StatEdu Studio

StatEdu Studio is a local Shiny application for assumption-guided statistical analysis and publication-ready result tables.

The app runs on the user's own Windows PC and opens in a local browser session. Data are analyzed locally and are not sent to an external server.

All statistical analyses use CRAN packages only.

Current Version

Current public version: 1.2.0

Current development version: 1.2.0

Version 1.2.0 is the public release line for local data import, data editing, assumption-guided statistical analyses, sample-size/effect-size calculators, latent-analysis workflows, and publication-ready HTML result output. It includes the 1.2.0 fixes for Excel import review, selected-data preview, Latent Mplus output folder handling, and packaged desktop startup behavior. Detailed development history is available in About > Version History.

Public 1.2.0 includes all analysis menus except the Mediation / Moderation Custom Model canvas. In the public build, direct analysis save controls remain visible but only HTML save is enabled by default; t-test / ANOVA is the exception and keeps HTML, figure, PDF, Excel, and Add result enabled. t-test / ANOVA results added to the Result collection can be exported from Result as Excel or Word.

Current Scope

Local Windows launcher through StatEdu_Studio.bat
Local data import for SPSS SAV, SAS, Stata, Excel XLS/XLSX, CSV, and DAT files
Cloud-synced file handling by copying data files to a temporary local read path before import
Data workflow with file loading, variable selection, measurement-level review, variable labels, and categorical value labels
Data Editor tools for coding error checks, Likert conversion, missing-value handling, reverse coding, calculated variables, formula-based variable transformation, recoding, and renaming
Frequencies / descriptives for categorical and continuous variables
Cross-tabulation analysis for binary, ordered, and categorical variables
t-test / ANOVA with normality checks, homoscedasticity checks, post-hoc options, effect sizes, trend options, and nonparametric fallbacks
ANCOVA with automatic or warning-only selection among standard ANCOVA, Robust ANCOVA (HC3), Ranked ANCOVA, and Interaction ANCOVA, plus Levene / Brown-Forsythe / Breusch-Pagan / fitted-value White-style variance checks, slope-homogeneity diagnostics, complete-case reporting, linearity plots, and influence sensitivity analysis
Standalone nonparametric tests using Mann-Whitney U and Kruskal-Wallis workflows
Paired tests and repeated-measures paired workflows for two or more repeated measurements
Standalone nonparametric paired tests using Wilcoxon signed-rank and Friedman workflows
Correlation analysis with automatic method selection, p-value / confidence interval output, method notes, reason notes, scatter plot matrices, and heatmaps
Factor analysis and principal component analysis with Pearson and polychoric matrix options, diagnostics, plots, and score-saving helpers
Reliability analysis for scale and item-level summaries
Linear regression with assumption-guided OLS, HC3 robust standard errors, bootstrap confidence intervals, and combined HC3 plus bootstrap output
Hierarchical regression with block-wise model comparison
Logistic regression for binary, ordered, and categorical dependent variables
Generalized linear models for independent-observation Gaussian, binary logistic, Gamma, and count outcomes, including Poisson versus negative-binomial screening, robust standard-error options, missing-data sensitivity engines, offset/exposure handling, SCI-style diagnostics, publication notes, reporting checklists, and suggested manuscript text
Penalized regression helpers for severe multicollinearity cases
Standalone sample size, power, and effect size calculators with method notes and references
Result saving to HTML, PDF, figures, and accumulated Result collections
Result collection export to HTML and PDF

For the full current method inventory, see docs/ANALYSIS_METHODS_EN.md.

Runtime Environment

Tested development environment: R 4.5.3 on Windows
App framework: Shiny local app
Package source: declared runtime and analysis dependencies are CRAN packages
Execution model: local browser session on 127.0.0.1; data remain on the user's PC

Some installed package binaries may have been built under a newer patch-level R version than the runtime R version. These build warnings are informational unless package loading or validation fails.

R Packages

Area	Packages	Role in StatEdu Studio
App UI	`shiny`, `DT`, `htmltools`, `markdown`	Shiny app shell, interactive tables, HTML helpers, and About documentation rendering
Data import	`haven`, `readr`, `readxl`, `openxlsx`	SAV, SAS, Stata, CSV, DAT, XLS, and XLSX import
Settings and data helpers	`jsonlite`, `xml2`, `rvest`, `callr`	Settings serialization, HTML/XML processing, and background R process support
Regression diagnostics	`car`, `lmtest`, `sandwich`, `nortest`, `boot`	Type II/III ANCOVA tables, Levene-style variance checks, Breusch-Pagan test, HC3 robust standard errors, Lilliefors normality test, and bootstrap inference
Linear / generalized models	`MASS`, `nnet`, `lmtest`, `sandwich`, `geepack`, `mice`, `lme4`, `lmerTest`, `plm`	GLM robust inference, ordered logistic, multinomial, and model-support utilities
Penalized regression	`glmnet`	Ridge, LASSO, and Elastic Net helper analyses
Post-hoc and group comparison	`agricolae`	Multiple-comparison procedures used in ANOVA-style workflows
Reliability, factor analysis, and correlations	`psych`, `polycor`	Reliability coefficients, factor/PCA helpers, polychoric/polyserial/tetrachoric correlation support
Sample size and power	`longpower`, `WebPower`, `TOSTER`	Cluster trial / SEM power and exact TOST equivalence calculations
Report export	`officer`, `flextable`, `openxlsx`	Report table support

Local Run

Install R.
Unzip the StatEdu Studio folder.
Double-click StatEdu_Studio.bat.

The app will open at 127.0.0.1 in the default browser. The launcher searches for Rscript.exe, installs missing runtime packages through R when needed, and starts the Shiny app on the local PC.

Validation

Version 1.2.0 carries forward the stabilization validation suite and adds release gating for public save controls and the custom model canvas. Public validation coverage includes calculators, data import, data editing, cross-tabulation, correlation auto-selection, factor analysis / PCA, logistic analysis and UI, paired guard handling, p-value formatting, regression coefficient output, GLM output, complex-sample workflows, longitudinal / panel workflows, latent-analysis workflow checks, and t-test / ANOVA guard handling. Effect-size comparisons use effectsize as a validation reference where its definitions match the app calculation; effectsize is not required at runtime.

Run the stabilization validation suite from the repository root before merging or packaging:

powershell -ExecutionPolicy Bypass -File scripts\validate_stabilization.ps1
powershell -ExecutionPolicy Bypass -File scripts\validate_stabilization.ps1 -Full

Run the Shiny and Electron smoke checks before preparing a release candidate:

powershell -ExecutionPolicy Bypass -File scripts\release_preflight.ps1
powershell -ExecutionPolicy Bypass -File scripts\smoke_shiny_app.ps1
powershell -ExecutionPolicy Bypass -File scripts\smoke_electron_release.ps1 -SkipUnpackedChecks

After Electron packaging is complete, run the full packaged-output preflight:

powershell -ExecutionPolicy Bypass -File scripts\release_preflight.ps1 -FullElectronSmoke

After automated checks pass, complete docs/RELEASE_MANUAL_QA.md and keep the completed QA record with the release notes and validation artifacts.

Use individual scripts/validate_*.R files only when iterating on a focused module before running the stabilization suite.

Citation

If you use StatEdu Studio in your research, please cite:

LEE, I. H. (2026). StatEdu Studio (Version 1.2.0) [Computer software]. https://doi.org/10.22934/statedu.studio

Product site: https://studio.statedu.com/

Development Model

This project is developed privately and released publicly after validation. Public releases should include source code, documentation, example data, and validation notes.

Feature overview