alphaN

• alphaN_report() now also states the scope of the "ES" and "moment" calibrations: exact for the normal linear model when p is supplied (evaluated at the effective sample size n - p), the conservative large-sample form otherwise, and asymptotic for other generalized linear models.
• New function alphaN_report() writes a preregistration-ready Markdown settings report: every input behind a calibrated alpha (sample size, evidence target, method, prior settings), the resulting alpha, the decision rule, and the references to cite. It mirrors the downloadable report of the companion Shiny application, can write straight to a file, and hard-wraps its lines (default 72 characters, see width) so the report reads well in consoles, files, and rendered documents.
• JAB() called without a covariate now returns a named vector with the Bayes factor of every coefficient except the intercept (which remains available on explicit request).
• alphaN_plot() and JAB_plot() have been restyled: colorblind-safe Okabe-Ito palette, light grid lines, open axes, and horizontal axis labels. alphaN_plot() additionally gains a log argument ("x", "y", "xy") which helps when the fast-falling "moment" curve is drawn next to the others, and its tick labels always use plain notation ("0.0001" and "10,000", never "1e-04").
• New function klauerBF() exports the effect-size and moment Bayes factors of Klauer, Meyer-Grant & Kellen (2025) that alphaN() inverts: from a t statistic (one-sample test or single regression coefficient) or from an F statistic for a joint test of q coefficients.
• alphaN() (methods "ES" and "moment") gains arguments q and p. With q > 1 the alpha level is calibrated for the joint F test of q coefficients through the exact regression-case Bayes factors of Klauer et al. (2025, Table 4), implemented natively including their Gaussian hypergeometric term; p sets the number of retained model parameters so that small-sample calibrations can use the effective sample size n - p (residual degrees of freedom). The defaults (q = 1, p = 0) reproduce the previous behavior exactly. The moment-prior default nu is now 5 + (q - 1) and the ES-prior scale recommendation generalizes to r = sqrt((nu - 2)/(nu * q)) * de, both following the paper (unchanged at q = 1).
• New function n_effective() computes the effective sample size n * (se/se_robust)^2 that Wulff & Taylor (2024) recommend as a sensitivity check when calibrating alpha with clustered (panel) data.
• alphaN_plot() gains a methods argument and can now draw the "ES" and "moment" curves alongside the prior-fraction methods.
• The regression-case implementation is validated against all printed Bayes factors in Table 8 of Klauer et al. (2025), in addition to the existing Table 7 anchors, and the q = 1 F form agrees with the validated t form to near machine precision.
• The quadrature and inversion machinery is additionally stress-tested against an independent oracle that integrates on the original effect scale over an infinite range with a plain density ratio (none of the package's substitution, windowing, or log-clamping choices); the normal-limit switch at n - p = 50,000 is measured directly by running both branches at the same effective sample size, and monotonicity of alpha in n and BF, and of the Bayes factors in the test statistic, is checked over grids that cross the switch, including joint tests and small residual degrees of freedom.

Documentation

• Klauer, Meyer-Grant & Kellen is now cited with its printed-issue details (2025, Psychonomic Bulletin & Review, 32, 1070-1094) throughout; it was previously cited by its online-first year, 2024.
• ?alphaN states the model scope of the "ES" and "moment" methods: exact under the normal linear model, asymptotic (like the prior-fraction methods) for other generalized linear models.
• citation("alphaN") now also lists Klauer et al. (2025) for users of the "ES" and "moment" methods.