ctable function

Cross-Tabulation

Cross-tabulation for a pair of categorical variables with either row, column, or total proportions, as well as marginal sums. Works with numeric, character, as well as factor variables.

ctable(
  x,
  y,
  prop = st_options("ctable.prop"),
  useNA = "ifany",
  totals = st_options("ctable.totals"),
  style = st_options("style"),
  round.digits = st_options("ctable.round.digits"),
  justify = "right",
  plain.ascii = st_options("plain.ascii"),
  headings = st_options("headings"),
  display.labels = st_options("display.labels"),
  split.tables = Inf,
  na.val = st_options("na.val"),
  rev = "none",
  dnn = c(substitute(x), substitute(y)),
  chisq = FALSE,
  OR = FALSE,
  RR = FALSE,
  weights = NA,
  rescale.weights = FALSE,
  ...
)

Arguments

• x: First categorical variable - values will appear as row names.

• y: Second categorical variable - values will appear as column names.

• prop: Character. Indicates which proportions to show: r

  (rows, default), c (columns), t (total), or n

  (none). Default value can be changed using st_options, option ctable.prop.

• useNA: Character. One of ifany (default), no , or always . This argument is passed on as is to table, or adapted for xtabs when weights are used.

• totals: Logical. Show row and column totals. Defaults to TRUE but can be set globally with st_options, option ctable.totals.

• style: Character. Style to be used by pander. One of simple (default), grid , rmarkdown , or jira . Can be set globally with st_options.

• round.digits: Numeric. Number of significant digits to keep. Defaults to 1. To change this default value, use st_options, option ctable.round.digits.

• justify: Character. Horizontal alignment; one of l (left), c (center), or r (right, default).

• plain.ascii: Logical. Used by pander; when TRUE, no markup characters are generated (useful when printing to console). Defaults to TRUE unless style = 'rmarkdown', in which case it is set to FALSE automatically. To change the default value globally, use st_options.

• headings: Logical. Show heading section. TRUE by default; can be set globally with st_options.

• display.labels: Logical. Display data frame label in the heading section. TRUE by default, can be changed globally with st_options.

• split.tables: Numeric. pander argument that specifies how many characters wide a table can be. Inf by default.

• na.val: Character. For factors and character vectors, consider this value as NA. Ignored if there are actual NA values or if it matches no value / factor level in the data. NULL by default.

• rev: Character. Dimension(s) to reverse for calculation of risk/odds ratios. One of rows / r , columns / c , both / b , or none / n

  (default). See details.

• dnn: Character vector. Variable names to be used in output table. In most cases, setting this parameter is not required as the names are automatically generated.

• chisq: Logical. Display chi-square statistic along with p-value.

• OR: Logical or numeric. Set to TRUE to show odds ratio with 95

  confidence interval, or specify confidence level explicitly (e.g., .90). CI's are calculated using Wald's method of normal approximation.

• RR: Logical or numeric. Set to TRUE to show risk ratio (also called relative risk with 95

  confidence level explicitly (e.g. .90). CI's are calculated using Wald's method of normal approximation.

• weights: Numeric. Vector of weights; must have the same length as x.

• rescale.weights: Logical. When TRUE, a global constant is applied so that the sum of counts equals nrow(x). FALSE by default.

• ...: Additional arguments passed to pander or format.

Returns

A list containing two matrices, cross_table and proportions. The print method takes care of assembling figures from those matrices into a single table. The returned object has classes summarytools and list , unless stby is used, in which case we have an object of class stby .

Details

For risk ratios and odds ratios, the expected structure of the contingency table is as follows (using No as reference):

Outcome
 Exposure      Yes     No
  Yes          a       b
  No           c       d

The rev parameter allows for different structures; use either one of rows , columns , or both to indicate which dimension(s) to reverse in order to match that structure. This does not affect display.

Note

Markdown does not fully support multi-header tables; until such support is available, the recommended way to display cross-tables in .Rmd documents is to use method=render. See package vignettes for examples.

Examples

data("tobacco")
ctable(tobacco$gender, tobacco$smoker)

# Use with() to simplify syntax
with(tobacco, ctable(gender, smoker))

# Show column proportions, without totals
with(tobacco, ctable(smoker, diseased, prop = "c", totals = FALSE))

# Simple 2 x 2 table with odds ratio and risk ratio
with(tobacco, ctable(smoker, diseased, totals = FALSE, headings = FALSE,
                     prop = "r", OR = TRUE, RR = TRUE))

# Grouped cross-tabulations
with(tobacco, stby(data = list(x = smoker, y = diseased), 
                   INDICES = gender, FUN = ctable))

## Not run:

ct <- ctable(tobacco$gender, tobacco$smoker)

# Show html results in browser
print(ct, method = "browser")

# Save results to html file
print(ct, file = "ct_gender_smoker.html")

# Save results to text file
print(ct, file = "ct_gender_smoker.txt")
## End(Not run)

See Also

table, xtabs

Author(s)

Dominic Comtois, dominic.comtois@gmail.com