Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: [Fwd: st: Kolmogorov Smirnov, groupvar as cumulative relativefrequencies or rawdata?]


From   Knag Anne-Christine <anne-christine.knag@imr.no>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: [Fwd: st: Kolmogorov Smirnov, groupvar as cumulative relativefrequencies or rawdata?]
Date   Mon, 09 Jul 2007 08:50:14 +0200

Title: Re: [Fwd: st: Kolmogorov Smirnov, groupvar as cumulative relative frequencies or rawdata?]

Hi,
Sorry about that, let me ask the question again, without the .dat:

Should groupvar in stata consist of raw data or already calculated cumulative relative frequencies..? I've done the latter, but now I see that other statistical-packages wants raw data and calculate the cumulative relative frequencies automatically.
 
I get the feeling from the help file that I have assumed correctly and that stata wants the cumulative rel. frequencies of the variable you want to test. Am I correct?
 
ksmirnov performs one- and two-sample Kolmogorov-Smirnov tests of the
    equality of distributions.  In the first syntax, varname  is the variable
    whose distribution is being tested, and exp must evaluate to the
    corresponding (theoretical) cumulative.  In the second syntax, groupvar
    must take on two distinct values.  The distribution of varname for the
    first value of groupvar is compared with that of the second value
 
Regards,
AC


On 7/7/07 11:25 PM, "Marcello Pagano" <pagano@biostat.harvard.edu> wrote:

Please, Anne-Christine,

Do not attach your winmail.dat to messages to the list.

Thanks,

m.p.

-------- Original Message --------   
 Subject:  st: Kolmogorov Smirnov, groupvar as cumulative relative frequencies or rawdata?  
 Date:  Sat, 7 Jul 2007 20:02:16 +0200  
 From:  Knag Anne-Christine <anne-christine.knag@imr.no> <mailto:anne-christine.knag@imr.no>   
 Reply-To:  statalist@hsphsun2.harvard.edu  
 To:  <statalist@hsphsun2.harvard.edu> <mailto:statalist@hsphsun2.harvard.edu>   
 References:  <002601c7c009$572c7b00$8f9e5680@a8js><E1E6452315B4DA40889EFA84D436380F045F4A@wpmailp5.rjr.com> E1E6452315B4DA40889EFA84D436380F045F4A@wpmailp5.rjr.com"><mailto:E1E6452315B4DA40889EFA84D436380F045F4A@wpmailp5.rjr.com> <001301c7c0a2$1bb0b930$8f9e5680@a8js> <fcffff0d0707070801v5f438f06p3c4d920815b67ad@mail.gmail.com> fcffff0d0707070801v5f438f06p3c4d920815b67ad@mail.gmail.com"><mailto:fcffff0d0707070801v5f438f06p3c4d920815b67ad@mail.gmail.com>   

Hi,
Should groupvar in stata consist of raw data or already calculated cumulative relative frequencies..? I've done the latter, but now I see that other statistical-packages wants raw data and calculate the cumulative relative frequencies automatically.
 
I get the feeling from the help file that I have assumed correctly and that stata wants the cumulative rel. frequencies of the variable you want to test. Am I correct?
 
ksmirnov performs one- and two-sample Kolmogorov-Smirnov tests of the
    equality of distributions.  In the first syntax, varname <https://webpost.imr.no/help.cgi?varname>  is the variable
    whose distribution is being tested, and exp <https://webpost.imr.no/help.cgi?exp>  must evaluate to the
    corresponding (theoretical) cumulative.  In the second syntax, groupvar <https://webpost.imr.no/help.cgi?varlist>
    must take on two distinct values.  The distribution of varname for the
    first value of groupvar is compared with that of the second value
 
Regards,
AC






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