similar to: predict.lm is broken in 1.1.0-patched (2000-August-7) (PR#627)

On 7 Aug 2000, Peter Dalgaard BSA wrote: > thomas@biostat.washington.edu writes: > > > > > coef.aov > > > function (object, ...) > > > { > > > z <- object$coef > > > z[!is.na(z)] > > > } > > > > coef.lm > > > function (object, ...) > > > object$coefficients > > > > Ok, this

predict.lm has been broken by recent changes to the patched branch. It fails for all singular fits. An example: library(MASS) data(quine) quine.hi <- aov(log(Days + 2.5) ~ .^4, quine) quine.nxt <- update(quine.hi, . ~ . - Eth:Sex:Age:Lrn) predict(quine.nxt) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 NA NA NA NA NA NA NA NA NA NA NA NA NA

Dear Martin, I think that your plan makes sense. It's too bad that aov() behaved differently in this respect from lm(), and thus created more work, but it's not be a bad thing that the difference is now explicit and documented. I expect that that other problems like this will surface, particularly with contributed packages (and I know that you're aware that this has already happened

>>>>> Fox, John <jfox at mcmaster.ca> >>>>> on Thu, 14 Sep 2017 13:46:44 +0000 writes: > Dear Martin, I made three points which likely got lost > because of the way I presented them: > (1) Singularity is an unusual situation and should be made > more prominent. It typically reflects a problem with the > data or the

>>>>> Martin Maechler <maechler at stat.math.ethz.ch> >>>>> on Thu, 14 Sep 2017 10:13:02 +0200 writes: >>>>> Fox, John <jfox at mcmaster.ca> >>>>> on Wed, 13 Sep 2017 22:45:07 +0000 writes: >> Dear Terry, >> Even the behaviour of lm() and glm() isn't entirely consistent. In both cases,

Hello, I've got what I'd expect to be a pretty simple issue: I fit an aov object using multiple error strata, and would like some significance tests for the contrasts I specified. In this contrived example, I model some test score as the interaction of a subject's gender and two emotion variables (angry, happy, neutral), measured at entry to the experiment (entry) and later

Hello, I've a problem with a self written routine taking a lot of memory (>1.2Gb). Maybe you can suggest some enhancements, I'm pretty sure that my implementation is not optimal ... I'm creating many linear models and store coefficients, anova p-values ... all I need in different lists which are then finally returned in a list (list of lists). The input is a matrix with 84 rows

Dear Madame, Dear Sir, I am able to obtain the coefficients from a 'summary' of 'lm', but NOT from a 'summary' of 'aov'. The following example shows my steps. ## Initialize rm(list = ls()) # remove (almost) everything in the working environment utils::data(npk, package="MASS") # get data model <- yield ~ block + N*P*K ## Using lm npk.lm <-

Hi R users, I sent a message yesterday about NA in model estimates ( http://r.789695.n4.nabble.com/How-set-lm-to-don-t-return-NA-in-summary-td3722587.html). If I use aov() instead of lm() I get no NA in model estimates and I use gmodels::estimable() without problems. Ok! Now I'm performing a lot of contrasts and I need correcting for multiplicity. So, I can use multcomp::glht() for this.

Hello all R-users _question 1_ I need to make a statistical model and respective ANOVA table but I get distinct results for the T-test (in summary(lm.object) function) and the F-test (in anova(lm.object) ) shouldn't this two approach give me the same result, i.e to indicate the same significants terms in both tests??????? obs. The system has two restrictions: 1) sum( x_i ) = 1 2) sum(

I am not sure whether there is a bug. When I tested the example given for "glht" in the help, I entered the following error: Running commands: amod <- aov(minutes ~ blanket, data = recovery) rht <- glht(amod, linfct = mcp(blanket = "Dunnett"), alternative = "less") Errors are: Error in try(coef.(model)) : could not find function

Hi, the function TukeyHSD gives incorrect results for balanced incomplete block designs, as the example below shows, but I can only half fix it. There are two problems, 1. It uses model.tables to estimate treatment means, 2. It uses the wrong standard error The first problem can be fixed using dummy.coef, if the lines > TukeyHSD.aov function (x, which = seq(along = tabs), ordered = FALSE,

I am not sure whether there is a bug. When I tested the example given for "glht" in the help, I entered the following error: Running commands: amod <- aov(minutes ~ blanket, data = recovery) rht <- glht(amod, linfct = mcp(blanket = "Dunnett"), alternative = "less") Errors are: Error in try(coef.(model)) : could not find function

Hallo! I have data of the following design: NSubj were measured at Baseline (visit 1) and at 3 following time points (visit 2, visit 3, visit 4). There is or is not a treatment. Most interesting is the question if there is a difference in treatment between the results of visit 4 and baseline. (The other time points are also of interest.) The level of significance is alpha=0.0179 (because of an

I am having trouble getting standard errors for contrasts using se.contrast() in what appears to be a simple case to me. The following test example illustrates my problem: Lab <- factor(rep(c("1","2","3"),each=12)) Material <- factor(rep(c("A","B","C","D"),each=3,times=3)) Measurement <-

Kia ora list members: I'm having a little difficulty getting the correct standard errors from a glm.object (R 1.9.0 under Windows XP 5.1). predict() will gives standard errors of the predicted values, but I am wanting the standard errors of the mean. To clarify: Assume I have a 4x3x2 factorial with 2 complete replications (i.e. 48 observations, I've appended a dummy set of data at the

# Your mailer is set to "none" (default on Windows), # hence we cannot send the bug report directly from R. # Please copy the bug report (after finishing it) to # your favorite email program and send it to # # r-bugs@biostat.ku.dk # ###################################################### The error occurs following the third Hit <Return> below: > y_rnorm(40) >

Hi all, I'm doing a Repeated-Measures ANOVA, but I don't know how to test its contrasts or where to find the p-values of its coefficients. I know how to find the coefficient estimates of a contrast, but not how to test these estimates. First I do something like: y.aov <- aov(y ~ fac1 * fac2 + Error(subj/(fac1 * fac2)), data=data) Then, with coef(y.aov) I get the coefficients

Is there code available to decompose interactions involving at least one nominal factor with more than 2 levels as described, e.g., by Tukey or by Mandel (1971, Technometrics, 13: 1-18)? Tukey's model: E(y[i,j]) = mu0 + a[i] + b[j] + c*a[i]*b[j], estimating a, b, and c so sum(a) = sum(b)= 0. Mandel essentially describes a singular value decomposition of the interaction. Thanks,

Hi all, I hope this question is not too trivial. I can't find an explanation anywhere (Stats and R books, R-archives) so now I have to turn to the R-list. Question: If you have a factorial design with two factors (say A and B with two levels each). What does the intercept coefficient with treatment.contrasts represent?? Here is an example without interaction where A has two levels A1 and