agreement: Raw and Chance-corrected Agreement

In dsnavega/osteometrika: osteometrika: an R package for human osteological data analysis

Description

Raw and Chance-corrected Agreement

Usage

agreement(x, weights = "linear", random = T, digits = 4)

Arguments

x	an n by data.frame containing n observations performed by m observers for a given attribute of interest.
weights	a character defining the weighting scheme for ordered factors. options are 'linear' or 'quadratic' weights. Default is 'linear'
random	a logical defining if the probability of expected agreement should be computed from totally random allocation model or from the marginals (as in Cohen's Kappa). Default is TRUE (totally random).
digits	number of decimal places (rounding)

Details

Current implementation generalizes the application of weighting schemes (ordered factors) to raw agreement computation and generalizes chance-corrected agreement to an arbitrary number of raters/rating sessions. Statistical significance is computed from a binomial test against expected probability of agreement and not assuming no agreement (0) as the null hypothesis.

Value

a list with the following components (ordered by relevance):

agreement	a vector with raw and chance-corrected agreement estimates
balanced	balanced agreement as the mean value of detailed agreement
detailed	level-specific agreement of a factor
expected	probability of expected agreement
p.value	significance probability of a binomial test
n	number of valid observations/ratings
m	number of observers/raters
k	number of levels
probability	a list of withprobabilities used to compute agreement
ordered	a logical indicating if ordered factors were analysed
scheme	weighting scheme applied
weighted	a logical indicating if analysis is weighted

Author(s)

David Navega

dsnavega/osteometrika documentation built on April 2, 2020, 9:36 a.m.