Date:         Thu, 6 Jan 2011 16:23:59 -0600
Reply-To:     Bhupinder Farmaha <bhupi80singh@YAHOO.CO.IN>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Bhupinder Farmaha <bhupi80singh@YAHOO.CO.IN>
Subject:      Re: CI and SE bars
Comments: To: Peter Flom <peterflomconsulting@mindspring.com>
In-Reply-To:  <8369769.1294350451359.JavaMail.root@mswamui-valley.atl.sa.earthlink.net>
Content-Type: text/plain; charset="UTF-8"

Yes, it is line plot with X as categorical variable. For few datasets, I have 4 X categories and for other I have 5 categories. I want to plot means and want to just join them linearly.

I don't think my audience would have trouble understanding CI and SEs. But that's just a guess.

Thanks
Bhupinder

-----Original Message-----
From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Peter Flom
Sent: Thursday, January 06, 2011 3:48 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Re: CI and SE bars

Bhupinder Farmaha <bhupi80singh@YAHOO.CO.IN> wrote <<<

I want to get opinions of you guys. Why few fields prefer to put and others SEs CI around means in the profile or X-Y plots. I saw few weeks go, someone mentioned in the group that it is not good idea to put bars in the plots? Any particular reason for not putting bars?
>>>>

I am not sure what kind of plot you mean.  If "X-Y plot" means "scatterplot" then where would the error bars go?

If you mean a line plot, where the mean Y at each X is plotted, and X is categorical, then there are better plots.  If there are not too many points at each X, then you can do a strip plot; if there are a lot, you can do parallel boxplots.

Peter

Peter L. Flom, PhD
Statistical Consultant
Website: http://www DOT statisticalanalysisconsulting DOT com/ Writing; http://www.associatedcontent.com/user/582880/peter_flom.html
Twitter:   @peterflom