Simple Generalizability Theory for Crossed and Nested Designs

Provides a small, beginner-friendly interface for estimating variance components in simple generalizability theory designs. The package currently supports a fully crossed persons-by-items design, generic balanced crossed designs with one or more additional facets such as raters, occasions, or forms, and a simple items-within-person nested design, along with design-study summaries for relative and absolute decisions. Includes data diagnostics, measurement error intervals, design comparison and cost planning, sensitivity analysis, Gaussian simulation and parametric bootstrap uncertainty estimates for balanced crossed designs.

gtheoryr

gtheoryr is a small R package for simple generalizability theory workflows. It is intentionally modest in scope so it is easy to understand, extend, and prepare for a first CRAN submission.

New in version 0.2.0

Nine functions now connect the G-study to assessment planning:

Read the planning guide for the research sources, assumptions and worked calculations. A runnable example is in inst/examples/planning.R.

The new planning functions reject negative variance estimates unless you explicitly request negative = "zero"; adjusted components are reported. All facets are random. Designs must be balanced, with one observation per cell. The bootstrap supports crossed designs and assumes independent Gaussian effects. No additional package dependencies are required.

The package currently includes:

• gstudy_pxi() for a fully crossed persons-by-items design
• gstudy_crossed() for generic balanced crossed designs with one or more facets
• gstudy_pxif() for a fully crossed persons-by-items-by-facet design
• gstudy_pxir() and gstudy_pxio() as convenience wrappers for raters and occasions
• gstudy_pxiro() as a convenience wrapper for persons-by-items-by-raters-by-occasions designs
• gstudy_nested_ip() for a simple balanced nested items-within-person design
• dstudy_pxi() for relative and absolute decision summaries
• dstudy_crossed() for generic crossed-design D-studies
• dstudy_pxif() for current-design or proposed-design summaries with a third facet
• dstudy_pxir() and dstudy_pxio() as convenience wrappers for raters and occasions
• dstudy_pxiro() for current-design or proposed-design summaries with raters and occasions together
• dstudy_nested_ip() for a simple nested-design D-study
• anova_table(), mean_squares_table(), and variance_components_table() for pulling tidy output tables from a G-study object
• variance_proportions_table() for showing how much each variance component contributes

Install locally

install.packages("path/to/gtheoryr_0.2.0.tar.gz", repos = NULL, type = "source")

Quick example

library(gtheoryr)

scores <- data.frame(
  person = rep(c("P1", "P2", "P3"), each = 3),
  item = rep(c("I1", "I2", "I3"), times = 3),
  score = c(8, 7, 9, 5, 4, 6, 7, 6, 8)
)

gs <- gstudy_pxi(scores, person = "person", item = "item", score = "score")
gs

dstudy_pxi(gs, n_items = 6)

Three-facet example

library(gtheoryr)

scores <- expand.grid(
  person = c("P1", "P2", "P3"),
  item = c("I1", "I2"),
  rater = c("R1", "R2"),
  stringsAsFactors = FALSE
)

scores$score <- c(8, 7, 7, 6, 5, 4, 6, 5, 7, 6, 8, 7)

gs3 <- gstudy_pxif(
  scores,
  person = "person",
  item = "item",
  facet = "rater",
  score = "score",
  facet_name = "rater"
)

gs3
variance_components_table(gs3)
dstudy_pxif(gs3)

Four-facet example

library(gtheoryr)

scores4 <- read.csv(
  system.file("extdata", "crossed_scores_rater_occasion.csv", package = "gtheoryr"),
  stringsAsFactors = FALSE
)

gs4 <- gstudy_pxiro(
  scores4,
  person = "person",
  item = "item",
  rater = "rater",
  occasion = "occasion",
  score = "score"
)

gs4
variance_components_table(gs4)
variance_proportions_table(gs4)
dstudy_pxiro(gs4)

CRAN readiness notes

Before submitting to CRAN, you should:

1. Verify that the maintainer details in DESCRIPTION are correct.
2. Run R CMD check --as-cran gtheoryr.
3. Add a cran-comments.md file summarizing check results.
4. Add tests and a vignette once the API settles down further.