Thanks to Kit Baum, a -stripplot- package
is now available from SSC. Stata 8 is required.
-stripplot- is an existing package renamed:
it was previously called -onewayplot-. The
.ado and .hlp are now revised in various ways.
The previous name -onewayplot- reflected
the name of the -graph, oneway- command available
in Stata for almost all of its history until Stata 8
and still available, half-hidden, as -gr7, oneway-.
In 1999, I wrote a -onewplot- program for
Stata 6, which offered functionality
overlapping with Stata's -graph, oneway-.
This is still available for users of Stata 6
or Stata 7 in the package -onewplot-. (The
name reflected the 8.3 filename.ext length constraint
that no longer bites.)
In 2003, I ported this to Stata 8 as -onewayplot-
and added several handles.
The renaming to -stripplot- reflects these facts
and guesses:
1. -stripplot- is one character and one syllable
shorter than -onewayplot- and probably easier
to remember.
2. The name "oneway plot", introduced in Stata
in 1985, is in several ways a good one, but in
statistical science as a whole I don't think
it is going anywhere. In contrast, strip plots
appear under that name in several recent books,
particularly those on or influenced by S-Plus and
especially R. My guess is that this is a name on the up.
(Most of the development of -onewayplot- was
innocent of what was being implemented
in S-Plus and R, as I don't have access to S-Plus
and only occasionally look at R.)
In contrast, the name -dotplot- seems popular in
many quarters and likely to remain so.
3. I doubt that anyone will really be confused
by the different meaning of strip plot in design
of experiments.
So much for the name. The main substantive
change in this -stripplot- is that what was the -by()-
option is now the -over()- option, and that
there is now support for a standard -by()- option.
This is in contrast to official Stata's -dotplot-,
which supports -over()- but not -by()-. (There
is no difficulty in principle here: changing
the -dotplot- code to support -by()- is a few minutes'
work, and has been done, unofficially,
but that still leaves all the hassle of
fixing the help, the dialog and the manual
that user-programmers are happy to leave to
StataCorp.)
In some ways, all this is not a big deal at all,
as after
. sysuse auto, clear
. scatter rep78 mpg, by(foreign)
produces something very like
. stripplot mpg, over(rep78) by(foreign)
But where -stripplot- comes into its own is whenever
you want the data points to be stacked, rather
like -dotplot-, except that -stripplot- offers ways
of controlling binning and/or height of displays that
I consider more intuitive than those of -dotplot-.
And also when you want several variables shown,
rather than several groups.
For those who might be interested further: the
help for -stripplot- offers a very detailed comparison
of -stripplot-, -dotplot- and -gr7, oneway-.
There are further enhancements to the code and the
help, but I'll not spell out what they are.
Separately, I'll comment on the revised version
of -beamplot-, which is a closely related program.
Nick
[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/
