similar to: glm.nb difficulties

Hi there, I have been having trouble running negative binomial regression (glm.nb) using library MASS in R v2.15.0 on Mac OSX. I am running multiple models on the variables influencing the group size of damselfish in coral reefs (count data). For total group size and two of my species, glm.nb is working great to deal with overdispersion in my count data. For two of my species, I am getting a

Hello, I have a question regarding estimation of the dispersion parameter (theta) for generalized linear models with the negative binomial error structure. As I understand, there are two main methods to fit glm's using the nb error structure in R: glm.nb() or glm() with the negative.binomial(theta) family. Both functions are implemented through the MASS library. Fitting the model using these

Dear R users, I am having problems trying to fit quasipoisson and negative binomials glm. My data set contains abundance (counts) of a species under different management regimens. First, I tried to fit a poisson glm: > summary(model.p<-glm(abund~mgmtcat,poisson)) Call: glm(formula = abund ~ mgmtcat, family = poisson) . . . (Dispersion parameter

I am running 9 negative binomial regressions with count data. The nine models use 9 different dependent variables - items of a clinical screening instrument - and use the same set of 5 predictors. Goal is to find out whether these predictors have differential effects on the items. Due to various reasons, one being that I want to avoid overfitting models, I need to employ identical types of

Full_Name: Ben Cooper Version: 1.5.0 OS: linux Submission from: (NULL) (134.174.187.90) The help page for glm.nb (in MASS package) says that it takes "Any other arguments for the glm() function except family" One such argument is start "starting values for the parameters in the linear predictor." However, when called with starting values glm.nb returns: Error in

I am trying to find a way to analyze a "simple" mixed model with two levels of a treatment, a random blocking factor, and (wait for it) negative binomial count distributions as the response variable. As far as I can tell, the currently available R offerings (glmmGibbs, glmmPQL in MASS, and Jim Lindsey's glmm code) aren't quite up to this. From what I have read (e.g.

Dear R users, I am trying to fully understand the difference between estimating overdispersion with glm.nb() from MASS compared to glm(..., family = quasipoisson). It seems that (i) the coefficient estimates are different and also (ii) the summary() method for glm.nb suggests that overdispersion is taken to be one: "Dispersion parameter for Negative Binomial(0.9695) family taken to be

In the optim() documentation, the control parameter "maxit" says "There is no other stopping criteria." That should be "are", or "criterion" ... While I'm at it -- poking at the code a little bit, it looks as if the initial simplex is set from the initial point by displacing each parameter value by max(0.1,0.1*pmax(fabs(Bvec))), where Bvec is the

Hi, I make an analysis sequence on R. In some cases the function glm.nb fail to ajust the model. Its Ok. The problem is that this error stop the program. I need treat this error and not stop the program. Something like this: ... model <- glm.nb(y~x,maxit=1000) if(glm.nb fail) { teste[i] <- 0 } else { teste[i] <- anova(modelo)$"P(>|Chi|)"[2] } ... I try this: ...

Sorry about re-posting this, it never went out to the mailing list when I posted this to r-help forum on Nabble and was pending for a few days, now that I am subscribe to the mailing list I hope that this goes out: I've been a viewer of this forum for a while and it has helped out a lot, but this is my first time posting something. I am running glm models for richness and abundances. For

Dear All, I've been using already for a year glm.nb() from the MASS package. But today, R gave me an error message when estimating one of my usual models: > depEsf.nb <- glm.nb(depE ~ manuf00E + corps00E + lngdp00E + lngdp00sqE + > lnpop00E + indshE + scishE + mechshE + elecshE + chemshE + drugshE + > urban_dummyE + aggl_dummyE + + eE1 + eE2 + eE3 + eE4 + eE5 + eE6 + eE7 +

When using test="F" in stat.anova() / anova.glm(), R uses the assumed dispersion parameter for the specified family (e.g. scale=1 for binomial), while S-PLUS automatically uses the estimated dispersion parameter (residual deviance/residual df). I think there are good reasons for the behavior in R -- it fits with the "you get what you actually asked for" philosophy -- and

I am using glm.nb, a ~b*c ( b is categorical and c is continuous). when I run this model I get the warning message: Warning messages: 1: In theta.ml(Y, mu, sum(w), w, limit = control$maxit, trace = control$trace > : iteration limit reached 2: In theta.ml(Y, mu, sum(w), w, limit = control$maxit, trace = control$trace > : iteration limit reached What does this mean? -- Graduate

Hi, I make some simulations with rnbinom and try to test with glm.nb. But in some data set the glm.nb fail. Look: pop <- rnbinom(n=1000,size=1,mu=0.05) > glm.nb(pop~1,maxit=1000) Error in while ((it <- it + 1) < limit && abs(del) > eps) { : missing value where TRUE/FALSE needed look some pop charactetistics: > summary(pop) Min. 1st Qu. Median Mean 3rd Qu.

optim ignores misspelled control parameters, so that trying to set (e.g.) "maxint=1000" in the control argument silently does nothing. The patch below (watch out for line breaks! also posted at http://www.zoo.ufl.edu/bolker/optim_patch.R , and http://www.zoo.ufl.edu/bolker/optim_new.R) adds three lines to optim.R that issue a warning if any names of elements of "control" fail

Hi, I have count data that have been meddled with enough to make them non integers. Using glm(poisson) returns a "non integer" error but glm(quasipoisson) does not. Just wondering if anyone knows if I am violating the assumptions of a quasipoisson error structure by using these non-integer response data? Thanks! I'd welcome your thoughts and/or references... Mark

Hi Folks, I'm playing with glm.nb() in MASS. Reference: the negative binomial distribution P(y) = (Gamma(theta+y)/(Gamma(theta)*y!))*(p^theta)*(1-p)^y y = 0,1,2,... in the notation of the MASS book (section 7.4), where p = theta/(mu + theta) so (1-p) = mu/(mu + theta) where mu is the expected value of Y. It seems from ?glm.nb that an initial value of theta is either supplied, or

Dear all, I am fitting a glm to count data using poison errors with the log link. My goal is to test for the significance of model terms by calling the anova function on two nested models following the recommendation in Michael Crawley's guide to Statistical Computing. Without going into too much detail, essentially, I have a small overdispersion problem (errors do not fit the poisson

Dear all, I have a quasipoisson glm for which I need confidence bands in a graphic: gm6 <- glm(num_leaves ~ b_dist_min_new, family = quasipoisson, data = beva) summary(gm6) library('VIM') b_dist_min_new <- as.numeric(prepare(beva$dist_min, scaling="classical", transformation="logarithm")). My first steps for the solution are following: range(b_dist_min_new)

Dear list, I have recently encountered an odd error when running glm(dep~indep, quasipoisson): while, with a subset of my data, I could get a perfectly reasonable model, once I include all of my data (17K+ observations, 29 variables), I get the following error: Error in if (any(y < 0)) stop("negative values not allowed for the quasiPoisson family") : missing value where