Descriptive vs Inferential Statistics in Medical Research
The difference between describing your sample and drawing conclusions about a population, with examples of what belongs in each part of a medical thesis.
Every results chapter has two jobs. First, describe what you found in your sample. Second, draw conclusions about the wider population. These two jobs use two families of statistics: descriptive and inferential.
Descriptive statistics: what did you observe?
Descriptive statistics summarise your sample. They make no claims beyond the people you studied.
| Variable type | Typical summary | Typical display |
|---|---|---|
| Categorical | Frequency and percentage | Table, bar chart |
| Numerical, symmetrical | Mean and standard deviation | Table, histogram |
| Numerical, skewed | Median and interquartile range | Table, box plot |
Example: "Of 220 participants, 118 (53.6%) were women. The mean age was 42.5 years (SD 11.2)."
The first table of most theses and papers (often called "Table 1") is descriptive. It shows the baseline characteristics of the participants, overall and by group.
Inferential statistics: what can you conclude?
Inferential statistics use your sample to make statements about the population, with an estimate of uncertainty. They include:
- Estimation: confidence intervals for a mean, proportion, difference or ratio.
- Hypothesis testing: tests such as chi-square, t-tests and ANOVA that give p-values.
- Modelling: regression to examine several factors together.
Example: "Of 220 participants, 63 had hypertension: a prevalence of 28.6% (95% CI 22.6% to 34.6%). Hypertension was more common in participants with a BMI of 25 or above (odds ratio 2.4, 95% CI 1.3 to 4.4; p = 0.005)."
How they fit together in a thesis
- Describe the participants (descriptive).
- For each objective, present the main estimate with a confidence interval (inferential).
- Present comparisons or associations with the appropriate test (inferential).
- Use tables for detailed numbers and a few well-chosen graphs for key findings.
A note on baseline comparisons in trials
In randomised trials, testing baseline differences between groups with p-values is generally discouraged, because any differences are due to chance by design. Describe the groups instead and comment on any important imbalance.
Common mistakes
- Reporting only percentages without the numbers (write 53.6%, n = 118).
- Using mean (SD) for skewed variables such as hospital stay.
- Presenting p-values without effect sizes or confidence intervals.
- Generalising sample findings to a population your sample does not represent.
Key takeaways
- Descriptive statistics summarise your sample; inferential statistics generalise with stated uncertainty.
- Choose the summary by variable type and distribution.
- Pair every key estimate with a 95% confidence interval.
- Your sampling method limits how far you can generalise.
References
- Kirkwood BR, Sterne JAC. Essential Medical Statistics. 2nd ed. Oxford: Blackwell Science; 2003.
- Mishra P, Pandey CM, Singh U, Gupta A, Sahu C, Keshri A. Descriptive statistics and normality tests for statistical data. Ann Card Anaesth. 2019;22(1):67–72. PubMed
- Hopewell S, Chan AW, Collins GS, et al. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ. 2025. PubMed
This article is for education. Always follow the requirements of your university, guide, ethics committee and target journal.