Research workflows
Analyze your research data with AI, inside Word.
Textmolt brings data analysis and statistical figures into Microsoft Word on Windows. Choose a CSV, describe your research question, review the analysis, and insert figures, captions, tables and research text into your document.
- Works in
- Microsoft Word on Windows
- Data input
- UTF-8 CSV · up to 2 MiB, 10,000 rows and 64 columns
- Figure output
- Up to four separate figures per analysis
How the workflow works
-
Choose your data
Open + → Analysis in the sidebar and select a CSV. Check the column names, missing values and unit of observation before describing your question.
-
Describe the comparison
Explain your groups, outcome and research design. Review what will be sent to AI: a locally generated summary by default, or the full CSV if you select that option.
-
Review before using
On Pro, including an active trial, advanced mode lets you discuss and confirm an analysis plan first. Check the generated figures, statistical sources and assumptions; resolve questions about the design before drawing conclusions.
-
Bring the result into your paper
Insert figures and captions into Word, with tables or methods and results text when available. Save the outputs and Python code if you want to inspect or reproduce the analysis.
Built-in statistical methods
| Method | Supported scope |
|---|---|
| Between-subjects ANOVA | One to three factors, with Type II or Type III sums of squares. |
| Welch ANOVA | One-way group comparisons allowing unequal variances. |
| Tukey / Tukey–Kramer | Pairwise comparisons using the pooled error of the fitted factorial model. |
| Games–Howell / Dunnett | Unequal-variance pairwise comparisons, or comparisons against one control. |
| Repeated-measures ANOVA | One within-subject factor, with Mauchly's test and Greenhouse–Geisser correction. |
| Multiple-testing correction | Holm, Bonferroni and Benjamini–Hochberg adjustments. |
Use methods appropriate to your study. Supported inputs and built-in tests have limits; review the analysis and source data before using results in a paper.
Try the appearance options
The figure below is computed and drawn on this page from public sample data. Switch the palette and the column width to see one result presented differently. In Word, appearance changes like these are applied on your computer, without another AI request.
Table 1. Descriptive statistics by group.
| Group | n | Mean | SD |
|---|---|---|---|
| Control | 10 | 5.03 | 0.58 |
| Treatment 1 | 10 | 4.66 | 0.79 |
| Treatment 2 | 10 | 5.53 | 0.44 |
Sample data: PlantGrowth (Dobson, 1983). Drawn and computed on this page as an illustration. It is not output from Textmolt; in Word, the analysis workflow builds figures from your own CSV.
A request to start from
Compare the outcome across these independent groups. Check whether the proposed analysis fits the design, show the individual observations, and prepare a figure with a caption and methods paragraph.
Example request only. It is not an executed analysis or a claim about results. Adapt the request to your data and research design.
Know where the statistics came from
Textmolt records calls to its built-in statistical functions and distinguishes them from AI-written calculations. Unsupported designs can be refused with an explanation.
- Check the reported method, sample sizes and missing-data handling.
- A recorded calculation does not independently verify every figure or sentence.
Check the analysis before preparing your paper
Check that the design, method, sample sizes and results agree before using the generated figures and research text. When the analysis changes, recheck the corresponding results and interpretation.
- Keep the Python analysis code and generated results for further review.
- Where an analysis archive supports it, figures can be redrawn later on your computer from the archive and the matching original CSV, without another AI request.
- For figure styling, captions, insertion and export, see the figures workflow.
Choose what AI receives
Analysis code runs in an isolated environment on your computer. AI requests still use the internet and pass through the Textmolt gateway to your chosen provider.
- A summary can contain column names and statistics; it is not automatically anonymous.
- Choosing full-data mode sends the CSV. An optional AI review may also send selected results or images after confirmation.
Questions about this workflow
Do I need Zotero or Viamolt for analysis?
No. The analysis workflow uses your selected CSV independently of the Zotero connection.
Can I select an Excel workbook?
Save the worksheet as CSV UTF-8 first. The current input is CSV, not an Excel workbook.
Where can I check access and costs?
See Pricing on the home page for Free and Pro access. AI usage is separate, through your own provider key or Textmolt credits. On Pro, runs beyond the daily number are confirmed one at a time and paid in Textmolt credits, also when you use your own key. On Free, wait for the monthly reset or upgrade.