FAQ

[R] Warning messages: not meaningful for factors

Dr Shrink
Sep 5, 2010 at 2:17 pm
Dear Experts,

I need to include the repeated structure in our data set object,
recall.sums.df, before using gls function.
Thus I used groupedData.
But I encountered error messages which may mean '*' is not not meaningful
factor.

Please let me know what I have to do.
Thanks,

Jeong
recall.sums.df[0:10, ]
recall.values recall.ind subj replication hemi region group
1 17.515 rL_GM_T1 1 1 1 1 1
2 18.830 rL_GM_T1 2 1 1 1 1
3 16.477 rL_GM_T1 3 1 1 1 1
4 20.905 rL_GM_T1 4 1 1 1 1
5 18.005 rL_GM_T1 5 1 1 1 1
6 19.533 rL_GM_T1 6 1 1 1 1
7 17.126 rL_GM_T1 7 1 1 1 1
8 19.910 rL_GM_T1 8 1 1 1 1
9 17.881 rL_GM_T1 9 1 1 1 1
10 21.159 rL_GM_T1 10 1 1 1 1
longs <- groupedData(recall.values~region* hemi* group* replication |
subj, data = recall.sums.df)
Warning messages:
1: In Ops.factor(region, hemi) : * not meaningful for factors
2: In Ops.factor(region * hemi, group) : * not meaningful for factors
3: In Ops.factor(region * hemi * group, replication) :
* not meaningful for factors
reply

Search Discussions

1 response

  • Dieter Menne at Sep 5, 2010 at 4:06 pm

    Dr Shrink wrote:
    I need to include the repeated structure in our data set object,
    recall.sums.df, before using gls function.
    Thus I used groupedData.
    But I encountered error messages which may mean '*' is not not meaningful
    factor.
    I know that when reading Pinheiro/Bates one has the expression that
    groupedData is required, but I always found that concept more confusing than
    helpful (and it has been abandoned in nlm4). Things work much more
    transparent without grouped data.

    I suggest that you first the numeric data to factors, e.g.

    recall.sums.df$subj = as.factor(recall.sums.df$subj)

    and then try again without using grouped data. However, it looks too me that
    you might be better of with lme anyway, probably with random= ~1|subj. gls
    is a very useful and underused tool, but probably not for this data set.
    I also suggest that you start getting the syntax right with + instead of *
    first. You model could easily be strongly underspecified, which also lead to
    nasty error messages.

    Dieter





    --
    View this message in context: http://r.789695.n4.nabble.com/Warning-messages-not-meaningful-for-factors-tp2527349p2527426.html
    Sent from the R help mailing list archive at Nabble.com.

Related Discussions

Discussion Navigation
viewthread | post

2 users in discussion

Dieter Menne: 1 post Dr Shrink: 1 post