India - Time Use Survey 1998

The sampling design adopted in the survey was three stage stratified design. The first, second and third stages were the district, villages/urban blocks and households. Proper stratification of the districts in the selected states were done using the population density and proportion of tribal population to ensure capturing of the variability in the population. In the villages/urban blocks also sub-stratification was adopted to ensure representation of all types of households in the Survey.

The total number of households covered in the sample was 18,628 planned.
The total sample size of 18,628 households were first distributed in proportion to the total number of estimated households as per the 1993-94 survey of the National Sample Survey Organisation. No. of first stage units (villages and sample blocks) were determined using the initial sample size to be allocated to each state and by assuming that in each f.s.u. , 12 households will be surveyed. The number of f.s.u. so arrived at was adjusted to be multiple of 8 as atleast 2 f.s.u. each may be covered in 4 sub-rounds.

Selection of villages : All the villages in the selected district were grouped in 3 categories namely large (population above 1200), medium (population between 400 to 1200) and small(population less than 400) . The total rural sample was distributed in three stratum in proportion to the population in the three stratum. In
case any stratum was not applicable in a particular district, the allocated sample was distributed in the remaining stratum only. If more than one village was to be selected in the particular stratum , then villages-were selected using circular systematic sampling with probability proportional to the population. If all the three strata were present then minimum sample size allotted in each stratum was 2.

Selection of urban sample blocks : All the towns in the selected district were grouped in 3 categories namely large(population more than 2 lakhs), medium(population between 50000 to 2 lakhs) and small (population less than 50000) . The total urban sample was distributed in three stratum in proportion to the population in the three stratum. In case any stratum was not applicable in a particular district, the allocated sample was distributed in the remaining stratum only. If more than one sample block was to be selected in the particular stratum, then ufs blocks in each of the towns were presented by investigator unit and ufs
blocks no. The requisite number of ufs blocks were then selected by using circular systematic sampling with equal probability. If all the three strata were present then minimum sample size allocated in each stratum was 2 due to this, in some cases, overall urban sample size allotted in a particular district might have increased.

As no previous survey was conducted on this topic and methodologies to be used were not firmed up, it was decided to conduct this survey on a pilot basis. However, to ensure the use of data collected in the pilot survey also, a proper sampling procedure was followed.

Refer the attached document named 'Report' attached under external resource


There was no deviation from the original sample deviation.


The total number of households covered in the sample was 18.591 as against 18,620 originally planned.
68 % of the households was in rural areas. Therefore, the non-response at 0.1 % was negligible.


For generating any estimate, one has to extract relevant portion of the data, and aggregate after applying the weights.

Weights (or multipliers) VARIABLES are given at the end of record of each dataset. The weights (multipliers) are
WGT_SS for Sub-sample-wise estimation and WGT_Combined for combined subsample estimates.

All records of a household will have same weight figure. In case of those Blocks/Levels, where Item/Person Sl.No. is not applicable the field is filled up with 00000.

Use of subsample-wise weights (multipliers)

For generating subsample-wise estimates based on data of all subrounds taken together, either Subsample-1 households or Subsample-2 households are to be considered at one time. Subsample code is available in the dataset For generating subsample-combined estimates based on data of all subrounds taken together all households are to be considered.