bfttest: Bayesian Sequential t-Test

In mrzdcmps/changeofevidence: Change of Evidence

Bayesian Sequential t-Test

Description

Calculates sequential Bayes Factors for t-tests (one-sample, paired, or independent samples), showing how evidence evolves as data accumulates. Uses the BFDA package for Bayes Factor calculations.

Usage

bfttest(
  x = NULL,
  y = NULL,
  formula = NULL,
  data = NULL,
  alternative = c("two.sided", "less", "greater"),
  mu = 0,
  prior.loc = 0,
  prior.r = 0.1,
  nstart = "auto",
  exact = TRUE
)

Arguments

x	A numeric vector of observations or a formula for independent samples test
y	Optional numeric vector for paired samples test
formula	Formula of form response ~ group where group has exactly 2 levels
data	Optional data frame containing the variables in formula
alternative	Direction of alternative hypothesis: "two.sided", "greater", or "less"
mu	Null hypothesis value (default = 0)
prior.loc	Location parameter for Cauchy prior (default = 0)
prior.r	Scale parameter for Cauchy prior (default = 0.1)
nstart	Minimum observations before first BF calculation ("auto" or >= 2)
exact	Logical. If TRUE, calculates BF for all points

Value

A list of class "seqbf" containing:

• t-value: Sequential t-statistics

• p-value: Sequential p-values

• BF: Sequential Bayes Factors

• test type: "one-sample", "paired", or "independent"

• prior: List with prior distribution details

• sample size: Number of observations

• alternative: Chosen alternative hypothesis

Examples

# One-sample test
x <- rnorm(30, 0.5, 1)
result1 <- bfttest(x, alternative = "greater")

# Independent samples test
group <- rep(c("A", "B"), each = 15)
values <- c(rnorm(15), rnorm(15, 0.5))
df <- data.frame(values = values, group = group)
result2 <- bfttest(values ~ group, data = df)

# Paired samples test
pre <- rnorm(20)
post <- pre + rnorm(20, 0.5)
result3 <- bfttest(pre, post)

mrzdcmps/changeofevidence documentation built on Feb. 27, 2025, 3:10 a.m.