rm_covsum {reportRmd}

Outputs a descriptive covariate table

Description

Returns a data frame corresponding to a descriptive table.

Usage

rm_covsum(
  data,
  covs,
  maincov = NULL,
  caption = NULL,
  tableOnly = FALSE,
  covTitle = "",
  digits = 1,
  digits.cat = 0,
  nicenames = TRUE,
  IQR = FALSE,
  all.stats = FALSE,
  pvalue = TRUE,
  effSize = FALSE,
  p.adjust = "none",
  unformattedp = FALSE,
  show.tests = FALSE,
  testcont = c("rank-sum test", "ANOVA"),
  testcat = c("Chi-squared", "Fisher"),
  full = TRUE,
  include_missing = FALSE,
  percentage = c("column", "row"),
  dropLevels = TRUE,
  excludeLevels = NULL,
  numobs = NULL,
  fontsize,
  chunk_label
)

Arguments

data

dataframe containing data

covs

character vector with the names of columns to include in table

maincov

covariate to stratify table by

caption

character containing table caption (default is no caption)

tableOnly

Logical, if TRUE then a dataframe is returned, otherwise a formatted printed object is returned (default).

covTitle

character with the names of the covariate (predictor) column. The default is to leave this empty for output or, for table only output to use the column name 'Covariate'.

digits

number of digits for summarizing mean data

digits.cat

number of digits for the proportions when summarizing categorical data (default: 0)

nicenames

boolean indicating if you want to replace . and _ in strings with a space

IQR

boolean indicating if you want to display the inter quantile range (Q1,Q3) as opposed to (min,max) in the summary for continuous variables

all.stats

boolean indicating if all summary statistics (Q1,Q3 + min,max on a separate line) should be displayed. Overrides IQR.

pvalue

boolean indicating if you want p-values included in the table

effSize

boolean indicating if you want effect sizes included in the table. Can only be obtained if pvalue is also requested. Effect sizes calculated include Cramer's V for categorical variables, Cohen's d, Wilcoxon r, or Eta-squared for numeric/continuous variables.

p.adjust

p-adjustments to be performed

unformattedp

boolean indicating if you would like the p-value to be returned unformatted (ie not rounded or prefixed with '<'). Best used with tableOnly = T and outTable function. See examples.

show.tests

boolean indicating if the type of statistical test and effect size used should be shown in a column beside the pvalues. Ignored if pvalue=FALSE.

testcont

test of choice for continuous variables,one of rank-sum (default) or ANOVA

testcat

test of choice for categorical variables,one of Chi-squared (default) or Fisher

full

boolean indicating if you want the full sample included in the table, ignored if maincov is NULL

include_missing

Option to include NA values of maincov. NAs will not be included in statistical tests

percentage

choice of how percentages are presented, one of column (default) or row

dropLevels

logical, indicating if empty factor levels be dropped from the output, default is TRUE.

excludeLevels

a named list of covariate levels to exclude from statistical tests in the form list(varname =c('level1','level2')). These levels will be excluded from association tests, but not the table. This can be useful for levels where there is a logical skip (ie not missing, but not presented). Ignored if pvalue=FALSE.

numobs

named list overriding the number of people you expect to have the covariate

fontsize

PDF/HTML output only, manually set the table fontsize

chunk_label

only used if output is to Word to allow cross-referencing

Details

Comparisons for categorical variables default to chi-square tests, but if there are counts of <5 then the Fisher Exact test will be used and if this is unsuccessful then a second attempt will be made computing p-values using MC simulation. If testcont='ANOVA' then the t-test with unequal variance will be used for two groups and an ANOVA will be used for three or more. The statistical test used can be displayed by specifying show.tests=TRUE.

Effect size can be obtained when p-value is requested.

Further formatting options are available using tableOnly=TRUE and outputting the table with a call to outTable.

Value

A character vector of the table source code, unless tableOnly=TRUE in which case a data frame is returned

References

Ellis, P.D. (2010) The essential guide to effect sizes: statistical power, meta-analysis, and the interpretation of research results. Cambridge: Cambridge University Press.doi:10.1017/CBO9780511761676

Lakens, D. (2013) Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4; 863:1-12. doi:10.3389/fpsyg.2013.00863

See Also

covsum,fisher.test, chisq.test, wilcox.test, kruskal.test, anova, cramer_v, eta_squared, and outTable

Examples

data("pembrolizumab")
rm_covsum(data=pembrolizumab, maincov = 'orr',
covs=c('age','sex','pdl1','tmb','l_size','change_ctdna_group'),
show.tests=TRUE)

# To Show Effect Sizes
rm_covsum(data=pembrolizumab, maincov = 'orr',
covs=c('age','sex'),
effSize=TRUE)

# To make custom changes or change the fontsize in PDF/HTML
tab <- rm_covsum(data=pembrolizumab,maincov = 'change_ctdna_group',
covs=c('age','sex','pdl1','tmb','l_size'),show.tests=TRUE,tableOnly = TRUE)
outTable(tab, fontsize=7)

# To return unformatted p-values
tab <- rm_covsum(data=pembrolizumab, maincov = 'orr',
covs=c('age','sex','pdl1','tmb','l_size','change_ctdna_group'),
show.tests=TRUE,unformattedp=TRUE,tableOnly=TRUE)
outTable(tab,digits=5)
outTable(tab,digits=5, applyAttributes=FALSE) # remove bold/indent