Features
Statistics software usually stops at the numbers. KyroStat is built for what comes after: the table in your manuscript, the sentence in your results section, the script your reviewer asks for. Nine capabilities, in depth.
Most researchers learned statistics on SPSS output, so KyroStat presents results in the same familiar shape: clearly labelled grids with the group statistics, test statistics, degrees of freedom, and significance columns where you expect them.
The format is deliberately conservative. When you are checking a result at 2 a.m. before a submission, familiarity is a feature: nothing needs decoding, and your advisor can read it without ever having seen KyroStat.
Alongside the working grid, every result is typeset as an APA 7 table ready for the manuscript, and as the exact APA result sentence, for example: t(48) = 2.81, p = .007, Hedges’ g = 0.62, 95% CI [0.21, 1.03].
The sentence is assembled deterministically from the computed result, following APA conventions precisely: p reported as "p < .001" below .001, leading zeros kept for statistics that can exceed 1 and dropped for those bounded by 1, and effect sizes appended with their confidence intervals.
APA 7 and most journals now require effect sizes, so KyroStat treats them as part of the result rather than an optional extra. Each test reports the effect size it calls for: Cohen’s d or Hedges’ g for t-tests, eta squared and omega squared for ANOVA, Cramer’s V for chi-square, R-squared for regression, and rank-based equivalents for nonparametric tests.
Where a confidence interval for the effect size is defined, it is computed and reported with the estimate, and the output includes a plain-language magnitude interpretation you can defend in a viva.
Journals increasingly run submitted manuscripts through statcheck, which recomputes every reported p-value from its test statistic and degrees of freedom and flags inconsistencies. KyroStat runs the same check on your results before you ever export them.
Each reported p-value for t, F, r, chi-square, and Z statistics is re-derived independently with scipy and compared against the reported value. A mismatch is flagged immediately, so a transcription error cannot survive into your manuscript.
Every analysis returns the exact code it ran: real Python (pandas, scipy, statsmodels) or R, not pseudocode. You can read it line by line, download it, and rerun it in your own environment to reproduce the result exactly.
This matters for more than transparency. Reviewers ask for analysis scripts, supervisors want to check methods, and pre-registration increasingly expects code. With KyroStat the script exists automatically, from your very first analysis.
Choosing the wrong test is the most common statistical error in student research. KyroStat reads your research question and your data together, then recommends 3 to 8 candidate tests and 2 to 5 plots, each with a rationale referencing your actual question.
The recommendations are bounded on purpose: the AI must justify every suggestion against your stated question and may not propose a test merely because the data contains numeric columns. You always see the reasoning, and you always make the final call.
Box plots, bar charts, scatter plots, and distribution plots are generated from your analysis with clean styling meant for figures in a paper, not exploratory sketches. Labels come from your actual variable names.
Plots export alongside the tables and code, so the figure in your manuscript and the statistics in your results section always come from the same run of the same data.
Writing the software paragraph of a Methods section is tedious and error-prone: which package versions did you use, and how do you cite them? KyroStat generates it for you: a paste-ready paragraph naming each statistical test, the engine (Python or R), and the exact package versions that actually ran your analysis.
Citations follow each package’s own recommended reference, formatted in APA, and a BibTeX file is included for LaTeX and reference-manager users. Versions are read from the live environment at run time, so the paragraph always matches reality.
Everything above assembles into a single formatted PDF: descriptives, test results in both table styles, plots, the APA sentences, the generated code, and the Methods paragraph with citations.
It is the artifact you hand an advisor at a supervision meeting, attach to an ethics amendment, or keep as the audit trail for your dissertation appendix.
New to KyroStat? Start with how it works or see how it compares to SPSS and GraphPad Prism.
Create an account and run your first analysis in minutes. No install, no statistics course required.