metis User Guide
A practical guide to installing metis, building PLS-SEM models, running analyses, and turning saved results into clean report tables with Tark.
1. What is metis?
metis is a free desktop app for researchers, lecturers, and students who work with Partial Least Squares Structural Equation Modelling (PLS-SEM). It brings the full workflow into one place: create a workspace, import your data, draw your model, run the analysis, and review the results without writing R code.
The statistical work is handled by R and seminr in the background. The app keeps the day-to-day experience visual and guided, so you can stay focused on your study instead of switching between scripts, folders, and separate tools.
Tark is the reporting helper inside metis. Once a model has saved results, Tark collects the important tables and an optional path diagram into a preview you can copy into Word or a thesis draft.
2. System Requirements
metis is currently available for Windows only. The following minimum specifications are recommended:
| Requirement | Minimum Specification |
|---|---|
| Operating system | Windows 10 or Windows 11 (64-bit) |
| Processor | 1.6 GHz dual-core (2 GHz+ recommended) |
| RAM | 4 GB (8 GB recommended for large datasets) |
| Disk space | Bundle installer: about 283 MB · Lite installer: about 78 MB |
| Internet | Needed for download, signup, and Lite package setup if packages are missing. The analysis itself runs locally. |
| R (Lite only) | R 4.0 or later, already installed system-wide |
3. Choosing Your Installer
metis comes in two Windows installers. Both give you the same app. The only difference is whether the installer brings R with it.
metis Bundle
Recommended for most users
- Includes the R support metis needs
- Best choice if you do not already use R
- Keeps metis separate from any R setup on your computer
- Good for students, first-time users, and shared computers
- Larger download because it carries more of the setup with it
Installation Steps
- Download the Bundle installer from the release page.
- Open the installer and choose where metis should be installed.
- Keep the desktop shortcut on if you want quick access.
- When setup finishes, launch metis and start your first workspace.
metis Lite
For users with R already installed
- Smaller download
- Uses the R version already installed on your computer
- Checks the required packages during setup
- Good for researchers and analysts who already work in R
- Does not include R
- If R or packages are missing, metis shows the next step clearly
Installation Steps
- Make sure R 4.0 or later is installed.
- Download the Lite installer from the release page.
- Open the installer and choose where metis should be installed.
- On first launch, metis checks for R and the packages it needs. If something is missing, it gives you a copyable command instead of a long error message.
3.3 Comparison at a Glance
| Feature | Bundle (~283 MB) | Lite (~78 MB) |
|---|---|---|
| Includes R support | Yes | No, uses your installed R |
| Requires prior R install | No | Yes (R 4.0+) |
| R package setup required | Handled by the app | Checked on first launch |
| Separate from your own R setup | Yes | No, shared |
| Suitable for non-R users | Yes | No |
| Download size | About 283 MB | About 78 MB |
| Recommended for | Most users | Existing R users |
4. First Launch & Setup
The first launch is a short setup check. Bundle users should be ready after the included R support is prepared. Lite users will see metis look for R and check the packages needed for analysis.
- Whether R is available, if you installed Lite.
- Whether the analysis packages are ready.
- Whether metis can save your workspace and launch normally.
If R is missing, metis shows the R setup screen. If packages are missing, it shows the missing packages and a single command you can copy. Your data is not uploaded during setup or analysis.
5. Getting Started
5.1 Application Overview
When metis opens, you start on Workspace Home. Think of a workspace as a study folder. It can hold datasets, models, and saved results, so the pieces of one project stay together.
You can create a new workspace, open an existing one, pin important workspaces, switch between grid and list view, and reopen recent models from the title bar. A model opens on the canvas, where you can work with multiple model tabs in the same session.
5.2 Importing Your Data
- Choose Import Dataset from the workspace or canvas.
- Select a CSV or Excel file. For CSV files, metis lets you confirm the delimiter and encoding.
- Review the preview, missing-value count, and detected variables.
- Open the dataset view if you need to clean small issues before modelling.
5.3 Reviewing and Cleaning Data
The dataset view is for quick, practical cleanup. You can rename columns, edit cells, delete rows or columns, add simple calculated columns such as mean or sum, and save the cleaned data back to the workspace.
A workspace can hold up to three datasets. From the Dataset Manager, choose which dataset a model should use, rename datasets, remove older versions, or add a new dataset when you need to compare versions.
5.4 Building Your Model
- Add constructs to the canvas and drag dataset variables onto them as indicators.
- Choose whether each construct is reflective or formative.
- Move indicators around the construct so the diagram stays readable.
- Draw arrows from predictor constructs to outcome constructs. You can also connect a construct to an existing arrow when you need a moderation relationship.
- Use tabs to keep more than one model open, and save manually whenever you want to lock in the latest version.
5.5 Running the Analysis
- Click the calculate button on the canvas to run PLS-SEM.
- Confirm the algorithm settings. The defaults are a good starting point for most models.
- metis checks that your model has indicators, paths, and matching dataset columns before running.
- When the run finishes, results open automatically and can be saved to the workspace.
6. Reading Results
The Results View keeps each analysis mode in its own clear view: PLS-SEM, Bootstrap, PLSpredict, and Advanced analysis. The left side lists the result groups, while the main area shows the selected table, chart, or path diagram.
Some panels include a chart above the table. You can copy visible tables, download tables for Excel, export a full HTML report, or copy the R script used to reproduce the model.
| Section | Contents |
|---|---|
| PLS-SEM | Path effects, loadings and weights, reliability, validity, R², f², VIF, model fit, and supporting data checks. |
| Bootstrap | Resampled path and measurement results with confidence intervals, t-values, and p-values. |
| PLSpredict | Predictive summaries, PLS vs LM comparisons, Q²predict, prediction errors, and CVPAT output when requested. |
| Advanced analysis | IPMA, NCA, and cIPMA outputs including priority maps, necessity checks, bottleneck tables, and cIPMA priorities. |
| Tark | Report-ready tables and an optional path diagram built from the saved results for a model. |
7. Bootstrap Analysis
Bootstrapping helps you judge whether paths and measurement results are stable enough to report. metis lets you set the number of subsamples, confidence interval type, confidence level, tails, and sign-change handling before the run starts.
- From the Model Canvas, choose Analysis > Run Bootstrap.
- Review the default settings. The current default is 500 subsamples.
- Run Bootstrap and wait for the results to open.
- Save the results if you want to reopen them later or use them in Tark.
8. PLSpredict
PLSpredict checks how well your model predicts data it has not seen during fitting. You can choose the number of folds and repetitions, and you can turn on CVPAT when you want the extra prediction test.
- Select Analysis > PLSpredict from the menu bar.
- Set folds and repetitions. The modal also shows the total validation cycles.
- Turn on CVPAT if you want that summary included.
- Review Q²predict, PLS vs LM comparison, prediction errors, and the histogram panels in the results view.
9. Advanced Analysis
Advanced analysis becomes available after you run PLS-SEM for the current model. It is meant for follow-up interpretation, especially when you want to understand which constructs deserve attention for a chosen target outcome.
- Run and save a PLS-SEM result for the model.
- Choose Analysis > Advanced analysis.
- Select a target construct and choose whether to include direct predecessors only or all upstream predecessors.
- Select IPMA, NCA, cIPMA, or keep all three selected.
- Run the analysis and review the priority map, construct table, necessity check, bottleneck table, and cIPMA priorities.
10. Tark Reports
Tark is the report helper inside metis. It takes saved results from a model and turns them into clean tables for writing. The goal is not to replace your interpretation, but to remove the busy work of collecting results, aligning labels, and rebuilding tables by hand.
What Tark creates
- Measurement model tables with loadings, reliability, validity, and VIF where available.
- Discriminant validity tables.
- Structural model tables for hypothesis testing, confidence intervals, effect sizes, and decisions.
- Explanatory and predictive power tables.
- Model fit tables.
- Optional PLSpredict and advanced-analysis tables when those saved results are included.
- An optional path diagram with the values you choose to show.
How to use Tark
- Run and save PLS-SEM, Bootstrap, and PLSpredict for the model you want to report.
- Click Tark it in the title bar.
- Choose the workspace and model. If a model is not ready, metis tells you which saved result is missing.
- Add a report title, choose full construct names or abbreviations, and decide whether to include the path diagram.
- Preview the report tables, then copy one table or copy everything into Word.
11. Exporting Results
Saves the current result set back into the workspace so you can reopen it later or use it in Tark.
Creates a browser report with the result sections, path diagram, and supported charts included.
Copies the visible table with formatting for Word, or downloads it for spreadsheet work.
Copies the matching R script to your clipboard for users who want to reproduce the model outside metis.
12. Data Privacy & Security
metis is built for research data that should stay close to you. Your workspace, dataset, model, and results are stored on your computer.
13. Known Limitations, v0.0.1
- Windows only. Linux and macOS support is planned for Phase 2, subject to sufficient user interest.
- CSV and Excel are the supported import formats. Other data formats may appear in the file picker, but use CSV or Excel for the smoothest workflow.
- Tark prepares tables, not the written argument. It gives you cleaner tables and diagrams, but you still write the interpretation for your study.
- Large calculations can take time. metis runs Bootstrap, PLSpredict, and advanced analysis locally through R and seminr. It automatically uses available CPU cores safely, but large models, 5,000-10,000 bootstrap samples, current workload, and hardware can still make calculations take several minutes. A longer run does not mean the calculation has failed; metis continues computing locally, and completed reports are generated from the finished seminr results.
- Saved results matter. Tark and later review workflows work best when you save PLS-SEM, Bootstrap, PLSpredict, and any Advanced analysis results you want to reuse.