expRse: Standard Errors of Retest- or Cross-Rater-Adjusted...

In funfield-lab/fancyr: Fancy Statistics for Correlational (r) Analyses

Standard Errors of Retest- or Cross-Rater-Adjusted Correlation

Description

From Rene Mottus. He writes:

'I got to this approach by applying some intuitions (like the mean of two parallel correlations is their individual standard errors' mean divided by sqrt(2), and then adjusted by the correlation) and then tinkering (the initial results based on intuition alone were slightly off, what needed tinkering was taking the roots and square roots at the right times). It agrees well with the simulation results. If someone can tell me why then that would be even better.'

Usage

expRse(rxy, rxxyy, n)

Arguments

rxy	the geometric mean of the two cross-item, cross-rater correlations (in your case, cross-item, cross-time correlations)
rxxyy	the geometric mean of the two cross-rater correlations of the same item (in your case, retest correlations)
n	number of observations associated with both rxy and rxxyy

Value

standard error of reliability-adjusted rxy correlation

funfield-lab/fancyr documentation built on Nov. 21, 2023, 2:42 p.m.