Garrett Ranking Analysis and Visualization

Performs Garrett ranking analysis of respondent-ranked items such as constraints, problems, factors, or priorities. The package converts respondent rankings into Garrett scores, calculates mean Garrett scores and final ranks and provides methods for summarizing,tabulating and visualizing ranking results. It also provides Kendall's coefficient of concordance for assessing the degree of agreement among respondents. Garrett ranking does not accommodate tied ranks and Kendall's coefficient of concordance is likewise computed for untied ranking data.For more details see Garrett and Woodworth (1969) < https://books.google.com/books?id=aoqSmQEACAAJ> and Buragohain and Dubey (2021) .

GarrettRank

Overview

GarrettRank provides functions for conducting Garrett ranking analysis. The package is designed to calculate Garrett scores from respondent-level ranking data and provide summarized ranking results in a convenient format.

The package also provides functions for generating tables, summaries, visualizations, and Kendall's coefficient of concordance for assessing agreement among respondents. Visualizations are built using ggplot2 and pheatmap.

Installation

The package can be installed using:

install.packages("GarrettRank")

Then load the package:

library(GarrettRank)

Functions

The main functions provided by GarrettRank are:

• garrett_rank() — performs Garrett ranking analysis.
• garrett_table() — produces the standard Garrett ranking/percent position conversion table used in the scoring calculation.
• kendall_w() — calculates Kendall's coefficient of concordance.

The package also provides S3 methods for:

• summary()
• plot()

Example Dataset

The package includes an example dataset called garrett_example.

The dataset contains:

• 50 respondents
• 22 constraints (C1 to C22)
• respondent names or serial numbers in the first column
• ranks assigned to each constraint
• no tied ranks

The dataset can be loaded using:

data("garrett_example")
head(garrett_example)

Basic Example

library(GarrettRank)
data("garrett_example")

# Perform Garrett ranking
result <- garrett_rank(garrett_example, respondent = "Respondent")

# View the result
result

# Summary
summary(result)

# Plot the results
plot(result)

data accepts a data.frame, matrix, CSV file path, or Excel file path. Each row should represent one respondent and each column one factor/constraint, with every respondent assigning each rank from 1 to the number of factors exactly once. respondent specifies the column name or position containing respondent IDs, so it can be excluded before analysis.

Plot types

plot() supports several type options, covering both item-level ranking summaries and respondent-level agreement patterns:

Additional arguments for type = "cluster":

• k — number of clusters to highlight (e.g., k = 3).
• scale — scaling applied before clustering (e.g., scale = "row").
• show_numbers — whether to display values on the cluster heatmap.

plot(result, type = "bar")
plot(result, type = "lollipop")
plot(result, type = "dot")
plot(result, type = "line")
plot(result, type = "heatmap")
plot(result, type = "cluster")
plot(result, type = "cluster", k = 3)
plot(result, type = "cluster", scale = "row", show_numbers = TRUE)
plot(result, type = "contribution")

Kendall's Coefficient of Concordance

Kendall's coefficient of concordance can be calculated using:

kendall_result <- kendall_w(garrett_example, respondent = "Respondent")
kendall_result
summary(kendall_result)

Citation

If you use GarrettRank in your research, please cite the package as:

Varghese, B. B., et al. GarrettRank: Garrett Ranking Analysis in R.

A formal citation will be added once the package is published. Once available, it can be retrieved with:

citation("GarrettRank")

Authors

• Blesson B. Varghese
• Adarsh V. S.
• Bhavana Sajeev
• Azhanuo Rutsa
• Joe Shiney M. A.

License

This package is licensed under the GPL-3 license.