lme4: Error in getResponseFormula(form) : "Form" must be a two sided formula

I'm sure I'm being stupid so flame away...

R2.2.1 on Windoze (boohoo) latest updates of packages.

I'm exploring a dataset (land) with three variables looking at an
narrowly unbalanced two group (GROUP) ANCOVA of a randomised
controlled trial analysing endpoint score (SFQ.LOCF.ENDPOINT) entering
the baseline score (SFQ.BASELINE) as covariate and the following work
fine:

res.same <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP,land)
res.diff <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP + SFQ.BASELINE*GROUP,land)
anova(res.same,res.diff)

I try:

lmList(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE | GROUP, land)

Call:
Error in getResponseFormula(form) : "Form" must be a two sided formula

I'm puzzled.  That looks like a two sided formula very like the one in
the help for lme4 (which had been loaded) and the data look OK:

table(land$SFQ.LOCF.ENDPOINT)

 3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 23
 1  1  2  4  8  5 16  9  7 14 18  7 16  9  6  8  4  6  2  3

table(land$SFQ.BASELINE)

 3  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 1  1  3  3  4 11  7  7 10 12  9 16 14  9  8  7  8  6  1  1

table(land$GROUP)

 1  2
87 89

Advice accepted gratefully and flames ruefully!

Chris

--
Chris Evans <chris at psyctc.org>
Consultant Psychiatrist in Psychotherapy, Rampton Hospital;
Research Programmes Director, Nottinghamshire NHS Trust,
Hon. Professor of Psychotherapy, Nottingham University,
Hon. SL Institute of Psychiatry
*** My views are my own and not representative of those institutions ***

Please check which packages you have attached when you call lmList. 
That error message looks as if it is coming from the version of lmList
that is in the nlme package, not the one in lme4.

On 2/6/06, Chris Evans <stats at psyctc.org> wrote:

I'm sure I'm being stupid so flame away...

R2.2.1 on Windoze (boohoo) latest updates of packages.

I'm exploring a dataset (land) with three variables looking at an
narrowly unbalanced two group (GROUP) ANCOVA of a randomised
controlled trial analysing endpoint score (SFQ.LOCF.ENDPOINT) entering
the baseline score (SFQ.BASELINE) as covariate and the following work
fine:

res.same <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP,land)
res.diff <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP + SFQ.BASELINE*GROUP,land)
anova(res.same,res.diff)

I try:

lmList(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE | GROUP, land)

Call:
Error in getResponseFormula(form) : "Form" must be a two sided formula

I'm puzzled.  That looks like a two sided formula very like the one in
the help for lme4 (which had been loaded) and the data look OK:

table(land$SFQ.LOCF.ENDPOINT)

3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 23
1  1  2  4  8  5 16  9  7 14 18  7 16  9  6  8  4  6  2  3

table(land$SFQ.BASELINE)

3  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
1  1  3  3  4 11  7  7 10 12  9 16 14  9  8  7  8  6  1  1

table(land$GROUP)

1  2
87 89

Advice accepted gratefully and flames ruefully!

Chris

--
Chris Evans <chris at psyctc.org>
Consultant Psychiatrist in Psychotherapy, Rampton Hospital;
Research Programmes Director, Nottinghamshire NHS Trust,
Hon. Professor of Psychotherapy, Nottingham University,
Hon. SL Institute of Psychiatry
*** My views are my own and not representative of those institutions ***

Hi, Doug and Chris:

          I just got the same error message with the "lmList" example in lme4:

 >      (fm1 <- lmList(breaks ~ wool | tension, warpbreaks))

Call:
Error in getResponseFormula(form) : "Form" must be a two sided formula

 > sessionInfo()

R version 2.2.1, 2005-12-20, i386-pc-mingw32

attached base packages:
[1] "methods"   "stats"     "graphics"  "grDevices" "utils"     "datasets"
[7] "base"

other attached packages:
      lme4   lattice    Matrix
"0.995-2" "0.12-11" "0.995-5"

library(lme4)

   (fm1 <- lmList(breaks ~ wool | tension, warpbreaks))

sessionInfo()

library(lme4)

(fm1 <- lmList(breaks ~ wool | tension, warpbreaks))

sessionInfo()

          I quit then restarted R and tried the same example in nlme:  It
seemed to work fine.

