Thanks very much -- & a clever approach too, I would say.
Eric
>Eric G. Wruck <[email protected]> asks about creating a stacked bar chart
>where one of the bars stacks three variables and another bar stacks two other
>variables. He then wants these two bars graphed for nine years.
>
>> I have five different variables (call them a, b, c, d, & e) for nine
>> years (i.e., 9 obs). I would like to construct a bar chart in the
>> following way:
>>
>> I would like to stack a, b, & c in a single bar, and d & e in
>> another & graph them both -over(year)-
>>
>> I have been able to graph d & e stacked over(year) & I've been able
>> to graph (a + b + c) and (d + e) over(year) but I can't figure out
>> how to get both in a single graph.
>
>
>-graph bar- is flexible, but not nearly so flexible as all of Stata. What is
>required here is a little good old-fashioned data manipulation.
>
>What Eric can do is duplicate each of his observations and set his 3 variables
>(a, b, and c) to 0 in the original observations and his 2 variables (d and e)
>to 0 in the duplicate observations. Then, all he need do is create a bar
>graph stacking all 5 variables -- half the bars will stack variables a, b, and
>c, the other two variables being 0, and the other half of the bars will stack
>variables d and e.
>
>Let's use the Klein data from the [R] reg3 entry, mainly because it has short
>names. Eric's a, b, and c will be w, p, and t; and his d, and e will be y and
>c. (I am too lazy to rename the variables.) We keep the first 9 observations
>to match Eric's data.
>
> . webuse kleinabr
> . keep in 1/9
>
>Now, expand the data to duplicate each observation and create a variable to
>label the original and duplicate observations,
>
> . expand 2
> . gen barname = cond(_n <= 9 , "my 2 var" , "my 3 vars")
>
>Replace the set of three variables to stack with 0 in half the observations.
>
> . replace w = 0 in 1/9
> . replace p = 0 in 1/9
> . replace t = 0 in 1/9
>
>Replace the set of two variables to stack with 0 in the other half of the
>observations.
>
> . replace y = 0 in 10/l
> . replace c = 0 in 10/l
>
>Graph the 5 variables over the bar identifier variable (barname) and year.
>(We will use horizontal bars, -hbar-, because they look better with so many
>labels).
>
> . gr hbar w p t y c , stack over(barname) over(year)
>
>The graph described above is prettier if we place all the 2-variable bars
>together and all the three variable bars together, though I think Eric wanted
>the first graph.
>
> . gr hbar w p t y c , stack over(year) over(barname)
>
>Finally, adorn graph to taste.
>
>
>-- Vince
> [email protected]
>
>*
>* For searches and help try:
>* http://www.stata.com/support/faqs/res/findit.html
>* http://www.stata.com/support/statalist/faq
>* http://www.ats.ucla.edu/stat/stata/
--
===================================================
Eric G. Wruck
Econalytics
2535 Sherwood Road
Columbus, OH 43209
ph: 614.231.5034
fax: 614.231.5039
eFax: 419.730.7809
eMail: [email protected]
website: http://www.econalytics.com
====================================================
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
