[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Re: Normalization in Stata
It should not be necessary to -reshape- the data because DHS data can
be downloaded in flat format, as Macro calls it, in which every
household member is one observation. For the benefit of other list
members let me mention that DHS means Demographic and Health Survey
By the way, I am sure that more than 203 households are listed in the
data. Not only would that be a very small sample for a DHS survey, I
also have not heard of a community where the average household has
184 members. In DHS data, the household ID (variable hv002) is
cluster specific (variable hv001), so that the actual household ID
for the entire sample is usually a combination of hv001 and hv002.
--- Phil Bardsley <firstname.lastname@example.org> wrote:
> What ORC Macro means by "normalize" in this case is to make
> each individual household member into a separate observation.
> That way the sampling weights are appropriate. You can do this
> with the -reshape long- command. The "Guide to DHS Statistics"
> by Rutstein and Rojas, available from ORC Macro, goes into
> detail about how to do this in SPSS. I can fax you the pages
> if you want to read it yourself.
> Date: Sat, 28 Jan 2006 19:34:50 +0000
> From: Sabah Abdullah <email@example.com>
> Subject: st: Re: Normalization in STATA
> How do we normalize variables in STATA? I have household
> survey from DHS and had simple output to describe the variables
> (see below). Note, the number of observations and the number of
> total household is problematic in running mlogit or ologit
> analysis. See, there were several members of the same household
> who took part in the sruvey (see the observation no. as 37,263
> as opposed to 203 household numbers.) I read in the DHS manual
> ' if the unit of analysis is the household members then the
> files need to be normalized". There are various normalization
> i.e. symetrical etc
> Any suggestion or comments?
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
* For searches and help try: