| Child marriage prevalence discrepancies [message #7083] |
Thu, 20 August 2015 19:38  |
kmcneill
Messages: 1
Registered: August 2015
|
Member |
|
|
Hello,
I am trying to assess the rates and changes in the prevalence of child marriage (under-15 and under-18) over time. However, in triangulating the data between the three rounds of DHS, the timing of the decline looks quite different according to the 2000 round than in the 2005 and 2011 rounds. (Please see attached graphs.)
I am trying to get a sense of which trends are more trustworthy, if possible. On the one hand, perhaps there were greater implementation challenges in the first round than the later rounds, and so the data is less reliable; on the other hand, the older cohorts of women are reporting their age at first marriage with less of a delay, and may be less subject to recall bias.
Any information that you can provide to contextualize these differences would be hugely helpful.
Thanks so much,
Kristen
|
|
|
|
| Re: Child marriage prevalence discrepancies [message #7084 is a reply to message #7083] |
Thu, 20 August 2015 20:24  |
Reduced-For(u)m
Messages: 292
Registered: March 2013
|
Senior Member |
|
|
My first thought would be that it could be an age-reporting issue in the data. Are you back-calculating their age at marriage from their age at survey time? One thing to try might be to drop women with age flags (they flag it if you can't remember certain parts of your age and they have to help you narrow in on it). You could also look at a histogram of reported ages and/or birth years to see if there are any funny spikes. In general I think that it is hard to trust some of the age reporting for people born more than 10 or 20 years ago - they just don't know very well. But maybe I'm totally off, so this is just a suggestion for something to check.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
