table_coef: Plotting Coefficients

In gentok/estvis: Visualize estimates from various models

Description

Drawing the coefficient table from model object(s). The wrapper function of texreg.

Usage

table_coef(
  m,
  format = "screen",
  file.name = NULL,
  dir = NULL,
  m.names = NULL,
  order.variable = "original",
  vcov.est = NULL,
  robust.type = "HC1",
  cluster.var = NULL,
  boot.sims = 300,
  boot.seed = 578,
  boot.ncores = 1,
  digits = 3,
  drop.intercept = FALSE,
  drop.intercept.names = "(Intercept)",
  drop.variable.names = NULL,
  caption = NULL,
  custom.variable.names = NULL,
  footnote.psymbol = TRUE,
  custom.footnote = NULL,
  show.table = TRUE,
  booktabs = TRUE,
  dcolumn = FALSE,
  use.packages = FALSE,
  stars = c(0.001, 0.01, 0.05, 0.1),
  symbol = "dagger",
  caption.above = TRUE,
  single.row = FALSE,
  float.pos = "ht!!",
  fontsize = "footnotesize",
  encoding_from = NULL,
  encoding_to = NULL,
  include.deviance = FALSE,
  include.smooth = FALSE,
  ...
)

Arguments

m	Single or the list of model object.
format	The format for the table output (characater). The default is "screen" (The R Output. Cannot be saved.). Other options are "tex" (LaTex table), "html" (HTML table) and "doc" (HTML table saved as Microsof Word file).
file.name	The name of the file to be saved (character). The default is NULL (The output not saved to file). If extension is provided, it will automatically overwrite format.
dir	The directory of the file to be saved. Only applicable when file.name is not NULL.
m.names	The set of names that identifies each element in m. Considered if m is a list of models. The length of the vector must correspond with the length of m. If NULL (default), each element i is temporarily named as Model i.
order.variable	Order of coefficients in the plot(character/numeric vector). "original" (default) preserves the original order of the variables. Alternatively, you can specify the order of variables by numeric vector (applied after drop.intercept and drop.variable.names are applied, thus you don't need the names for dropped variables).
vcov.est	Single or a list of alternative variance-covariance matrix. Each element must be one of raw variance-covariance matrix, "robust", "boot", or NULL (default). If "robust", robust standard error (also see robust.type). If "cluster", cluster robust standard error (also see cluster.var). if "boot", bootstrapped standard error calculated by Boot function is used. Ignored if NULL (default) or the m is not model object. Must have the same length as m if it is a list.
robust.type	The type of robust standard error (applied only when vcov.est=="robust").
cluster.var	Single or a list of vector, matrix, or data.frame of cluster variables, where each column is a separate variable. Alternatively, a formula specifying the cluster variables to be used (see details in cluster.vcov. Applied only when vcov.est=="cluster".)
boot.sims	Number of iterations if bootstrap is used.
boot.seed	Random number seed if bootstrap is used.
boot.ncores	Number of cores to parallelize bootstrap. The default is 1. Use "auto" to automatically detect number of cores in computer.
digits	Rounding decimal points for values in the table.
drop.intercept	Drop the intercept from the plot (boulean). If FALSE (default), intercept included in the plot.
drop.intercept.names	The name(s) of intercept (character/character vector). Needed if drop.intercept is TRUE. This is used to identify and eliminate intercept variables from the output. Default value is '(Intercept)'.
drop.variable.names	The name(s) of additional variables to drop (character/character vector) from the ouput. The default is NULL.
caption	Table title caption (character). The default is to include no title caption. (Not applied if format=="screen")
custom.variable.names	List of alternative variable names in the output (character vector). The default is NULL. This is applied AFTER drop.intercept and drop.variable.names are applied, thus you don't need the names for dropped variables.
footnote.psymbol	Whether to add p-value symbols in the footnote.
custom.footnote	Custom footnote (character). The default is NULL. If assigned, footnote are added to the plot by plot_footnote function, and the function exports gtable object. Note that gtable object is less customizable than ggplot object. If it is also the case that footnote.gof == TRUE, custom footnote will be added as the new line after the GOF footnote.
show.table	Return the "screen" table at the end of function (boulean). Always returned if format=="screen". The default is TRUE.
booktabs	See texreg.
dcolumn	See texreg.
use.packages	See texreg.
stars	See texreg.
symbol	See texreg.
caption.above	See texreg.
single.row	See texreg.
float.pos	See texreg.
fontsize	See texreg.
encoding_from	If encoding conversion is required, what is the orignal encoding? (Only used when format=="tex".)
encoding_to	If encoding conversion is required, what is the encoding to be coverted to? (Only used when format=="tex".)
include.deviance	See extract.
include.smooth	See extract.
...	Other options in texreg or extract function.

