| Flag for age at death (b13) - unexpected results [message #1155] |
Tue, 14 January 2014 09:07  |
schoumaker
Messages: 66
Registered: May 2013
Location: Belgium
|
Senior Member |
|
|
Hello,
I am looking at the flag variable (b13) for age at death.
In a few DHS data files (AMIR54, MBIR53), the percentage with code 6 (Reported age at death is outside the range expected for the units given) is unusually high. In addition, the age at death seems to be recorded correctly for these cases in b6. I do not underestand why b13 is coded as 6.
Has anyone an explanation for this ?
Thank you.
Bruno
Bruno Schoumaker
Centre for Demographic Research
Université catholique de Louvain
|
|
|
|
| Re: Flag for age at death (b13) - unexpected results [message #1157 is a reply to message #1155] |
Tue, 14 January 2014 19:10  |
Reduced-For(u)m
Messages: 292
Registered: March 2013
|
Senior Member |
|
|
Just because if Bruno wants to know, I want to know - I checked a couple of datasets I have handy (most recent Ghana and Kenya, just "randomly") and found a similar pattern.
tab b13 *(displays values of flag and frequency)
* gives me 91 (Ghana) and 135 (Kenya) obs listed as "outside range", which is almost all of the missing ones (a few in the other categories).
tab b6 if b13!=0 *(displays ages at death if there is a flag)
* Ghana - 91 hits for an age listed as "years: 1"
* Kenya - 125 hits (all but 10) listed as "301"
I realize this doesn't help Bruno, but just wanted to point out that I see it other places too, and now I am curious about it. So...
|
|
|
|
|
|
|
|
| Re: Flag for age at death (b13) - unexpected results [message #1167 is a reply to message #1157] |
Wed, 15 January 2014 13:51 |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
Dear User,
Here is a reponse from one of our programming experts, Guillermo Rojas:
B13 is a variable that gives an indication of the quality and imputation needed for children's age at death. Code 6 in that variable indicates that the information on age at death doesn't comply with the instruction that the interviewers are given to collect that variable, these instructions are:
220
IF DEAD:
How old was (NAME) when he/she died?
IF '1 YR', PROBE:
How many months old
was (NAME)?
RECORD DAYS IF
LESS THAN 1
MONTH; MONTHS IF
LESS THAN TWO
YEARS; OR YEARS.
If those instructions aren't followed variable B13 is set to code 6 (i.,e., if the age at death was given as : 0 months, or 1 year, or more than 30 days or more than 23 months).
If this has not answered your question adequately, please reply
Thank you!
[Updated on: Thu, 06 March 2014 11:04] Report message to a moderator |
|
|
|
| Re: Flag for age at death (b13) - unexpected results [message #1168 is a reply to message #1167] |
Wed, 15 January 2014 14:24 |
Reduced-For(u)m
Messages: 292
Registered: March 2013
|
Senior Member |
|
|
Thanks Liz. So practical question then:
Suppose I'm computing U1 infant mortality rate by year of birth (just in the simplest, Ndeaths/Nbirths, way) - include or exclude the flagged observations? It seems like this is mostly an issue with enumerators following a difficult set of instructions (when to record days v. months, for example), and may not be an indication of bad data. But it also seems like if the answers are off by, say, a few months or so, you've got a reasonable chance of ascribing the person to the wrong cohort/birth year.
Just curious what DHS does with these observations when computing infant/child mortality rates. I don't think the DHS reports usually compute the year-on-year mortality rates, but what do they do with these observations in general?
Thanks.
|
|
|
|
| Re: Flag for age at death (b13) - unexpected results [message #1169 is a reply to message #1168] |
Wed, 15 January 2014 15:41 |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
Dear User,
Here is another response from Guillermo Rojas:
We don't exclude any cases, even in the case of missing data for age at death we impute an age at death in months in variable B7. For the flagged cases we use whatever information is provided and convert to age at death in months in B7. We use B7 for all mortality rates.
If this has not answered your question adequately, please reply
Thank you!
|
|
|
|
| Re: Flag for age at death (b13) - unexpected results [message #1170 is a reply to message #1169] |
Wed, 15 January 2014 15:54 |
Reduced-For(u)m
Messages: 292
Registered: March 2013
|
Senior Member |
|
|
That is definitely the answer I needed. Thanks Liz for being so quick on this, and thank Guillermo for the response too. And thanks Bruno for bringing up the problem (it is not the first time you've brought an interesting data problem to my attention, I assume it won't be the last).
Thanks all.
j
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
