|
On Tue, 18 Apr 2006 19:02:39 -0300, Rogerio Porto <rdporto1@TERRA.COM.BR>
wrote:
>Peter,
>
>it seems you are biasing your data at the following truncation step:
>
>data draws&J.;
>set draws&J.;
>if HDD < 0 then HDD = 0;
>if CDD < 0 then CDD = 0;
>if P_TTL < 0 then P_TTL = 0;
>if P_STD < 0 then P_STD = 0;
>run;
>
>HTH,
>
>Rogerio.
Hi Again-
A couple of additional insights about my problem. First, I am assuming
for the simulation that all distributions are normal (and this is
correct). Second, I am aware that truncating the variables at zero is a
slight bias, though, I'm not sure if this is the cause of my main source
of bias. Aside from the truncation method I used in the code, is there
another way to ensure that no negative values get drawn? (i.e. it doesn't
make sense to have negative temperature or rainfall).
Any other thoughts? OR... some example code to ensure that no negative
values are getting drawn?
Thanks.
PHL
| 