          Chris, have you tried this in nlme?  If you absolutely need some
feature of lme4, at least you could do this kind of preliminary work in
nlme, then switch to lme4 (after quitting and restarting R to avoid
potential conflicts between nlme and lme4).

          hope this helps.
          spencer graves

Douglas Bates wrote:

Please check which packages you have attached when you call lmList.
That error message looks as if it is coming from the version of lmList
that is in the nlme package, not the one in lme4.

On 2/6/06, Chris Evans <stats at psyctc.org> wrote:

I'm sure I'm being stupid so flame away...

R2.2.1 on Windoze (boohoo) latest updates of packages.

I'm exploring a dataset (land) with three variables looking at an
narrowly unbalanced two group (GROUP) ANCOVA of a randomised
controlled trial analysing endpoint score (SFQ.LOCF.ENDPOINT) entering
the baseline score (SFQ.BASELINE) as covariate and the following work
fine:

res.same <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP,land)
res.diff <- lm(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE + GROUP + SFQ.BASELINE*GROUP,land)
anova(res.same,res.diff)

I try:

lmList(SFQ.LOCF.ENDPOINT ~ SFQ.BASELINE | GROUP, land)

Call:
Error in getResponseFormula(form) : "Form" must be a two sided formula

I'm puzzled.  That looks like a two sided formula very like the one in
the help for lme4 (which had been loaded) and the data look OK:

table(land$SFQ.LOCF.ENDPOINT)

3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 23
1  1  2  4  8  5 16  9  7 14 18  7 16  9  6  8  4  6  2  3

table(land$SFQ.BASELINE)

3  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
1  1  3  3  4 11  7  7 10 12  9 16 14  9  8  7  8  6  1  1

table(land$GROUP)

1  2
87 89

Advice accepted gratefully and flames ruefully!

Chris

--
Chris Evans <chris at psyctc.org>
Consultant Psychiatrist in Psychotherapy, Rampton Hospital;
Research Programmes Director, Nottinghamshire NHS Trust,
Hon. Professor of Psychotherapy, Nottingham University,
Hon. SL Institute of Psychiatry
*** My views are my own and not representative of those institutions ***

Hi all,
I have a basic question. how can i visualize two or more density curves on
the same plot?

ex:
x1<-runif(100,10,80)
x2<-runif(100,1,100)
kernelgraf<-density(x1,kernel = "gaussian", width= 20)
plot(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,1,0), main="")
kernelgraf<-density(x2,kernel = "gaussian", width= 20)
points(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,0,1), main="")

here i am using plot + points, but i do not like the graphical output for
points.
  

can anybody help me?

thanks in advance

Roberto Furlan
University of Turin
  

Le 11.02.2006 19:44, statistical.model at googlemail.com a ??crit :

Hi all,
I have a basic question. how can i visualize two or more density curves on
the same plot?

ex:
x1<-runif(100,10,80)
x2<-runif(100,1,100)
kernelgraf<-density(x1,kernel = "gaussian", width= 20)
plot(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,1,0), main="")
kernelgraf<-density(x2,kernel = "gaussian", width= 20)
points(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,0,1), main="")

here i am using plot + points, but i do not like the graphical output for
points.
 

You are almost there, replace your last command by :

points(kernelgraf, type="l", col="blue")

Romain

PS : Maybe a function points.density setting type="l" would be a good 
thing ?

can anybody help me?

thanks in advance

Roberto Furlan
University of Turin
 

Romain Francois wrote:

Le 11.02.2006 19:44, statistical.model at googlemail.com a ??crit :

Hi all,
I have a basic question. how can i visualize two or more density 
curves on
the same plot?

ex:
x1<-runif(100,10,80)
x2<-runif(100,1,100)
kernelgraf<-density(x1,kernel = "gaussian", width= 20)
plot(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,1,0), main="")
kernelgraf<-density(x2,kernel = "gaussian", width= 20)
points(kernelgraf, xlab="Probability", xlim=c(0,100), ylim=c(0,.1),
col=rgb(0,0,1), main="")

here i am using plot + points, but i do not like the graphical 
output for
points.
 

You are almost there, replace your last command by :

points(kernelgraf, type="l", col="blue")

Romain

PS : Maybe a function points.density setting type="l" would be a good 
thing ?

Or maybe use

lines(kernelgraf, type="l", col="blue")

As I always say, when you want lines, use lines().

Duncan Murdoch