Timeline for R lmer model: degree of freedom and chi square values are zero
Current License: CC BY-SA 4.0
19 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 31, 2020 at 15:19 | vote | accept | SuperDuperMario | ||
| Aug 23, 2020 at 18:00 | history | tweeted | twitter.com/StackStats/status/1297594473212510209 | ||
| Aug 23, 2020 at 9:18 | answer | added | Robert Long | timeline score: 3 | |
| Aug 23, 2020 at 0:11 | history | edited | SuperDuperMario | CC BY-SA 4.0 |
added 318 characters in body
|
| Aug 22, 2020 at 23:50 | comment | added | SuperDuperMario | @StatsStudent Sure! dropbox.com/sh/88m8h6blow2xbn5/AABiNccsUlu3AlfPyamQP4n_a?dl=0 I have put in the data file and the R markdown file. I just realised because I subsetted the data, I only have 288 observations. Well, although after doing some search, now I'm not sure whether what I thought was 'observations' (i.e. data points) is the actual 'observations' you meant. I also asked a question about the procedures I'm using in this post stats.stackexchange.com/questions/484238/… It describes what I did in the R code. Thank you a million! | |
| Aug 22, 2020 at 23:06 | comment | added | StatsStudent | @RoroMario, any chance you could post your data somewhere that we could read it in to help diagnose? | |
| Aug 22, 2020 at 22:14 | comment | added | SuperDuperMario | @JeremyMiles Thanks for the suggestions! I'll update my questions. Thank you! | |
| Aug 22, 2020 at 22:13 | comment | added | SuperDuperMario | @StatsStudent Thanks for your answer! I think you must be right, because I also got a warning after I run the interaction only model [AB <- lmer(DV~ A:B + (1|speaker), data, REML=FALSE)] saying "fixed-effect model matrix is rank deficient so dropping 1 column / coefficient". But I don't understand why since I have 576 observations; A has 4 levels; B has 2 levels. What should I do if your explanation is true? | |
| Aug 22, 2020 at 22:09 | comment | added | SuperDuperMario | @RobertLong I have 576 observations. A has 4 levels and B has 2 levels. | |
| Aug 22, 2020 at 6:19 | comment | added | Robert Long |
How many observations do you have, and how many levels do A and B have ?
|
|
| Aug 22, 2020 at 5:02 | comment | added | StatsStudent | Is it possible you have fewer observations than variables in your model matrix when you include the interaction term into your model? For example if A and B are both factor level variables with, say 3 levels each, then your model with just A+B would require 2+2 = 4 observations for fitting, while the model A:B would require 2+2+2$\times$2=8 observations for fitting? I'm wondering if the number of observations you have is sufficient to fit the additive models, but then, when you fit the interaction model, you have too few observations? | |
| Aug 22, 2020 at 0:14 | comment | added | Jeremy Miles | Also, it can be overwhelming to people who want to answer if you ask multiple questions. I would consider asking one question at a time. First, to understand what's happening with the models. Then to discuss what comparisons should be done. (Someone might only want to answer one question, but feel like they shouldn't answer only half). | |
| Aug 22, 2020 at 0:10 | comment | added | Jeremy Miles | Look at your parameter estimates (e.g. using summary(full) and compare them. Perhaps post them? | |
| Aug 22, 2020 at 0:00 | comment | added | SuperDuperMario | sorry, C was supposed to be 'interaction'. I have changed it now. Thanks! | |
| Aug 21, 2020 at 23:59 | history | edited | SuperDuperMario | CC BY-SA 4.0 |
added 10 characters in body
|
| Aug 21, 2020 at 23:56 | comment | added | Jeremy Miles | It's better to copy and paste output with commands, for that reason. | |
| Aug 21, 2020 at 23:55 | comment | added | Jeremy Miles | What is C? I see models full, A, B and interaction. But then you do anova(full, C) | |
| Aug 21, 2020 at 23:55 | review | First posts | |||
| Aug 22, 2020 at 3:27 | |||||
| Aug 21, 2020 at 23:53 | history | asked | SuperDuperMario | CC BY-SA 4.0 |