Counting 1.2 billion Indians is a herculean task. Rrishi Raote describes the incredible logistics of Census 2011
Over the night of February 28 and March 1, a small and businesslike army will spread across the land. To locate their targets, this army will look into municipal pipes, under flyovers and bridges, on footpaths, along railway lines, up dark staircases, around places of worship — everywhere the legions of the “houseless” come to rest. Other elusive targets will have been captured on the preceding day, including tourists in hotels and every human on every ship that will be in Indian territorial waters at midnight.
For 00:00 on March 1, 2011, is the moment of reckoning. The Office of the Registrar General & Census Commissioner has to find out exactly how many human beings there are in the territory of India at that moment, Indians as well as foreigners (except diplomats because they reside in enclaves that are not a part of India; by the same logic, Indian diplomats abroad are included).
The Indian census is the biggest exercise of its kind, performed once a decade since 1881. The first, though partial, Indian census was in 1872; the Census of 2011 is the 15th in modern times, and the seventh since Independence. There are too many pieces in the jigsaw puzzle that need to fall into place.
The headcount will be done by 2.7 million enumerators — teachers drawn from government-run primary schools — in three weeks between February 9 and 28. From March 1 to 5, they will do the rounds one more time to ensure that the data is accurate. Three weeks after that, around the end of March, Registrar General & Census Commissioner C Chandramouli will declare the provisional results. Full and final results will come a year and a half later. The plethora of data generated by Chandramouli’s team — numbers, maps, atlases et al — will be purchased by policy planners, research institutes, analysts and corporations. “Hardly any of the books that we had come out with after the 2001 Census is left,” says Chandramouli with quiet pride.
Teachers are chosen for enumeration because they are the only people who can get inside every household in the country and ask personal questions. About half the enumerators are women. In towns and cities, women enumerators often face harassment. So they are required to go from door to door with a colleague or family. It’s a fair bit of work though. There are more than 300 million households in the country, which means each enumerator has to make 125-150 calls in three weeks. For this work, each of them will get paid up to Rs 6,500.
The 2011 Census has three parts. The first was the housing census done between April and September 2010 when the same enumerators collected data for the National Population Register, a record of the citizens of India. About 30 million houses were counted then. This data will be given to the Unique Identity Authority of India which will remove all duplication and generate a unique identity number. Finally, an identity card for all Indians will be readied. The second is the headcount. The third will be the caste census, the first since 1931. The Union Cabinet has said that it should be done between June and September this year; now it is up to the states to decide when they want to get it done. The housing census has been linked to the headcount (people will be asked if they participated in that round too), which will help answer questions like how many PhDs live in houses without latrines.
The government has allocated Rs 2,200 crore for the 2011 Census, which brings the per capita cost to Rs 18. Chandramouli says this is the lowest in the world; in the United States, it costs almost $50 (around Rs 2,500). The size, of course, is unmatched — the enumerators will have to go to 8,011 towns and 649,989 villages in 640 districts of the country. Aboriginal islands will be accessed with help from some designated “contact” people. In Sentinel Islands in the Andamans, where the tribes are still hostile, the enumerators will throw coconuts and red cloth into the sea. Once these hit the shores, the tribesmen will come out to gather these “presents”. From afar, the enumerators will take video films, come back and freeze the frames, and count their numbers. Forest dwellers, those who do not live in villages, will be counted separately by forest officers.
States frequently change the boundaries of districts and tehsils, often for political reasons. This can create havoc with the census. So, on December 31, 2009, an advisory was sent out by Chandramouli to all states to desist from altering boundaries. With the housing census done, Chandramouli’s office has digitised the map of every village in the country and every town. Within towns, every ward has been mapped. In addition, the maps of all the 33 state capitals have been done using satellite imagery — down to the last structure. Any construction that didn’t look like a house from above was physically checked by the enumerators. This is invaluable information for town-planners and others like metro service operators and road builders.
