Re: st: svyset

[Steve Samuels <sjsamuels@gmail.com>]

What analyses are you planning, Rebecca? If the purpose is analystic,
e.g.. to develop predictions, estimate odds ratios, or to otherwise test
hypotheses or quote p-values, omit the fpc. 

The reasoning is interesting (Cochran, 1977, p.39): For a finite
population it is seldom of scientific interest to ask if a null hypothesis
(e.g. that two proportions are equal) is exactly true. Except by a very rare
chance, it will not be, as one would discover this by enumerating the entire
population. This leads to the adoption of a "superpopulation" viewpoint,
which is taken by almost all statisticians these days. See also
Deming(1966) pp 247-261 "Distinction between enumerative and analystic
studies"; Korn and Graubard (1999), p. 227.

If the purpose is to describe aspects of the sampled population, you may
include the fpc.


References
Cochran, W. G. (1977). Sampling techniques (3ded.). New York: Wiley.

Deming, W. E. (1966). Some theory of sampling. New York: Dover Publications.

Korn, E. L., & Graubard, B. I. (1999). Analysis of health surveys 
(Wiley series in probability and statistics). New York: Wiley.


Steve


On Nov 2, 2012, at 8:36 AM, Rebecca Pietrelli wrote:

Thank you very much Stas Kolenikov!

So, if the systematc procedure had worked with replacement, I could simply use:

svyset psu_var [pw=hhweights_var], strata(strata1_var)

I don't need to create strata2.

I was wondering if I have to use the fpc. I know that it is necesary
to use it with systematic procedure. Now I am confused because the
first stage is a combined procedure (PPS + systematic) and the second
stage is a systematic procedure.

thank you
Rebecca
2012/11/2 Stas Kolenikov <skolenik@gmail.com>:
> This design took a lot of compromises. Systematic sampling is THE
> worst sampling procedure, and the only excuse for using it is that you
> only had access to paper, and could only do a long addition on the
> margin of the printout of your list of EAs with cumulative population
> sizes. Still, systematic sampling could hit some large units (larger
> than {the total population size}/{# of units sampled}) more than once,
> so it could have worked effectively with replacement (which is what
> you want, without replacement is tedious to work with).
> 
> The finite population correction is only relevant for SRS. It does not
> generalize well to PPS type designs, where, technically speaking, you
> should use double probabilities of selection (and if you find yourself
> doing that, you would recall that the systematic sampling does not
> allow unbiased variance estimation... that's why I said it is the
> worst method).
> 
> You are correct regarding strata2: you'd have to create an indicator
> for the type of the HH.
> 
> --
> -- Stas Kolenikov, PhD, PStat (SSC)  ::  http://stas.kolenikov.name
> -- Senior Survey Statistician, Abt SRBI  ::  work email kolenikovs at
> srbi dot com
> -- Opinions stated in this email are mine only, and do not reflect the
> position of my employer
> 
> On Fri, Nov 2, 2012 at 6:06 AM, Rebecca Pietrelli
> <rebecca.pietrelli@gmail.com> wrote:
>> Hi,
>> 
>> I really hope anyone can help me.
>> I am using the Stata command svyset in order to declare the survey
>> design to files of my dataset (Uganda Migration Housolds Survey 2010).
>> 
>> The sampling design is two-stage stratified.
>> In the first stage: enumeration areas - EAs - (PSU) were selected,
>> separately for rural and
>> urban areas. The applied procedure is a PPS (proportionally done based
>> on the number of households in the respective stratum according to the
>> 2006 Uganda household survey) combined with a systematic approach.
>> In the second stage: households were selected in each EA, using a
>> systematic procedure (4 hhs with international migrants, 3 with
>> internal migrants and 3 without migrants).
>> 
>> I think to use the following command:
>> 
>> svyset psu_var [pw=hhweights_var], strata(strata1_var) fpc(fpc1) ||
>> _n, strata(strata2_var) fpc(fpc2).
>> 
>> I have the following doubts:
>> 
>> 1) I am not sure if the first stage is with or without replacement (It
>> is not mentioned in any part of the report!!!). I suppose that a PPS
>> combined with a systematic procedure is without replacement. Is this
>> assumption correct?
>> 
>> 2) I have the strata1_var (rural or urban) but I don't have the
>> strata2_var. In that case, should I create it? (for ex. strata2 = 1 if
>> hh does not have migrants, = 2 if it has internal migrants and = 3 if
>> it has international migrants).
>> 
>> 3) I don't have fpc in the data. I think to create them as
>> [(N-n)/(N-1)]^(1/2). So in the first stage, is N the total number of
>> EAs in Uganda or is it the total number of hhs living in Uganda?
>> 
>> Thank you very much for your help and time.
>> Rebecca
>> 
>> 
>> 
>> 
>> Thank you very much for any help.
>> Rebecca
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/