ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Accuracy of the 2001 Census: Highlights of Post-Enumeration Survey

The results of the Post- Enumeration Survey of the 2001 Census (conducted to ascertain the accuracy of the census) do not answer a number of questions regarding the content errors of the census. This article analyses the survey and points out the urgent need to look anew at the census methodology for assessing its accuracy.

COMMENTARY

Accuracy of the 2001Census: Highlights of Post-Enumeration Survey

Ashish Bose

in respect of a number of questions canvassed in the census. For example, if we match the marital status as reported in the census and the PES, we can find out the extent of response error, assuming that the the PES figure is correct. The PES thus seeks to answer two questions of great relevance to census methodology, namely,

(1) how accurate was the headcount in the census, and (2) how accurate were the characteristics of individuals, reported by the heads of households (or some responsible member in the absence of the head of the household)? In short, the PES throws light on the quantitative as well as the qualitative aspects of census enumeration.

At the outset we must mention that the census organisation does not “correct” the final figures as reported in the census on the basis of the PES results. This would be an enormous exercise and by the time such analysis is available, preparations would be on for the next census! We may mention that while the preliminary headcount figures of the 2001 Census were made available to the public within a few weeks of the census enumeration (and the final population figures within a few

The results of the Post-Enumeration Survey of the 2001 Census (conducted to ascertain the accuracy of the census) do not answer a number of questions regarding the content errors of the census. This article analyses the survey and points out the urgent need to look anew at the census methodology for assessing its accuracy.

Ashish Bose (ashishb@vsnl.com) is with the Institute of Economic Growth, Delhi.

A
fter a long wait of over seven years, one has now access on the census web site to the results of the Post-Enumeration Survey (PES) conducted soon after the 2001 Census. It may be recalled that the household enumeration for the 2001 Census took place during February 9-28, 2001. The PES was conducted during April-July 2001 in 3,000 enumeration blocks in all the 29 states and six union territories of India.

No census operation in the world can claim 100 per cent accuracy. In a massive operation like the Indian census, there is bound to be considerable undercount because of failure to enumerate every single household. Likewise there may be some overlap leading to double counting. We have to look for the net omission rate in the

Table 1: Net Omission Rate , 1991 and 2001 (in %)

census count to ascer-

Census Year Total Population

Rural Population

Urban Population

tain how accurate the

P M F

P M F

P M F

census was. Apart from

1991 1.76 1.73 1.79 1.68 1.60 1.77 1.98 2.11 1.83

this coverage error, the 2001 2.33 2.35 2.31 1.68 1.62 1.75 3.98 4.15 3.79

The PSE report gives the net omission rate per 1,000 persons enumerated.

Source: Registrar General, India, Report on Post Enumeration Survey (Census of India, tain the response error 2001), Delhi, 2008, p 12.

PES also seeks to ascer-

May 31, 2008

EPW
Economic & Political Weekly

COMMENTARY

Zone Total Rural Urban

India 2.33 1.68 3.98

Northern 4.86 3.42 7.61

Southern 3.37 2.53 4.98

Eastern 1.98 1.27 4.90

Central 1.29 1.17 1.71

Western 1.12 0.75 1.66

North-eastern 0.76 0.85 0.28

We have arranged the zones in descending order of net omission rate for the total population. Source: Registrar General, India, Report on Post Enumeration Survey (Census of India, 2001), pp 28-30.

Table 3: Net Omission Rate in 0-4 Age Group, 2001 (in %)

Zone Total Population Rural Population Urban Population

India 3.21 2.61 5.88
Northern 6.11 4.97 11.39
Southern 4.97 3.78 7.86
Eastern 2.05 1.73 5.30
Central 1.73 1.65 2.33
Western 1.66 1.34 2.25
North-eastern 1.16 1.24 0.37

We have arranged the zones in descending order of net omission rate for the total population in 0-4 age group. Source: Same as in Table 2.

Table 4: Net Omission Rate in the Age Group 20-24 Years, 2001 (in %)

Zone Total Population Rural Population Urban Population

India 3.21 2.54 5.28

Northern 5.47 4.31 9.81

Southern 4.53 3.40 6.84

Eastern 2.30 1.83 4.94

Central 1.68 1.58 2.16

Western 1.61 1.10 2.27

North-eastern 1.03 1.12 0.45

The zones are arranged in descending order of net omission rate. Source: Same as Table 2.

months), the report on the PES was ready in June 2006 but was not released till mid-April 2008 (and that too on the web site and no hard copy is available for the common reader). It may be recalled that the report of the Post Enumeration Check (as it was called in 1991) for the 1991 Census was released in 1994.1

Headcount Accuracy

An important technical point to note here is that the figures for net omission for undercount are not available for each state but only for the six regions of the country grouping states and UTs. These regions are: southern, eastern, north-eastern, northern, western and central. Assuming that the Delimitation Commission wanted to find out the “correct” population of say, Uttar Pradesh, they would have to know the net omission rate for Uttar Pradesh. But this was not possible as Uttar Pradesh was clubbed with Madhya Pradesh,

Economic & Political Weekly

EPW
may 31, 2008

Chhattisgarh and Uttaranchal under the central zone.

In earlier censuses also, the pretext for not giving statewise PES figures referred to sampling. We feel that sampling experts could have sorted out this problem by, say, increasing the sample size.

Does it make much sense to club Delhi with states like Jammu and Kashmir, Himachal Pradesh, Punjab and Haryana under the northern zone? Obviously, the net omission rate in a big city like Delhi would be much higher than, say, in Himachal Pradesh. Besides, census-taking in Jammu and Kashmir poses special problems. The net omission rate was 4.86 per cent for the northern zone, the highest among all the zones in India. In the urban areas it was 7.61 per cent, also the highest in India.

Omission Rate in South

Turning to the more literate southern zone (comprising Andhra Pradesh, Karnataka, Kerala, Tamil Nadu and a few UTs) it is somewhat surprising to find a net omission rate of 3.37 per cent compared to the all-India figure of 2.33 per cent. Were the literate people reluctant to get counted in the census or were they at work when the census enumerator called on them?

Family planning is far more successful in the southern states, compared to the northern states. The PES report gives the net omission rate for five-year age groups. It is very surprising that in the southern zone, the net omission rate was as high as

4.97 per cent for the age group 0-4 years. In India, as a whole it was 3.21 per cent and in the central zone (which includes Uttar Pradesh) the figure was only 1.73 per cent. Do the PES figures for net omission rate in the southern zone cast a shadow on the shining examples of Kerala and Tamil Nadu, in particular, on the family planning front? No doubt the total fertility rate (TFR) is a better indicator than the proportion of population in the 0-4 age group. It is just possible that both the census and PES results are wrong!

Urbanisation and Net Omission

There is no doubt that with increasing urbanisation and growing mobility of labour, census taking is becoming increasingly difficult. We present a few tables derived from the data given in the PES report for the 2001 Census.

In 2001, the urban net omission rate shot up to 3.98 per cent from 1.98 per cent in 1991. The overall impact was that the net omission rate in India as a whole increased from 1.76 per cent to 2.33 per cent between 1991 and 2001.

In Table 2, we give the figures for the six zones of India. The northern zone has the

Young Research Ambassador’s Award

Celebrating 20 years of Katha An exciting opportunity for research on education and reading trends in India.

Katha will offer up to 10 short-term research awards in 2008 to support work in the areas of children’s literature and education. Katha Young Research Ambassador’s Award will be open to independent scholars, advanced graduate students, and doctoral candidates below 35 years of age. Researchers would be given the opportunity of presenting their study at the Katha International Conference, 2008. Each chosen researcher will be given a travel and hospitality grant for participation at the Conference. To go through the research topics and to download the application form please visit: www.katha.org Send your application with relevant documents to: The Research Team, Katha, A3, Sarvodaya Enclave, Sri Aurobindo Marg, New Delhi-110017 either by post or Email at: akshaya@katha.org. For further information, contact: 011-26524511, 26524350, 26868193

Last date for receipt of applications: 30th June, 2008

COMMENTARY

Agewise Omission

As mentioned before, the PES report gives data for all the five yearly age groups. We have chosen the 0-4 age group which has, by and large, the highest omission rate.

Another age group which is marked by high omission rate is 20-24 years, where there is considerable mobility of labour. Table 4 (p 15) gives the zonewise figures.

A detailed examination of the net omission rate for all the age groups in different zones shows that the figures are more than 10 per cent in the northern zone for the following age groups relating to urban males:

11.26 per cent for 0-4, 10.93 per cent for 20-24, 11.65 per cent for 25-29 and 11.04 per cent for 30-34 age group. The omission rate for urban females in the northern zone was highest for 0-4 age group: 11.55 per cent.

It is very difficult to interpret the figures for the different zones. For example, the figures are very low even in the urban areas of north-eastern zone. Is there a cultural factor? Or, is there something wrong in grouping of states into six zones or are there problems with the sampling procedure?

We will not go into the details of response or content error in regard to a number of questions canvassed in the 2001 Census as in our view, in the absence of statewise figures, the clubbed data for zones are not very meaningful. Besides, we know nothing about non-sampling errors. The time has come when we should have a serious look at the census methodology as well as the PES to assess the accuracy of the census. By way of consolation for the census organisation, one may note that (as stated in the PES report), compared to India’s 2.3 per cent, in UK the omission rate was 2.2 per cent (2001 Census) and in China

1.8 per cent (2000 Census).

Note

1 The pretext for the delay in the 2001 Census as I understood from census commissioner D K Sikri was that in view of the deliberations of the Delimitation Commission’s work adjusting seats for Parliament and state assemblies, it was thought prudent not to release the PES figures to avoid needless political controversies. So the delay was in spite of the hard work put in by the census commissioner and his colleagues.

May 31, 2008

EPW
Economic & Political Weekly

Dear reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Comments

(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top