Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Incorrect Sub-Population Cases: What could be Wrong


From   Chao Yawo <Yawo1964@yahoo.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Incorrect Sub-Population Cases: What could be Wrong
Date   Mon, 13 Jul 2009 02:57:47 -0400

Hello,

I am running an svy logistic regression model and noticed a sudden
decrease in the number of cases in my sub-population, from a subpop
size of 3130 observed during the bivariate analysis, to a subpop size
of 1968 produced by the multivariate logistic model.  I am wondering
what could be causing the issue.  Here are twoo fragments:  the first
from the bivariate and the second, the multivariate.

Beyond missing values on some variables in the multivariate model,
what could account for this disparity in cases?  thanks in advance for
your help in resolving this.

I have applied all the latest updates.

rgds, CY



*** Bivariate Results  ***

. svy, subpop(SexActiveHIVNegMale): logit V781_R Year2534
(running logit on estimation sample)

Survey: Logistic regression

Number of strata   =       200                  Number of obs      =     10687
Number of PSUs     =       357                  Population size    = 11252.784
                                                Subpop. no. of obs =      3130
                                                Subpop. size       = 3276.3349
                                                Design df          =       157
                                                F(   1,    157)    =      4.19
                                                Prob > F           =    0.0424

------------------------------------------------------------------------------
             |             Linearized
      V781_R |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    Year2534 |   1.060853   .5184369     2.05   0.042     .0368424    2.084864
       _cons |  -2.539335   .1933592   -13.13   0.000    -2.921256   -2.157414



*** Multivariate ***

. svy, subpop(SexActiveHIVNegMale): logit V781_R Year2534 Year35
NonChristians Primary SecondaryHigherEduc wealth1  wealth2  wealth4
wealth5 ///
>                                       TwoRiskyBehav WantMoreKids
(running logit on estimation sample)

Survey: Logistic regression

Number of strata   =       200                  Number of obs      =      9525
Number of PSUs     =       357                  Population size    = 9969.9646
                                                Subpop. no. of obs =      1968
                                                Subpop. size       = 1993.5153
                                                Design df          =       157
                                                F(  11,    147)    =      3.93
                                                Prob > F           =    0.0001

------------------------------------------------------------------------------
             |             Linearized
      V781_R |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    Year2534 |   1.291047   .6174052     2.09   0.038     .0715545    2.510539
      Year35 |   .6223602   .4980958     1.25   0.213    -.3614732    1.606194
NonChristi~s |  -.2284399   .2398172    -0.95   0.342    -.7021241    .2452443
 PrimaryEduc |   .5837684    .329269     1.77   0.078    -.0666002    1.234137
SecondaryH~c |    .354775   .2689387     1.32   0.189    -.1764297    .8859798
     wealth1 |  -.4953019   .3694412    -1.34   0.182    -1.225018    .2344144
     wealth2 |  -.5417754   .2990481    -1.81   0.072    -1.132452    .0489011
     wealth4 |   .8858961   .3466728     2.56   0.012     .2011517     1.57064
     wealth5 |   .9065355   .2816133     3.22   0.002     .3502959    1.462775
TwoRiskyBe~v |   .4633936   .3322442     1.39   0.165    -.1928516    1.119639
WantMoreKids |   .4557333   .3714754     1.23   0.222    -.2780009    1.189468
       _cons |  -3.720843   .7599802    -4.90   0.000    -5.221947   -2.219738
------------------------------------------------------------------------------
Note: strata with single sampling unit centered at overall mean.

.
end of do-file
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index