Date: Wed, 10 Dec 1997 11:17:13 -0800
Reply-To: Martin J Hillyer <hillyer@JUNO.COM>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: Martin J Hillyer <hillyer@JUNO.COM>
Subject: Re: Probability scale on SAS graph
I've been nomail for a while, so I don't know whether this solution has
been proposed by someone else, but just in case... You can put
probability scales on graphs without needing the QC module, as long as
you have GRAPH module (perhaps more available than QC).
Simply put ticks on the probability axis at (say) +/-1.96, etc (i.e., +/-
various standardized deviates at convenient probability levels) using
order= in an axis statement, and label those tickmarks using value=.
Then plot the (sorted) data against the appropriate plotting position,
i.e., probit(i/(n+1)) - the quantile corresponding to each ordered
observation. You can use other formulas for the deviates - (i-0.5)/n,
(i-0.375)/(n+0.5),... - whatever you like.
Most elementary statistics texts have sections on probability plotting,
and what you are doing is the same, putting whatever ticks you like using
order= and value= on your axis statement.
Here's a little demo program that shows how:
do i=1 to 100;
proc sort data=xdata; by x;
data xdata1; set xdata;
prob= _n_/101 ;
/*find the number of points in the data set by some means*/
/*this plot uses 1/(n+1) to determine the plotting position*/
proc gplot data=xdata1;
value=("1" "5" "10" "25" "50" "75" "90" "95" "99")
(also email@example.com for large messages (>64K)
or messages with attached files)
On Tue, 9 Dec 1997 16:01:54 -0500 Digital Equipment Corp
> Hello Bruce, I'm not completely certain about this, but I think
> that Proc Capability (if you have the SQC module) will produce
> these graphs with probability scale. If you are just using
> like Proc Gplot, then you probably can still do this using the
> axis statement.
> Regards, Randy
>Bruce S Libman wrote:
>> I am trying to make a SAS graph where one of the axes is in
>> scale. Anyone know how to do this?
>> Bruce Libman
>> Department of Biology
>> University of Mississippi
>> University, MS 38677