similar to: Posthoc test for 3-way interaction (log-linear model)

Hi all, I've been running loglinear models for three-way tables: one of the variables having three levels, and the other two having two levels each. An example looks like below: > yes.no <- c("Yes","No") > switch <- c("On","Off") > att <- c("BB","AA","CC") > L <- gl(2,1,12,yes.no) > T <-

Hi, I'm analyzing my data using GEE, which looks like below: > interact <- geeglm(L ~ O + A + O:A, + data = data1, id = id, + family = binomial, corstr = "ar1") > summary(interact) Call: geeglm(formula = lateral ~ ontask + attachment + ontask:attachment, family = binomial, data = firstgroupnowalking, id = id, corstr = "ar1") Coefficients:

Hello. We have some questions concerning the statistical analysis of a dataset. We aim to compare the sample means of more than 2 independent samples; the sample sizes are unbalanced. The requirements of normality distribution and variance homogeneity were not met even after transforming the data. Thus we applied a nonparametric test: the Kruskal-Wallis-test (H-Test). The null hypothesis was

Hi there, I was hoping to post this message. Im hoping I;ve got the right spot?! Hi there, sorry for basic question but im very new to R. Im trying to run a lme model with two categorical variables, each having 6 (for the explanatory variable C.f) and 5 levels (for expl. variable D.f) respectively. The lme runs fine, summary and anova commands are ok. But now i've tried running posthoc

dear all, I want to perform a posthoc test for my ANCOVA: a1<-aov(seeds~treatment*length) With summary(glht(a1, linfct = mcp(treatment = "Tukey"))) R tells me: "covariate interactions found -- please choose appropriate contrast" How do I build these contrasts? Ideally, I would like to have the posthoc test for the ANCOVA including a block-effect

Dear List, are there post-hoc tests like Scheffe, LSD, etc. available after ANOVA test is performed with significant F-statistic? I have tried help.search("Scheffe"), but "No documentation found" (and I have most of packages installed). Probably there are such tests in R, and I am just searching badly... My second question is: Which test/method I should use for ANOVA-like

Hi All, I have performed a 3-way ANOVA analysis for my experimental data using aov function. My simple R funtion for this is: 3aof <- function(x){ m <- data.frame(R,S,T, x); anova(aov(x ~ R+S+T+R*S+R*T+S*T+R*S*T, m) ) } Now, I am getting P values for all the main and interactions effects. If I want to perform postdoc test on one of my main effects, say T, what method I should use (i have

Dear R experts I am running a negative-binomial GLM (glm.nb) to test the null hypotheses that species 1 and 2 are equally abundant between site 1 and site2, and between each other. So, I have a 2x2 factorial design with factors Site (1,2) and Taxon (1,2). Since the Site:Taxon interaction is significant, I need to do the equivalent to a "post-hoc test" for ANOVA, however, the same tests

The null loglinear model is an intercept-only model for log frequency, log(f) = \mu For a one-way table the test of the null model is the same as the chisq.test. This can be fit using loglin(), but I don't think there is any way to specify this using MASS::loglm > t1<- margin.table(Titanic,1) > t1 Class 1st 2nd 3rd Crew 325 285 706 885 > loglin(t1, NULL) 0

loglm() is a port of an original written for S-PLUS. The fact that it carries only a reference to the data frame is neither intentional nor unintentional, but an unnoticed side-effect. I can see advantages both ways. (I'm not so sure, either, that what you say is standard behaviour for model fitting functions really is so universal.) It would not be too hard to come up with a version that

Dear R-help community, I am working with multidimensional contingency tables and I am having trouble getting loglm to run on all dimensions without typing out each dimension. I have generated random data and provided output for the results I want below: d1.c1 <- rnorm(20, .10, .02) d1.c2 <- rnorm(20, .10, .02) d2.c1 <- rnorm(20, .09, .02) d2.c2 <- rnorm(20, .09, .02) d3.c1 <-

Bug report: there is a problem with loglm (or some underlying code) when used with updating within a function. For example, this code works as expected, when used outside a function: ------------------------------------------------- library(MASS)

Dear all, sorry to bother you all with this but I've been trying to use the loglm in MASS package (v2.8.0) and cannot get any sensible output. I'm wondering am I doing something very foolish or missing something obvious. For example, I tried the documentation help(loglm) example - here's the code # Case 1: frequencies specified as an array. sapply(minn38,

Hello All R Users, Function loglm() in library MASS can be cajoled to accomodate structural zeros in a cross-classification table. An example from Fienberg demonstrates how this can be done. My question is: Can the function glm() perform the same task? Can glm() estimate a log-linear model with fixed zeros like loglm()? Thanks for your help, Andrew ## Fienberg, The Analysis of Cross-Classified

Summary: loglin will fail if there is a global variable called c. I have fixed this for 0.64.2 and 0.65. On Sat, 22 May 1999, Cor en Aylin wrote: > Dear Brian, > > Using the mass library for R (the most recent port dated 16 may 1999) I Could you give version numbers in such reports please: at the moment there is an official version (5.3pl037) of MASS for R, plus a pre-release of MASS

I'd like to fit and explore a collection of hierarchical loglinear models that might range from the independence model, ~ 1 + 2 + 3 + 4 to the saturated model, ~ 1 * 2 * 3 * 4 I can use add1 starting with a baseline model or drop1 starting with the saturated model, but I can't see how to get the model formulas or terms in each model as a *list* that I can work with further. Consider

I have numerous replicated goodness-of-fit experiments (observed compared to expected counts in categories) and these replicates are nested within a factor. The expected counts in each cell are external (from a scientific model being tested). The calculations I need within each level of the nesting factor are a heterogeneity G test, with the total G and the pooled G across replicates. Then I

Last week Giovanni Parrinello posted a message asking why various packages were loaded when he loaded an .Rdata file. Brian Ripley replied saying he thought it was because the saved workspace contained a reference to the namespace of ipred. (Correspondence copied below). This begs the question: how did the reference to the namespace of ipred come to be in the .Rdata file? Brian did say it is

Dear all, I have a problem to make figures with two columns in package vcd. Here an example code I take from "\library\vcd\html\plot.loglm.html" What I need, I want to make two figures in one plot. How could I do that. I have tried with layout(rbind(c(1, 1, 2, 2))) but the same result, two plot. Best wishes, Muhammad Subianto library(vcd) oldpar <- par(mfrow=c(1, 2)) ## mosaic

Dear R-helpers, Im trying to fit a Log-linear model on a dataset with bird counts from 60 sites over 14 years for 12 months each (factors for the models). One of the aims is to predict the missing values in this dataset with model predictions. Ive first tried to work with GLM's, that worked fine except for models with one or more interaction-terms. The GLMs run, run .. for hours. So I