Sniff the Electoral Roll

“Chikpet, Balepet, Sultanpet and Cottonpet have heaps of garbage on the roadside,” is a news item, though stink in the city has ceased to be news. BBMP Officers and Employees Association, with a demand that their corrupt practices should not be questioned by BMTF, exposed a lot of stink, showing the deeper root of stink.

I am discussing another kind of smell, that the nose does not sense.

Kent Beck  coined the phrase “Code Smell” to indicate a symptom in the source code of a software program that possibly indicates a deeper problem. We can uncover the real problems by investigation and avert future failures. Here we discuss some smells in Electoral Roll.

In majority of houses we expect two ro more voters. A house with only one voter could mean that some voters are not enrolled or are excluded by deletion from the list. Of the 23,71,710 houses in the 27 constituencies under study 7,75,282 (31.72%) houses have only one voter each. We also have 80 houses with more than 100 members each, which too is a smell.

In a typical family with husband, wife and an adult son or daughter, we find four names – three members as voters, husband appearing thrice – as voter, husband, and father – and his father as relative. Thus we expect four names. In a typical family of four voters, we would find five names. In these two cases, a house has one non-voter name each. At one extreme, we can have the count of non-voter names equal to the number of houses. At other extreme, none of the relatives in the list may be a voter, resulting in as many non-voters as voters. If we find a large number of non-voters in the list, that could be because a person’s name is spelt differently as a voter and as relative. We have seen such examples in a previous post. Such inconsistencies make analysis difficult. Another reason for large non-voter names is that the person who appears as relative is not in the voter list – sometimes, incorrectly. The voter lists of 23,71,710 houses have 52,63,073 non-voter names, which is an average of 2.22 non-voters per family. 

When a voter’s relative is shown as husband, it is likely that the woman is living with her husband. Where relatives are shown as husbands, names of husbands do not appear as voters in 15,81,059 women voter records. I have considered the voters below 60 years of age, to reduce the possibilities of deceased husbands.

In 1,216 records, full name of voters match full name of relative, character by character. This can be an error.

We expect certain range of age difference between relatives.

  1. 844 mothers and less than 19 years older than their children. In many cases, the age difference is less than 10 years. 18 years being the legally minimum age for marriage for a girl, if the age difference between mother and son / daughter is less than 19 years, there may be an error. If the difference is lower than 13 or 14, then we are almost certain about an error.
  2. 14,566 fathers are less than 22 years older than their children and in many cases the difference is less than 10.
  3. It is difficult to say what the age difference between wife and husband should be. However, the list has 3,906 husbands who are more than 20 years older than their wives.

In the blog post of August 20, we discussed that sex of voters was changed in 2,877 records. If the relative of a voter is husband and the relationship was healthy heterosexual, the CEO’s office has to change the sex of both spouses or none to keep the balance. If they change the sex of only one of the spouses, leaving out the other, we have a problem of homosexual record. By this count, we find 1,789 lesbian relationships and 142 gay relationship in voter records.

You can object that if the name of a relative matches the name of another voter in the same house, it does not prove them to be the same person. It does not. We are discussing possibilities and leads for further analysis with an intent to improve the voter lists. With deeper investigation we will find more error patterns and proofs of errors. We can also correct many errors with help of software. 

Superficial corrections will not remove system errors. With a sensitive nose we have to sniff and reach the real origin of the stink. Some of the smells may not lead to any error – e.g., the large age difference between spouses may all be correct. 

We need more sniffers – human sniffers, with concern for quality.

Advertisements

About pgbhat

A retired Naval Officer and an educationist. Has experience with software industry. A guest faculty at different institutes and a corporate trainer with software development companies.
This entry was posted in eGovernance, Profession, Social Issue, Technology. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s