BayesAiken: Bayesian Content Validity Coefficients with Credibility
Intervals
Bayesian inference for the six rating coefficients proposed by
Aiken (1985) <doi:10.1177/0013164485451012> and Aiken (1989)
<doi:10.1177/0013164489492003>: V (content validity), H (homogeneity),
C (congruence), R (repeatability), A (rater agreement) and I (item
coherence). All six are expressed as functionals of the population
distribution of rating categories, so a single Dirichlet-Multinomial model
yields their full posterior by push-forward. This treats each judge as one
categorical observation and avoids the pseudo-replication of Beta-Binomial
formulations. R is modelled on the c-by-c joint table because it is not a
functional of the marginal distribution. Provides classical estimates with
the correct parity correction, highest density and equal-tailed
credibility intervals, posterior threshold probabilities, prior
sensitivity analysis and a friendly step-by-step verbose report.
Documentation:
Downloads:
Linking:
Please use the canonical form
https://CRAN.R-project.org/package=BayesAiken
to link to this page.