Methods Text Auto-generation

How Licklider generates a methods text draft for each figure, what it contains, and how to use it before publication.

Every figure in Licklider has an associated methods text draft. The draft is generated automatically from the analysis record — the statistical test used, the data processing steps applied, the disclosures confirmed, and the quality checks resolved — and is intended as a starting point for the methods section of a manuscript.

The draft is a template, not a finished text. It should be reviewed and edited before inclusion in a submission.

That design is deliberate. A methods section has to reflect what the figure actually did, but it also has to match the study context, journal requirements, and the claims you intend to make. Licklider can draft the figure-specific statistical description from the record; it does not replace author review.

Where to find it

The methods text draft appears in the Inspector alongside the figure, in the Methods template section of the Action Bar. The section shows the generated text alongside a note that it should be edited before publication.

What the draft contains

The draft is organized into sections that reflect the analysis record:

Data

The dataset used, the number of observations, and the group structure.

Preprocessing

Any data cleaning, imputation, or transformation steps that were applied before analysis. If imputation was used, the method and the number of imputed values are included.

Statistical test

The test that was run, the software used, and the key parameters. If the test was selected automatically based on normality and variance checks, the basis for the selection is noted.

Effect size

The effect size measure and its value, where applicable.

Caveats

Any quality check findings that were acknowledged — for example, a violated assumption, an outlier sensitivity result, or a pseudoreplication concern that was resolved with a disclosure.

Transparency

The disclosure status of the analysis intent, hypothesis direction, and other confirmatory elements.

Because the draft is record-based, it is only as accurate as the analysis record itself. If the data contract, preprocessing record, test selection, or disclosure state is incomplete or incorrect, the generated text will carry that same incompleteness forward in polished language.

That is also why the draft includes caveats and transparency fields. The goal is not only to describe what test ran, but to keep acknowledged limitations and confirmatory choices visible in the written record rather than letting them disappear during manuscript preparation.

Methods summary line and supported tests

The compact methods summary (when present in the figure record) uses the same canonical method names as the rest of the product. Licklider can automatically assemble that summary for analyses that include, among others:

  • Repeated measures ANOVA (with sphericity correction when applicable)
  • Kruskal-Wallis test
  • Friedman test
  • Pearson correlation
  • Spearman correlation
  • Fisher's exact test (2x2 and R×C / Freeman-Halton extension)

Other common entries include Welch t-test, Student's t-test, paired t-test, Mann-Whitney U, Wilcoxon signed-rank, one-way ANOVA, chi-square, and linear regression, depending on the selected analysis. The exact wording follows the figure metadata and the label maps used for methods display in the Inspector.

What it does not cover

The draft covers the statistical analysis for a single figure. It does not write:

  • Study design or recruitment sections
  • Sample justification or power analysis rationale (though the power analysis result is available separately)
  • Discussion or interpretation text
  • Citation formatting
  • Cross-figure harmonization of wording across an entire manuscript
  • Journal-specific reporting structure or house style

Using the draft

Copy the draft text from the Inspector into your manuscript editor. Review each section:

  • Check that the test name and parameters match what you intend to report
  • Expand any placeholders with the specific values from your analysis
  • Edit the preprocessing section to reflect only the steps that are relevant to report
  • Remove caveats sections that do not apply
  • Add study-level details that are outside the figure record, such as recruitment, randomization, blinding, sample sourcing, or protocol-specific wording
  • Check whether citations, software version wording, and journal-required phrasing still need to be added manually

The draft uses neutral, passive-voice language consistent with scientific writing conventions. It is designed to be edited, not used verbatim.

The neutral style is intentional. It makes the output easier to adapt across journals and reduces the chance that the generator adds interpretation or promotional language that does not belong in methods text.

Licklider does not determine automatically whether the generated wording is sufficient for your target journal, whether all required citations have been added, whether study-level protocol details are missing, or whether a figure-specific methods paragraph is consistent with the rest of your manuscript.

That limitation matters because auto-generated text can sound complete even when important context is missing. Use the draft as a structured starting point, not as a publication-ready guarantee.

What this page does not cover