Skip to contents

Getting Started with StatsPackage

Source: vignettes/getting-started.Rmd
getting-started.Rmd

StatsPackage is a reusable inference toolkit for confidence intervals, hypothesis tests, power, sample size planning, and categorical-data analysis.

Load the package

Confidence intervals

Use ci_mu() for one-sample or two-sample mean intervals:

ci_mu(xbar = 12.4, n = 15, s = 3.2, quiet = TRUE)
ci_mu(xbar = c(65, 63), n = c(40, 38), s = c(3.0, 2.5), method = "welch", quiet = TRUE)

Use ci_p() for one-sample exact intervals or two-sample difference intervals:

ci_p(x = 56, n = 100, quiet = TRUE)
ci_p(x = c(42, 30), n = c(100, 100), quiet = TRUE)

Hypothesis testing

p_test() supports one-sample exact and two-sample z tests.

p_test(x = 21, n = 100, p0 = 0.20, quiet = TRUE)
p_test(x = c(35, 24), n = c(100, 100), p0 = 0, quiet = TRUE)

Variance-family tests use tail-based two-sided rejection regions:

var_test_chisq(s = 4.2, n = 12, sigma0 = 5, alternative = "two.sided", quiet = TRUE)

Categorical data 

obs <- c(18, 35, 31, 16)
chisq_gof_probs(observed = obs, p = c(0.2, 0.3, 0.3, 0.2), quiet = TRUE)
tab <- matrix(c(22, 18, 14, 26), nrow = 2, byrow = TRUE)
chisq_table(tab, type = "independence", quiet = TRUE)
table_props(tab, margin = "row", quiet = TRUE)

Power and required sample size 

power_z_mu(mu_a = 103, mu0 = 100, sigma = 10, n = 64, quiet = TRUE)
n_required_z_mu(mu_a = 103, mu0 = 100, sigma = 10, beta_target = 0.1, quiet = TRUE)

For local validation before publishing:

devtools::test("StatsPackage -1.0")
devtools::check("StatsPackage -1.0")
source("2.CheatsheetV5.r")