Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Cluster Survey DHS Parameterization


From   Steve Samuels <sjsamuels@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Cluster Survey DHS Parameterization
Date   Thu, 21 Oct 2010 09:50:15 -0400

Andew Lover:

It really helps if you give all references. I assume that the list
discussion you referred to is the thread at
http://www.stata.com/statalist/archive/2009-07/msg00906.html. I've
little experience with DHS, so take my comments below with that in
mind.

It looks to me like you should define eight strata, as shown in Table
A1 p. 42 of the report you reference. These are presumably formed from
(v023 or v024 -I don't know which ) and v025.  Compare these to the
v022 stratum variable.

To check: Make sure that the frequency counts of PSUs match the totals
in that table. You should also be able to reproduce the means and
proportions in Table B2, page 49 (they are all ratio estimates) Your
standard errors by Taylor Series linearization should be the same if
you use the first-stage fpcs.

Your statement that ORs vary with the cluster specification puzzles
me. This shouldn't happen. If it does, then something  is very wrong.

Also, you seem to be working with the HH file, since you refer to HVxx
variables. ORs would usually be computed on individual data. It's
possible that individual weights differ from HH weights if there was
some correction for individual non-reponse. The MIS2 document says
nothing about that, but you should compare the two weights.

Steve

Steven J. Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax:    206-202-4783

>
> (apologies to those I spam'ed earlier with private emails- FAQ! FAQ!).
>
> I'm attempting to use a 2006 Malaria Indicator cluster survey,
> administered by the DHS in Angola; Stata 11.1 on a Mac.
>
> It appears that a quite complex stratification and clustering process
> was used, and we are uncertain of the svy:set necessary. We've applied
> the sample weight correction (divide by 1e6), and have played with
> various other strata besides the "stratum" var in DHS, which other
> posts here have suggested is a vestigial data field. We are getting
> wildly different ORs with different cluster settings, and so....
>
> The current settings, almost certainly not correct:
>
> pweight: sampwgt (= HV005 "sample weight")/1e6)
>          VCE: linearized
>  Single unit: missing
>     Strata 1: stratum (= HV0022 "sample stratum number", or HV025
> "rural/urban" or HV024 "Region")
>         SU 1: PSU (= HV021 "Primary sampling unit")
>        FPC 1: <zero>
>
> Other fields like HV001 (Cluster number), HV002 (Household number) are
> not being utilized (as per DHS suggestion).
> The full DHS report (sampling info in Appendix A) at
> measuredhs(dot)com/pubs/pdf/MIS2/MIS2(dot)pdf
>
> Any ideas or suggestions?
> cheers-
> Andrew
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index