This is a blog post from
How many urban constituencies does India really have? How many Muslim-dominated constituencies do we have? These are key questions around our elections that we have simply not been able to answer – until now.
[A brief aside on a frustrating data issue, before I get to the meat of the matter. Feel free to scroll down to the charts if this does not interest you.]
Most demographic data for India comes from the Census, which splits its data down from the state and district level to the block, village and town level. The smallest unit of Census data is an Enumeration Block (EB) which consists of around 150 households.
Electoral data, on the other hand, comes from the Election Commission of India (ECI), and is centred around parliamentary and assembly constituencies, the smallest unit of which is a polling booth which covers around 1,000 voters.
The problem is that the ‘twain don’t meet. Theoretically, if the ECI simply listed all the Census EBs in an assembly or parliamentary constituency, journalists and researchers would be able to combine electoral and demographic data, and talk of how rural or urban a constituency is, how young or old it is, what its caste and religious composition is. The problem is that it doesn’t.
I spoke to Srinivasan Ramani, senior assistant editor at the Economic and Political Weekly, a data and elections expert who is part of a community of data enthusiasts called Data Meet. Srinivasan is among the people who share my pain of being unable to fully and satisfactorily resolve demographic and electoral data. “The ECI…does it in a way (which they usually do) in which it is difficult to extract information,” he said, adding that there is no uniformity across states. This has meant that intrepid data scientists have tried themselves, some times using GIS mapping, to build constituency data themselves, up from the polling booth level, and fitting it to census boundaries.
Last week I met the people who have now cracked this enormous problem. Datanet India is a private firm (which runs among other things indiastat.com and electionsinindia.com) based out of Delhi. They publish their electoral analysis in a proprietary publication called India Elects, which they shared with me.
During a long conversation at their office last week, where we commiserated with each other about the difficulty of dealing with ECI data, the affable R. K. Thukral who set up and heads Datanet India took me through their process, which includes extensive GIS mapping and field visits to between 150-200 constituencies. These field visits are necessary to resolve some of the state-level issues Srinivasan alluded to earlier.
Turning to his copy of the Delimitation Commission’s book on Haryana, Mr. Thukral opened to Panchkula constituency as an example and showed me the problem. Along with the names of towns and villages listed for that constituency, there are “KCs” and “PCs”. What are these? “Kanungo Circles and Patwari Circles, British-era revenue divisions still used in Haryana. But what villages or EBs do they refer to?” Mr. Thukral points out. I have no idea. Asking District Magistrates to decode such idiosyncrasies becomes part of Datanet’s job.
At the end of all of this, I got some interesting numbers. For instance, India has just 21 fully urban constituencies. It has 54 constituencies that are more than 75% urban. It has 108 constituencies that are more than half urban.
Which are our only fully urban constituencies? Maharashtra has the most – Mumbai, Pune, Thane and Nagpur.
And what of Muslims in electoral constituencies?
As it turns out, India has 15 constituencies with more than 50% Muslims. It has four constituencies with 80-100% Muslims – Baramulla, Srinagar and Anantnag in Jammu & Kashmir, and Lakshadweep. Another eleven have 50-80% Muslims: Hyderabad (Andhra Pradesh/ Telangana), Karimganj, Dhubri and Barpeta (Assam), Kishanganj (Bihar), Malappuram and Ponnani (Kerala), Maldaha Dakshin, Jangipur, Baharampur and Murshidabad (West Bengal). There are 80 constituencies with 20-50% Muslims and 448 with 0-20% Muslims. Interestingly, Gujarat, Rajasthan, Chhattisgarh and Tamil Nadu are among the states without a single constituency that has more than 20% Muslims.
So what does this mean for Muslim political representation? I looked at the Muslim population of those 23 constituencies from which the winning MP was a Muslim
Yes, the chances of a Muslim MP winning rise with the proportion of Muslim population, but it’s the same for Hindu MPs as well. Parties tend to give tickets by religious composition, but then their justification is that people vote along religious lines. Ultimately, if we care about raising Muslim political representation, this is a cycle both voters and parties might need to break out of.