A Prism alternative for statistics beyond the figure

The honest part

Prism is built around the graph — Magic Stat is built around the result

Prism earned its place in life-science labs because the figure is the product: a graph editor that people actually like using, with curve fitting and the conventions of biology built in. If your day is dose-response curves and figures for a manuscript, and Prism serves you, there is no reason to change on this page's account.

Magic Stat overlaps on the analyses that matter in those labs — dose-response with IC50, nonlinear curve fitting such as Michaelis-Menten, ROC and AUC, Bland-Altman, Kaplan-Meier survival with the log-rank test, and Cox regression — but it is organised differently. Every analysis, not only the graph, ends in an APA-formatted table and a report, and the method coverage reaches well beyond the life sciences into ecology, econometrics, psychometrics and SEM.

So the split is roughly this: if the figure editor is the reason you open the program, that is Prism's territory. If you need the analysis, the table and the write-up alongside the figure, that is where Magic Stat is aimed.

What it does

The life-science analyses, plus the manuscript

Dose-response curves — 3PL, 4PL, 5PL and two-site models, reporting IC50 with its confidence interval.
Nonlinear regression for mechanistic models (Michaelis-Menten and others), fitted by nonlinear least squares.
Medical statistics: ROC/AUC with DeLong confidence intervals and the Youden index, Bland-Altman agreement with limits and bias tests, Kaplan-Meier with the log-rank test, and Cox proportional hazards.
APA tables — mean (SD), exact p-values, compact letter displays — copied into Word as tab-separated text.
A report for every analysis with the method, the settings and the interpretation, exportable to .docx, .html or .md.
Range beyond the life sciences if the project needs it: the ANOVA and regression families, GLM/GAM, mixed models, SEM/PLS, meta-analysis, time series, ordination, ecology and econometrics.
Local and cross-platform — Windows, macOS and Linux; the data never leaves the machine.

How it works

From the data table to a written result

Open the data. CSV or Excel, in a spreadsheet layout rather than a Prism-style table.
Choose the analysis. dose-response, nonlinear fit, ROC, survival, or a standard test.
Set the model. e.g. 4PL for screening data; the fit and its parameters are reported with confidence intervals.
Read the result. table and figure together, with the interpretation next to them.
Export. figure as PNG/PDF/SVG, table to Word, report to .docx/.html/.md.

Options

Which models, exactly

Dose-response: 3 parameters (3PL), 4 parameters (4PL), 5 parameters (5PL, with an asymmetry factor) and two-site models — each reporting IC50.

Curve fitting: nonlinear regression by least squares, including mechanistic models such as Michaelis-Menten.

Medical tests: ROC/AUC (DeLong CI, Youden), Bland-Altman (limits of agreement, trend and proportional-bias tests), Kaplan-Meier (product-limit) and Cox PH.

What is not here: Prism's graph editor is its own thing, and Magic Stat does not attempt to reproduce it. Figures are produced for publication; they are not edited the way Prism lets you edit them.

Frequently asked

Prism questions, answered honestly

Should I leave Prism?

Not if the graph editor is why you use it. Prism is good at figure-making and at the life-science analyses that go with it; that is not a weakness to correct.

Does Magic Stat do dose-response and IC50?

Yes — 3PL, 4PL, 5PL and two-site models, each reporting IC50 with a confidence interval.

And curve fitting?

Yes: nonlinear regression by least squares, including mechanistic models such as Michaelis-Menten.

What about survival and ROC?

ROC/AUC with DeLong intervals, Bland-Altman, Kaplan-Meier with the log-rank test, and Cox proportional hazards are all included.

Is it free?

No — a free 48-hour trial with no card, then US$990 per year.

Which systems?

Windows, macOS and Linux.