I am trying to run an analysis and I its unclear to me if I have to include random factors.

I am working with microbiome data and have OTU tables (e.g. list of species per sample) for 5 different groups/samples. I also have measurements for the home range overlap between these 5 groups. I am interested in the question if range overlap predicts the OTU overlap. So do for instance groups that overlap a lot have more similar OTU's.

I have 5 groups and will have to compare 10 pairs:

A - B, A - C, A - D, A - E, B - C, B - D, ect.

My question is if I can run a simple linear model like:

lm(otu_sharing ~ range_sharing)

or do I need to control for the fact that I use the same groups multiple time by including them as a random factor:

lmer(otu_sharing ~ range_sharing + (1|group1)  + (1|group2)) 
  • $\begingroup$ This is very unclear, at least for somebody who is not familiar with your research topic. Can you describe your data in more details? So you have 5 groups and two measurements ("otu" and "range") for each of them. How many observations (samples?) is there per group? I.e. how many numbers do you have in your dataset, is it only 5*2=10 numbers? $\endgroup$ – amoeba Oct 19 '16 at 21:03
  • 1
    $\begingroup$ An OTU is an Operational Taxonomic Unit, in our study they are bacterial species in the gut. Per group I have one table of OTUs that are present in a group (OTU table). We then calculated how many of the OTUs present in a group are also present in another group. This would be the % OTU overlap (OTU sharing). We also know the home ranges of the groups and can calculated how much these home ranges overlap (range_sharing). So we have 5 OTU tables, but have 10 values of OTU sharing and 10 values of range_sharing. $\endgroup$ – user572549 Oct 20 '16 at 13:12
  • $\begingroup$ This makes it clearer. So if I understood correctly, 10 values are there because this is the number of all pairs among the 5 groups, correct? $\endgroup$ – amoeba Oct 20 '16 at 13:30
  • $\begingroup$ That is correct. $\endgroup$ – user572549 Oct 20 '16 at 14:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.