Feature :
Go to Page : 1 - 23 - 4

Post Enumeration Surveys in Census

The data collected through any field inquiry is invariably subject to certain amount of error, that normally creep in due to the error committed by the investigator or even the respondent. A massive operation like population census is no exception. A large number of countries carry out a Post Enumeration Survey (PES) after the completion of census to systematically measure the degree of error or in other words the degree of accuracy. The systematic attempt to estimate the accuracy of the count by sample surveys has become a part of the census operations in India since 1951 census.

The PES field work for Census 2001, earlier known as Post Enumeration Check, has already been completed. The main objective of the PES is to quantify the likely omission or duplication in census enumeration in terms of coverage of census houses/households or individuals, in other words, to determine the coverage error. In conducting this survey, opportunity is also taken to measure error in responses, or the content error in the recorded household characteristics, with regard to certain items canvassed in the census.

The PES is usually intended to provide answer to the questions:

    1. How accurately the census houses /households/persons have been enumerated in the census and
    2. How precisely have certain characteristics of the household/ individual been recorded in the census.

The PES also serves the important purpose of providing feedback regarding operational matters like concepts and procedures, which would help, to some extent, in improving the future census operations.

In the earlier censuses, the PES used to be conducted only after the population enumeration. However, for the first time in Census 2001, in addition to the usual PES of the population enumeration, a PES of the Houselisting Operations was also conducted after completion of the Houselisting Operations.This was carried out in 16 major States (having population of 10 million and above as per 1991 population census) including National Capital Territory of Delhi. These 16 states and Delhi together comprise about 97 percent of the population of the country. The PES for Houselisting was aimed at measuring the coverage and content errors to permit better interpretation of data collected during the Houselisting Operations of Census of India 2001. Thus two Post Enumeration Surveys have been conducted in the Population Census 2001, one after the Houselisting Operations in the year 2000 and the other after the Population Enumeration in 2001.

Sampling frame and sample selection :

The PES of Houselisting Operations has been conducted in a sub-sample of the existing Sample Registration System (SRS) units as a new sampling frame was not available. The SRS units of the 1991 Census available in the Office of the Registrar General, India for rural and urban areas separately have been used as the sampling frame. Wherever, the SRS units had population in excess of the average size of the Houselisting Blocks (about 650 persons), two or more Houselisting Blocks were carved out within the SRS units. A sub-sample of the existing SRS units was selected for the rural and urban areas in each of the States included for the PES of the Houselisting Operations. A sample size of 25 percent of the rural SRS units and 50 percent of the urban SRS units has been selected for the same. The urban SRS units are divided into city (population more than 0.1 million or 1 lakh) and non-city (population less than 0.1 million or 1 lakh). The required number of sample units was selected linear systematically. The objective of the PES Houselisting was to estimate both the coverage error - (in terms of omission of census houses) and also content error (housing, household amenities and asset characteristics).

The PES of the population enumeration pertaining to the Census of India 2001 was conducted in all the States and Union territories of the country for the first time. In earlier censuses, it used to be conducted only in selected States. In 1991, it was conducted in all the major States having population of 10 million and above and the states of Himachal Pradesh, Tripura and Union Territory of Delhi. For the PES of population enumeration-Census of India 2001, a sample of 3,000 enumeration blocks (EBs) has been distributed among all the States and Union territories in proportion to their projected population of 2001 subject to minimum of 100 and 4 EBs for major States and Union territories respectively. The list of EBs used at the time of the population enumeration separately for cities, non-city urban and rural areas has been used as the sampling frame for conducting the PES. For selecting the sample, the required number of EBs are selected linear systematically for the above mentioned three areas separately. In the last census, a second stage sampling was also undertaken by selecting a 10 percent sub-sample of the residential households enumerated in the census from each of the selected EBs for the purpose of conducting the PES. But during the current PES - 2001, all the residential households in the 3,000 selected EBs are enumerated for estimating the coverage error. This is an improvement which is expected to greatly enhance the confidence in the estimation of coverage error. The institutional and houseless households are excluded from the PES. Due to operational considerations, the enquiry on content error was limited in nature and its scope was restricted to few items like age, literacy, marital status, economic activity and fertility etc.

For the survey on content error, a 20 percent sub-sample of the 3,000 EBs selected for estimating the coverage error has been taken linear systematically and all the residential households in those 20 percent or 600 EBs were enumerated for that purpose.

PES involves three stages of operations namely,
(i) independent listing and enumeration in the selected 3,000 EBs
(ii) desk matching of those with the census records and
(iii) field reconciliation for those households or individuals for whom the entries in the two records viz., the PES and the census do not match. These are reconciled by a field visit.