arsenal (version 3.3.0)

tableby.control: Control settings for tableby function

Description

Control test and summary settings for the tableby function.

Usage

tableby.control(test = TRUE, total = TRUE, test.pname = NULL,
  numeric.simplify = FALSE, cat.simplify = FALSE,
  ordered.simplify = FALSE, date.simplify = FALSE,
  numeric.test = "anova", cat.test = "chisq", ordered.test = "trend",
  surv.test = "logrank", date.test = "kwt", test.always = FALSE,
  numeric.stats = c("Nmiss", "meansd", "range"), cat.stats = c("Nmiss",
  "countpct"), ordered.stats = c("Nmiss", "countpct"),
  surv.stats = c("Nmiss", "Nevents", "medSurv"),
  date.stats = c("Nmiss", "median", "range"), stats.labels = list(Nmiss
  = "N-Miss", Nmiss2 = "N-Miss", meansd = "Mean (SD)", medianrange =
  "Median (Range)", median = "Median", medianq1q3 = "Median (Q1, Q3)", q1q3
  = "Q1, Q3", iqr = "IQR", range = "Range", countpct = "Count (Pct)",
  Nevents = "Events", medSurv = "Median Survival", medTime =
  "Median Follow-Up", medianmad = "Median (MAD)"), digits = 3L,
  digits.count = 0L, digits.pct = 1L, digits.p = 3L,
  format.p = TRUE, conf.level = 0.95, chisq.correct = FALSE,
  simulate.p.value = FALSE, B = 2000, times = 1:5, ...)

Arguments

test

logical, telling tableby whether to perform tests of x variables across levels of the group variable.

total

logical, telling tableby whether to calculate a column of totals across group variable.

test.pname

character string denoting the p-value column name in summary.tableby. Modifiable also with modpval.tableby.

numeric.simplify, date.simplify

logical, tell tableby whether to condense numeric/date output to a single line. NOTE: this only simplifies to one line if there is only one statistic reported, such as meansd. In particular, if Nmiss is specified and there are missings, then the output is not simplified.

cat.simplify, ordered.simplify

logical, tell tableby whether to remove the first level of the categorical/ordinal variable if binary. If TRUE, only the summary stats of the second level are reported (unless there's only one level, in which case it's reported). NOTE: this only simplifies to one line if there is only one statistic reported, such as countpct. In particular, if Nmiss is specified and there are missings, then the output is not simplified.

numeric.test

name of test for numeric RHS variables in tableby: anova, kwt (Kruskal-Wallis). If no LHS variable exists, then a mean is required for a univariate test.

cat.test

name of test for categorical variables: chisq, fe (Fisher's Exact)

ordered.test

name of test for ordered variables: trend

surv.test

name of test for survival variables: logrank

date.test

name of test for date variables: kwt

test.always

Should the test be performed even if one or more by-group has 0 observations? Relevant for kwt and anova.

numeric.stats

summary statistics to include for numeric RHS variables within the levels of the group LHS variable. Options are N, Nmiss, Nmiss2, mean, meansd, median, q1q3, medianq1q3, iqr, range, medianrange, or other R built-in or user-written functions.

cat.stats

summary statistics to include for categorical RHS variables within the levels of the group LHS variable. Options are N, Nmiss, Nmiss2, count, countpct, countrowpct, countcellpct, or other R built-in or user-written functions.

ordered.stats

summary statistics to include for categorical RHS variables within the levels of the group LHS variable. Options are N, Nmiss, count, countpct, or other R built-in or user-written functions.

surv.stats

summary statistics to include for time-to-event (survival) RHS variables within the levels of the group LHS variable. Options are Nevents, medsurv, NeventsSurv, NriskSurv, medTime, rangeTime.

date.stats

stats functions to perform for Date variables: Nmiss, median, range, medianrange, q1q3, medianq1q3, or other R built-in or user-written functions.

stats.labels

A named list of labels for all the statistics function names, where the function name is the named element in the list and the value that goes with it is a string containing the formal name that will be printed in all printed renderings of the output, e.g., list(countpct="Count (Pct)").

digits

Number of decimal places for numeric values.

digits.count

Number of decimal places for count values.

digits.pct

Number of decimal places for percents.

digits.p

Number of decimal places for p-values.

format.p

Logical, denoting whether to format p-values. See "Details", below.

conf.level

Numeric, denoting what confidence level to use for confidence intervals. (See, e.g., binomCI)

chisq.correct

logical, correction factor for chisq.test

simulate.p.value

logical, simulate p-value for categorical tests (fe and chisq)

B

number of simulations to perform for simulation-based p-value

times

A vector of times to use for survival summaries.

...

additional arguments.

Value

A list with settings to be used within the tableby function.

Details

All tests can be turned off by setting test to FALSE. Otherwise, test are set to default settings in this list, or set explicitly in the formula of tableby.

If format.p is FALSE, digits.p denotes the number of significant digits shown. The p-values will be in exponential notation if necessary. If format.p is TRUE, digits.p will determine the number of digits after the decimal point to show. If the p-value is less than the resulting number of places, it will be formatted to show so.

See Also

anova, chisq.test, tableby, summary.tableby, tableby.stats.

Examples

# NOT RUN {
set.seed(100)
## make 3+ categories for Response
mdat <- data.frame(Response=c(0,0,0,0,0,1,1,1,1,1),
                   Sex=sample(c("Male", "Female"), 10,replace=TRUE),
                   Age=round(rnorm(10,mean=40, sd=5)),
                   HtIn=round(rnorm(10,mean=65,sd=5)))

## allow default summaries in RHS variables, and pass control args to
## main function, to be picked up with ... when calling tableby.control
outResp <- tableby(Response ~ Sex + Age + HtIn, data=mdat, total=FALSE, test=TRUE)
outCtl <- tableby(Response ~ Sex + Age + HtIn, data=mdat,
                  control=tableby.control(total=TRUE, cat.simplify=TRUE,
                  cat.stats=c("Nmiss","countpct"),digits=1))
summary(outResp, text=TRUE)
summary(outCtl, text=TRUE)
# }