# RE: st: creating a difficult graph (not sure what to call it)

 From "Nick Cox" To Subject RE: st: creating a difficult graph (not sure what to call it) Date Fri, 29 Nov 2002 09:24:33 -0000

```Ulrich Kohler
>
> > I am hoping to create a graph which is an X by Y matrix.
> Each cell is
> > shaded according to the percentage of observations in
> that cell (in some
> > sense this is a table in graphical form).  Does anyone
> have any ideas how
> > to approach this in stata?  Maybe an example will help:
> >
> > X is industries, Y is occupations:  the graph would look
> like this where
> > letters represent shadings based on the % in each cell
> >
> > 	   1	    2	    3            4
> > 	---------------------------------
> >    1	|  lllll	|  lllll	|  mmm	| mmm	|
> > 	---------------------------------
> >    2	| dddd	|  ddd	|  lllll	| mmm	|
> > 	---------------------------------
> >    3	| llllll	| mmm	| llll	| ddd	|
> > 	---------------------------------
> >    4	| dddd	|  mmm	| dddd	| llllll	|
> > 	---------------------------------
> >
> > Or is the easier to do in excel?
>
> What is excel? As far as I understand this, you have two
> variables (x and y)
> with just a few categories. If so, you might try heavy
> jittering. Something
> like
>
> . graph y x, yline(0.5(1)4.5) xline(0.5(1)4.5) xlab(1(1)4)
> ylab(1(1)4) /*
> */ jitter(20) xscale(0,5) yscale(0,5)
>
> (Fiddle around with the number in jitter. It can be between
> up to 30. -help
> grtwoway- states that jitter(5) is already large)

Like Ulrich, I cannot advise on excel,
or even MS Excel.

I played with a similar problem a while
back; the result was -tabplot-, which is
on SSC. It yields a series of bar charts
in table form. It doesn't implement
shadings; I'm not in favour of using