st: Complex Survey (microdata), meff and inference

["Patrick W�hrle Guimar�es" <probabilidade@gmail.com>]

  I have been working with complex samples (weight, strata and primary
sampling unit) and I have some subjects in relation to applicability
in Stata. In a complex sample whenever the inference is made - it is
not necessary to measure the effect of the survey design (for example:
MEFF - Misspecification effect) If MEFF goes larger than 1, the
inference should not be made with the normal distribution but with a
type of normal corrected (t0~N[0;MEFF(�;v0)].

  In income distribution and poverty, I have been reading some
algorithms- none of them has that concern in relation to the effect of
the survey design in inference (at least they should be tested the
effect of the survey design can be despised).

  I began to read the manual of DaD (and his conversion in Stata -
Dasp 1.4) and the same problem. When the statistical packages tested
unit root in the beginning there was the same problem - the researcher
had to appeal to a book because the generated statistics was mistaken.

  The inference that is done in those cases is "nominal" and "no
real". For each procedure and for every year it should be possible to
take that in bill - did anybody already come across that making
analyses in income distribution and poverty and as it solved? With
best regard,...


-- 
Patrick

*
*   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/