Your thesis compares two treatments, or tests a new diagnostic marker, or follows patients for two years. Which statistics will the committee expect? In medicine and the health sciences, the answer is set less by habit than by study design and by published reporting guidelines. Statistics for a medical thesis therefore start with two questions: what kind of study is this, and which guideline applies to it?
This article is part of our guide How to Write a Master’s Thesis by the Mastermind PhD team and relates to our Statistical Analysis service. Sources last checked: November 2026.
This guide matches common questions to tests, explains the clinical measures examiners look for, and summarizes five reporting standards that structure the methods and results chapters.
What is statistical analysis in a medical thesis? It is the planned use of descriptive and inferential methods to answer a clinical or public health question from patient or population data: describing the sample, comparing groups or time points, estimating measures of association or diagnostic accuracy with confidence intervals, adjusting for confounders, and reporting everything according to the guideline for the study design (for example CONSORT for randomized trials or STROBE for observational studies).
Start with the design and the sample size
In statistics for a medical thesis, the design decides most of the analysis. A randomized controlled trial compares outcomes between arms; a cohort study estimates risk over time; a case-control study estimates odds ratios; a cross-sectional study estimates prevalence and associations; a diagnostic accuracy study compares a test with a reference standard.
The first numerical step in statistics for a medical thesis is the sample size. Calculate it before data collection from the primary outcome, the clinically important difference, the expected variability or event rate, alpha (usually .05) and power (usually 80% or 90%). Report all of these inputs in the methods chapter, together with the software used. Ethics committees and examiners both look for this calculation.
Which test fits which question?
| Question | Data | Common test or model |
|---|---|---|
| Compare a continuous outcome between two independent groups | Normal / skewed | Independent t-test / Mann-Whitney U |
| Compare a continuous outcome before and after in the same patients | Normal / skewed | Paired t-test / Wilcoxon signed-rank |
| Compare three or more groups | Normal / skewed | One-way ANOVA / Kruskal-Wallis |
| Compare proportions between groups | Categorical | Chi-square, or Fisher’s exact test for small expected counts |
| Compare paired proportions (before and after) | Binary, same patients | McNemar’s test |
| Relationship between two continuous variables | Normal / skewed or ordinal | Pearson / Spearman correlation |
| Predict a binary outcome adjusting for confounders | Binary outcome | Multiple logistic regression (odds ratios) |
| Predict a continuous outcome | Continuous outcome | Multiple linear regression |
| Time to an event (death, relapse) | Censored time data | Kaplan-Meier curves, log-rank test, Cox regression (hazard ratios) |
| Accuracy of a diagnostic test | Test result vs reference standard | Sensitivity, specificity, predictive values, ROC curve and AUC |
| Agreement between two raters | Categorical / continuous | Cohen’s kappa / ICC and Bland-Altman plot |
| Repeated measurements over time | Continuous, several time points | Repeated-measures ANOVA or mixed models |
📱 Ask a statistician about your data — Send a short description of your instrument and sample size on WhatsApp and get a suitable analysis plan. Every analysis comes with the raw SPSS file and a results report.
Describing patients correctly
Descriptive tables are where statistics for a medical thesis are first checked. Table 1 of most medical theses describes baseline characteristics. Report mean and standard deviation for approximately normal variables and median with interquartile range for skewed ones such as length of stay or laboratory values; report categorical variables as counts and percentages. In randomized trials, do not test baseline differences between arms with p values. CONSORT discourages this because any difference is, by design, due to chance.
Effect measures clinicians expect
Statistics for a medical thesis are judged on effect estimates with confidence intervals, not on p values alone. The ICMJE recommendations ask authors to quantify findings with measures of uncertainty such as confidence intervals and to avoid relying solely on hypothesis testing.
- Risk ratio (RR) and risk difference for cohorts and trials.
- Odds ratio (OR) for case-control studies and logistic regression.
- Hazard ratio (HR) for survival analysis.
- Number needed to treat (NNT), the reciprocal of the absolute risk reduction, when the audience is clinical.
- Mean difference with its 95% CI for continuous outcomes, plus a standardized effect size where useful.
An illustrative sentence: “Mortality at 30 days was lower in the intervention arm (8.0% vs 12.5%; RR 0.64, 95% CI 0.42 to 0.97).” The numbers are an example, not data from a study.
