Skip to content
Scintria
ToolsGrant writerFunding
HomeLiteratureClinical trialsEvidence LabProtocol packData analysisLearnProjectsLibraryReferencesApplications
Start free
ScintriaStart free
Scintria Learn

Biostatistics

Ten chapters, thirty lessons: describe data honestly, understand probability and uncertainty, test hypotheses, model outcomes, plan sample sizes and read clinical trials critically. Every lesson has worked examples and a quiz.

0 of 30 lessons read

10 chapters · 30 lessons, each with a quiz

1

Data and description

Know what kind of data you have, summarise it honestly, and show it without misleading anyone.

  • Types of data and scales of measurement12 min
  • Summarising data: centre, spread and shape15 min
  • Graphs that tell the truth12 min
2

Probability and distributions

The language of chance: how likely things are, how tests mislead, and the three distributions you will meet most often.

  • Probability rules and why a positive test can mislead14 min
  • The binomial and Poisson distributions13 min
  • The normal distribution and z-scores13 min
3

Sampling and estimation

How a sample of a few hundred people tells you about millions, and how to say honestly how sure you are.

  • Populations, samples and the ways studies go wrong13 min
  • Sampling variation, standard error and the central limit theorem13 min
  • Confidence intervals for means and proportions14 min
4

Hypothesis testing

How to decide whether a difference is real or chance, what a p-value does and does not say, and how tests are abused.

  • The logic of a significance test14 min
  • Comparing means: the t-tests15 min
  • Rank tests, multiple comparisons and the misuse of p-values14 min
5

Analysing categorical data

Risks, odds and ratios; chi-square and exact tests; and why a crude association can be completely wrong.

  • Risk, odds and the measures of association15 min
  • Chi-square and Fisher's exact tests13 min
  • Confounding, stratification and the Mantel-Haenszel estimate14 min
6

Correlation and linear regression

Describing relationships between measurements, predicting one from another, and adjusting for several things at once.

  • Correlation: strength of a straight-line relationship12 min
  • Simple linear regression15 min
  • Multiple regression: adjusting for several variables14 min
7

Logistic and Poisson regression

Regression for yes/no outcomes and for counts and rates: the workhorses of epidemiology and trials.

  • