Hi Stat. experts,
I have a query on the formulation of lognormal errors for a simulation study.
Suppose a lognormal observational error, E, is introduced to a deterministic
model in the following way:
Nobs = N predicted *E
I read in a book that to make log(E) normally distributed with a mean 0 and
a standard deviation sigma, one should set E = exp(Z*sigma-(sigma^2)/2)
where Z is normally distributed with a mean 0 and a standard deviation 1.
Is this formula correct? If so, can someone enlighten me on the derivation of
this formula? Many thanks. Siddeek
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
