[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: RE: graph the percent missing fairly efficiently (forget the pie)
> Thank you for the responses, which were very useful.
> Forgive me for moving to
> a variant question.
> What would be the simplest way to create a graph in
> which the x-axis is the
> name of many variables (from 10 to 100; but that is a
> visual question), and
> the y asix is the percent of the observations that are
> missing. One aspect of
> the question (as in the prior form of the question) is how
> to limit the degree
> of changing the data (recoding), although it may necessary.
> The other aspect
> is how to graph the percent missing.
Sorry, I think that way lies frustration only.
Consider the results of
. u auto
. sort make
. graph mpg in 1/50, by(make) bar
noting that with this form you can _only_
have 50 bars, but that is not the main
problem. The main problem is that the
labels are completely unreadable,
and this remains true unless you
were to truncate the labels (in your
case it would be the variable names)
to say 2 characters each, and a lot
of help that would be.
Of course, this is also an indictment
of Stata, which understandably and admirably refuses
to produce giraffe graphics in which
you have twist your neck 90 degrees
to read vertical text beneath a
This example is of a bar chart,
but I believe that you would have exactly
the same problem with any other graph
of the form you sketch.
Previous postings by Richard Goldstein
and myself indicate that there is no
need to do any recoding of the data
to tackle this question. And this is
why I recommended the form of the
graph I used earlier, although
it would still be difficult
to read with ~ 100 variables.
* For searches and help try: