Re: st: How to graph

[Roberto Ferrer <refp16@gmail.com>]

Kisil,

You're welcome.

My guess is you typed in a wrong quote mark for the local macro.  But unless you
show the exact error,  it's difficult to assess the problem. (That code was
tested.)


Regards,
Roberto

On Fri, Mar 21, 2014 at 4:50 PM, Kisil ktl <kisilktl@gmail.com> wrote:
> Roberto, I applied your codes but received an error related with nq()
> in which I couldn't understand the reason. Anyway, Carlos's solution
> worked well for my data. The graph was really annoying me, it is done
> finally. Thank you very much again for your help.
>
> Greetings
> Kisil
>
> On Fri, Mar 21, 2014 at 5:07 AM, Roberto Ferrer <refp16@gmail.com> wrote:
>> Another way:
>>
>> *------------------------ begin code ---------------------
>>
>> *clear all
>> set more off
>>
>> webuse nlswork
>>
>> keep idcode year ln_wage
>> keep if inlist(year, 68, 88)
>>
>> local q = 10 // number of quantiles
>>
>> pctile pct68 = ln_wage if year == 68, nq(`q')
>> pctile pct88 = ln_wage if year == 88, nq(`q')
>> gen diff = pct88 - pct68
>> gen xvar = _n*`=100/`q'' in 1/`=`q'-1'
>>
>> label variable pct68 "1968"
>> label variable pct88 "1988"
>> label variable xvar "percentile"
>>
>> twoway connected pct68 pct88 xvar, ytitle(Ln(wage)) msize(small small)
>> //additional graph
>> twoway connected diff xvar, ytitle(wage growth) msize(small)
>>
>> *------------------------ end code -------------------------
>>
>> On Thu, Mar 20, 2014 at 5:11 PM, Carlos Avellaneda Suárez
>> <carlos.avellaneda8@gmail.com> wrote:
>>> I am sure that more advanced users may have better solutions, but this
>>> can work for your purposes. Suppose year 68 = 2000 and year 88 = 2005:
>>> *Begin*
>>> webuse nlswork
>>> preserve
>>>  collapse (p10) p10=ln_wage (p20) p20=ln_wage ///
>>>  (p30) p30=ln_wage (p40) p40=ln_wage (p50) p50=ln_wage ///
>>>  (p60) p60=ln_wage (p70) p70=ln_wage (p80) p80=ln_wage ///
>>>  (p90) p90=ln_wage , by(year)
>>>  keep if year==68 | year==88 // You'll have to replace this line with
>>> the years you are interested in
>>>  foreach v of varlist p10-p90 {
>>>  gen growth_`v'=`v'-`v'[_n-1] if _n>1
>>>  drop `v'
>>>  }
>>>  reshape long growth_p, i(year) j(percentile)
>>>  graph bar (mean) growth_p, over(perc)
>>> restore
>>> *End*
>>> Hope this helps!
>>> Carlos
>>>
>>> 2014-03-20 16:02 GMT-05:00 Kisil ktl <kisilktl@gmail.com>:
>>>> Thank you for your reply Carlos. Sorry for being late to respond,
>>>> since I don't have stata on my own computer.
>>>> I applied what you suggested, however this is not the case I actually
>>>> want to graph .
>>>> I want to have percentiles (10th, 20th,..,90th) ordered on the x-axis
>>>> and difference of the log wage in 2000 from the log wage in 2005 for
>>>> the respective percentile on the y-axis, I'm not interested in the
>>>> years between 2000 and 2005.
>>>>
>>>> Greetings
>>>> Kisil
>>>>
>>>>
>>>> On Tue, Mar 18, 2014 at 9:44 PM, Carlos Avellaneda Suárez
>>>> <carlos.avellaneda8@gmail.com> wrote:
>>>>> Hi Kisil,
>>>>> One solution is to collapse your dataset. Here is an example you can
>>>>> reproduce in a do-file:
>>>>> *****Begin Example*****
>>>>> webuse nlswork
>>>>> preserve
>>>>> collapse (p10) p10=ln_wage (p20) p20=ln_wage ///
>>>>> (p30) p30=ln_wage (p40) p40=ln_wage (p50) p50=ln_wage ///
>>>>> (p60) p60=ln_wage (p70) p70=ln_wage (p80) p80=ln_wage ///
>>>>> (p90) p90=ln_wage , by(year)
>>>>> tsset year
>>>>> tsline p*
>>>>> restore
>>>>> *****End*****
>>>>> By the way, why would you estimate a percent change in a log variable?
>>>>> Remember that the first difference of the logarithm of a variable
>>>>> approximates to the percentage change of the unaltered variable. With
>>>>> the previous example, you just have to generate new variables of the
>>>>> first difference of each percentile.
>>>>> Hope this helps.
>>>>> Carlos
>>>>>
>>>>> 2014-03-18 13:41 GMT-05:00 Kisil ktl <kisilktl@gmail.com>:
>>>>>> Dear statalist,
>>>>>>
>>>>>> I work on a data set that has log wage information of workers through six years.
>>>>>> In order to show how log wage at different percentiles changed between
>>>>>> the years 2000 and 2005, I want to make a graph that has
>>>>>> 10th,20th,..,90th percentiles on the x-axis and corresponding
>>>>>> percentage log wage changes on the y-axis.
>>>>>> For example, the graph should show how many percent did the log wage
>>>>>> of a worker at 90th or 10th percentile of the wage distribution change
>>>>>> between 2000 and 2005.
>>>>>> egen rank = rank(wage)
>>>>>> This command computes the percentile ranks of workers with respect to
>>>>>> their log wages, but I could not do the rest of it. How can I make
>>>>>> this kind of a graph?
>>>>>>
>>>>>> Best regards
>>>>>> Kisil
>>>>>> *
>>>>>> *   For searches and help try:
>>>>>> *   http://www.stata.com/help.cgi?search
>>>>>> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/