[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Error-bars---again

From   Maarten buis <>
To   stata list <>
Subject   RE: st: Error-bars---again
Date   Tue, 23 Jun 2009 17:16:27 +0000 (GMT)

--- moleps islon wrote:
> Looking at the number of postings requesting an option of adding
> error-bars with ci,sd or se to bar graphs I'm sure one exists but I
> still havent found an ado handling this. Is there one?
> Or is there a way to do ci(b)plot over multiple categories?
> ie. something akin to graph bar x, over(z) over(y)

I am not a big fan of these "detonator plots" as it encourages the
practice of "looking whether confidence intervals overlap" to test
whether means are equal. Also I don't like the representation of
means by a bar (it requires a meaningful zero for it to make sense,
and that is often lacking). Also, there are a lot better ways of 
representing the distributions of a variable across categories, 
like Nick Cox's -stripplot- (type -findit stripplot- to find it).

So my real answer is: "Don't do that". However, if you really 
really want to, you can use Roger Newson's -eclpot-. Notice that
this command is much more versatile and can create lots of 
really useful graphs besides detonator plots. It is discussed in:

Roger Newson (2003) Confidence intervals and p–values for 
delivery to the end user. The Stata Journal, 3(3):245--269.

Hope this helps,

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen


*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index