measureR 0.0.7

Focus: the new HTML interface shared with projectLSA 0.2.0.

• Fully redesigned interface: collapsible icon sidebar, top view tabs (grouped where there are many), compact Excel-like tables, a decimal-separator switch and a colour / black-and-white switch at the top right.
• Every table has Export Excel and Copy for Word (APA style, highlights kept); every plot can be copied as an image.
• Black-and-white mode uses line types and hatching; plots use rounded bars.
• Word reports (.docx) for all five modules (content validity, CTT, EFA, CFA/SEM, IRT): objectives, method, results, APA 7 tables, figures and references built from what was cited.
• Report form with optional AI drafting; AI model settings and a floating assistant with a guide for each view.
• R console as a drawer at the top right; a single progress indicator replaces the pop-ups.
• IRT module regrouped, with item- and theta-invariance tests, local dependence and further views.
• Version and date badge on the home page.

measureR 0.0.6

Focus: redesigned Classical Test Theory (CTT) module, plus an update to the CFA/SEM analysis menu.

CFA/SEM module: data analysis menu

• The estimator settings of the CFA/SEM analysis menu now include an ordinal data option. When a categorical estimator is selected (WLS, DWLS, ULS, PML, WLSM, WLSMV, WLSMVS, ULSM, ULSMV, or ULSMVS), the indicators can be declared as ordered categorical variables, so that the model is estimated from polychoric correlations and thresholds. The option is switched on by default for these estimators, all selected indicators are treated as ordinal unless a subset is chosen, and it is hidden for continuous-data estimators (ML, MLR, …).
• Heterotrait-Monotrait ratios (HTMT) are computed from polychoric correlations when the indicators are ordinal.
• Multi-group CFA and the Auto Invariance Test support ordinal indicators: the scalar level constrains thresholds, and all levels use the theta parameterization so that configural, metric, scalar, and strict models are directly comparable.
• With ordinal indicators, selecting FIML for missing data applies pairwise deletion.
• The model comparison table, the path diagram legend, the HTML report, and the AI assistant context state when a model was estimated with ordinal indicators, and the Methodological Guide describes the new option.
• Saved analysis sessions keep the ordinal-data setting.

Layout

The CTT module now follows the analysis workflow: Data → Item Analysis → Distractors → Reliability → Scores → Report, followed by Settings and About. Summary statistics are shown as one calm strip instead of coloured cards, the item table is more compact, and references moved to About.

• Distractors is a tab of its own (response with key): an overview of all items with non-functioning options (< 5%) and options with a positive option-total correlation, the option table of the selected item, and a plot of option choice by total-score group.
• Reliability combines alpha, SEM, the 95% band, and split-half reliability with the alpha-if-deleted chart.
• Scores groups the score distribution, person scores with the SEM band, and scoring of new data.

CTT module

• Item Analysis is now a dashboard: summary cards (alpha, SEM, mean score, items to review), a full item table, and an item-detail panel (item selector, ICC, distractor analysis) beside it. Clicking a table row selects the item.
• Item table now reports difficulty (p), discrimination (item-total correlation), alpha-if-deleted, a flag for items whose removal raises alpha, and a Retain / Revise / Eliminate recommendation.
• Interpretation cut-offs follow published sources and are shown in the app and the report: discrimination (Ebel & Frisbie, 1991), difficulty (Allen & Yen, 1979; Crocker & Algina, 1986), Cronbach’s alpha (George & Mallery, 2003; Nunnally & Bernstein, 1994), and distractor functioning (Haladyna & Downing, 1993).
• Discrimination for polytomous items now uses the item-total correlation (previously the biserial coefficient).
• Built-in polytomous data replaced: the old example was multiple-choice letters recoded A=1..D=4, which are not ordinal and gave meaningless results (alpha ~ 0.03). It is now a simulated 200 x 20 rating-scale dataset (1-4, graded response model).
• New “Analysis by Data Type” guide explaining how CTT differs for dichotomous, polytomous, and response-with-key data.
• Prepare Data: step-by-step sidebar, data summary cards (examinees, items, missing %), and tabs for preview, response frequencies, and the required data format.
• Score Distribution & Reliability: reliability card with SEM and 95% band, split-half (Spearman-Brown) cards, an alpha-if-deleted chart, histogram with mean line, and extended descriptives (skewness, kurtosis, SEM, alpha).
• Person scores now include an SEM-based confidence band (68/90/95/99%) and Kelley’s estimated true score, in Score New Data and in a new Person Scores table.
• Score New Data now scores response-with-key data using the key, reports how many items matched, and shows a score histogram.
• Distractor analysis flags non-functioning distractors (< 5%) and distractors with positive item-total correlation.
• New Settings tab with the decimal separator (dot or comma) and decimal places. It applies to tables, cards, plots, the HTML report, and the exported score file (comma gives a semicolon-separated .csv). Semicolon-delimited uploaded .csv files are read with a decimal comma automatically.
• Report rewritten: the old report referenced fields that did not exist in the CTT result. The new report has an overview with an automatic summary, reliability (alpha, SEM, split-half, alpha-if-deleted chart), a colour-coded item table with cut-offs, a difficulty-discrimination plot, distractor analysis (response with key), score distribution, person scores with SEM band, the console output and references. The HTML download now renders on demand and no longer requires generating the preview first.
• The R console and AI assistant now receive a compact results summary instead of the raw result object.

measureR 0.0.5

This release moves directly from 0.0.3 to 0.0.5 (version 0.0.4 was not released). The application remains a Shiny app launched with run_measureR().

New features

• AI assistant in every module: ask free-form questions about the current results or generate a results summary. Supports Google Gemini, OpenAI, Groq, OpenRouter, and Anthropic; users supply their own API key in the Settings on the homepage. Reference documents (including PDF) can be attached as context.
• R console panel: a floating panel showing the R code behind each analysis, so results can be reproduced and validated outside the app.
• HTML report export for the Content Validity, CTT, EFA, CFA/SEM, and IRT modules, with an optional AI-generated summary embedded in the report.
• IRT module replaces the earlier LTA module, with item/test information visualisation and Excel/RDS export of results.
• CFA/SEM: variable aggregation (parceling) builder, scoring of new data, robust fit-index option, editable model-comparison table, and a method guide.
• Excel (.xlsx) export of results and data templates across modules.

Other changes

• Added dependencies: httr, jsonlite, pdftools, writexl.
• Updated the homepage and styling.

measureR 0.0.3

• CFA/SEM Module Major Upgrade:
  • Added Structural Equation Modeling (SEM) support within the CFA module. SEM models are automatically detected via the ~ operator in lavaan syntax.
  • Expanded estimator options from 6 to 18 (ML, GLS, WLS, DWLS, ULS, DLS, PML, MLM, MLMVS, MLMV, MLF, MLR, WLSM, WLSMVS, WLSMV, ULSM, ULSMVS, ULSMV).
  • Added decimal separator option (dot/comma) for international compatibility.
  • Added Variances tab with Heywood case detection and warning alerts.
  • Implemented persistent model counter for accurate model tracking across deletions.
  • Added auto-generated model notes that detect syntax changes between model runs.
  • Added model deletion (trash icon) in the fit comparison table.
  • Expanded color schemes from 4 to 12 (added Pastel, Greyscale, Earth, Vibrant, Monochrome, Sunset, Rose, Mint).
  • Added collapsible plot settings panels (General & Layout, Node & Edge Sizes, Measurement Model, Colors & Display, Fit Indices).
  • Added Measurement Model sub-controls: Curve, Split Layout, SubScale width/height.
  • Added fit indices selector for customizing which fit indices appear on the path plot.
  • Added fit indices embedded in path plot as a legend box.
  • Added Download Plot (PNG) button.
  • Added Export Report (HTML) button with auto-generated HTML reports using RMarkdown.
  • Improved HTMT calculation to handle duplicate factor names in model syntax.
  • Relaxed acceptable fit status criteria for more realistic model evaluation.
  • Added SPSS (.sav) file import support via the haven package.
• Added new dependencies: shinyBS, flextable, kableExtra, knitr, rmarkdown, officer, magick, haven, scales.

measureR 0.0.2

• Enhanced the Content Validity module with several methodological improvements:
  • Integrated official critical values of Aiken’s V (Aiken, 1985) for rating scales ranging from 3 to 7 categories.
  • Implemented scale-based Aiken’s V computation using user-selected theoretical rating ranges instead of observed minimum–maximum values.
  • Added dynamic selection of the item ID column for uploaded datasets.
  • Improved visual decision indicators for CVR, Aiken’s V, and CVI tables (color-coded validity status).
  • Disabled inferential evaluation of Aiken’s V for dichotomous (0/1) data.
• Improved data validation and robustness for uploaded content validity datasets.
• Minor UI refinements in Content Validity and CTT modules.

measureR 0.0.1

• Initial release to CRAN.
• Includes a full Shiny-based graphical user interface for:
  • Content Validity (CV)
  • Exploratory Factor Analysis (EFA)
  • Confirmatory Factor Analysis (CFA)
  • Classical Test Theory (IRT)
  • Item Response Theory (IRT)
• Includes interactive visualizations, downloadable outputs, and built-in example datasets.
• Provides run_measureR() as the main entry point for launching the application.