Re: st: Confidence intervals for survey data from the estimation matrices

[Steven Samuels <sjsamuels@gmail.com>]

-svy: tab- calculates SEs and confidence intervals on logit transforms of the proportions, then transforms the interval-endpoints back as "lb" and "ub".  This guarantees lb>0 and ub<1. You can see the formula relating the SE of the proportion and the SE of the logit transform. on page 127 of the Stata 11 Survey manual.  Apparently e(V) contains the SE of the proportion.

Steve
sjsamuels@gmail.com

On Jul 19, 2011, at 10:55 PM, Darren White wrote:

Hello,

I am attempting to compile estimate tables directly from STATA's temporary estimate matrices, however none of the listed matrices after command: ereturn list (below) contain the confidence intervals that are displayed using the svy: command in STATA (table below). Also the estimate variance matrix e(V) and the square root e(V) x 1.96 +/- give the estimates in the automatic estimate display ( below at bottom) after some renaming of the e(b) and e(V) matrices but these are different to those returned form the svy: tab command.

Can anyone tell me how STATA arrives at the confidence interval estimates using the svy: tab command and where they could be found in the matrix environment? Alternatively how would I replicate the calculation routine for survey data using the variance provided in the e(V) matrix? The e(b) matrix has the same percentage estimates as returned form the svy: tab agecat bmicat, ci command 

Any help or suggestions would be greatly appreciated

Darren White
Population Epidemiology Unit
Queensland Health
Brisbane
Australia

___________________________________________________________________
. svy: tab  agecat bmicat, ci per
(running tabulate on estimation sample)

Number of strata   =         1                  Number of obs      =         7
Number of PSUs     =         7                  Population size    =         7
                                               Design df          =         6

-------------------------------------------------
  bmicat | percentages           lb           ub
----------+--------------------------------------
       1 |       42.86               9.06        84.95
       2 |       57.14             15.05        90.94
          | 
 Total |         100                          
-------------------------------------------------
_____________________________________________________________________
. ereturn list

scalars:
            e(stages) =  1
            e(census) =  0
         e(singleton) =  0
     e(N_strata_omit) =  0
              e(df_r) =  6
                 e(N) =  7
          e(N_strata) =  1
             e(N_psu) =  7
             e(N_pop) =  7
                 e(r) =  1
                 e(c) =  2
             e(total) =  7
            e(mgdeff) =  .
           e(cvgdeff) =  .
              e(rank) =  1

macros:
           e(cmdline) : "svy : tab  agecat bmicat, ci per"
               e(cmd) : "tabulate"
            e(prefix) : "svy"
           e(cmdname) : "tabulate"
      e(marginsnotok) : "_ALL"
           e(predict) : "_no_predict"
         e(estat_cmd) : "svy_estat"
            e(colvar) : "bmicat"
            e(rowvar) : "agecat"
            e(setype) : "cell"
           e(vcetype) : "Linearized"
               e(vce) : "linearized"
             e(wtype) : "pweight"
              e(wvar) : "weight"
              e(wexp) : "= weight"
        e(singleunit) : "missing"
            e(depvar) : "Mean"
        e(properties) : "b V"

matrices:
                 e(b) :  1 x 2
                 e(V) :  2 x 2
         e(V_srs_col) :  2 x 2
          e(Deft_col) :  1 x 2
          e(Deff_col) :  1 x 2
         e(V_srs_row) :  1 x 1
          e(Deft_row) :  1 x 1
          e(Deff_row) :  1 x 1
             e(V_srs) :  2 x 2
              e(Deft) :  1 x 2
              e(Deff) :  1 x 2
             e(V_col) :  2 x 2
             e(V_row) :  1 x 1
               e(Col) :  1 x 2
               e(Row) :  1 x 1
              e(Prop) :  1 x 2
               e(Obs) :  1 x 2
 e(_N_strata_certain) :  1 x 1
  e(_N_strata_single) :  1 x 1
         e(_N_strata) :  1 x 1
_________________________________________________________________________________________
.	ereturn post b V

.	ereturn display
						
	Coef.	Std. Err.	z	P>z	[95% Conf.	Interval]
						
	p11    .4285714	.2020305	2.12	0.034	.0325989	.8245439
	p12    .5714286	.2020305	2.83	0.005	.1754561	.9674011
						
________________________________________________________________________________________

********************************************************************************
This email, including any attachments sent with it, is confidential and for the sole use of the intended recipient(s). This confidentiality is not waived or lost, if you receive it and you are not the intended recipient(s), or if it is transmitted/received in error.
Any unauthorised use, alteration, disclosure, distribution or review of this email is strictly prohibited.  The information contained in this email, including any attachment sent with it, may be subject to a statutory duty of confidentiality if it relates to health service matters.
If you are not the intended recipient(s), or if you have received this email in error, you are asked to immediately notify the sender by telephone collect on Australia +61 1800 198 175 or by return email.  You should also delete this email, and any copies, from your computer system network and destroy any hard copies produced.
If not an intended recipient of this email, you must not copy, distribute or take any action(s) that relies on it; any form of disclosure, modification, distribution and/or publication of this email is also prohibited.
Although Queensland Health takes all reasonable steps to ensure this email does not contain malicious software, Queensland Health does not accept responsibility for the consequences if any person's computer inadvertently suffers any disruption to services, loss of information, harm or is infected with a virus, other malicious computer programme or code that may occur as a consequence of receiving this email.
Unless stated otherwise, this email represents only the views of the sender and not the views of the Queensland Government.
**********************************************************************************


*
*   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/