eq.tost.CI: Independent samples equivalence test using confidence...

In cribbie/equivalencetests: Equivalence Tests

Description

The two one-sided test for independent samples can be expressed using the confidence inclusion principle. If the confidence interval around a mean difference is within the equivalence bounds, then the mean difference is considered to be practically meaningless.

Usage

eq.tost.CI(x, y, ei, alpha = 0.05, na.rm = FALSE)

## S3 method for class 'eq.tost.CI'
print(x, ...)

## S3 method for class 'eq.tost.CI'
plot(x, y = NULL, ...)

Arguments

x	object of class eq.tost.CI
y	a NULL object
ei	equivalence interval
alpha	desired alpha level
...	additional arguments
dat	an N x 3 matrix or data.frame containing raw data used to compute the correlation matrix between variables. The input may also be a 1 x 3 vector of correlations (r12, r13, and r23, respectively) and requires a sample size input (N)
n	sample size when dat input is a vector of correlations
x	object of class eq.tost.CI

Value

returns a list containing the p-value, confidence interval, and statistical decision

Author(s)

Rob Cribbie [e-mail] and Phil Chalmers [e-mail]

Examples

## Not run: 
#raw data
set.seed(1234)
x <- rnorm(100)
y <- rnorm(100)
dat <- data.frame(dv=c(x,y),  group=c(rep('g1', length(x)), rep('g2', length(y)) ))

# Plot raw data distribution

ggplot(dat,aes(x=dv)) + 
  geom_histogram(data=subset(dat,group == 'g1'),fill = 'red', alpha = 0.2) +
  geom_histogram(data=subset(dat,group == 'g2'),fill = 'blue', alpha = 0.2)
  
# Run test
ei <- 0.50
res <- eq.tost.CI(x,y, ei=ei, alpha=.05)
plot(res)

## End(Not run)

cribbie/equivalencetests documentation built on May 14, 2019, 11:33 a.m.