RE: st: RE: pattern-fills in stacked bar graph

["Nick Cox" <n.j.cox@durham.ac.uk>]

Sorry, but that's not my point. -graph bar- by default plots means, so you are stacking means on top of each other. The question is whether that matches the science of what you are doing. 

In essence, I see a case logically for stacking whatever can be combined additively, so sums and frequencies are cases in point. That could apply to means too, but it seems less common. As you're new to Stata, I wanted to check that you understood that -graph bar- does that by default. I know a bit about Stata but I don't know what income gaps are precisely, so that's what you have to spell out. 

Nick 
n.j.cox@durham.ac.uk 

Natalie Doss

Yes, the variables should be called mean_gap_ed, etc..I used shortened
names to simplify.

On Wed, May 26, 2010 at 12:05 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:

> So, you are stacking _means_ ? Actually, that's not a question.
> You are stacking means. Is that what you want to do?

Natalie Doss

> Thank you. Here is an example similar to my data, but simpler:
>
> x-variable: gender, taking values 0 and 1.
> y-variables: income gap, divided into 3 components:
> gap_ed,  gap_exp,  gap_unknown
>
> I am using Stata 10.0. Without trying to use pattern fills, this is
> very easy to do using the "bar graph" command. As in:
>
> graph bar gap_ed gap_exp gap_unknown, over(gender) stack
>
> On Wed, May 26, 2010 at 2:56 AM, Martin Weiss <martin.weiss1@gmx.de>
> wrote:
>
>> It is difficult to advise you on this, as we do not have your data.
> Maybe
>> give an excerpt. Also, you should say that -tw parea- is user-written
> by
>> Sergiy.
>
> Natalie Doss
>
>> I am trying to do a stacked bar graph with different pattern fills. I
> have 4
>> outcomes and 7 components to each outcome. I am confused about how to
> use
>> twoway
>> parea to make a bar graph. I believe I have to reorganize my data a
> fair
>> amount, but am unsure of what exactly I need. An earlier thread on
> this said
>> that the collapse and contract commands would be most useful, but I'm
> pretty
>> new to stata and am wondering if anyone has more specific ideas.
>

*
*   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/