The Chief Electoral Officer published a new set of voters’ lists on 28 Jan 2013. Download of the voter lists using a software program, and data extraction from them, was since disabled by converting the files to image files, but Prajalytics, a software company, had already downloaded the files for 27 Bangalore constituencies manually and extracted voter records. The data quoted in succeeding paragraphs is based on the voter lists of 27 constituencies of Bangalore city downloaded on 28th January 2013.
Violation of Data Model
The Instruction Manual for Standardization of Electoral Roll Database (Ver 1.0) published by Election Commission of India on 29 January 2010 defines database schema and naming conventions. Voter lists of Karnataka violate these standards.
1. ID Card Number
The Hand Book for Electoral Registration Officers states, "EPIC Number is designed to act as the permanent unique identity for every elector and thus it is very crucial that, once created, this information is not lost."
- About 11,000 EPIC numbers are duplicated – the same number given to more than one person, or the same record is repeated with another serial number or in another file.
- 22,879 IDs are arbitrarily numbered without complying with the new or old standard of numbering.
- 91,453 EPIC number fields are blank.
2. House Number
- 66,443 house numbers are longer than 10 characters, violating the guidelines.
- 1,603 records do not have house numbers.
- 53,844 house numbers are written as ‘.’,’-‘ etc.
- 39,656 houses have more than 10 voters each. 112 houses show more than 100 voters each. One has 452 voters and another has 347 voters.
- By contrast, 8,88,555 houses have only one voter each.
3. Voter Name and Relative Name
Prefixes like Dr., Prof., etc. are not allowed as per EC’s data standards. However, these are used in thousands of records.
- 1,911 voter names have ‘Late’ prefixed.
- Aliases of names are used in hundreds of places. E.g., (Ammu) Shanthi, (Cheluvaraju) Gangamma, etc.
- 1,431 records do not have voter names. 157 names are one character long.
- 5,647 records do not have a relative’s name or the name is one character long.
- Standard conventions are not used in entering names, addresses, and house numbers.
- Name of the same person is spelt differently as a voter, as father and as husband. Often, a voter’s father’s name is different from that of his mother’s husband.
4. Relation Type Suspects
- Voters with more than one wife: 13044
- Age difference between parent and child less than 13 years: 10,002
- Husband shown as female: 1,829
- Male’s relative is husband: 30
- Same name of voter and relative : 1,482
5. Age of Voter
By EC’s standards, age of a voter should be between 18 and 120.
- 502 voters are less than 18, including 39 voters who are 0 years old.
- 96 voters are above 120. 17 are above 200 years. Voter with EPIC# XUL4087888 is 4,818 years old.
From the records it is difficult to guess if a wrong sex is assigned, but a sample survey has shown some records where males have been shown as females and females as males. One person has even been shown as male and female at two serials – STZ0267013 and STZ3778412.
7. Address of Voter
The information is incomplete in most cases. No standard is followed. The same word is spelt differently at different places.
Good use of technology can reduce efforts by order of magnitude and also improve quality and scalability of operations. Do the maximum possible using technology before collecting data or processing information manually. The CEC stated on June 11th, 2012 that the EC would give priority to providing a clean electoral roll and hassle free registration and will use technology to achieve the objective.
It is very important that the CEO partners with professional organisations and demand highest quality of products and services.
Impractical rules need to be changed rather than ignored. Many tasks currently manually done can be done with less effort and more accuracy with software.
Many instructions are oral and not traceable. Written instructions and accountability should be practiced. Audit trails should be recorded in the database for all the actions required as per rules.⊕