VVPAT count in the chosen sample of ‘n’ EVMs, the hand counting of VVPAT slips will have to be done for all the remaining (N-n) EVMs forming part of that ‘population’.

Lacuna from the ECI’s side regarding VVPAT-based audits of EVMs:

  • ECI has not specified the ‘population’ to which its sample size relates. 
  • It has not explained how it arrived at its sample size. 
  • It has been silent on the 'next steps' in the event of a mismatch between the EVM count and the VVPAT count in the selected sample, and it has ignored reported occurrences of mismatch.
  • A system of VVPAT-based audits of EVMs in which these three vital issues have been left vague or unaddressed is categorically unacceptable.

Why ECI’s sample size is erroneous:

  • The hypergeometric distribution model should form the basis of the sampling plan for the VVPAT-based audit of EVMs because it is an exact fit. 
  • In the discussion that follows, we assume the percentage of defective EVMs in the population (P) to be 1% and calculate sample sizes for various population sizes for a 99% probability of detecting at least one defective EVM. 
    • We also compute the probability that the ECI-prescribed sample size of “five EVMs per Assembly constituency” will fail to detect a defective EVM for different population sizes. 
    • The great merit of the hypergeometric distribution model is that the sample size is the greatest when P is very close to zero (which is what the ECI claims it is), and it becomes lesser as P increases.

  • As seen from Table 1, when the population size (N) of EVMs is 100, the sample size (n) required is 99; that is, it is nearly as big as the population size. 
    • As N increases, n also increases but at a much slower rate and ‘hits a plateau’ beyond some point so that further increases in population size have no effect on the sample size.

  • As seen from Table 2, if we define the EVMs deployed in an Assembly constituency or Parliamentary constituency as the ‘population’, then in view of the smaller population sizes (N), the sample sizes (n) required are rather big.
    • Hence, both these choices for ‘population’ are administratively unviable.

  • As seen from Table 3, if we define the EVMs deployed in a State as a whole or India as a whole as the ‘population’, then in view of the bigger population sizes (N), the sample sizes (n) required are very small. 
  • However, the workload involved in hand counting the VVPAT slips for all the remaining (N-n) EVMs of the population, in the event of a mismatch, is very large and administratively unviable for India as a whole and for all States except the smaller States.

Way Forward:

  • The ‘plateau effect’ of sample sizes is used to divide the bigger States into ‘regions’ (an integral number of districts) with EVM population sizes of about 5,000 each. 
  • EVMs deployed in the region are treated as the ‘population’. On average, there would be about 20 Assembly constituencies in a region. The sample size required is 438, and the average number of EVMs per Assembly constituency whose VVPAT slips are to be hand-counted is 22. 
    • For example, U.P., which has 1,50,000 EVMs, can be divided into 30 regions, each with roughly 5,000 EVMs. 
    • If a defective EVM turns up, hand counting of VVPAT slips of the remaining EVMs will be confined to the region. This option is statistically robust and administratively viable.
  • Over the years, the Supreme Court (SC) has been indulgent towards the ECI due to its plenipotentiary role in the conduct of elections under Article 324 of the Constitution of India. However, the SC cannot continue to turn a blind eye to the ECI, making a mockery of the VVPAT-based audit of EVMs and defeating the very purpose of introducing the VVPAT.
    • SC must compel the ECI to make public how it has defined the population, how it has arrived at its sample size, and most importantly, its decision rule in the event of a mismatch. Only then will the SC’s order of 2013 on VVPAT be implemented faithfully in letter and spirit.

Book A Free Counseling Session