Sample Microdata of House Listing of India 2011 - 1 percent sample dataset
India, 2011
Get MicrodataCreated on
Apr 08, 2021
Last modified
Jul 04, 2022
Page views
1255
Downloads
4
Identification
Survey ID Number
IND_2011_HLIST_1pc_v02
Title
Sample Microdata of House Listing of India 2011 - 1 percent sample dataset
Country
| Name | Country code |
|---|---|
| India | IND |
Series Information
The Sample Microdata of Census of India is a series that currently contains the following datasets:
- Census of India 2001, Housing Microdata Sample (CIHMS), 5 percent subset
- Census of India 2011, House Listing Microdata Sample (HLIST), 5 percent subset
- Census of India 2011, House Listing Microdata Sample (HLIST), 1 percent subset
- Census of India 2011, Sample Microdata of Census (SMCI), 5 percent subset
- Census of India 2011, Sample Microdata of Census (SMCI), 1 percent subset
- Census of India 2001, Housing Microdata Sample (CIHMS), 5 percent subset
- Census of India 2011, House Listing Microdata Sample (HLIST), 5 percent subset
- Census of India 2011, House Listing Microdata Sample (HLIST), 1 percent subset
- Census of India 2011, Sample Microdata of Census (SMCI), 5 percent subset
- Census of India 2011, Sample Microdata of Census (SMCI), 1 percent subset
Abstract
Census of India is one of the largest operations for collecting data on various items. It involves considerable efforts right from the development of the questionnaire and instructions, recruitment and training thousands of enumeration staff, actual enumeration work, and compilation and tabulation of data to publishing the report. The census provides a mine of valuable data on housing, population and their characteristics, which are not otherwise available at lower geographic levels. Census organization makes all efforts to make available the tabulations required by various Governmental and non-governmental data users by providing data in a set of standard tables. It is obvious that it cannot satisfy the requirements of all users. There would be some cross tabulations not available in the standard set of tables that are required for specific planning/monitoring purpose or for research. There have been regular demands for making available the census data at the individual record level, subject to satisfying the statutory confidentiality clauses, from users to prepare their own cross tabulations.
Providing Public Use Sample data is one of the ways in which use of census data can be increased. With this view, it has been decided to make available to users on a sample basis, data records relating to Population Enumeration Census 2011. The confidentiality of individual level information is maintained by excluding all data items leading to individual identity. The Office of the Registrar General, India (ORGI) has made available Public Use Sample view data from the first phase of Census, known as House listing as Housing Census for Census 2001 and 2011, known as “Census of India Housing Micro data Sample (CIHMS)”. Now, the sample data from the Population Enumeration phase of both Census 2001 and 2011 is made available, which will be known as Sample Microdata of Census of India (SMCI). Documentation of the sampling procedure and estimation methodology is provided here to the users to enable them to properly understand and analyze the sample data.
Two datasets are being made available for SMCI - one at national level (one per cent sample) and the other at State level (five per cent or ten per cent sample). The basic difference between these two sets is that the State sample would allow the user to generate estimates at national, state and district levels, while, the one per cent sample would allow tabulation only at national and state levels. It may be mentioned here that while selecting the 5 per cent sample, it was felt necessary to provide larger sample for small districts of the country. Hence a 10 per cent sample has been selected from such districts for inclusion in the state sample. Only records relating to households are part of the file, i.e., the non-residential houses are not included in the data files.
Providing Public Use Sample data is one of the ways in which use of census data can be increased. With this view, it has been decided to make available to users on a sample basis, data records relating to Population Enumeration Census 2011. The confidentiality of individual level information is maintained by excluding all data items leading to individual identity. The Office of the Registrar General, India (ORGI) has made available Public Use Sample view data from the first phase of Census, known as House listing as Housing Census for Census 2001 and 2011, known as “Census of India Housing Micro data Sample (CIHMS)”. Now, the sample data from the Population Enumeration phase of both Census 2001 and 2011 is made available, which will be known as Sample Microdata of Census of India (SMCI). Documentation of the sampling procedure and estimation methodology is provided here to the users to enable them to properly understand and analyze the sample data.
Two datasets are being made available for SMCI - one at national level (one per cent sample) and the other at State level (five per cent or ten per cent sample). The basic difference between these two sets is that the State sample would allow the user to generate estimates at national, state and district levels, while, the one per cent sample would allow tabulation only at national and state levels. It may be mentioned here that while selecting the 5 per cent sample, it was felt necessary to provide larger sample for small districts of the country. Hence a 10 per cent sample has been selected from such districts for inclusion in the state sample. Only records relating to households are part of the file, i.e., the non-residential houses are not included in the data files.
