s-news
[Top] [All Lists]

Re: 101 statistics question

To: kamiltoth@yahoo.com, s-news@lists.biostat.wustl.edu
Subject: Re: 101 statistics question
From: Rolf Turner <rolf@math.unb.ca>
Date: Sat, 20 May 2006 20:17:39 -0300 (ADT)
Kamil Toth wrote:

> I have a simple question. Suppose I have a sequence of integers
> k[i],i=1,...,N drawn from the Poisson distribution. What is the
> distribution of random values x[i]=A*k[i], where A is a constant? Any
> hints or references are greatly appreciated. Note, I am not asking
> how to find distribution empirically, I would like to know what is
> the analytical form of distribution.

        Do not seek mystical enlightenment.

        The ``analytical'' form of the distribution is just

        P(X = Ak) = P(Y = k) (where Y ~ Poisson(lambda))
                  = exp(-lambda) * (lambda^k)/k!

        or, if you insist:

        P(X = x) = exp(-lambda) * (lambda^(x/A))/((x/A)!)  .

        That is *all* you can say, and all you need to say.  I.e.
        ``That is all ye know on earth and all ye need to know.''

        There is no magic.

                                cheers,

                                        Rolf Turner
                                        rolf@math.unb.ca

<Prev in Thread] Current Thread [Next in Thread>