Re: Mismatch in the partner's age given in the IR file vs MR file [message #29624 is a reply to message #29621] |
Thu, 11 July 2024 11:40 |
Bridgette-DHS
Messages: 3190 Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
I believe you did the merge correctly, but it would have been simpler to use the CR files for these surveys. They already have matched the partners using the same criteria. I opened the CR file for the NFHS-5 (IACR7EFL.dta) and then entered "gen delta=v730-mv012". I entered "tab delta" to get the unweighted distribution of delta. delta is 0 for 77% of the cases; +/-1 for 12%, and +/- 2 for 4%. The remaining cases are spread over a very wide range, from -37 to +66!
The differences indicate reporting errors, and they reflect on the quality of all the age data. Reports of age in many surveys, especially in South Asia, show high levels of heaping at multiples of 5, and that alone can shift a true age by one or two years. Very large differences are difficult to explain, but fortunately there are very few of them. The great majority of these pairs of ages are within 2 years of each other.
|
|
|