Kind of Data
Census/enumeration data [cen]
Unit of Analysis
Individual, Household (not including institutional households)
Version
Version Date
2021-04
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Office of the Registrar General, India | Ministry of Home Affairs |
Sampling
Sampling Procedure
The Sample Microdata of Census of India (SMCI) has been prepared by the ORGI based on the data collected in the Population Enumeration during 9-28 February 2011 throughout India. The SMCI files are the text (ASCII) files, containing records of individual members of the households for national or the state sample. By pooling the district level samples drawn independently, the state level sample has been prepared. Like the CIHMS, where there may be some districts with a 5-percent sample and some with a 10-percent sample depending upon the size of the districts, the SMCI sample is a uni-stage stratified sample and the size of the sample is same as CIHMS multiplied by the average household size. The record for each individual contains a field showing the weight of the record. These weights for each individual, when applied to the records, expand the sample to the relevant total.
According to Census 2011, there are 26 districts with population less than 1,00,000 with average population of 61851 or 12,940 households and 614 districts with average population of 19,68,380 or 4,11,795 households. Instead of estimating sample size separately for each of the state, if we take 5 percent sample for each of the state population more than 1,00,000 the average number of 5 percent comes out to be close to 20,000 houses/households. For states less than 1,00,000 population, average household is 12940, as such, we may go for 10 percent level of significance and the sample size may be close to 5000 households.
It is therefore proposed to adopt 5 percent sample of households of the state with population more than 1, 00,000 satisfying the condition of 5 percent level of error and 5000 households of the state population less than 1,00,000 satisfying the condition of 10 percent level of error.
To improve the sampling efficiency, the household records within each district were first stratified into homogeneous groups on the basis of place of enumeration, size of village/towns and geography as stratification variables. In all, 6strata's are formed in each district. The household records were further sorted by all the geographic codes: Tahsil/ Taluk/ P.S. / Development Block/ Circle/ Mandal, Town/ Village, Ward (for urban), Enumeration Block Number, and Household Number within each of the 6 strata. This sorted frame provides a high level of implicit stratification. Finally, a systematic random sample from this sorted frame has been drawn within each district with a sampling interval of 20 or 10 as the case may be. A detailed description of variables included in the sampled households and recodes of SMCI is an important input for the data users which may be seen in Annex 2 of Chapter 3.
The National SMCI is an aggregate of 20 percent or 10 percent sub samples of 5 percent sample or 10 percent sample respectively selected at district level for state sample using the same implicit stratification, which effectively give the 1 percent sample of total households at national level. For example, in the case of the districts with a 5 percent sample of households, a random systematic sub sample of 1 in 5 households was selected from the 5 percent sample, keeping the original order of selection based on the stratification criteria. A random systematic sub sample of 1 in 10 households was selected from the 10 percent sample for the smaller districts. The national SMCI will enable us to generate the reliable estimates at state level only. For this purpose, the identity of the districts has been concealed by replacing their location codes by 99 in the 1 percent national level SMCI file.
According to Census 2011, there are 26 districts with population less than 1,00,000 with average population of 61851 or 12,940 households and 614 districts with average population of 19,68,380 or 4,11,795 households. Instead of estimating sample size separately for each of the state, if we take 5 percent sample for each of the state population more than 1,00,000 the average number of 5 percent comes out to be close to 20,000 houses/households. For states less than 1,00,000 population, average household is 12940, as such, we may go for 10 percent level of significance and the sample size may be close to 5000 households.
It is therefore proposed to adopt 5 percent sample of households of the state with population more than 1, 00,000 satisfying the condition of 5 percent level of error and 5000 households of the state population less than 1,00,000 satisfying the condition of 10 percent level of error.
To improve the sampling efficiency, the household records within each district were first stratified into homogeneous groups on the basis of place of enumeration, size of village/towns and geography as stratification variables. In all, 6strata's are formed in each district. The household records were further sorted by all the geographic codes: Tahsil/ Taluk/ P.S. / Development Block/ Circle/ Mandal, Town/ Village, Ward (for urban), Enumeration Block Number, and Household Number within each of the 6 strata. This sorted frame provides a high level of implicit stratification. Finally, a systematic random sample from this sorted frame has been drawn within each district with a sampling interval of 20 or 10 as the case may be. A detailed description of variables included in the sampled households and recodes of SMCI is an important input for the data users which may be seen in Annex 2 of Chapter 3.
The National SMCI is an aggregate of 20 percent or 10 percent sub samples of 5 percent sample or 10 percent sample respectively selected at district level for state sample using the same implicit stratification, which effectively give the 1 percent sample of total households at national level. For example, in the case of the districts with a 5 percent sample of households, a random systematic sub sample of 1 in 5 households was selected from the 5 percent sample, keeping the original order of selection based on the stratification criteria. A random systematic sub sample of 1 in 10 households was selected from the 10 percent sample for the smaller districts. The national SMCI will enable us to generate the reliable estimates at state level only. For this purpose, the identity of the districts has been concealed by replacing their location codes by 99 in the 1 percent national level SMCI file.
