similar to: lmer specification for random effects: contradictory reults

Dear Thierry, thank you for the quick reply. I have only one question about the approach you proposed. As you suggested, imagine that the model we end up after the model selection procedure is: mod2.1 <- lmer(dT_purs ~ T + Z + (1 +T+Z| subject), data =x, REML= FALSE) According to the common procedures specified in many manuals and recent papers, if I want to compute the p_values relative to

Dear R-help members, Apologies - I am posting on behalf of a colleague, who is a little puzzled as STATA and R seem to be yielding different survival estimates for the same dataset when treating a variable as ordinal. Ordered() is used to represent an ordinal variable) I understand that R's coxph (by default) uses the Efron approximation, whereas STATA uses (by default) the Breslow. but we

Hi Dears, When I introduce an interaciton in a piecewise model I obtain some quite unusual results. If that would't take u such a problem I'd really appreciate an advise from you. I've reproduced an example below... Many thanks x<-rnorm(1000) y<-exp(-x)+rnorm(1000) plot(x,y) abline(v=-1,col=2,lty=2) mod<-lm(y~x+x*(x>-1)) summary(mod) yy<-predict(mod)

Hello R users, I'm working with a time-series of several years and to analyze it, I?m using GAM smoothers from the package mgcv. I?m constructing models where zooplankton biomass (bm) is the dependent variable and the continuous explanatory variables are: -time in Julian days (t), to creat a long-term linear trend -Julian days of the year (t_year) to create an annual cycle - Mean temperature

Dear R help, I am trying to work out if I am justified in log-transforming data and specifying Gamma in the same glm. Does it have to be one or the other? I have attached an R script and the datafile to show what I mean. Also, I cannot find a mixed-model that allows Gamma errors (so I cannot find a way of including random effects). What should I do? Many thanks, Pete --------------

Dear List-members, I found a strange behaviour in the lqs function. Suppose I have the following data: y <- c(7.6, 7.7, 4.3, 5.9, 5.0, 6.5, 8.3, 8.2, 13.2, 12.6, 10.4, 10.8, 13.1, 12.3, 10.4, 10.5, 7.7, 9.5, 12.0, 12.6, 13.6, 14.1, 13.5, 11.5, 12.0, 13.0, 14.1, 15.1) x1 <- c(8.2, 7.6,, 4.6, 4.3, 5.9, 5.0, 6.5, 8.3, 10.1, 13.2, 12.6, 10.4, 10.8, 13.1, 13.3, 10.4, 10.5, 7.7, 10.0, 12.0,

Greetings, I am interested in creating a stepwise fixed order regression function. There's a function for this already called add1( ). The F statistics are calculated using type 2 anova (the SS and the F changes don't match SPSS's). You can see my use of this at the very end of the email. What I want: a function to make an anova table with f changes and delt R^2. I ran into

I can use stepAIC on an lme object in 1.6.2, but I get the following error if I try to do the same in 1.7.0: Error in lme(fixed = resp ~ cov1 + cov2, data = a, random = structure(list( : unused argument(s) (formula ...) Does anybody know why? Here's an example: library(nlme) library(MASS) a <- data.frame( resp=rnorm(250), cov1=rnorm(250), cov2=rnorm(250),

Hi I am comparing four different linear mixed effect models, derived from updating the original one. To compare these, I want to use anova(). I therefore do the following (not reproducible - just to illustration purpose!): dat <- loadSPECIES(SPECIES) subs <- expression(dead==FALSE & recTreat==FALSE) feff <- noBefore~pHarv*year # fixed effect in the model reff <-

Hello, Any advice or pointers for implementing Sobel's test for mediation in 2-level model setting? For fitting the hierarchical models, I am using "lme4" but could also revert to "nlme" since it is a relatively simple varying intercept model and they yield identical estimates. I apologize for this is an R question with an embedded statistical question. I noticed that a

Dear R-list, I would like to use the lattice library to show several groups on the same graph. Here's my example : ## the data f1 <- factor(c("mod1","mod2","mod3"),levels=c("mod1","mod2","mod3")) f1 <- rep(f1,3) f2 <-

Hi I would like to shorten mod1 <- nls(ColName2 ~ ColName1, data = table, ...) mod2 <- nls(ColName3 ~ ColName1, data = table, ...) mod3 <- nls(ColName4 ~ ColName1, data = table, ...) ... is there something like cols = c(ColName2,ColName3,ColName4,...) for i in ... mod[i-1] <- nls(ColName[i] ~ ColName1, data = table, ...) I am looking forward to help Christof

Hello there, I am trying to fit an exponential model using nls to some data. #data t <- c(0,15,30,60,90,120,240,360,480) var <- c(0.36,9.72,15.50,23.50,31.44,40.66,59.81,73.11,81.65) df <- data.frame(t, var) # model # var ~ a+b*(1-exp(-k*t)) # I'm looking for values of a,b and k # formula # mod <- nls(formula = var ~ a+b *(1-exp((-k)*t)), start=list(a=0, b=10,

Greetings, 1) Is there a nice way of extracting the variance estimates from an lme fit? They don't seem to be part of the lme object. 2) In a series of simulations, I am finding that with ML fitting one of my random effect variances is sometimes being estimated as essentially zero with massive CI instead of the finite value it should have, whilst using REML I get the expected value. I guess

Sorry for the last email, sent too early. I have a small data set that has a hierarchical structure. It has both temporal (year, months) and spatial (treatment code and zone code). The following explains the data: WSZ_Code the water supply zone code (1 to 8) Treatment_Code the treatment plant which supplies each water supply zone (1 to 4)

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

I have a general question about using thin plate splines in the surf and ordisurf routines. My rudimentary knowledge of a gam is that with each predictive variable there is a different smooth for each one and then they are added together with no real interaction term (because they don't handle this well?). Now, If I have two variables that have a high D^2 score and a low GCV score (I am

Hi R Users, After some research I haven't find what I want. I'm manipulating a dataframe with 70k rows and 30 variables, and I run out of memory when exporting this in a *.txt file after some computing I have used : > memory.size()/1048576.0 [1] 103.7730 and I make my export : > write.table(cox,"d:/tablefinal2.txt",row.names=F,sep=';') >

Hello, In the past I have tended to reside more in the ANOVA camp but am trying to become more familiar with regression techniques in R. I would like to get the F change from a model as I take away factors: SO... mod1<-lm(y~x1+x2+x3).......mod2<-lm(y~x1,x2).......mod3<-lm(y~x1) I can do this by hand by running several models in R and taking the MSr1/MSe1, MSr2/MSe2... This is

I am examining the following nlme model. asymporig<-function(x,th1,th2)th1*(1-exp(-exp(th2)*x)) mod1<-nlme(fa20~(ah*habdiv+ad*log(d)+ads*ds+ads2*ds2+at*trout)+asymporig(da.p,th1,th2), fixed=ah+ad+ads+ads2+at+th1+th2~1, random=th1+th2~1, start=c(ah=.9124,ad=.9252,ads=.5,ads2=-.1,at=-1,th1=2.842,th2=-6.917), data=pca1.grouped) However, the two random effects (th1 and th2)