| Re: Apply weight to merge data file [message #19224 is a reply to message #19061] |
Mon, 11 May 2020 22:49  |
Trevor-DHS
Messages: 805
Registered: January 2013
|
Senior Member |
|
|
The simple answer is none of these three. If you are analyzing women and men the easiest thing to do is to analyze women and men separately and then you can use v005 for women and mv005 for men.
However if you need combined estimates for both sexes, then you need to adjust the weights for women and men to take into account the fact that there were almost 7 times as many women interviewed than men, as the men's interviews were only in a subsample of the survey. The simplest way to do this is to take a reliable estimate of the population of women and men age 15-49 for India around 2015, either from census estimates or from sources such as the UN World Population Prospects and divide the population of women by the number of women age 15-49 interviewed from table 3.1 in the NFHS report (699,686) and similarly for men divide the population of men by the number of men age 15-49 interviewed (103,411). This will give you a factor to multiply the women's weight (v005) by and similarly a factor to multiply the men's weight (mv005) by. Then when you analyze the women and men age 15-49 together, apply these factors to the weights as part of your analysis.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
