st: "Too many weights" in Stata -svy- command

[davidj miller <davjmiller@mac.com>]

Dear Statalisters: 

I am trying to combine two downloadable files from the US's National Center for Health Statistics and use these to estimate  various percentiles and standard errors for chemical  measurements.  The psu (sdmvpsu) and strata (sdmvstra)  information are contained in one file and the sample weight the tI want to use wtsc2yr)  info is contained in the second file.  This is the do file I set up:


fdause "ftp://ftp.cdc.gov/pub/Health_Statistics/nchs/nhanes/2007-2008/DEMO_E.XPT";, clear

describe

save "E:\DEMO 2007_2008.dta", replace 

fdause "ftp://ftp.cdc.gov/pub/Health_Statistics/nchs/nhanes/2007-2008/OPD_E.XPT";

merge m:1 seqn using  "E:\DEMO 2007_2008.dta"

des

summarize wtsc2yr


svyset sdmvpsu[pw=wtsc2yr], strata[sdmvstra]



After the -svyset- command in my do file, Stata stops and returns:

 "too many weights
r(198);"

So I tried to simplify the problem by 'svyset-ing just the demographic file (DEMO_E.XPT") to see if the problem re-occurred using a different set of sampling weights that occur in that file (wtint2yr) and avoid  potential problem introduced by -merge- and I ended up with the same "too many weights" message.  


. fdause "ftp://ftp.cdc.gov/pub/Health_Statistics/nchs/nhanes/2007-2008/DEMO_E.XPT";, clear

. summarize sdmvpsu wtint2yr sdmvstra

    Variable |       Obs        Mean    Std. Dev.       Min        Max
-------------+--------------------------------------------------------
     sdmvpsu |     10149    1.510986    .4999039          1          2
    wtint2yr |     10149    29277.38     26009.7   2359.374   186295.5
    sdmvstra |     10149    65.51306    4.253306         59         74


. svyset sdmvpsu[pweight=wtint2yr], strata[sdmvstra] 
too many weights
r(198)



Simplifying further using the auto data set with foreign, mpg, and rep78 to represent psu, weight, and strata, respectively,  the same thing happens: 

. sysuse auto
(1978 Automobile Data)

. svyset foreign[pweight=mpg], strata[rep78] 
too many weights
r(198);


Interestingly, the example in the Stata -svy- command help file works fine:  

. webuse nhanes2f

.     . svyset psuid [pweight=finalwgt], strata(stratid)

      pweight: finalwgt
          VCE: linearized
  Single unit: missing
     Strata 1: stratid
         SU 1: psuid
        FPC 1: <zero>

.     . svy: mean zinc
(running mean on estimation sample)

Survey: Mean estimation

Number of strata =      31       Number of obs    =       9189
Number of PSUs   =      62       Population size  =  104176071
                                 Design df        =         31

--------------------------------------------------------------
             |             Linearized
             |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
        zinc |   87.18207   .4944827      86.17356    88.19057
--------------------------------------------------------------

.     . mean zinc

Mean estimation                     Number of obs    =    9189

--------------------------------------------------------------
             |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
        zinc |   86.51518   .1510744      86.21904    86.81132
--------------------------------------------------------------

. 



There was a "too many weights" problem addressed on Statalist earlier, but I did not see that as being applicable. 
 
I have done this kind of -merge-ing and -svyset-ing before (using Stata 11) and didn't seem to have problems, and I didn't see anything in the Stata 13 manual indicating any changes in this (at least from Stata 12)


I am  using Stata 13 SE for PC

Thank you for any help you can provide.  

David Miller





davjmiller@mac.com





*
*   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/