For the headcount, 5.4 million instruction manuals were printed in 18 languages, and 340 million schedules (forms for the headcount) in 16 languages. For this, about 12,000 tonnes of paper was required. This is no ordinary paper — it has to be scanned and read by software. “We didn’t have specifications for this kind of paper in the country,” says Chandramouli. Finally, a clutch of paper mills such as Hindustan Paper Corporation, West Coast Paper Mills and Bindal Paper came forward. The other problem was to locate presses which could print this volume (340 million prints in three months) digitally. Each form is unique — it has a number and bar code plus the name of the district and tehsil it is meant for. Five or six such presses were assigned the work. Chandramouli’s team labelled the packs at the presses and handed them over to the Department of Posts for final delivery to 17,000 addresses across the country. (Once the enumeration is over, the department of posts will collect these forms at every village and town, and bring them to the offices of the Registrar General & Census Commissioner.)
Chandramouli says that most of the schedules have either reached their destination or are on the move — not without some goof-ups and accidents. Thus, addresses got mixed up in Uttar Pradesh, the consignment fell into a river near Kalimpong and the recent strike in Karnataka brought truck movement to a standstill in the state. Protestors in Jammu & Kashmir destroyed the forms for the National Population Register and the house census. Those records have to be recreated. The headcount in snow-bound villages of Jammu & Kashmir, Himachal Pradesh and Uttarakhand was completed between September 11 and October 5, 2010. Births and deaths after that will go unrecorded, though Chandramouli thinks the numbers won’t be significant.
The real problem is with the around 200 villages in Chhattisgarh and West Bengal where Naxals have given a call to boycott the Census. “In Jharkhand, Orissa [the other Naxal-affected states] and Jammu & Kashmir, we expect 100 per cent coverage,” says Chandramouli. In all probability, the villages in Chhattisgarh and West Bengal will go uncounted. The numbers cannot be extrapolated from 2001 because birth and death rates in these villages are not known. To encourage people to come out and get counted, Chandramouli’s office has appointed Mudra to do a media campaign. Sachin Tendulkar and Priyanka Chopra will support it.
Significant changes have been made in the 2011 Census schedules. For the first time, apart from male and female, respondents can tick a third option — others. They will have to give their date of birth and not just the age. “People have the tendency to report ages around some favourite numbers that end with five, zero or even digits. This results in heaping around these numbers,” says Chandramouli. “To make the data rich, we are asking the date of birth as well as the age.” In marital status, separated and divorced have different codes. The codes for disability have been raised from five to eight. In the work category, a third slab has been introduced between those who work for more than six months and those who work less than six months — those who work between three and six months. This will capture the beneficiaries of redistributive programmes such as the Mahatma Gandhi National Rural Employment Guarantee Scheme which offers jobs for 100 days in a year. And prostitutes have been taken out of the category of beggars; a new category titled others has been created where people whose income cannot be categorised under legal means will be included.
To scan the schedules, the census office has bought about 60 high-speed duplex Kodak i780 scanners of the kind that are usually used in Western postal departments and hospitals. Each scanner costs about $30,000 (around Rs 14 lakh) and can handle 130 A3-size, double-sided forms per minute (or 130,000 a day). Kodak says the order was for machines to capture “270 million forms in less than four months, working three shifts a day” in Phase 1, and the same number in Phase 2. To turn handwritten entries into data at such speed, the census office has bought eFlow 4.5 Intelligent Character Recognition software from Top Image Systems of Israel. Operators from HCL Infosystems of India will run the machines. State-owned BSNL has been asked to identify a call centre where the uncounted can call to get included.
How accurate is the headcount? Once the final numbers are out, each state goes back to recheck the numbers. The omission rate was 1.8 per cent in 1981, 1.9 per cent in 1991 and 2.3 per cent in 2001. (It jumped in 2001, says Chandramouli, because it was the first time that forms were digitised.) This year, too, he expects the omission rate to be around 2 per cent — the internationally-accepted norm.
Data users, on their part, make two points. One, the census office takes a lot of time to release the final numbers; as a result, they have to make do with old data. And two, the headcount in the urban areas is not accurate. “The census is not so good at the urban level. It hasn’t been integrated with PIN codes,” says Laveesh Bhandari of Indicus Analytics. Chandramouli admits that this is a problem — working couples are invariably not at home, gatekeepers often turn the enumerators away. Still, the count is on.