Many thanks Stas.
On Thu, Oct 1, 2009 at 10:07 PM, Stas Kolenikov <[email protected]> wrote:
>
> Stata uses Rao-Scott corrections for the scale and degrees of freedom
> of the independence test chi^2. It might have happened that something
> went negative because of this corrections. For instance, in your
> extended case, you lost 1.02 degrees of freedom to the design (compare
> chi2(9) and F(7.98, something) ).
>
> On Thu, Oct 1, 2009 at 8:19 AM, Mihir <[email protected]> wrote:
> > Dear all:
> >
> > Anybody has any idea what can be done in the case below?
> >
> > Thanking you,
> > Mihir
> >
> > On Wed, Sep 30, 2009 at 1:29 AM, Mihir <[email protected]> wrote:
> >>
> >> Dear all:
> >>
> >> I am stuck with a problem in pearson chi-square test with survey
> >> command. I have setup survey setting for Nationwide Inpatient Survey
> >> (NIS) data. I used hospital id as sampling unit, discwt as sampling
> >> weight and stratum variable as strata as suggested on NIS website.
> >> When I am trying to run chi-square test for independence between age
> >> category and year of data collection from 1998 to 2007, it gives
> >> proper result, but when I run the same command by retaining only two
> >> years (1998 and 2007), it does not produce p-value. I don't understand
> >> why it is happening? Can anybody help me - what I should do? I am also
> >> getting similar results with other variables that same command works
> >> with more year but doesn't work with two years.
> >>
> >> I have pasted my output for your ready reference.
> >>
> >> Thanking you in advance,
> >> Mihir
> >>
> >> svyset hospid [pweight=discwt], strata(stratum) vce(linearized)
> >>
> >> pweight: discwt
> >> VCE: linearized
> >> Strata 1: stratum
> >> SU 1: hospid
> >> FPC 1: <zero>
> >>
> >> svy: tabulate year age2cat, count column row obs percent format(%10.0g) pearson
> >> (running tabulate on estimation sample)
> >>
> >> Number of strata = 60 Number of obs = 157222
> >> Number of PSUs = 3270 Population size = 759816.12
> >> Design df = 3210
> >>
> >> -------------------------------------------
> >> | age2cat
> >> YEAR | <65 year >=65 yea Total
> >> ----------+--------------------------------
> >> 1998 | 27548.394 10497.654 38046.048
> >> | 72.408029 27.591971 100
> >> | 4.9658285 5.1193873 5.0072705
> >> | 5452 2102 7554
> >> |
> >> 1999 | 25591.467 9586.1825 35177.649
> >> | 72.749224 27.250776 100
> >> | 4.613076 4.6748901 4.6297582
> >> | 5265 1991 7256
> >> |
> >> 2000 | 25067.72 9839.9109 34907.631
> >> | 71.811576 28.188424 100
> >> | 4.5186663 4.7986258 4.5942209
> >> | 5163 2032 7195
> >> |
> >> 2001 | 24673.646 9882.7968 34556.443
> >> | 71.401001 28.598999 100
> >> | 4.4476311 4.8195399 4.5480007
> >> | 4983 1976 6959
> >> |
> >> 2002 | 83751.067 30869.413 114620.48
> >> | 73.068152 26.931848 100
> >> | 15.096831 15.054075 15.085292
> >> | 17682 6513 24195
> >> |
> >> 2003 | 80686.437 27375.013 108061.45
> >> | 74.66718 25.33282 100
> >> | 14.544405 13.349963 14.222053
> >> | 17002 5737 22739
> >> |
> >> 2004 | 77406.999 26898.738 104305.74
> >> | 74.211641 25.788359 100
> >> | 13.953259 13.117698 13.72776
> >> | 16052 5540 21592
> >> |
> >> 2005 | 74552.231 27032.143 101584.37
> >> | 73.389467 26.610533 100
> >> | 13.438663 13.182755 13.3696
> >> | 15263 5507 20770
> >> |
> >> 2006 | 68720.522 26875.807 95596.329
> >> | 71.886152 28.113848 100
> >> | 12.387449 13.106515 12.581508
> >> | 14234 5538 19772
> >> |
> >> 2007 | 66760.785 26199.193 92959.979
> >> | 71.816696 28.183304 100
> >> | 12.03419 12.776551 12.234536
> >> | 13813 5377 19190
> >> |
> >> Total | 554759.27 205056.85 759816.12
> >> | 73.012306 26.987694 100
> >> | 100 100 100
> >> | 114909 42313 157222
> >> -------------------------------------------
> >> Key: weighted counts
> >> row percentages
> >> column percentages
> >> number of observations
> >>
> >> Pearson:
> >> Uncorrected chi2(9) = 91.4953
> >> Design-based F(7.98, 25611.84)= 3.0966 P = 0.0017
> >>
> >> keep if year==1998|year==2007
> >>
> >> svy: tabulate year age2cat, count column row obs percent format(%10.0g) pearson
> >> (running tabulate on estimation sample)
> >>
> >> Number of strata = 60 Number of obs = 26744
> >> Number of PSUs = 1205 Population size = 131006.03
> >> Design df = 1145
> >>
> >> -------------------------------------------
> >> | age2cat
> >> YEAR | <65 year >=65 yea Total
> >> ----------+--------------------------------
> >> 1998 | 27548.394 10497.654 38046.048
> >> | 72.408029 27.591971 100
> >> | 29.210724 28.60642 29.041449
> >> | 5452 2102 7554
> >> |
> >> 2007 | 66760.785 26199.193 92959.979
> >> | 71.816696 28.183304 100
> >> | 70.789276 71.39358 70.958551
> >> | 13813 5377 19190
> >> |
> >> Total | 94309.179 36696.848 131006.03
> >> | 71.988428 28.011572 100
> >> | 100 100 100
> >> | 19265 7479 26744
> >> -------------------------------------------
> >> Key: weighted counts
> >> row percentages
> >> column percentages
> >> number of observations
> >>
> >>
> >> Pearson:
> >> Uncorrected chi2(1) = 0.9557
> >> Design-based F(., .) = . P = .
> >>
> >> *
> >> * 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/
> >
>
>
>
> --
> Stas Kolenikov, also found at http://stas.kolenikov.name
> Small print: I use this email account for mailing lists only.
>
> *
> * 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/
