R help - Mar 2015 - Help w/ variable names in loop with lmer

Hi, R users-

I'm estimating random effects models with cross-level interactions; I want
to interact each of a vector of level-1 variables with each of a vector of
level-2 variables.  Here's the code:

===================#create data frame with level-1 variables
k <- as.data.frame(cbind(robo, asalto, secuestro, asesinato))

#create data frame with level-2 variables
l <- as.data.frame(cbind(IDH_IDH, IDH_ingpc, eco_pb, IM_indice, tasa_robo,
hom_tasa, totdelitos1, totdelitos2, total, pri, pan, prd))

#get cross-level interactions

for (i in 1:length(k)) {
for (j in 1:length(l)) {
print(summary(lmer(hrprotcrim ~ k[,i]*l[,j] + (k[,i] | Municipio))))
}
}
=====================
The code works and produces 48 (4 level-1 x 12 level-2) sets of output.
The problem is, the output is illegible because instead of the variable
names, I get the indices:

[output]
=================================Linear mixed model fit by REML
['lmerMod']
Formula: hrprotcrim ~ k[, i] * l[, j] + (k[, i] | Municipio)

REML criterion at convergence: 8801.4

Scaled residuals:
    Min      1Q  Median      3Q     Max
-2.4447 -0.7017 -0.2639  0.6766  3.0835

Random effects:
 Groups    Name        Variance Std.Dev. Corr
 Municipio (Intercept) 1.067868 1.0334
           k[, i]      0.005387 0.0734   1.00
 Residual              2.976150 1.7252
Number of obs: 2163, groups:  Municipio, 180

Fixed effects:
               Estimate Std. Error t value
(Intercept)    2.710847   0.101715  26.651
k[, i]        -0.056720   0.355802  -0.159
l[, j]         0.002701   0.002289   1.180
k[, i]:l[, j]  0.006510   0.006340   1.027

Correlation of Fixed Effects:
            (Intr) k[, i] l[, j]
k[, i]      -0.048
l[, j]      -0.514  0.028
k[,i]:l[,j]  0.034 -0.566 -0.072
=================================
Two questions:

1)  How can I get variable names instead of indices in the above output
2)  How can I estimate this with "mapply" instead of the double loop?

Here's the code for "mapply"

M4 <- mapply(function(k,l){summary(lmer(hrprotcrim ~ k*l + (k |
Municipio)))})

And here's what I get:

list()

I'd be grateful for any pointers.

Best,
David



-- 
Personal Web site:
http://investigadores.cide.edu/crow/

Web site for M?xico, las Am?ricas y el Mundo:
http://mexicoyelmundo.cide.edu/

===================================David Crow, Ph.D.
Profesor-Investigador/Assistant Professor
Director General, *Las Am?ricas y el Mundo*
Divisi?n de Estudios Internacionales
Carretera M?xico-Toluca 3655
Col. Lomas de Santa Fe 01210  M?xico, D.F.
Tel.:  5727-9800, ext. 2152
Fax:  5727-9872
===================================
Conmutador: 5727-98-00 Lada sin costo: 01 800 021 2433 (CIDE) |?

	[[alternative HTML version deleted]]

David Crow <david.crow <at> cide.edu> writes:
> 
> Hi, R users-
> 
> I'm estimating random effects models with cross-level interactions; I
want
> to interact each of a vector of level-1 variables with each of a vector of
> level-2 variables.  Here's the code:
> 
> ===================> #create data frame with level-1 variables
> k <- as.data.frame(cbind(robo, asalto, secuestro, asesinato))
> 
> #create data frame with level-2 variables
> l <- as.data.frame(cbind(IDH_IDH, IDH_ingpc, eco_pb, IM_indice,
tasa_robo,
> hom_tasa, totdelitos1, totdelitos2, total, pri, pan, prd))
> 
> #get cross-level interactions
> 
> for (i in 1:length(k)) {
> for (j in 1:length(l)) {
> print(summary(lmer(hrprotcrim ~ k[,i]*l[,j] + (k[,i] | Municipio))))
> }
> }
> =====================> 
> The code works and produces 48 (4 level-1 x 12 level-2) sets of output.
> The problem is, the output is illegible because instead of the variable
> names, I get the indices:
> 
> [output]
> =================================> Linear mixed model fit by REML
['lmerMod']
> Formula: hrprotcrim ~ k[, i] * l[, j] + (k[, i] | Municipio)
> 
[snip]
> Two questions:
> 
> 1)  How can I get variable names instead of indices in the above output
> 2)  How can I estimate this with "mapply" instead of the double
loop?
> 
> Here's the code for "mapply"
> 
> M4 <- mapply(function(k,l){summary(lmer(hrprotcrim ~ k*l + (k |
> Municipio)))})
> 
> And here's what I get:
> 
> list()
> 
I would suggest appropriate use of ?reformulate

Assume dd is a data frame containing all variables

lev1vars <- c("robo", "asalto", "secuestro",
"asesinato")
lev2vars <- c("IDH_IDH", "IDH_ingpc", ...)
ffun <- function(L1var,L2var) {
    ff <- reformulate(paste(L1var,L2var,sep="*"),
                      paste(L1var,"Municipio",sep="|"),
                      response="hrprotcrim")
    environment(ff) <- parent.frame()  ## trickiness
    return(summary(lmer(ff,data=dd)))
}
mapply(ffun,lev1vars,lev2vars)

?

If you had given a reproducible example I could have tested this ...