[R-sig-ME] Help!

[Ben Bolker]

Eric ADJAKOSSA <ericadjakossah at ...> writes:

>  [snip]
> I am about to
> use the book "Mixed-effects models in S and S-plus" as part of my
> master's thesis. I adjusted a linear mixed effects model, but I can
> not make some graphs to validate this model. I use the software R
> (version 2.13.0). In fact R displays errors that I can not correct for
> several days. 

  [SNIP]

> groupCSP <- groupedData(conCSP ~ NBDORM_f | VILLAG, data =
> subset(Tori_util_sd, select = c(conCSP,TOTTDR_f, TOTTDR, VILLAG
> ,IND_VEGET_f , NBDORM_f, EAU, TMUR, NO, LONG, LAT)))
> 
> cm <- lme(log(conCSP) ~ NBDORM_f + EAU*IND_VEGET_f + TOTTDR , data =
> groupCSP, random = ~ 1 | VILLAG, na.action = na.omit)
> 
> > plot( cm, VILLAG~resid(.), abline = 0 )
> Erreur dans `[<-.data.frame`(`*tmp*`, , ".y", value = c(2L, 2L, 2L, 2L, 8L,  :
>  replacement has 479 rows, data has 

  Hard to answer without a reproducible example, but I suspect the
problem is with NAs in your data that are being automatically
removed in one place and not in the other.  I would try subsetting
and na.omit()ing your data first, i.e. 

subdat <- subset(Tori_util_sd, select = c(conCSP,TOTTDR_f,
    TOTTDR, VILLAG ,IND_VEGET_f , NBDORM_f, 
    EAU, TMUR, NO, LONG, LAT)

subdat <- na.omit(subdat)

groupCSP <- groupedData(...,data=subdat)

etc.