Who am I and what do I do?
I am currently a PostDoc at the Terrestrial Ecology lab in Gent (personal webpage), working on the spatial dimension of the relationship between diversity and ecosystem function across the food chain. I will be helped in this task by data collected within the TREEWEB project which is a collection of forest patches of different tree diversity and density scattered through an urban-agricultural landscape in eastern Flanders (Belgium).
I was (until 01.2017) a PhD student at the chair for Terrestrial ecology by the Technical University of Munich (DE). The thesis looked at how variations in plant diversity and community composition affect arthropod communities and arthropod-mediated ecosystem function. This work was within the framework of a very cool, very long-running grassland biodiversity experiment located in Jena.
What do I blog about?
I aimed to blog both about stats/R stuff as well as ecological issues arising in my field of research, however a quick look through the article I wrote would show you that I mainly write article about stats/R stuff. I have a constant desire to spend more time writing ecologically-orientated articles … If you have any wishes/suggestions don’t hesitate to email me (mail available in here)
Why do I blog?
Sometimes people ask me questions and it is nice to just tell them: “Have a look at this post”. Sometimes I find myself re-searching for a specific hack or a way to do a certain analysis and on the 2nd or 3rd time I usually write a post about it for future reference. Sometimes I think that blogging is giving me an important experience in writing clear, short messages that is going (hopefully) to be relevant to my career.
biologyforfun 
Dear Lionel,
I just found your plot_fit function, which is super great, many thanks for that. I have been using ezPredict previously but got really tired of adapting the loops for summing the booted values.
However, the function gives me an arrow message: “object ‘name_f’ not found” (without the inter_var condition its working; I tried changing the factor and level names).
Here the specifics:
‘data.frame’: 1843 obs. of 4 variables:
$ FEED.sq : num 4.47 3.16 3.87 4.47 4.47 …
$ L : Factor w/ 2 levels “GS”,”PS”: 1 1 1 1 1 1 1 1 1 1 …
$ fP.ID : Factor w/ 40 levels “1”,”2″,”3″,”4″,..: 1 1 1 1 1 1 1 1 1 1 …
$ POS.rel.sc: num [1:1843, 1] -1.71 -1.61 -1.51 -1.41 -1.32 …
Linear mixed model fit by REML [‘lmerMod’]
Formula: FEED.sq ~ L * POS.rel.sc + (1 | fP.ID)
pred78<-plot_fit(m.fe.83,focal_var="POS.rel.sc",n=100, inter_var= “L”,RE="fP.ID")
I hope you mend you comment regarding contacting you if issues like that occur and hope you have some time for the problem.
Many thanks in advance
Jörg
PS: in general I think it would be super cool if one could specify more than one variable e.g. inter_var=c(“L”,”G”) (however I guess refitting the model with:
fe.0.S$inter<-with(fe.0.S,interaction(L, G, sep="x")) and FEED.sq ~ inter * POS.rel.sc + (1 | fP.ID) would work also?)
Hi Jörg,
Thanks for your interest, I changed a bit the code which should resolve the error with “name_f” that you were getting, get the new code from here: http://wp.me/p2SacB-7V. Also you can pass more than one interacting variables as a character vector to the inter_var argument so c(“L”,”G”) should work if L and G are variable names in your original dataframe. Let me know if you have further questions/issues. Also note that you may comment directly on the post itself.