Hi,
Venables and Ripley, commenting on the use of glm with binomial family
(MASS book, page 190):
"If the response is a numeric vector it is assumed to hold the data in
a ratio form, y[i] = s[i]/a[i], in which case tha a[i]s must be given as
a vector of weights using the weights argument".
So, if
All,
I wonder if glm with a quasibinomial option would work. The variance
would depend qualitatively on
the mean in a seemingly reasonable way, but would be adjusted using a
factor determined by the data.
David Farrar,
National Center for Environmental Assessment, U.S.EPA, Cincinnati
r-sig
Hi Bálint,
Here are my two cents.
By using LM with transformed data (which transformation can also be
logit, loglog, cloglog, probit) you loose the Binomial error
structure, because you won't follow the trial/success experiment
scheme. But percent cover is not that kind of [0,1] data where this
s
Bálint Czúcz wrote:
Dear List,
does anyone know a good way to perform GLM on ratio data (i.e. data
between 0 and 1)? Binomial GLM is quite straightforward to use if you
have integer numbers for successes/failures. But how to proceed if you
only have the ratio? This can occur in a multitude of wa
Dear List,
does anyone know a good way to perform GLM on ratio data (i.e. data
between 0 and 1)? Binomial GLM is quite straightforward to use if you
have integer numbers for successes/failures. But how to proceed if you
only have the ratio? This can occur in a multitude of ways, e.g the
response v