Weighting
As the SMCI is a sample data from the Census of India 2011 Population Enumeration records, it is necessary to apply a weight or expansion factor to generate the estimates for making inferences for the entire population at districts/state/national levels. The selection of the weight would depend upon the level of aggregation. The weight is calculated as the inverse of the sampling fraction. Therefore, the data records for districts with a 5 percent sampling rate are assigned a weight of 20, and those in districts with a 10 percent sampling rate are assigned a weight of 10. In the case of the 1 percent sample for the National SMCI file, the weight is 100. The weights are included in the data file for use in the Census and Survey Processing (CSPro) application or with any other software for generation of estimates of population parameter. The weighted estimates of the total number of households with a particular characteristic will be the sum of the estimates for the households with that characteristic. In the case of ratios and proportions, both the numerator and denominator would be estimated separately using corresponding weights. Software such as CSPro and SPSS can automatically weight the survey data once the weight is specified in the application. The SMCI file contains individual weights for each household, which when applied to the individual records, expands the sample to the relevant total.
Data Collection
Dates of Data Collection
| Start | End |
|---|---|
| 2011-02-09 | 2011-02-28 |
Data Collection Mode
Face-to-face [f2f]
Data Appraisal
Estimates of Sampling Error
Most of the estimates that users will make from the SMCI will be the weighted total number of households with particular characteristics. Since the total number of households in the frame for the district or state is known, the weighted estimate of the total number of households with a particular characteristic would be equivalent to:
Y = N × p,
Where, Y = weighted total number of households in the geographic domain (for example, district or state) with a particular characteristic as estimated from SMCI
N = total number of households in the frame for the geographic domain, from the Census Population Enumeration, 100 percent data.
p = proportion of households in geographic domain with a particular characteristic as estimated from SMCI.
The weighted estimates produced from the SMCI files are subject to two types of errors: sampling errors and non-sampling errors. The sampling errors are due to the fact that the estimates are based on a sample of the Census of India 2011 Population Enumeration data, not on the 100 percent data used to produce the published results. Another sample would give different results. The sampling error is a measure of the average variability among the estimates from all possible samples, so it is an indicator of the precision of the sample estimate. The non-sampling errors include all other types of error, which affect the Census Population Enumeration data, including response error, coding and processing errors. The quality control procedures in the census operations were designed to reduce the non-sampling errors as much as possible.
The sampling error of an estimate is measured statistically by the standard error, or square root of the variance of the estimate. The standard error decreases according to the square root of the sample size, so it depends on the size of the population and the sampling rate.
The sampling procedures used for selecting the sample household records for the SMCI file is actually slightly more efficient than a simple random sample. The stratification in the sample design results in design effects, which are generally less than one. The design effect defined as the ratio of the variance of an estimate based on the actual sample design to by the corresponding variance from a simple random sample of the same size and is a measure of the relative efficiency of the design. Therefore, the standard errors calculated using the formula above can be considered to be slight overestimate of the actual standard errors, but it provides a good approximation for obtaining the standard error. The formulas above were used to produce a generalized table of standard errors, which can be used to obtain the approximate standard error for a particular estimate, given in Anenex-1, from the SMCI.
While estimating ratios where both the numerator and denominator would be estimated separately using corresponding weights, then the calculation of standard error includes variance and covariance of both the variables.
Y = N × p,
Where, Y = weighted total number of households in the geographic domain (for example, district or state) with a particular characteristic as estimated from SMCI
N = total number of households in the frame for the geographic domain, from the Census Population Enumeration, 100 percent data.
p = proportion of households in geographic domain with a particular characteristic as estimated from SMCI.
The weighted estimates produced from the SMCI files are subject to two types of errors: sampling errors and non-sampling errors. The sampling errors are due to the fact that the estimates are based on a sample of the Census of India 2011 Population Enumeration data, not on the 100 percent data used to produce the published results. Another sample would give different results. The sampling error is a measure of the average variability among the estimates from all possible samples, so it is an indicator of the precision of the sample estimate. The non-sampling errors include all other types of error, which affect the Census Population Enumeration data, including response error, coding and processing errors. The quality control procedures in the census operations were designed to reduce the non-sampling errors as much as possible.
The sampling error of an estimate is measured statistically by the standard error, or square root of the variance of the estimate. The standard error decreases according to the square root of the sample size, so it depends on the size of the population and the sampling rate.
The sampling procedures used for selecting the sample household records for the SMCI file is actually slightly more efficient than a simple random sample. The stratification in the sample design results in design effects, which are generally less than one. The design effect defined as the ratio of the variance of an estimate based on the actual sample design to by the corresponding variance from a simple random sample of the same size and is a measure of the relative efficiency of the design. Therefore, the standard errors calculated using the formula above can be considered to be slight overestimate of the actual standard errors, but it provides a good approximation for obtaining the standard error. The formulas above were used to produce a generalized table of standard errors, which can be used to obtain the approximate standard error for a particular estimate, given in Anenex-1, from the SMCI.
While estimating ratios where both the numerator and denominator would be estimated separately using corresponding weights, then the calculation of standard error includes variance and covariance of both the variables.
Metadata production
Date of Metadata Production
2021-04-06