[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Re: RE: Error bars on bar plots
At a risk of inflaming the situation and being branded a Luddite, I'll add
On Fri, 27 Jun 2003, george hoffman wrote:
> 1. report means and sd's of groups and means comparison s(anova with
> 2. graph the means and sd's with bar (or other symbols) and error
> 3. options to graph se's or ci's.
> -currently using joanne garrett's predxcat.ado with modifications to
> do the above, but bars are not an easy option.
Request 1 seems to ask for a command that incorporates the analysis with
drawing the graphic. It has always seemed to me a strength of Stata that
it distinguished the operations: eg, you did fit / predict / graph to draw
a scatterplot with regression line. Recent and current trends seem to be
reversing this policy, which is fine for those who need to do repeated
standard operations, but dangerous whenever *real* data are analysed.
Hence I would rather stick to and articulate the old policy, and point to
the portmanteau procedures as conveniences.
Hence request 3 should be interpreted as providing an option for hi-lo
lines, which in specific cases may be SEs or CIs. That also reminds the
user to *note in the caption* which they are.
R. Allan Reese Email: email@example.com
* For searches and help try: