Question

I’ve had some success using mtable from the memisc package to pretty-print results (APSR-style) for several regressions side-by-side, but I’m having trouble getting the same command to work for models fit with lrm() from the MASS package and lmrob() from the robustbase package.

Works great:

lm0 <- lm(sr ~ pop15 + pop75,              data = LifeCycleSavings)
lm1 <- lm(sr ~                 dpi + ddpi, data = LifeCycleSavings)
lm2 <- lm(sr ~ pop15 + pop75 + dpi + ddpi, data = LifeCycleSavings)

mtable123 <- mtable("Model 1"=lm0,"Model 2"=lm1,"Model 3"=lm2)

Fails at the mtable() command ("Error in qt(p, df, lower.tail, log.p): Non-numeric argument to mathematical function"):

rlm0 <- rlm(sr ~ pop15 + pop75,              data = LifeCycleSavings)
rlm1 <- rlm(sr ~                 dpi + ddpi, data = LifeCycleSavings)
rlm2 <- rlm(sr ~ pop15 + pop75 + dpi + ddpi, data = LifeCycleSavings)

mtable123 <- mtable("Model 1"=rlm0, "Model 2"=rlm1, "Model 3"=rlm2)

Fails at the mtable() command ("Error in UseMethod("getSummary"): no applicable method for 'getSummary' applied to an object of class "lmrob""):

lmrob0 <- lmrob(sr ~ pop15 + pop75,              data = LifeCycleSavings)
lmrob1 <- lmrob(sr ~                 dpi + ddpi, data = LifeCycleSavings)
lmrob2 <- lmrob(sr ~ pop15 + pop75 + dpi + ddpi, data = LifeCycleSavings)

mtable123 <- mtable("Model 1"=lmrob0, "Model 2"=lmrob1, "Model 3"=lmrob2)

I’m open to suggestions that don’t involve mtable, but I’m looking for plain-text or tab-separated output (in other words, not LaTeX).

Answer 1

The documentation at ?getSummary in the memisc package describes how you can extend the mtable function to be used on other model types.

In particular, you need to create an appropriate method for getSummary and then create a summary template using setSummaryTemplate. So, for instance, something like this works:

getSummary.rlm <- function(obj, alpha = 0.5,...){
    obj_summary <- summary(obj)

    coefficients <- cbind(coef(obj_summary),dt(coefficients[,3],df = obj_summary$df))
    #You can add the CI code yourself, this is a placeholder:
    coefficients <- cbind(coefficients,matrix(NA,nrow(coefficients),2))
    colnames(coefficients) <- c('est','se','stat','p','lwr','upr')

    sumstat <- c(sigma = obj_summary$sigma,df = obj_summary$df[2],stddev = obj_summary$stddev)

    return(list(coef = coefficients,sumstat = sumstat))
}

setSummaryTemplate(rlm = c(sigma = "($sigma:#)",df = "($df:#)",stddev = "($stddev:#)"))

mtable123 <- mtable("Model 1"=rlm0,"Model 2"=rlm1,"Model 3"=rlm2)
mtable123

Calls:
Model 1: NULL
Model 2: NULL
Model 3: NULL

=======================================
             Model 1  Model 2  Model 3 
---------------------------------------
(Intercept)  28.528*   6.497** 28.945**
             (7.535)  (1.225)  (7.602) 
pop15        -0.434**          -0.473**
             (0.149)           (0.150) 
pop75        -1.595            -1.655  
             (1.059)           (1.120) 
dpi                    0.001*  -0.000  
                      (0.001)  (0.001) 
ddpi                   0.462    0.385  
                      (0.206)  (0.203) 
---------------------------------------
sigma          3.739    4.383    3.523 
df            47.000   47.000   45.000 
stddev         3.998    4.098    3.931 
=======================================