[R] Problems using estimable() (gmodels) on lme()-objects

CG Pettersson cg.kavlinge at gmail.com
Mon Dec 14 11:23:07 CET 2015


R version 3.2.3 (2015-12-10) -- "Wooden Christmas-Tree", W7-32



Dear all,

I just upgraded R and downloaded a new version of “gmodels”, to use
estimable() for extracting fixed effects from lme()-objects.

Now it seems like estimable() have changed its view of dimensions of my
lme-objects and contrast matrix dimensions, as I get this error message:



> estimable(y_15,Mat15)

Error in estimable.default(y_15, Mat15) :

 Dimension of structure(c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,
0, : 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
: 25x26, not compatible with no of parameters in y_15: 25

Dimension of 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,

>



Mat15 is made as a 25x25 contrast matrix, but the updated estimable() sees
it as a 25x26.

Why, and what should I do about it?



I do think the change has occurred in estimable() as I tested one year old
lme() objects and contrast matrix who worked perfectly then, giving the
same error message when run with updated R.
Does estimable() see the row-names as the first column nowdays if I don´t
specify anything else?

All the best
/CG Pettersson

	[[alternative HTML version deleted]]



More information about the R-help mailing list