Re: st: RE: calculating proportions

[Nick Cox <njcoxstata@gmail.com>]

I don't see a precise question here. The fraction earning 500 exactly
has no direct relation to the associated relative rank.  I recommend
that you use a simpler example and draw graphs to see what is going
on.

Nick

On Mon, Feb 20, 2012 at 11:29 PM, Ekaterina Hertog
<ekaterina.hertog@sociology.ox.ac.uk> wrote:
> Hi Nick,
> thank you for the advice. i looked at all the 3 packages you recommended and
> thought that relrank could do what I want, but having run it I realise that
> I am not sure how rank was calculated.
> I used the following code:
> relrank income if sex==1, ref(income if sex==2) g(rank)
>
> My dataset contains 4132 women. 268 of them have income=0. That means that
> in case of a man whose income is 0(zero) 6.48% of women earn as little as he
> does. My new variable rank takes the value 0.0648 in this case, seemingly
> corresponding to the 6.48% of women whose income equals 0. However, if I
> look at different income level (say 500). 6% of all women earn 500, but the
> value of rank in this case is 0.93. I have looked at the help file, but am
> still no quite clear what is going on.
> Ekaterina
>
>
> On 20/02/2012 12:19, Nick Cox wrote:
>>
>> Look at these modules on SSC:
>>
>> relrank (Ben Jann)
>> invcdf (Ben Jann)
>> ppplot (Nick Cox)
>>
>> Nick
>> n.j.cox@durham.ac.uk
>>
>> Ekaterina Hertog
>>
>> I use Stata 12. I have a dataset which includes id number, sex and
>> income information for every individual. I would like to construct for
>> every man 3 variables: what is the proportion of the total female
>> population that earns as much as he does, that earns less than he does,
>> and that earns more than he does. I have tried creating a table which
>> gives me frequencies of individuals by sex by income, but am getting
>> stuck at that stage.
*
*   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/