The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Merging data files » discrepancies after merging PR and IR/MR
discrepancies after merging PR and IR/MR [message #6938] Wed, 05 August 2015 12:10 Go to next message
fabienne is currently offline  fabienne
Messages: 6
Registered: March 2015
Hello everyone

I am working with the DHS VI from Liberia.

I appended the women file (IR) and the men file (MR) and merged it to the household member file (PR), which resulted in 48,219 observations and 13,357 matches (merged on: v001 v002 v003).

To check the success of the merging process, I looked at the consistency of variables which are given in both the individual (IR/MR) datasets and the household (PR) dataset:
- hv105 and (m)v012 (Age of respondent)
- hv106 and (m)v106 (highest educational level)

Unfortunately, I found 2990 age and 691 education discrepancies. Commonly, the age discrepancies are small (1 year). Does this indicate a failed merge (which I do not believe) or is there a methodological explanation for this effect? Which age and education variable is more trustworthy and should be used for further analysis?

Thank you and best regards, Fabienne
Re: discrepancies after merging PR and IR/MR [message #6980 is a reply to message #6938] Fri, 07 August 2015 10:52 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 1770
Registered: February 2013
Senior Member
Response from Senior Data Processing Specialist, Noureddine Abderrahim:

You need to ensure that the file constructed from the men and the women's data is complete. You need also to keep in mind that the person's file does not include only the individuals who were interviewed but also members who were not interviewed such as children, non defacto members, etc.

Also, be aware that the data in the household roster (such as HV105, HV106, ..) might be different from the individual data (V012, MV012, V106, MV106) because the respondent to the household interview might be different from the respondent to the individual interview.

To check that the merge was done correctly, you need to see if the data of the women and men in the new file, matches the data in the individual data files (such as V106 in the new data file matches V106 in the individual data file, the same for MV106 in the men's data file).
Previous Topic: Weighting after Merging
Next Topic: merging children's file and wealth index
Goto Forum:

Current Time: Fri Apr 3 07:02:17 Eastern Daylight Time 2020