Critique of EVM-VVPAT judgment

Source-This post on Critique of EVM-VVPAT judgment has been created based on the article “The EVM-VVPAT case judgment is disappointing” published in “The Hindu” on 30 April 2024.

UPSC Syllabus–GS Paper-2– Salient Features of the Representation of People’s Act.

Context– The Supreme court in a recent case has rejected all pleas that were seeking 100% EVM-VVPAT verification. This article provides the critique of the Supreme Court’s Order on this judgement.

Read more- Supreme Court VVPAT judgement

What are the arguments provided by the author to criticize the judgement?

1) Arbitrary sample size-The author criticizes the Supreme Court’s order mandating a uniform sample size of “5 EVMs per Assembly Constituency” for VVPAT-based audit of EVMs, stating that it does not conform to statistical sampling theory principles.

2) Limitation of lot acceptance sampling – VVPAT audit is a case of “lot acceptance sampling” where a random sample is drawn from a defined “population” of EVMs, and the entire population is accepted or rejected based on the number of “defective EVMs” (EVM count mismatching VVPAT count)

Note- Lot acceptance sampling-This is a quality control method where a sample is taken from a batch or “lot” of products, and based on the inspection of that sample, a decision is made about whether to accept or reject the entire batch.

3) Lack of clarity- It highlights that the Court did not compel the ECI to disclose the defined “population” and “next steps” in case of mismatch.Further,it did not ask the ECI to clarify these points itself after seeking expert opinion.

4) Statistical Analysis of Court-Mandated Sample Size– The Court-mandated sample size has a 95% probability of failing to detect a defective EVM if an Assembly Constituency is the “population”, and 70% if a Parliamentary Constituency is the “population”.

What are the recommendations given by the author to ensure robust VVPAT-based EVM audit system?

1) There is a need to implement a statistically sound, VVPAT-based EVM audit system that can detect mismatches with 99% or 99.9% accuracy.

2) There should be matching of the votes at the start of the counting day. If they match perfectly, the results should be declared based on the EVM count. In case of a mismatch, there should be manual counting of VVPAT slips for all remaining EVMs of the particular “population” and declaring results based on the VVPAT count.

Question for the practice

Why is EVM-VVPAT case judgment criticized by the author?