Five reporting standards for statistics for a medical thesis
The EQUATOR Network hosts reporting guidelines for most designs. Five are used most often in theses:
| Guideline | Study design | What it asks for in the statistics |
|---|---|---|
| CONSORT | Randomized trials | Flow diagram, sample size, randomization, analysis population (intention to treat), effect sizes with CIs |
| STROBE | Cohort, case-control, cross-sectional | Confounder handling, missing data, sensitivity analyses, unadjusted and adjusted estimates |
| STARD | Diagnostic accuracy | Flow of participants, 2 × 2 table, sensitivity and specificity with CIs |
| PRISMA | Systematic reviews and meta-analyses | Search, selection flow, risk of bias, synthesis methods, heterogeneity |
| SAMPL | Statistical reporting across designs | Which test for which outcome, assumptions checked, exact p values, CIs, software |
CONSORT was updated in 2025; check EQUATOR for the current version of each checklist, include the completed checklist as an appendix if your faculty allows, and cite it in the methods chapter.
Confounding, missing data and multiple outcomes
Three issues cause many corrections in statistics for a medical thesis. Observational studies need adjustment for confounders chosen in advance from clinical knowledge, not only from significant univariable tests. State how missing data were handled (complete case analysis, multiple imputation) and how many cases each analysis used. Define one primary outcome; secondary outcomes are reported as such, and many comparisons call for a correction or for cautious interpretation.
Software and documentation
SPSS, Stata, R and MedCalc are all common in medical theses. Name the software and version, keep the syntax or script, and make sure the numbers in the abstract, tables and text agree. Statistics for a medical thesis are often reviewed by a statistician on the committee, who will check the design, the test and the reporting guideline together.
📱 Ask on WhatsApp now — 5 years of experience across all disciplines and 14 academic services. Two-stage payment (80/20); revisions are free until supervisor approval, within 30% of the chapter.
Frequently asked questions
Do I need a statistician for statistics for a medical thesis?
Many medical faculties recommend consulting one, especially for the sample size calculation and for survival or multivariable models. The analysis remains your responsibility, so make sure you understand and can defend each choice.
Should I report p values or confidence intervals?
Both, with priority on effect estimates and their 95% confidence intervals. Report exact p values rather than “significant” or “NS”.
When do I use Fisher’s exact test instead of chi-square?
When expected counts are small; a common rule is more than 20% of cells with expected counts below 5, or any expected count below 1. SPSS reports Fisher’s test automatically for 2 × 2 tables.
What is the difference between sensitivity and positive predictive value?
Sensitivity is the proportion of patients with the disease who test positive. Positive predictive value is the proportion of patients who test positive who actually have the disease, and it depends on prevalence.
Is intention-to-treat analysis required in a trial?
CONSORT asks authors to state which participants were included in each analysis and whether groups were analyzed as randomized. Intention to treat is the standard primary approach, with per-protocol analysis often added as a sensitivity analysis.
Sources
- Schulz, K. F., Altman, D. G., & Moher, D. (2010). CONSORT 2010 statement: Updated guidelines for reporting parallel group randomised trials. BMJ, 340, c332. https://doi.org/10.1136/bmj.c332
- von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gøtzsche, P. C., & Vandenbroucke, J. P. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. The Lancet, 370(9596), 1453–1457. https://doi.org/10.1016/S0140-6736(07)61602-X
- Bossuyt, P. M., Reitsma, J. B., Bruns, D. E., et al. (2015). STARD 2015: An updated list of essential items for reporting diagnostic accuracy studies. BMJ, 351, h5527. https://doi.org/10.1136/bmj.h5527
- Lang, T. A., & Altman, D. G. (2015). Basic statistical reporting for articles published in biomedical journals: The “Statistical Analyses and Methods in the Published Literature” or the SAMPL Guidelines. International Journal of Nursing Studies, 52(1), 5–9. https://doi.org/10.1016/j.ijnurstu.2014.09.006
- International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. https://www.icmje.org/recommendations/
- EQUATOR Network. Enhancing the QUAlity and Transparency Of health Research: reporting guidelines. https://www.equator-network.org/
Sources last checked: November 2026. This article is general educational guidance and does not replace your university regulations or your supervisor’s instructions. How we prepare our content.