Thats great thanks very much for your help
On 29 Sep 2010, at 17:30, Ben Bolker wrote:
[I'm a little confused: are you "Sam Smith" or "Chris Mcowen" ... ?]
This is admittedly a bit confusing, but the best scale on which to
compute standard errors is the link scale.
It turns out (I hadn't r
[I'm a little confused: are you "Sam Smith" or "Chris Mcowen" ... ?]
This is admittedly a bit confusing, but the best scale on which to
compute standard errors is the link scale.
It turns out (I hadn't realized this) that predict.glm does give
you not-crazy answers when you ask for
se.fit=T
Dear List and Ben
( I apologise if this has been sent twice, but it is not showing in my sent
folder and i have been having trouble with my email of late)
Right, that makes sense, thanks
The reason i used type= response was i wanted to convert the predicted
probabilities to the response scale,
Right, that makes sense, thanks
The reason i used type= response was i wanted to convert the predicted
probabilities to the response scale, as surely this is the scale at which a
95CI value is most useful for?
I.e
>> pp <- predict(model1,se.fit=TRUE, type = "response")
1 0.68
Probability
On 10-09-29 10:04 AM, Sam wrote:
> Hi Ben and list,
>
> Sorry to be a pain! I have followed your code, and modified it -
>
**You should not use type="response" here.**
The point is that the (symmetric) confidence intervals are computed on
the link/linear predictor
scale, and then inverse-link-
Hi Ben and list,
Sorry to be a pain! I have followed your code, and modified it -
> pp <- predict(model1,se.fit=TRUE, type = "response")
>> etaframe <-
> + with(pp,cbind(fit,lower=fit-1.96*se.fit,upper=fit+1.96*se.fit))
>> pframe <- plogis(etaframe)
>> pframe
My response variable is 0 or 1, the
Hi Ben and list,
Sorry to be a pain! I have followed your code, and modified it -
> pp <- predict(model1,se.fit=TRUE, type = "response")
>> etaframe <-
> + with(pp,cbind(fit,lower=fit-1.96*se.fit,upper=fit+1.96*se.fit))
>> pframe <- plogis(etaframe)
>> pframe
My response variable is 0 or 1, the
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
## from ?glm
counts <- c(18,17,15,20,10,20,25,13,12)
outcome <- gl(3,1,9)
treatment <- gl(3,3)
d.AD <- data.frame(treatment, outcome, counts)
glm.D93 <- glm(counts ~ outcome + treatment, family=poisson,
data=d.AD)
## predict on 'link'
I am looking to do the same but am a bit confused
> and apply the inverse link function for your model.
i understand up to this point and i understand what this means, however i am
unsure why it needs to be done and how you do it - i.e i use family="binomial"
is this wrong if i use this method
zozio32 gmail.com> writes:
>
>
> Hi
>
> I had to use a glm instead of my basic lm on some data due to unconstant
> variance.
>
> now, when I plot the model over the data, how can I easily get the 95%
> confidence interval that sormally coming from:
>
> > yv <- predict(modelVar,list
Hi
I had to use a glm instead of my basic lm on some data due to unconstant
variance.
now, when I plot the model over the data, how can I easily get the 95%
confidence interval that sormally coming from:
> yv <- predict(modelVar,list(aveLength=xv),int="c")
> matlines(x
11 matches
Mail list logo