Value

Regression table to be shown in console if format=="screen" or show.table==TRUE, NULL if else.

Examples

 
## Load Data
library(pscl)
data(vote92)

## Recode Variables
vote92$voteBush <- as.numeric(
  factor(vote92$vote,levels=c("Clinton","Bush")))*1 - 1
vote92$bushdis <- sqrt(vote92$bushdis)
vote92$clintondis <- sqrt(vote92$clintondis)

vote92$w <- abs(rnorm(nrow(vote92)))

## Estimate Logistic Regression
fm <- formula(voteBush ~ dem + rep +
               clintondis + bushdis +
               persfinance + natlecon)
m <- glm(fm, data = na.omit(vote92[,c(all.vars(fm),"w","perotdis")]),
         family = binomial("logit"), weights=w)

## Basic Table
table_coef(m)

## Export in Tex
table_coef(m, format = "tex")

## Show in Single Row
table_coef(m, single.row = TRUE)

## Try Different Standard Errors
table_coef(list(m,m,m,m), 
           m.names = c("Standard","Robust(HC1)","Cluster Robust","Bootstrapped"),
           vcov.est = list(NULL,"robust","cluster","boot"),
           cluster.var = na.omit(vote92[,c(all.vars(fm),"w","perotdis")])$perotdis)

## Estimate Model by Male and Female Subset
m_male <- glm(fm, data = vote92[vote92$female==0,],
              family = binomial("logit"))
m_female <- glm(fm, data = vote92[vote92$female==1,],
                family = binomial("logit"))

## Table with Two Models
table_coef(list(m_male,m_female))

## Add Custom Variable Names
vn <- c("(Intercept)",
        "Democrat","Republican",
        "Ideological Distance from Clinton",
        "Ideological Distance from Bush",
        "Retrospective Personal Finance",
        "Retrospective National Economy")
table_coef(list(m_male, m_female), 
           m.names = c("Male", "Female"),
           custom.variable.names = vn)

## Add Title (Only Tex/HTML)
table_coef(list(m_male, m_female), format="tex",
           m.names = c("Male", "Female"),
           caption = "Vote for Bush (1992)",
           custom.variable.names = vn)

## Omit Some Variables
table_coef(list(m_male, m_female), 
           m.names = c("Male", "Female"),
           drop.variable.names = c("dem","natlecon"),
           custom.variable.names = vn[-c(2,7)])

## Reorder Variables So That Intercept Comes at the end
table_coef(list(m_male, m_female), 
           m.names = c("Male", "Female"),
           drop.variable.names = c("dem","natlecon"),
           order.variable = c(5, seq(1,4,1)),
           custom.variable.names = vn[-c(2,7)])


## Interaction Model
fm2 <- formula(voteBush ~ dem + rep +
                 clintondis + bushdis +
                 persfinance + natlecon + 
                 dem*clintondis)
m2 <- glm(fm2, data = vote92,
          family = binomial("logit"), weights=w)
table_coef(list(m,m2), 
           custom.variable.names = c(vn,"Democrat * Distance from Clinton"),
           order.variable=c(1,2,4,5,6,7,8,3))

gentok/estvis documentation built on April 2, 2020, 1:58 p.m.