Exploring Alternative Stratification Variables for Sampling of Households in the PLFS

Soon after the results based on the “first round” (2017–18) of the annual series of the Periodic Labour Force Survey (PLFS) got released, a lot of hue and cry was made about their comparability with the employment–unemployment estimates available through the past quinquen­nial series of Employment–Unemployment Survey (EUS) with the latest one conducted during 2011–12. One of the major arguments put forth by some of the critics in support of the said “non–comparability” was the deviation in the use of number of members in the household with secondary or above level of general education as the stratification variable for stratification and sampling of households in the PLFS instead of income–based criterion (approximated by the house­hold monthly per capita expenditure in the urban areas) used earlier in the EUS.

Apparently, the confusion regarding comparability of the PLFS results with the past EUS series has not completely evaporated. On this particular issue, we firmly believe that the adjustment of sample data at the household level with corresponding design–based weights—which differ from stratum to stratum—would nullify the effect of change in the stratification variable on the final estimate. And, thus, it may not be realistic to challenge the comparability of PLFS estimates with the past series on the grounds of such a change in the stratification variable.