similar to: polr question

I have experimented with the cheese data example from McCullagh&Nelder, page 175. With a proportional odds model they obtain a residual deviance of 20.31. Estimating the same model with polr(MASS) gives a residual deviance of 762.11 !, while using ordglm(gnlm) gives a deviance of 523.94. Can anybody explain these differences? The data frame with the data are: > cheese Cheese N

I am new to the R. I managed to download and install and got some basic summary work done. I need to do ORDERED LOGIT model. I searched web-site and mail archives and I think that I need to use "ordglm" function. However, I am not sure what package has it. I tried several (like RMUTIL, ORDINAL, REPEATED) etc. but no avail. Any help is useful. I still haven't subscribed to the

Dear Peter. I guess you know that Jim Lindseys code include nordr and ordglm in library gnlm - I attach the htmls which do various linear and nonlinear ordinal regressions - exemplified with just the data mentioned, McCullagh (1980) JRSS B42, 109-142. I had it work very fine. -----Oprindelig meddelelse----- Fra: Peter Malewski <p.malewski at tu-bs.de> Til: Troels Ring <tring at

My attempt to test a model using ordglm code is running into problems, and I thought if you have a moment you might illucidate the situation. Here is the data: http://www.people.fas.harvard.edu/~corr/6.4.05.RData Here is the code: # I coerce tcn8 matrix data to a vector, because ordglm will not accept matrix data. y<-as.vector(tcn8[,62]) x<-as.vector(tcn8[,60])

I fit a proportional odds model with the polr-function of the MASS package from Venables and Ripley Applying the confint method to calculate confidence intervals for the parameters I get the following error message Waiting for profiling to be done... Re-fitting to get Hessian Error in X[, -i, drop = FALSE] : incorrect number of dimensions Can someone explain the error-message? (The

Hi (1)In polr(), is there any way to calculate a pseudo analogue to the R^2. Just for use as a purely descriptive statistic of the goodness of fit? (2) And: what are the assumptions which must be fulfilled, so that the results of polr() (t-values, etc.) are valid? How can I test these prerequisites most easily: I have a three-level (ordered factor) response and four metric variables. many

I'm running polr() and getting warning messages from glm.fit(). It seems reasonable to use glm.control() to turn on the trace and follow what glm.fit() does when called by polr(); or is it? glm.control(maxit=10, trace=TRUE) polr(act~., data=mm) The glm.control() sets the trace TRUE, but there's no change in the output from polr(). Many thanks in advance for any help/pointers.

Dear community, I am currently trying to fit an ordinal logistic regression model with the polr function. I often get the same error message : "attempt to find suitable starting values failed", for example with : require(MASS) data(iris) polr(Species~Sepal.Length+Sepal.Width+Petal.Length+Petal.Width,iris) (I know the response variable Species should be nominal but I do as levels

Hi, I'm getting an error message using polr(): Error in optim(start, fmin, gmin, method = "BFGS", hessian = Hess, ...) : initial value in 'vmmin' is not finite The outcome variable is ordinal and factored, and the independant variable is continuous. I've checked the source code for both polr() and optim() and can't find any variable called

Hi all, The problem is that when you try to use the function summary of a polr object in a function, it does not work. The problem is not related to the formula or the structure of data involved. It is probably related to the use of the function "vcov" in the code of summary for polr, and the iterative procedure to estimate the Hessian. Anyway, here there is an example extracted from

Hi, I obtained all the coefficients that I need from polr. However, I'm wondering how I can obtain the standard error of each estimated coefficient? I saved the Hessian and do something like summary(polrObj), I don't see any standard error like when doing regression using lm. Any help would be really appreciated. Thank you! - adschai

Dear all, To modelize the abundance of fish (4 classes) with a set of environmental variables, I used the polr and predict.polr functions. I would like to know how to bring the cumulated probabilities back to a discrete ordinal scale. For the moment I used the predict.polr function with the argument "class". Is there an other way? polrf <- polrf <- polr_mod(formula =

Dear Sir, I get an error message when I use polr() in MASS package. My data is "ord.dat". I made "y" a factor. y y1 y2 x lx 1 0 0 0 3.2e-02 -1.49485 2 0 0 0 3.2e-02 -1.49485 3 0 0 0 1.0e-01 -1.00000 4 0 0 0 1.0e-01 -1.00000 5 0 0 0 3.2e-01 -0.49485 6 0 0 0 3.2e-01 -0.49485 7 1 1 0 1.0e+00 0.00000 8 0 0 0 1.0e+00 0.00000 9 1 1 0

For the following model, library(vcd) arth.polr <- polr(Improved ~ Sex + Treatment + Age, data=Arthritis) summary(arth.polr) where Improved is an ordered, 3-level response I'm looking for a *simple* way to test the validity of the proportional odds assumption, typically done via a score test for equal slopes/effects over the predictors. I do find a po.test= option in the repolr package

Dear all, I have used polr in MASS but I am uncertain about the summary(polr.object) interpretation and would be happy for help on that. This is my summary: > summary(shade.polr) Re-fitting to get Hessian Call: polr(formula = as.ordered(shade) ~ as.factor(objekt), data = sof, weights = as.numeric(frek)) Coefficients: Value Std. Error t value 2.1699520 0.3681840 5.8936612

Hi, I'm trying to fit a ordered logistic regression. The response variable (y) has three levels (0,1,2). The command I've used is: /ordlog<-polr(y~x1+x2+x3+x4, data=finalbase, subset=heard, weight=wt, na.action=na.omit) / (There are no NA's in y but there are NA's in X's) The error I'm getting is: /Warning messages: 1: non-integer #successes in a binomial glm! in:

Hi, I am searching for methods to compare regression models with an ordered categorical response variable (polr versus multinom). The pattern of predictions of both methods (using the same predictor variables) is quite different and the AIC is smaller for the multinom approach. I guess polr has more strict premises for the structure of the response variable, which methods can be used to test for

Dear members of the list, I'm fitting ordinal regressions using polr, and in some models I get the error copied below. Dependent variable is an ordered factor of bird abundance categories, and predictors are continuous habitat variables. > ro6 <- polr(formula = abun ~ InOmbrot + Oliva.OC + ToCultAr + DivCulArb + AltitMax + COORXY) > summary(ro6) Re-fitting to get Hessian

Hello all, I am trying to use polr (the ordered logistic model from MASS) but I am getting the following error message: Error in if (all(pr > 0)) -sum(wt * log(pr)) else Inf : missing value where TRUE/FALSE needed My response variable is a factor with 3 levels and I have 2 independent variables. I am not sure if I guessed the starting parameters right, which I imagine could be a source of

Hi all, I am trying to do a ordered probit regression using polr(), replicating a result from SAS. >polr(y ~ x, dat, method='probit') suppose the model is y ~ x, where y is a factor with 3 levels and x is a factor with 5 levels, To get coefficients, SAS by default use the last level as reference, R by default use the first level (correct me if I was wrong), The result I got is a