| Type: | Package |
| Title: | Krippendorff's Alpha for Multi-Valued Data |
| Version: | 0.6.0 |
| Description: | Calculate Krippendorff's alpha for multi-valued data using the methods introduced by Krippendorff and Craggs (2016) <doi:10.1080/19312458.2016.1228863>. Nominal, ordinal, interval, and ratio data types are supported, with option to create bootstrapped estimates of alpha. |
| License: | AGPL (≥ 3) |
| Encoding: | UTF-8 |
| URL: | https://github.com/therealcfdrake/mvalpha |
| BugReports: | https://github.com/therealcfdrake/mvalpha/issues |
| Depends: | R (≥ 4.2.0) |
| RoxygenNote: | 7.3.3 |
| LazyData: | true |
| Imports: | stats, utils, rlang, Rdpack, Rcpp, arrangements, stringr |
| Suggests: | testthat (≥ 3.0.0) |
| RdMacros: | Rdpack |
| LinkingTo: | Rcpp |
| Config/testthat/edition: | 3 |
| SystemRequirements: | OpenMP |
| NeedsCompilation: | yes |
| Packaged: | 2026-06-16 02:35:37 UTC; corie |
| Author: | Corie Drake  [aut,
cre, cph] |
| Maintainer: | Corie Drake <therealcfdrake@gmail.com> |
| Repository: | CRAN |
| Date/Publication: | 2026-06-16 02:50:02 UTC |
Published Examples
Description
These data represent examples found in the original paper describing the calculation of multi-valued Krippendorff's alpha by Krippendorff and Craggs (2016).
Usage
ex_table3
ex_table8a
ex_table8b
ex_table8c
ex_table9a
ex_table9b
ex_table9c
Format
Each is a multi-valued nominal set with observers as columns and units as rows.
Source
doi:10.1080/19312458.2016.1228863
References
Krippendorff K, Craggs R (2016). “The Reliability of Multi-Valued Coding of Data.” Communication Methods and Measures, 10(4), 181–198. doi:10.1080/19312458.2016.1228863.
Generate Multi-Valued Data Sets
Description
Generate Multi-Valued Data Sets
Usage
generate_mv_data(
type = "nominal",
n_units = 10,
n_observers = 3,
n_labels = 5,
tpr = 0.8,
fpr = 0.01,
card_pmf = c(0.3, 0.4, 0.2, 0.1),
p_missing = 0
)
Arguments
type |
Data type. One of "nominal", "ordinal", "interval", or "ratio". |
n_units |
Number of units (rows) in data. |
n_observers |
Number of observers (cols) in data. |
n_labels |
Number of possible labels which could be applied to the data. |
tpr |
True Positive Rate. To generate the data, first a latent set of labels. |
fpr |
False Positive Rate. To generate the data, first a latent set of labels. |
card_pmf |
Probability mass function describing the cardinality of observed label sets. Length should be less than or equal to |
p_missing |
Proportion of observations that are randomly missing, indicated by |
Value
A matrix with n_units rows and n_observers columns with list elements.
Examples
generate_mv_data()
Estimate Multi-Valued Krippendorff's Alpha
Description
mvalpha() calculates Krippendorff's alpha statistic when multi-valued observers are
allowed to apply multiple values to an observation.
Usage
mvalpha(data, type = "nominal", verbose = TRUE, n_boot = NULL, n_threads = 1)
Arguments
data |
a data frame containing a list column for each observer. Each row represents
an observation unit, and each cell contains a vector of 0 to |
type |
a string describing the data type of the label set. This can be "nominal", "ordinal", "interval", or "ratio" and is used to select the appropriate distance metric. |
verbose |
a logical value which toggles whether status updates are printed to the console while alpha is being calculated. |
n_boot |
an integer representing the number of bootstrap estimates to calculate
for mvDo. The default, |
n_threads |
an integer describing the number of cores to allocate to parallelization. |
Value
An object of class mvalpha
References
Krippendorff K, Craggs R (2016). “The Reliability of Multi-Valued Coding of Data.” Communication Methods and Measures, 10(4), 181–198. doi:10.1080/19312458.2016.1228863.
Examples
library(mvalpha)
### replicate example from Table 3 in Krippendoff and Craggs (2016) with bootstrapped estimates
# View data
ex_table3
# # Estimate alpha
# x <- mvalpha(ex_table3, verbose = TRUE, n_boot = 500)
#
# # View result
# x
#
# # View the unique values observed in the data
# x$values
#
# # View the unique labels used to code the data
# x$labels
#
# # Histogram of bootstrapped estimates
# hist(x$bootstrap_mvalpha)
Create new mvalpha class object
Description
Wrapper for creating mvalpha class object.
Usage
new_mvalpha(
mvalpha,
type,
mvDo,
mvDe,
bootstrap_mvalpha,
unique_cardinalities,
units,
observers,
labels,
values,
values_by_unit,
dist_CK,
p_CK,
data
)
Arguments
mvalpha |
Multi-valued alpha estimate |
type |
a string describing the data type of the label set. This can be "nominal", "ordinal", "interval", or "ratio" and is used to select the appropriate distance metric. |
mvDo |
Observed disagreement |
mvDe |
Expected disagreement |
bootstrap_mvalpha |
Bootstrap estimates of mvalpha |
unique_cardinalities |
Numeric vector of the unique cardinalities observed in the data |
units |
Names of units |
observers |
Names of observers |
labels |
Unique labels used in data |
values |
Unique values used in data |
values_by_unit |
Table of values by unit |
dist_CK |
Distance matrix for label sets C and K |
p_CK |
Probability matrix for label sets C and K |
data |
a data frame containing a list column for each observer. Each row represents
an observation unit, and each cell contains a vector of 0 to |
Value
an mvalpha object
Print mvalpha class object
Description
Print generic
Usage
## S3 method for class 'mvalpha'
print(x, ...)
Arguments
x |
mvalpha object |
... |
additional parameters |
Value
invisibly returns the alpha estimate of an mvalpha object
Efficient Set Operations
Description
Find the intersection and set difference(s) of two sets all at once and more
efficiently than calling base::intersect() and base::setdiff() separately.
Based on this stackoverflow answer https://stackoverflow.com/a/72631719
Usage
set_ops(A, B, type)
Arguments
A, B |
sets (vectors) of elements |
type |
a string describing the data type of the label set. This can be "nominal", "ordinal", "interval", or "ratio" and is used to select the appropriate distance metric. |