24 Feb
2003
24 Feb
'03
8:01 p.m.
On the beta binomial distribution, are there boundaries for
gamma and mu in the alternative parametization? For example, I
plugged in something greater than 1 for gamma, and I got a
negative beta...and NAs for my beta distribution of pies. 8(
Thanks, Olivia.
24 Feb
24 Feb
8:56 p.m.
Hey all,
One question based on Olivia's email: is it OK if we use alpha and beta as
our parameterization for the beta-binomial? What advantage do we get by
using mu and gamma?
Best,
Dan
On Mon, 24 Feb 2003, Olivia Lau wrote:
On the beta binomial distribution, are there
boundaries for
gamma and mu in the alternative parametization? For example, I
plugged in something greater than 1 for gamma, and I got a
negative beta...and NAs for my beta distribution of pies. 8(
Thanks, Olivia.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
10:07 p.m.
If you can, use the alternative parameterization which is directly related
to mean and variance and hence is relatively easier to interpret. If
you are not sure, just go ahead with the standard parameterization.
Note that you need to use rbeta() for this question. Bounds on mu and
gamma are implied by bounds on alpha and beta.
Kosuke
On Mon, 24 Feb 2003, Daniel Jacob Hopkins wrote:
Hey all,
One question based on Olivia's email: is it OK if we use alpha and beta as
our parameterization for the beta-binomial? What advantage do we get by
using mu and gamma?
Best,
Dan
On Mon, 24 Feb 2003, Olivia Lau wrote:
On the beta binomial distribution, are there
boundaries for
gamma and mu in the alternative parametization? For example, I
plugged in something greater than 1 for gamma, and I got a
negative beta...and NAs for my beta distribution of pies. 8(
Thanks, Olivia.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
10:59 p.m.
New subject: [gov2001-l] negative binomial & co
A few questions with respect to problem 4...
1) When is meant by "write down the joint density..." - should we just
copy the formulation of Negative Binomial distribution on page 53?
2) Should the R function compute the probability of one y_i value, or
create a matrix of probabilities over a range of y, or plot them?
3) Where can we learn more about "numerical integration via the grid
method"?
4) Why is Negative Binomial distribution called "density" in question 4
if it's discrete?
hope everybody is enjoying the evening ;)
s t a n
****************************
Stanislav Markus
Ph.D. Candidate
Harvard University
Department of Government
e: smarkus(a)fas.harvard.edu
t: 617.513.5407
-----Original Message-----
From: gov2001-l-admin(a)fas.harvard.edu
[mailto:gov2001-l-admin@fas.harvard.edu] On Behalf Of Daniel Jacob
Hopkins
Sent: Monday, February 24, 2003 8:56 PM
To: Olivia Lau
Cc: gov2001-l(a)fas.harvard.edu
Subject: Re: [gov2001-l] boundaries on gamma and mu for beta-binomial
distribution?
Hey all,
One question based on Olivia's email: is it OK if we use alpha and beta
as
our parameterization for the beta-binomial? What advantage do we get by
using mu and gamma?
Best,
Dan
On Mon, 24 Feb 2003, Olivia Lau wrote:
On the beta binomial distribution, are there
boundaries for
gamma and mu in the alternative parametization? For example, I
plugged in something greater than 1 for gamma, and I got a
negative beta...and NAs for my beta distribution of pies. 8(
Thanks, Olivia.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
11:23 p.m.
New subject: [gov2001-l] negative binomial & co
1) When is meant by "write down the joint
density..." - should we just
copy the formulation of Negative Binomial distribution on page 53?
I'm asking you to write down the joint density of y and lambda (before
integrating out lambda), not the marginal density of y.
2) Should the R function compute the probability of
one y_i value, or
create a matrix of probabilities over a range of y, or plot them?
You should compute the probabilities over a range of y.
3) Where can we learn more about "numerical
integration via the grid
method"?
If you understand the basic concept, you are fine. don't worry about it.
4) Why is Negative Binomial distribution called
"density" in question 4
if it's discrete?
It should be "mass function", but often "density" is used for both
discrete and
continuous distributions.
hope everybody is enjoying the evening ;)
I'm going to bed!
Kosuke
7749
days inactive
7749
days old
gov2001@lists.gking.harvard.edu
4 comments
4 participants
participants (4)
-
dhopkins＠fas.harvard.edu -
kimai＠fas.harvard.edu -
olau＠fas.harvard.edu -
smarkus＠fas.harvard.edu