vcmeta (version 1.5.0)

replicate.agree: Compares and combines G-index of agreement in original and follow-up studies

Description

This function computes adjusted Wald confidence intervals from an original study and a follow-up study where the effect size is a G-index of agreement. Adjusted Wald confidence intervals for the difference and average effect size are also computed. The confidence level for the difference is 1 – 2*alpha, which is recommended for equivalence testing. As a measurement of agreement, the G-index is usually preferred to Cohen's kappa.

Usage

replicate.agree(alpha, f1, n1, f2, n2, k)

Value

A 4-row matrix. The rows are:

• Row 1 summarizes the original study

• Row 2 summarizes the follow-up study

• Row 3 estimates the difference in G-indicies

• Row 4 estimates the average G-index

The columns are:

• Estimate - MLE of G-index (single study, difference, average)

• SE - standard error of adjusted estimate

• LL - lower limit of the adjusted confidence interval

• UL - upper limit of the adjusted confidence interval

Arguments

alpha

alpha level for 1-alpha confidence

f1

number of objects rated in agreement in original study

n1

sample size (number of objects) in original study

f2

number of objects rated in agreement in follow-up study

n2

sample size (number of objects) in follow-up study

k

number of rating categories

References

Bonett2022vcmeta

Examples

replicate.agree(.05, 85, 100, 160, 200, 2)

# Should return:
#                       Estimate         SE         LL        UL
# Original:                 0.70 0.07252105  0.53093828 0.8152156
# Follow-up:                0.60 0.05661961  0.47726289 0.6992077
# Original - Follow-up:     0.10 0.09159681 -0.05844824 0.2428784
# Average:                  0.65 0.04579840  0.55040374 0.7299302