Date:         Thu, 15 Oct 2009 09:58:50 +0200
Reply-To:     Anders Mørup Jensen <amo@SSI.DK>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Anders Mørup Jensen <amo@SSI.DK>
Organization: TDC Totalloesninger
Subject:      Re: how to simulate clinical trial data which has lognormal
Comments: To: sas-l@uga.edu

As I understand the OP, he wanted the lognormal sample to have Mean=2 and
Std=1 (not the underlying normal distribution).

With a litle algebra applied to wellknown formulas for the mean and variance
of a lognormal distribution (see e.g. Wikipedia) this means that the
underlying normal distribution should have:

Mean value = 2*log(2) - log(5)/2
Variance  = log(5/4);

resulting in the need1 dataset below.

data need;    /* ynorm has mean 2 and standard deviation 1  */

 do i=1 to 200;
 ynorm=rannor(0)+2;
 ylognorm=exp(ynorm);
 output;
 end;
run;

proc means data=need;
     var   ylognorm;
run;

data need1;   /* ylognorm has mean 2 and standard deviation 1  */

 do i=1 to 20000;
 ynorm=sqrt(log(5/4))*rannor(0)+(2*log(2)-log(5)/2);
 ylognorm=exp(ynorm);
 output;
 end;
run;

proc means data=need1;
     var   ylognorm;
run;

/ Anders




"Ya Huang" <ya.huang@AMYLIN.COM> skrev i en meddelelse
news:200910142008.n9EF4ebt011509@malibu.cc.uga.edu...
> /* If X is a random variable with a normal distribution,
>   then Y = exp(X) has a log-normal distribution;
>   likewise, if Y is log-normally distributed,
>   then log(Y) is normally distributed. */
>
> data need;
> do i=1 to 200;
> ynorm=rannor(0)+2;
> ylognorm=exp(ynorm);
> output;
> end;
> run;
>
> ods listing close;
> ods pdf file="c:\temp\junk.pdf";
>
> goptions reset=all device=sasprtc;
>
> proc gchart data=need;
> vbar ylognorm;
> vbar ynorm;
> run;
>
> ods _all_ close;
>
>
> On Wed, 14 Oct 2009 11:09:48 -0700, jingtailan@gmail.com
> <jingtailan@GMAIL.COM> wrote:
>
>>all:
>>
>>please advise on
>>1. how to use SAS8.2 to create a data set which has
>>N=200
>>mean=2
>>std=1
>>distribution=lognormal
>>
>>2 After creation of these data, what procedure I should use to get PDF
>>graph?
>>
>>thanks a lot.
>>jingtailan