R devel - Oct 2024 - Error message when calling t.test() and aov() with a factor variables

Hi all,

Just noticed that the error that arises when calling t.test() with factors could
be a bit clearer:
> t.test(factor(c(3,1,2,4,3,5,4,5)), factor(c(2,1,2,3,4,5)))
Error in var(x) : Calling var(x) on a factor x is defunct.
  Use something like 'all(duplicated(x)[-1L])' to test for a constant
vector.
In addition: Warning message:
In mean.default(x) :
  argument is not numeric or logical: returning NA

Obviously, using factors as input is nonsense, but this might happen on accident
and then the error message could be a bit more on point. Similar for aov():
> aov(factor(c(3,1,2,4,3,5,4,5)) ~ factor(c(2,1,2,2,2,1,2,1)))
Call:
   aov(formula = factor(c(3, 1, 2, 4, 3, 5, 4, 5)) ~ factor(c(2,
    1, 2, 2, 2, 1, 2, 1)))
Error in levels(x)[x] :
  only 0's may be mixed with negative subscripts
In addition: Warning messages:
1: In model.response(mf, "numeric") :
  using type = "numeric" with a factor response will be ignored
2: In Ops.factor(y, z$residuals) : '-' not meaningful for factors

Not a big deal and trying to catch all of the silly things users may do is of
course impossible, but for this one adding a check that the (response) variable
is actually numeric could be useful.

Best,
Wolfgang

--
Wolfgang Viechtbauer, PhD, Statistician | Department of Psychiatry and
Neuropsychology | Maastricht University | PO Box 616 (VIJV1) | 6200 MD
Maastricht, The Netherlands | +31(43)3884170 | https://www.wvbauer.com

>>>>> Viechtbauer, Wolfgang (NP) writes:
> Hi all,
> Just noticed that the error that arises when calling t.test() with factors
could be a bit clearer:
>> t.test(factor(c(3,1,2,4,3,5,4,5)), factor(c(2,1,2,3,4,5)))
> Error in var(x) : Calling var(x) on a factor x is defunct.
>   Use something like 'all(duplicated(x)[-1L])' to test for a
constant vector.
> In addition: Warning message:
> In mean.default(x) :
>   argument is not numeric or logical: returning NA
> Obviously, using factors as input is nonsense, but this might happen on
accident and then the error message could be a bit more on point. Similar for
aov():
>> aov(factor(c(3,1,2,4,3,5,4,5)) ~ factor(c(2,1,2,2,2,1,2,1)))
> Call:
>    aov(formula = factor(c(3, 1, 2, 4, 3, 5, 4, 5)) ~ factor(c(2,
>     1, 2, 2, 2, 1, 2, 1)))
> Error in levels(x)[x] :
>   only 0's may be mixed with negative subscripts
> In addition: Warning messages:
> 1: In model.response(mf, "numeric") :
>   using type = "numeric" with a factor response will be ignored
> 2: In Ops.factor(y, z$residuals) : '-' not meaningful for factors
> Not a big deal and trying to catch all of the silly things users may
> do is of course impossible, but for this one adding a check that the
> (response) variable is actually numeric could be useful.
Indeed.

As always, the question is whether we want to give an error unless
is.numeric, or ensure via as.numeric?

Best
-k
> Best,
> Wolfgang
> --
> Wolfgang Viechtbauer, PhD, Statistician | Department of Psychiatry and
> Neuropsychology | Maastricht University | PO Box 616 (VIJV1) | 6200 MD
> Maastricht, The Netherlands | +31(43)3884170 | https://www.wvbauer.com
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel