Following the vignette in rstanarm fo computing loo, I am using the following code, which works just fine
fits <- list()
fits[] <- stan_glm(PV1READ ~ 1, data = PISAdata, seed = 832762) # This is the baseline model 1
fits[] <- update(fits[], formula = PV1READ ~ Female + ESCS + HOMEPOS + ICTRES) # model 1 variables
fits[] <- update(fits[], formula = PV1READ ~ JOYREAD + PISADIFF + SCREADDIFF + SCREADCOMP) # model 2 variables
fits[] <- update(fits[], formula = PV1READ ~ METASUM + GFOFAIL + MASTGOAL + SWBP + WORKMAST + ADAPTIVITY + COMPETE) # model 3 variables
fits[] <- update(fits[], formula = PV1READ ~ PERFEED + TEACHINT + BELONG) # model 3 variables
My question is whether it is necessary to compute fits[] the baseline model, or was this model in the vignette considered a reasonable model for the applied Milk example. Once the weights are obtained, are the weights attached to model 1 (the baseline model) are they interpretable. For this example, by the way, the weights are 0.000.
To include mathematical notation in your post put LaTeX syntax between two
$ symbols, e.g.,
p(\theta | y) \propto p(\theta) p(y | \theta).
Don’t forget to add relevant tags to your topic (top right of this form) for application area and/or class of models you work with.