Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

RE: st: Quintiles

From   Leonardo Jaime Gonzalez Allende <>
To   "" <>
Subject   RE: st: Quintiles
Date   Thu, 9 Aug 2012 14:49:57 -0400

Thank you Austion,
It is useful to shorten my programming (I didn't find any command to do that directly).  

I had in mind pweight.

The point raised for you was exactly about what we were concerned. If we cut the weight factor of the sample to obtaining 5 groups of exactly same size, and as you correctly says, "respondents tend to round their answers", we face the problem of having households with the same income in different quintiles. 

Using xtitle command we avoid the situation of having people with the same income in the same quintil, but at the same time we have to accept slights differences in the number of population of each quintil.

Thank you for your comments. 


-----Mensaje original-----
De: [] En nombre de Austin Nichols
Enviado el: Jueves, 09 de Agosto de 2012 14:00
Asunto: Re: st: Quintiles

Leonardo Jaime Gonzalez Allende <>:
Try this example out:

sysuse auto, clear
keep price weight
ren price y
sort y
g s=sum(weight)
g c=(s)/(s[_N])
g qbreak=0
g w2=.
qui foreach l in 20 40 60 80 {
 replace qbreak=1 if c>=`l'/100 & c[_n-1]<`l'/100
 replace w2=s-(`l'/100*s[_N]) if c>=`l'/100 & c[_n-1]<`l'/100
g w1=weight-w2
g id=_n
expand 2 if w2<.
g wt=weight
bys id: replace wt=w1 if _n==1 & w1<.
by id: replace wt=w2 if _n==2
g sum=sum(wt)
g cum=sum/sum[_N]
l if qb==1

On Wed, Aug 8, 2012 at 3:44 PM, Leonardo Jaime Gonzalez Allende
<> wrote:
> Yes, Maarten, you are right, sorry for writing you directly.
> I don't was planning to cut a person or household in many parts. The question was about a possible adjustment to the weight factor, if the observation of the sample is the cut point of the quintile.
> If I sort the households of a sample by their incomes, a household "x" could represents 300 households but the accumulated frequency of the population is e.g. 20,02%.
> My question was if there is an efficient way (command) to repeat the observation and adjust weight factor as follow:
> the same household "xa" now represents 280 households and now the accumulated frequency of the population is e.g. 20% (exactly) (leaving to the first quintile).
> the same household "xb" (the other part) represents 20 households accumulating a frequency of 20,02% of the population (changing to the second quintile this part).
> Regards,
> Leonardo
> -----Mensaje original-----
> De: Maarten Buis []
> Enviado el: Miércoles, 08 de Agosto de 2012 12:41
> Para:
> Asunto: Re: st: Quintiles
> --- Leonardo Jaime Gonzalez Allende asked:
>>>> I'm trying to divide a sample of households (expanded)
>>>> into quintiles using the xtile command. I want to create
>>>> 5 groups with the exactly same quantity of population,
>>>> but using the xtitle command, the quantity of households
>>>> in each quintil is very slightly different to 20% when the
>>>> number of observations isn't exactly dividable by 5.
>>>> Do you know any command to divide the population
>>>> (sample expanded) into 5 groups of exactly same weight?
> --- I answered:
>>> That is logically impossible.
> -- Leonardo Jaime Gonzalez Allende wrote me privately:
>> sorry for write you directly, but I like to know, why is
>> logically impossible separate the population (by
>> incomes) in 5 groups of the same weith?
> Don't sent such follow-up questions privately. If you find my answer
> puzzling, than chances are that someone else who is following this
> discussion finds that too. This is explained in the Statalist FAQ.
> Think of it this way: How can you divide 6 persons in 5 equally sized
> groups? You could assign one person to each group, and than you are
> left with one person. If you could split the remaining 1 person up
> into 5 1/5th persons, than we could create 5 equally sized groups.
> However, that is impossible (or rather bloody, if we take that too
> literally). So given the inherently discrete nature of the number of
> observations you cannot divide your data up into 5 groups of exactly
> the same size if the number of observations in not dividable by 5.
> -- Maarten

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index