the summary() of a model.avg made with MuMIn in R, give a lot of interesting results, in particular model averaged coefficients (estimate, standard error, adjusted standard error and a z value with a relative P value). My question is, which kind of test has been done to obtain this P value and in particular what is the relation between the variables turning out to be significant from this test and the relative variable importance obtained from the model average?
What is "significant"? P0.05 is usually not a good reason!
Dear Tiago!
Let us denote mm the model, you can get the parameters:
model$coefArray
Or try mm[1], mm[2], nad so on
Best wishes: Zoli
You need to present all the variables whether significant or not that were included in the model
I concur with Jochen Wilhelm above. Report all the variables. Also, do not use stepwise regression for datamining (i.e., to figure out which theoretically unmotivated predictors have an impact on the outcome variable). Statistics exists for hypothesis testing, so for each predictor included in your model, there should be a theoretical motivation as to why the predictor SHOULD have an impact on the outcome.
Anyway, I wrote a script a function to set all MuMIn parameters correctly and to dredge models in parallel, drastically reducing the time it takes to dredge a model. Here's the script, you can run it by entering the following code (with the correct parameters; notice the quotes around the data and model object names).
source("http://goo.gl/S5vMNv"); par_dredge (data="data_object_name", model="model_object_name")
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling