Academy series · Methodology & analytics

Statistical reporting series

Twelve practical guides on reporting statistical methods and results in theses and journal articles, written by a statistician.

Each guide takes one family of analyses, shows what examiners, reviewers and reporting guidelines expect to see, and works through an illustrative example. Read them in order or go straight to the analysis you ran.

The guides

Read in order, or jump to your analysis

Each guide has a worked example, a "what to report" table, common mistakes and a full reference list.

  1. 01Reporting statistical methods: how to write a methods section that others can check and repeatA results section is only as credible as the methods behind it. This guide explains what to report about sample size, pre-specified analyses, data preparation, assumptions, missing data, multiplicity and software, so that an examiner or reviewer could repeat your analysis from your description.
    • Choose the reporting guideline and protocol standard that fit your design, from SAMPL and APA JARS-Quant to CONSORT 2025, SPIRIT 2025 and STROBE
    • Justify a sample size and report a power calculation in full
    Mr John, M.Sc.7 min read19 references
  2. 02Reporting statistical results in academic manuscripts: principles common to every analysisWhatever test or model you used, examiners and reviewers ask the same questions: what was done, to whom, with what uncertainty, and can it be checked? This overview sets out the shared principles behind the major reporting guidelines and points you to the detailed guide for each family of analyses.
    • Choose the reporting guideline that matches your study design, using SAMPL and APA JARS-Quant as general baselines
    • Report descriptive statistics with sensible precision, the right measure of spread and complete denominators
    Mr John, M.Sc.8 min read14 references
  3. 03Reporting hypothesis tests in academic manuscriptsA p-value on its own tells readers very little. This guide shows how to report a hypothesis test so that others can check it, judge the size and precision of the effect, and see how many tests were run.
    • State null and alternative hypotheses, the alpha level and the sidedness of each test before analysis
    • Report the test statistic, degrees of freedom, exact p-value, effect size and confidence interval together
    Mr John, M.Sc.7 min read14 references
  4. 04Reporting correlation analyses in research manuscriptsA correlation coefficient is easy to compute and easy to misreport. This guide shows how to choose the coefficient, check what it assumes, and report it with its sample size, confidence interval and a plot, so that readers can judge both strength and precision.
    • Choose between Pearson, Spearman and Kendall coefficients according to the data and the question
    • Check linearity, monotonicity, outliers and independence with a scatter plot rather than tests alone
    Mr John, M.Sc.7 min read12 references
  5. 05Reporting regression analyses in research manuscriptsA regression table is only as useful as the description around it. This guide sets out what readers and reviewers need in order to understand, check and reproduce a regression analysis, following SAMPL, APA JARS-Quant and, for prediction models, TRIPOD+AI.
    • State whether a regression model is explanatory or predictive and choose the matching reporting guideline
    • Describe the outcome, predictors, coding and functional form so that the model can be rebuilt
    Mr John, M.Sc.7 min read12 references
  6. 06Reporting ANOVA and ANCOVA in academic researchAn F statistic and a p-value tell readers very little on their own. This guide explains how to report ANOVA and ANCOVA so that the design, assumptions, effect sizes and covariate adjustment are clear, following SAMPL, APA JARS-Quant and CONSORT 2025.
    • Describe an ANOVA or ANCOVA design fully, including factors, levels, covariates and repeated measures
    • Report the assumptions that matter and choose Welch's F when group variances differ
    Mr John, M.Sc.7 min read13 references
  7. 07Reporting association analyses in biomedical researchAssociation analyses link exposures, biomarkers or policies to outcomes, but a test result alone says little. This guide shows how to report the question, the data, the measure of association, its precision and its adjustment so that readers can judge what the association means.
    • Frame an association question with a defined population, exposure, comparator and outcome
    • Distinguish tests of association from measures of association and report both correctly
    Mr John, M.Sc.6 min read12 references
  8. 08Reporting risks, rates and ratios in health research manuscriptsRisks, rates, odds and hazards answer different questions, and reviewers notice when they are mixed up. This guide shows how to choose the right measure, report it with its denominator, time frame and confidence interval, and pair every relative effect with an absolute one.
    • Distinguish risks, rates, odds and prevalence and state the numerator, denominator and time frame of each
    • Choose between the risk ratio, odds ratio, rate ratio and hazard ratio according to study design
    Mr John, M.Sc.7 min read13 references
  9. 09Reporting survival (time-to-event) analyses in research manuscriptsTime-to-event data are incomplete by design, and that shapes everything about how they should be reported. This guide covers defining time zero and censoring, Kaplan-Meier estimates and plots, Cox models, competing risks and the absolute measures that should accompany every hazard ratio.
    • Define the event, time origin, end of follow-up and censoring rules for a time-to-event outcome
    • Explain why censoring is handled by Kaplan-Meier estimation and what non-informative censoring means
    Mr John, M.Sc.6 min read13 references
  10. 10Confirmatory factor analysis: specifying, evaluating and reporting a measurement modelConfirmatory factor analysis tests whether survey items measure the constructs your theory says they do. This guide covers specification, estimation, fit assessment and reporting, with examples relevant to SPSS AMOS and other SEM software.
    • Distinguish confirmatory from exploratory factor analysis and specify a measurement model from theory
    • Check identification, choose an estimator suited to the data and plan an adequate sample size
    Mr John, M.Sc.7 min read12 references
  11. 11Reporting Bayesian analyses in research manuscriptsA Bayesian result can only be judged if readers can see the prior, the model, how the posterior was computed and how much the conclusion depends on those choices. This guide follows the Bayesian Analysis Reporting Guidelines (BARG) and shows what to report at each step.
    • Explain why the prior, likelihood and computation must all be reported for a Bayesian result to be reproducible
    • Specify and justify priors, including weakly informative and data-derived priors, and check them with prior predictive simulation
    Mr John, M.Sc.7 min read11 references
  12. 12How to present PhD data in tables and figuresGood tables and figures let examiners see your evidence at a glance; poor ones hide it or mislead. This guide shows how to choose between a table and a figure, build each to APA 7 conventions, show the data rather than just summaries, use colour that everyone can read, and meet thesis and journal requirements.
    • Choose between a table, a figure and plain text for each piece of evidence
    • Build tables and figures with the elements APA 7 expects: number, title, clear headings or axes, and notes
    Mr John, M.Sc.7 min read14 references
For every analysis

The universal reporting checklist

Whatever test you ran, these eight points apply. The guides show how each one looks in practice.

Before you submit
  • Name the test or model, and say why it fits your design and data
  • Give the test statistic, degrees of freedom and exact p-value (write p < .001 only below that)
  • Report an effect size with its 95% confidence interval (or credible interval), in units readers understand
  • Say which assumptions you checked, how, and what you did when they were not met
  • Give the sample size for every analysis and explain how missing data were handled
  • Correct for multiple comparisons, or explain why you did not
  • Separate pre-specified analyses from exploratory ones
  • Name the software, version and packages, and share code where you can
Your next step

Want a statistician to check your results?

  1. 1Share your analysis output and draft write-up
  2. 2A statistician reviews the tests and the reporting with you
  3. 3You get clear corrections and a quote first
Academic Statistics course