Echantillonnage
Méthode d'échantillonnage
Two separate samples were drawn: one in districts where only rural areas are surveyed (123 districts), and another in districts where rural areas and urban centres are surveyed (21 districts).
The sampling frame used to draw units in rural areas only is the 1998 Population Census. A two-stage sample design was adopted:
- In the first stage, 30 villages are selected. Every year, 20 villages from the previous year are retained and 10 new villages are added, using probability proportional to size. This gives a "rotating panel" of villages, which generates better estimates to assess changes.
- In the second stage, 20 households are sampled in each of the 30 selected villages. The village is divided into four parts. In each of the four parts, volunteers start from the central location and pick every 5th household on the left hand-side in a circular fashion until 5 households are selected from each part. This allows the selection of 600 households.
To avoid bias in the sampling frame, the sampling of 21 rural and urban districts was done by Pakistan Bureau of Statistics. This way, it was ensured that the boundaries of rural and urban areas do not overlap with each other and selected blocks/villages are different for the urban districts and same rural districts. The sampling frame used to draw units in the 21 districts where urban centres where also surveyed is the 2011 Economic Census and the housing listing for 2011 census. A stratified two-stage sample design was adopted. Each urban area was divided into block of 200-250 households, to form the Primary Sampling Units (PSU) for urban domain. Villages are PSUs for rural domain. Then, stratification for urban areas was performed based on the type of area (large city, other urban area), and larger cities were further stratified according to income (high, middle, low). In rural domain, each administrative district has been treated as independent and separate stratum.
- In the first stage, PSUs are selected using probability proportional to size method.
- In the second stage, 20 households are sampled in each selected PSUs using systematic sampling technique.
Télécharger la documentation