mtable: Comparative Table of Model Estimates

In memisc: Management of Survey Data and Presentation of Analysis Results

Comparative Table of Model Estimates

Description

mtable produces a table of estimates for several models.

Usage

mtable(...,coef.style=getOption("coef.style"),
    summary.stats=TRUE,
    signif.symbols=getOption("signif.symbols"),
    factor.style=getOption("factor.style"),
    show.baselevel=getOption("show.baselevel"),
    baselevel.sep=getOption("baselevel.sep"),
    getSummary=eval.parent(quote(getSummary)),
    float.style=getOption("float.style"),
    digits=min(3,getOption("digits")),
    sdigits=digits,
    show.eqnames=getOption("mtable.show.eqnames",NA),
    gs.options=NULL,
    controls=NULL,
    collapse.controls=FALSE,
    control.var.indicator=getOption("control.var.indicator",c("Yes","No"))
  )
## S3 method for class 'memisc_mtable'
relabel(x, ..., gsub = FALSE, fixed = !gsub, warn = FALSE)

## S3 method for class 'memisc_mtable'
format(x,target=c("print","LaTeX","HTML","delim"),
    ...
    )

## S3 method for class 'memisc_mtable'
print(x,
    center.at=getOption("OutDec"),
    topsep="=",bottomsep="=",sectionsep="-",...)

write.mtable(object,file="",
             format=c("delim","LaTeX","HTML"),...)

## S3 method for class 'memisc_mtable'
toLatex(object,...)

Arguments

...	as argument to mtable: several model objects, e.g. of class lm; as argument to print.memisc_mtable, toLatex.memisc_mtable, write.memisc_mtable: further arguments passed to format.memisc_mtable; as argument to format.memisc_mtable: further arguments passed to format.default; as argument to relabel.memisc_mtable: further arguments passed to dimrename.
coef.style	a character string which specifies the style of coefficient values, whether standard errors, Wald/t-statistics, or significance levels are reported, etc. See coef.style.
summary.stats	if FALSE, no summary statistics are repored. If TRUE then for each object in ... either all summary statistics are reported, or those specified by the option "summary.stats.<cls>", where <cls> is the class of the respective object. This argument may also contain a character vector with the names of the summary statistics to report, or a list of character vectors with names of summary statistics for each object passed as argument in ....
signif.symbols	a named numeric vector to specify the "significance levels" and corresponding symbols. The numeric elements define the significance levels, the attached names define the associated symbols.
factor.style	a character string that specifies the style in which factor contrasts are labled. See factor.style.
show.baselevel	logical; determines whether base levels of factors are indicated for dummy coefficients
baselevel.sep	character that is used to separate the base level from the level that a dummy variable represents
getSummary	a function that computes model-related statistics that appear in the table. See getSummary.
float.style	default format for floating point numbers if no format is specified by coef.style

.

digits	number of significant digits if not specified by the template returned from getCoefTemplate getSummaryTemplate
sdigits	integer; number of digits after decimal dot for summary statistics.
show.eqnames	logical; if TRUE, left-hand sides of equations are (always) shown in the table header; if FALSE, left-hand sides of equations are not shown; if NA, left-hand sides of equations are shown only if left-hand sides differ among models or one of the models has multiple equations.
gs.options	an optional list of arguments passed on to getSummary
controls	an optional formula or character vector that designates "control variables" for which no coefficients are reported, but only whether they are present in the model.
collapse.controls	a logical values; should the report about inclusion of control variables collapsed to a single value? If yes, models should either contain none or all of the control variables.
control.var.indicator	a character vector with to elements; the first element being used to indicate the presence of a control variable or all control variables (if collapse.controls=TRUE), the second element being used otherwise. By default these elements are "Yes" and "No".
x, object	an object of class mtable
gsub, warn, fixed	logical values, see relabel
target	a character string which indicates the target format. Currenlty the targets "print" (see mtable_format_print), "LaTeX" (see mtable_format_latex), "HTML" (see mtable_format_html), and "delim" (see mtable_format_delim) are supported.
center.at	a character string on which resulting values are centered. Typically equal to ".". This is the default when forLaTeX==TRUE. If NULL, reported values are not centered.
topsep	a character string that is recycled to a top rule.
bottomsep	a character string that is recycled to a bottom rule.
sectionsep	a character string that is recycled to seperate coefficients from summary statistics.
file	name of the file where to write to; defaults to console output.
format	character string that specifies the desired format.

