nonparboot {npboottprm}

Nonparametric Bootstrap Test with Pooled Resampling for Small Sample Sizes

Description

This function performs a nonparametric bootstrap test with pooled resampling for small sample sizes, as described in Dwivedi et al. (2017). It supports t-tests (independent and paired) and F-tests (one-way ANOVA), with a user-specified number of bootstrap resamples.

Usage

nonparboot(
  data,
  x,
  y = NULL,
  grp = NULL,
  nboot,
  test = c("t", "pt", "F"),
  conf.level = 0.95,
  seed = NULL,
  na_rm = FALSE
)

Arguments

data

A data frame containing the variables to be analyzed.

x

A character string specifying the column in 'data' to be used as the primary variable.

y

An optional character string specifying the column in 'data' to be used as the second variable for paired t-tests. Default is NULL.

grp

An optional character string specifying the column in 'data' to be used as the grouping variable for independent t-tests and F-tests. Default is NULL.

nboot

An integer specifying the number of bootstrap resamples to perform.

test

A character string specifying the type of test to perform. Must be one of "t", "pt", or "F" for independent t-test, paired t-test, or F-test, respectively. Default is "t".

conf.level

A numeric value between 0 and 1 specifying the confidence level for confidence intervals. Default is 0.95.

seed

An optional value interpreted as an integer to set the seed for the random number generator, for reproducibility. Default is NULL (no seed).

na_rm

Remove observations with missing values. Default is FALSE.

Value

A list with the following components:

• p.value: The p-value of the test.

• orig.stat: The test statistic calculated from the original data.

• ci.stat: The confidence interval for the test statistic from the bootstrap distribution.

• bootstrap.stat.dist: The distribution of the test statistic values from the bootstrap resamples.

• effect.size: The effect size (mean difference or eta-squared) calculated from the original data.

• ci.effect.size: The confidence interval for the effect size from the bootstrap distribution.

• bootstrap.effect.dist: The distribution of effect size values from the bootstrap resamples.

References

Dwivedi AK, Mallawaarachchi I, Alvarado LA (2017). "Analysis of small sample size studies using nonparametric bootstrap test with pooled resampling method." Statistics in Medicine, 36 (14), 2187-2205. https://doi.org/10.1002/sim.7263

Examples

# Example usage of nonparboot
np_res <- nonparboot(iris, x = "Sepal.Length", grp = "Species", nboot = 1000, test = "F")
print(np_res$p.value)