[R-sig-ME] [FORGED] Re: Export several lme outputs to a single excel file

[Rolf Turner]

On 8/10/19 9:28 PM, Mario Garrido wrote:

> Dear Emmanuel Curis,
> your approach was working perfectly, but at some point w gives me the
> error. when introduced the new column
> I have no problem in running model, the errors appears when introducing
> extra $Md column. I wonder whether the problem is the * of the
> significance, but is not, also is not due to character strings since my
> variables are recognized as factors. I have no missing data in my
> matrix,... And, as I said, problem only arise when I introduced the new
> column
> 
> Thanks in advanxe
>> anova(TempLight_StdzDiff_3trt_earlypeak)-> r15
>> r15
> Analysis of Variance Table
> 
> Response: StdzDiff
>            Df  Sum Sq Mean Sq F value  Pr(>F)
> Trtmnt     2   0.991  0.4953  0.2382 0.78891
> sp         2  13.781  6.8904  3.3134 0.04407 *
> Trtmnt:sp  4   4.123  1.0306  0.4956 0.73898
> Residuals 53 110.217  2.0796
> ---
> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
>> r15$Md<- "TempLight_StdzDiff_3trt_earlypeak"
>> r15
> Analysis of Variance Table
> 
> Response: StdzDiff
>            Df  Sum Sq Mean Sq F value  Pr(>F) Md
> Trtmnt     2   0.991  0.4953  0.2382 0.78891
> sp         2  13.781  6.8904  3.3134 0.04407
> Trtmnt:sp  4   4.123  1.0306  0.4956 0.73898
> Residuals 53 110.217  2.0796
> Warning message:
> In data.matrix(x) : NAs introducidos por coerción

This is a "generic" problem; it is not peculiar to your model nor to 
models fitted using lme, or other mixed modelling software.

Consider the following example:

set.seed(42)
x <- 1:20
y <- rnorm(20)
fit <- lm(y ~ x)
m   <- anova(fit)
m$newColumn <- "yeeeeks"
m

This produces:

> Analysis of Variance Table
> 
> Response: y
>           Df Sum Sq Mean Sq F value  Pr(>F) newColumn
> x          1  4.113  4.1130  2.5865 0.12518          
> Residuals 18 28.624  1.5902                          
> Warning message:
> In data.matrix(x) : NAs introduced by coercion

The "reason" is that m is (in the first instance) of class "anova" and 
there are (not unreasonably) certain restrictions as to how you can 
treat an object of this class.

A work-around to get something like what you appear to want could be:

set.seed(42)
x <- 1:20
y <- rnorm(20)
fit <- lm(y ~ x)
m   <- anova(fit)
m   <- cbind(m,newColumn=c("yeeeks",rep("",nrow(m)-1)))
m

However m is now of class "data.frame" whence it is printed by the 
method print.data.frame() rather than print.anova().  Consequently NAs 
show up in the output of the print method:

           Df      Sum Sq    Mean Sq    F value    Pr(>F) newColumn
x          1  0.04176282 0.04176282 0.03784897 0.8479256    yeeeks
Residuals 18 19.86132473 1.10340693         NA        NA

You could just live with those NAs, or you could convert the "F value" 
and "Pr(>F)" columns from numeric to character mode and replace the NAs 
by null strings "".

HTH

cheers,

Rolf Turner

-- 
Honorary Research Fellow
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276