Problem with xp.boot.par.est for bootscm

[rikardn]

In the UPSS script for bootscm task 9 is using xp.boot.par.est, which displayed an empty plot. I think that the problem is related to the levels function call here:

xpose4/R/bootgam.plots.R

Line 111 in f79acae

lev.ord <- unlist(sapply(lev.ord,function(x) levels(pl.dat$cov)[grep(x,levels(pl.dat$cov))]),use.names = F)

It returns NULL since there is no levels attribute set on the object. Perhaps an earlier behaviour of R levels was to return the elements themselves in this case.

I managed to patch it, by replacing levels with unique

lev.ord <- unlist(sapply(lev.ord,function(x) unique(pl.dat$cov)[grep(x,unique(pl.dat$cov))]),use.names = F) 

This isn't a 100% solution since there will now be an extra .1 in the covariate effect labels as can be seen here: