similar to: Survey package GLM vs. Linear Mixed Models?

Hi, All Thanks to Robert Baskin, Thomas Lumley, and Spencer Graves for the valuable helps. I have learned a lot from this discussion. I put all discussions together without editing, so we can see how things are evolved. Likely, I have a lot of articles to read. As in the discussion, mixed modeling approach is a poosible but may be over-kill in my posted data analyses. I will explore other

Han-Lin I don't think I have seen a reply so I will suggest that maybe you could try a different approach than what you are thinking about doing. I believe the current best practice is to use the weights as a covariate in a regression model - and bytheway - the weights are the inverse of the probabilities of selection - not the probabilities. Fundamentally, there is a difficulty in making

Hi folks I have some data to which I've been fitting linear mixed effects models. I am currently using a lme model in the nlme package, with terms for random effects due to repeated measures on individuals and the corCAR1 serial correlation structure. However, there is some suggestion in the data (and from theory) that a breakpoint (change point) model may be more appropriate. Scott, Norman

Hello, I want to specify a linear regression model in which the metric outcome is predicted by two factors and their interaction. glm() computes effects for each factor level and the levels of the interaction. In the case of singularities glm() displays "NA" for the corresponding coefficients. However, svyglm() aborts with an error message. Is there a possibility that svyglm()

Hello All, I am looking for an R library/function that allows the specification of a custom link function in a linear mixed model. I've been using glm() in library MASS to fit fixed-effect models with a custom link but my study design demands mixed models. Any suggestions on the best R library/function to achieve this would be greatly appreciated. I have tried, to no avail, to

Hi experts, I have just about started to use R (after using SAS for more than 5 years) and still finding my way...I have been trying to replicate PROC MIXED results in LMER but noticed that the estimates are coming different. My SAS code is as follows (trying to randomise X2 and Intercept): PROC MIXED DATA = <DATASET NAME> NAMELEN=100 METHOD=REML MAXITER=1000; CLASS GEOGRAPHY; MODEL y

I am trying to fit a linear mixed model of the form y_ij = X_ij \beta + delta_i + e_ij where e_ij ~N(0,s^2_ij) with s_ij known and delta_i~N(0,tau^2) I looked at the ecme routine in package:pan, but this routine does not allow for different Vi (variance covariance matrix of the e_i vector) matrices for each cluster. Is there an easy way to fit this model in R or should I bite the bullet and

Hi, I have created a global model using lmer knowing it contains at least one fixed effect which has missing values. I add the term na.action=na.omit to the model formula as shown below, and the summary output is produced fine, until I wish to simplify the model and compare the resulting model with the previous one using anova. As soon as the covariate containing the missing values is removed,

Hello all, I would like to fit a mixed effects model, but my response is of the negative binomial (or overdispersed poisson) family. The only (?) package that looks like it can do this is glmm.ADMB (but it cannot run on Mac OS X - please correct me if I am wrong!) [1] I think that glmmML {glmmML}, lmer {Matrix}, and glmmPQL {MASS} do not provide this "family" (i.e. nbinom, or

Hello folks, I have recently finished a pilot study of a survey and am working to complete the statistical analysis of the results in R. My Phd is technically in computer science (not statistics), although I teach basic stats and have a "decent" working knowledge of the area. With that said, my expertise in psychometrics theory and factor analysis is weaker, so I thought I would send an

Hello all, I am using AddHealth data to fit a cure, aka split population model using nltm. I am not sure how to account for the complex survey design - does anyone have any suggestions? Any help would be greatly appreciated! Sincerely, Sam

I'm new to R and am trying to extract the factors of a dataframe using numeric indices (e.g. df[1]) that are input to a function definition instead of the other types of references (e.g. df$out). df[1] is a list(?) whose class is "dataframe". These indexed lists can be printed successfuly but are not agreeable to the plot() and lm() functions shown below as are their df$out

Hello, I'm fitting linear mixed models to gene-expression data from microarrays, in a data set where 4608 genes are studied. For a sample of 5 subjects and for each gene we observe the expression level (Intensity) in four different tissues: N, Tp, Tx and M. I want to test whether the expression level is different accross tissues. Between-subject variability is modeled with a random

This question comes up periodically, probably enough to give it a proper thread and maybe point to this thread for reference (similar to the 'conservative anova' thread not too long ago). Moving from lme syntax, which is the function found in the nlme package, to lmer syntax (found in lme4) is not too difficult. It is probably useful to first explain what the differences are between the

I have looked through the new "Complex Surveys" book and the documentation for the "survey" package and it appears to me that there are no functions in "survey" that help one to design a sampling scheme. For example, in the book section 2.8 discusses the design of stratified samples, but there is no mention of any functions in the "survey" package that

Dear all, I'm using the latest version of the package lme4 and R version 2.7.2 (2008-08-25). After I run the model, I get the results of the model (cf. below). Then, I run an ANOVA using the "anova" function and I get the following message "Error in anova(lmer1) : Calculated PWRSS for a LMM is negative". I went trough the R-mailing list and a similar error message was

Boxplot and bxp seem to have changed behaviour a bit of late (R 2.4.1). Or maybe I am mis-remembering. An annoying feature is that while at=3:6 will work, there is no way of overriding the default xlim of 0.5 to n+0.5. That prevents plotting boxes on, for example, interval scales - a useful thing to do at times. I really can see no good reason for bxp to hard-core the xlim=c(0.5, n+0.5) in the

Hi all, I have data collected from a survey administered on a subset of the population. I also have the population proportions of variables such as gender, race and housing type. I would like to combine the weights from each separate cross tab (of gender, race and housing type) such that the weighted proportions of my survey data matches that of the population. I have tried the following:

Hi R users, I am trying to use the aggregate function with a survey design object and survey functions, but get the following error. I think I am incorrectly using the syntax somehow, and it may not be possible to access variables directly by name in a survey-design object. Am I right? How do I fix this problem? I have used aggregate with "mean" and "weighted.mean", and

Dear member, in cooperation with the software engineering research group at Free University of Berlin (Freie Universit?t Berlin), we are currently conducting an empirical survey on time consumption of participants in Open Source/Free Software projects. In this survey we want to find out how much time is consumed in the different activities and aspects of Open Source/Free Software