Hi Mike,
the model you consider is a special case of the four-parameter logistic model
where the
lower and upper asymptotes are fixed at 0.5 and 1, respectively.
Therefore, this (dose-response) model can fitted using the R package 'drc':
library(drc)
xy.m <- drm(y~x, fct = L.4(fixed=c(NA,0.5,
Ah, perfect! Thanks so much Ken.
In the meantime I played with developing an optim() driven ML search
(included below for posterity), but the glm() + mafc approach is faster and
apparently yields identical results.
# First generate some data
# Define a modified logistic functio
Mike Lawrence thatmike.com> writes:
> Where f(x) is a logistic function, I have data that follow:
> g(x) = f(x)*.5 + .5
> How would you suggest I modify the standard glm(..., family='binomial')
> function to fit this? Here's an example of a clearly ill-advised attempt to
> simply use the standard
Mike Lawrence wrote:
On Sat, Nov 8, 2008 at 3:59 PM, Rubén Roa-Ureta <[EMAIL PROTECTED]> wrote:
...
The fit is for grouped data.
...
As illustrated in my example code, I'm not dealing with data that can be
grouped (x is a continuous random variable).
Four points:
1) I've showed y
On Sat, Nov 8, 2008 at 3:59 PM, Rubén Roa-Ureta <[EMAIL PROTECTED]> wrote:
> ...
> The fit is for grouped data.
> ...
As illustrated in my example code, I'm not dealing with data that can be
grouped (x is a continuous random variable).
--
Mike Lawrence
Graduate Student
Department of Psychology
Mike Lawrence wrote:
Hi all,
Where f(x) is a logistic function, I have data that follow:
g(x) = f(x)*.5 + .5
How would you suggest I modify the standard glm(..., family='binomial')
function to fit this? Here's an example of a clearly ill-advised attempt to
simply use the standard glm(..., famil
Hi all,
Where f(x) is a logistic function, I have data that follow:
g(x) = f(x)*.5 + .5
How would you suggest I modify the standard glm(..., family='binomial')
function to fit this? Here's an example of a clearly ill-advised attempt to
simply use the standard glm(..., family='binomial') approach:
7 matches
Mail list logo