descr: Create a descriptive statistics table

In imbi-heidelberg/DescrTab: Creates a table with descriptive statistics

Description

Generate a table of descriptive statistics with p-values obtained in tests for difference between the groups.

Usage

descr(
  dat,
  group,
  var.names,
  percent.vertical = T,
  data.names = T,
  nonparametric = c(),
  landscape = F,
  pos.pagebr = NULL,
  paired = F,
  var.equal = T,
  correct.cat = F,
  correct.wilcox = T,
  silent = T,
  p.values = T,
  group.min.size = F,
  group.non.empty = F,
  cat.non.empty = F,
  n.or.miss = "n",
  adaptive.miss = T,
  group.miss = F,
  t.log = c(),
  index = T,
  create = "knitr",
  digits.m = 1,
  digits.sd = 2,
  digits.qu = c(),
  digits.minmax = 1,
  digits.p = c(1),
  q.type = 2,
  default.unordered.unpaired.test = "Chisq"
)

Arguments

dat	Data frame. The data set to be analyzed. Can contain continuous or factor (also ordered) variables.
group	Vector of the grouping variable.
var.names	Optional. Vector of names to be used in the table for the analyzed variables.
percent.vertical	Logical. Should "vertical" percentages for categorical variables be provided?
data.names	Logical. If var.names are specified, should names as saved in dat be added in brackets?
nonparametric	Logical or vector of indices. If logical / vector of indices then all / only these continuous variables will be tested using non-parametric methods.
landscape	Logical. Should the table be in landscape? Only useful if you want create a "pdf"- or "knitr"-document in the following. (see pos.pagebr)
pos.pagebr	Vector of positions of the pagebreak in tex (or pdf). This is a bit fuzzy. It is the number of lines after a pagebreak should be done. If it is not specified, 45 will be used for "landscape=FALSE" and 30 will be used for "landscape=TURE". Only useful if you want know the number for a pagebreak when you create a "pdf"- or "knitr"-document in the following.
paired	Logical. Should paired tests be applied? The groups must have the same length.
var.equal	Logical. Should variances be assumed to be equal when applying t-tests?
correct.cat	Logical. Should correction be used in chi-sqared tests (see chisq.test)
correct.wilcox	Logical. Should correction be used in wilcoxon tests (see wilcox.test)
silent	Logical. Should intermediate stages be shown (more for technical reasons)?
p.values	Logical. Should calculate p-values? If you won't p-values index were set to FALSE.
group.min.size	For each variable, a p-value is only calculated if each non-empty group contains at least group.min.size observations for that variable.
group.non.empty	For each variable, a p-value is only calculated if each group contains at least one observation for that variable.
cat.non.empty	For categorical variables a p-value is only calculated if each category is non-empty.
n.or.miss	Should the number of observations, missings for continuous variables, and/or missings for categorical variables be provided ("n", "miss", "miss.cat")? Combinations are allowed.
adaptive.miss	Should the missing row be automatically omitted if there are not missings?
group.miss	Logical. Schould add a column for the Missings in group?
t.log	Vector of indices: The variables for which the log of the original data should be used when testing for a difference between the groups.
index	Logical. Should the tests used be labeled by footnotes? Only usefull if "p-values" in which.col.
create	Which output document should be produced in the following step (one of "pdf", "tex", "knitr", "word" or "R").
digits.m	Number of digits for presentation in the table: For mean.
digits.sd	Number of digits for presentation in the table: For standard deviation.
digits.qu	Vector of numbers of digits for presentation in the table: For quantiles (if no value is specified it will be tried to provide a reasonable presentation).
digits.minmax	Number of digits for presentation in the table: For minimum and maximum.
digits.p	Vector with numbers of digits for presentation in the table: For percentages. First vector element is number of digits for the first variable, second element for second variable and so on.
q.type	Integer between 1 and 9 that selects a quantile algorithm.
default.unordered.unpaired.test	Any of c("Chisq", "Fisher_exact", "Fisher_boschloo"). Chooses the default test for categorical, unordered, unpaired variables.

Value

Depending on the value of the create parameter either pdf, word, tex, R or an file optimized for use in connection with knitr will be created containing the descriptive statistics table with the speak for the document to create in the following. For example you choose create="pdf" then the table is written in TeX-Code. Attention: the table has no caption and numbers of observations per group.

Author(s)

Lorenz Uhlmann, Csilla van Lunteren

See Also

med.new
inqur
minmax
f.r
formatr
m.cat
m.cont
p.cat
p.cont

Examples

## Not run: 
##Dataset with two groups
infert
attach(infert)

#is.factor(education)
#is.factor(age)
#is.factor(parity)
#is.factor(induced)
#is.factor(case)
#infert$case<-as.factor(case)
#case<-as.factor(case)

#is.factor(spontaneous)
#is.factor(stratum)
#is.factor(pooled.stratum)

#we use case as Grouping variable

#Version 1
descr(dat = infert, group = case,
var.names = c("education", "age", "parity", "induced", "spontaneous", "stratum", "pooled.stratum"),
create = "word")
#Version 2
group <- case
dat <- infert[,-5]
descr(dat = dat, group = group,
var.names = c("education", "age", "parity", "induced", "spontaneous", "stratum", "pooled.stratum"),
create = "word")


##Dataset with more then two groups

ChickWeight
attach(ChickWeight)
#is.factor(weight)
#is.factor(Time)
#is.factor(Chick)
#is.factor(Diet)

#we use Diet as Grouping variable

#Version 1
descr(dat = ChickWeight, group = Diet, var.names = c("weight", "Time", "Chick"), create = "word")

#Version 4
group <- Diet
dat <- ChickWeight[,-4]
descr(dat = dat, group = group, var.names = c("weight", "Time", "Chick"), create = "word")

## End(Not run)

imbi-heidelberg/DescrTab documentation built on June 29, 2021, 2:07 p.m.