Details

mtable constructs a table of estimates for regression-type models. format.memisc_mtable formats suitable for use with output or conversion functions such as print.memisc_mtable, toLatex.memisc_mtable, or write.memisc_mtable.

Value

A call to mtable results in an object of class "mtable" with the following components:

coefficients	a list that contains the model coefficients,
summaries	a matrix that contains the model summaries,
calls	a list of calls that created the model estimates being summarised.

Examples

#### Basic workflow

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

options(summary.stats.lm=c("R-squared","N"))
mtable("Model 1"=lm0,"Model 2"=lm1,"Model 3"=lm2)

options(summary.stats.lm=c("sigma","R-squared","N"))
mtable("Model 1"=lm0,"Model 2"=lm1,"Model 3"=lm2)

options(summary.stats.lm=NULL)

mtable123 <- mtable("Model 1"=lm0,"Model 2"=lm1,"Model 3"=lm2,
    summary.stats=c("sigma","R-squared","F","p","N"))

(mtable123 <- relabel(mtable123,
  "(Intercept)" = "Constant",
          pop15 = "Percentage of population under 15",
          pop75 = "Percentage of population over 75",
            dpi = "Real per-capita disposable income",
           ddpi = "Growth rate of real per-capita disp. income"
  ))

# This produces output in tab-delimited format:
write.mtable(mtable123)

## Not run: 
# This produces output in tab-delimited format:
file123 <- "mtable123.txt"
write.mtable(mtable123,file=file123)
file.show(file123)
# The contents of this file can be pasted into Word
# and converted into a Word table.

## End(Not run)

## Not run: texfile123 <- "mtable123.tex"
write.mtable(mtable123,format="LaTeX",file=texfile123)
file.show(texfile123)
## End(Not run)

#### Examples with UC Berkeley data

berkeley <- Aggregate(Table(Admit,Freq)~.,data=UCBAdmissions)

berk0 <- glm(cbind(Admitted,Rejected)~1,data=berkeley,family="binomial")
berk1 <- glm(cbind(Admitted,Rejected)~Gender,data=berkeley,family="binomial")
berk2 <- glm(cbind(Admitted,Rejected)~Gender+Dept,data=berkeley,family="binomial")

mtable(berk0,summary.stats=c("Deviance","N"))
mtable(berk1,summary.stats=c("Deviance","N"))

mtable(berk0,berk1,berk2,summary.stats=c("Deviance","N"))

mtable(berk0,berk1,berk2,
          coef.style="horizontal",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="stat",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="ci",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="ci.se",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="ci.se.horizontal",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="ci.p.horizontal",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="ci.horizontal",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="all",
          summary.stats=c("Deviance","AIC","N"))
mtable(berk0,berk1,berk2,
          coef.style="all.nostar",
          summary.stats=c("Deviance","AIC","N"))

mtable(by(berkeley,berkeley$Dept,
  function(x)glm(cbind(Admitted,Rejected)~Gender,
        data=x,family="binomial")),
      summary.stats=c("Likelihood-ratio","N"))


mtable(By(~Gender,
  glm(cbind(Admitted,Rejected)~Dept,
        family="binomial"),
        data=berkeley),
      summary.stats=c("Likelihood-ratio","N"))

berkfull <- glm(cbind(Admitted,Rejected)~Dept/Gender - 1,
                      data=berkeley,family="binomial")
relabel(mtable(berkfull),Dept="Department",gsub=TRUE)

#### Array-like semantics

mtable123 <- mtable("Model 1"=lm0,"Model 2"=lm1,"Model 3"=lm2,
    summary.stats=c("sigma","R-squared","F","p","N"))

dim(mtable123)
dimnames(mtable123)
mtable123[c("dpi","ddpi"),
          c("Model 2","Model 3")]

#### Concatention
mt01 <- mtable(lm0,lm1,summary.stats=c("R-squared","N"))
mt12 <- mtable(lm1,lm2,summary.stats=c("R-squared","F","N"))
c(mt01,mt12) # not that this makes sense, but ...
c("Group 1"=mt01,
  "Group 2"=mt12)

memisc documentation built on Oct. 24, 2024, 5:07 p.m.