 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: How to calculate mortality ratios
Gaby, I had overlooked this post when I stated on the List that you
hadn't answered my questions. You had indeed. I am very sorry that I
stated otherwise.
-Steve
On Dec 4, 2007, at 4:36 AM, Ana Gabriela Guerrero Serdan wrote:
Dear Steve,
The frame was taken from a census, which i dont have.
The survey was done in two stages. PSUs were selected
with linear systematic pps sampling. Stratification
was done at the regional level and urban/rural areas.
I have information in the survey on the households per
psu, psu, strata, regions, pweights, and all hhs
characteristics, etc..
The main/second respondents in each houshold provided
information on who has died in their household and
what was the cause of death. In the questionaire, the
interveiwer had several options to mark from.
If there would be a one-member household that is
death and had lived alone, certainly this is not
reflected in the hh survey.
The survey is like a LSM survey , so the aim is to
know about living conditions. It is not a mortality
survey although some quesions were asked.
I want to calculate mortality ratios per region by
different causes, this would be sort of a CMR (*1000),
so I would be able to compare them.
thanks,
Gaby
--- Steven Joel Hirsch Samuels
<[email protected]> wrote:
Okay, Gaby. Next, please describe how the sample
was taken. What
was the sampling 'frame'? What were the strata?;
what was the
'first-stage' of sampling? What were later stages?
In most household
surveys, 'household' is rarely the first stage of
sampling. Usually
the strata are divided into smaller areas and a
sample of these is
taken. Only at a later stage are HH drawn.
How was the information collected? how did the
survey ascertain
characteristics of people who had died? What if
they had lived
alone? Who provided the causes of death?
-Steven
On Dec 3, 2007, at 6:02 AM, Ana Gabriela Guerrero
Serdan wrote:
Dear Steve,
I have two datasets.
1) One dataset has individual information for
each
individual in the household (individual
characteristics) and also hh id, regions,pweights,
psu, etc..
2) However, deaths are in a separate dataset which
includes only those individuals that died, it
specifies: household id, region, gender, pweights,
psu, and the causes of death.
I want to calculate mortality ratios for each of
the
causes of death (e.g. disease, traffic accident)
per
region.
So basically the hh id is the variable to know
which
individuals died in each household.
So I think I need to first joinby both datasets,
then
calculate the population estimates per region and
then
mortality ratios?
thanks,
Gaby
--- Steven Joel Hirsch Samuels
<[email protected]> wrote:
Gaby, why don't you lay out exactly what your
files
are, and what
analysis variables are in them. Do you want to
do
'proportional'
mortality ratios, single mortality rates? Do you
want to test
differences among regions (AI02)?
-Steven
On Dec 2, 2007, at 10:13 AM, Ana Gabriela
Guerrero
Serdan wrote:
Steve,
thanks. Yes sorry my ratio should have been x/y.
I missed to calculate the total population first
as
this is in a separate file. I was doing
calculations
only among those that died from different
causes.
So from file 2 I need to calculate my total
population. I have individual information for
each
household in the survey. Not sure if I should
first
generate a count variable?
svyset AI06 [pw= expweigh], strata( AI05)
gen count=_n
svy: total count, over(AI02)
then use this to calculate the mortality ratio
in
file
1 as you indicated.
thanks again,
Gaby
--- Steven Joel Hirsch Samuels
<[email protected]> wrote:
Gaby:
1. Ratio requires a numerator and denominator:
svy: ratio myratio = x/y
svy: ratio dead should return an error
message
in Stata 10.
What are your individual observations? people
with
dead indicating
status yes/no, or some other unit? If you have
an
'area' as your
observation, with 'dead' counting deaths and
'pop'
giving the
population total for the area, then
svy: ratio (mr= dead/pop)
would work.
2. svy: prop death will give proportions dead
and
not dead. This
will be correct if your analytic unit is
person.
3. I suggest that you form new strata by
pooling
singletons into
neighboring regions.
-Steven
On Dec 2, 2007, at 8:25 AM, Ana Gabriela
Guerrero
Serdan wrote:
Dear Stata Users,
I have two related questions:
1) Im trying to calcuate mortality ratios for
different sub-populations using survey data.
Is it very naive from my side to use the
following
commands?
svyset AI06 [pw= expweigh], strata( AI05)
svy: ratio death, over (Province)
I also get similar results with prop.
2) I dont get any SE because I have strata
with
one
sampling unit. I see that Stata suggests (as
also
some
books) to delete or collapse the strata with
one
sampling unit. Do you know what are the
implications
for this? Would I need to collapse the strata
according to region?
thanks,
Gaby
Gaby Guerrero Serdan
Deparment of Economics
Royal Holloway, University of London
TW20 OEX
Egham, Surrey
England, UK
http://www.rhul.ac.uk/economics/About-Us/postgrads.html
http://www.flickr.com/photos/49939890@N00/
Tel: +44 7912657259
=== message truncated ===
Gaby Guerrero Serdan
Deparment of Economics
Royal Holloway, University of London
TW20 OEX
Egham, Surrey
England, UK
http://www.rhul.ac.uk/economics/About-Us/postgrads.html
http://www.flickr.com/photos/49939890@N00/
Tel: +44 7912657259
______________________________________________________________________
______________
Get easy, one-click access to your favorites.
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/