New answers tagged

1

I think there are two problems going on here. First, I'll hazard a guess that your model is far too complex—all of these nestings are probably the source of the errors. Second, and more importantly, you are fitting different data to the same outcome variables many times over. Your outcome (please correct me if I am wrong here) is drug overdose deaths in ...


0

I'm the developer of the DHARMa package. Frans Rodenburg is right. Just to summarise The plot to the right is clustered because you have so many data = many residuals. Ff you want to reduce the number of residuals, you can take a subset of the data, or aggregate residuals via the recalculateResiduals function. Given your large number of data points, some ...


0

I'm the developer of the DHARMa package. It's hard to say what's going on without seeing your data, but I suspect that the issue arises as follows: As one can see in the DHARMa plots, your fixed effects design just creates 4 predictions (I assume both A,B only have 2 levels, so you have 4 values from A*B) In your binned plots, however, there are a lot more ...


0

Now the estimated variances of the random factors (N.mussel, the mussels identity) are really small, suggesting I should omit them. No, you shouldn't, this would be p-hacking. You should leave the model like it is. Moreover, how many observations do you have per group? Depending on your sample size, this could be completely random. In general, if you ...


2

Leaving regression outcomes and inputs in their original units is desirable because the association measure can be interpreted geometrically as a slope relating them. For instance, if you related miles driven by a car to the gas fueled between stops, the slope describes fuel efficiency as miles per gallon. It's intuitive. The most popular approach to making ...


0

My recommendation, if you want to test for collinearity and have categorical variables as well as continuous ones and are using R is to use the perturb package. The idea here is to add small amounts of random noise to the variables - by adding uniform or normal noise to the continuous variables and by shifting some categorical ones - and seeing what happens ...


1

I'd like to answer the more 'initial' question. If you suspect any sort of heterogeneity in variance among either the dependent variable due some some factors, you should go ahead and plot the data using scatter and box plots. Some common patterns to check for, I put this list below from various sources on the web. In addition, plot your dependent ...


0

In most cases, I’d recommend displaying the estimated marginal means and their confidence intervals — e.g., via emmeans() — or displaying them graphically. And also show the pairwise comparisons among them, with the Tukey adjusted P values. (Just show the P values, let people decide for themselves what’s important.) If you do regard one particular factor ...


5

Function glmer() uses by default the Laplace approximation, which is not optimal for dichotomous data. A better alternative is the adaptive Gaussian quadrature. You can use this method by setting argument nAGQ of glmer() to a higher number (e.g., 11 or 15) or alternatively using the GLMMadaptive package. In your example it gives: library("GLMMadaptive") ...


1

I tried with the glmmADMB package, an alternative of lme4 for linear mixed modelling. You can install this package with this code: install.packages("R2admb") install.packages("glmmADMB", repos=c("http://glmmadmb.r-forge.r-project.org/repos", getOption("repos")), type="source") Then you go: library(glmmADMB) helpmeobiwan <-list(NestPlot = c(1, ...


Top 50 recent answers are included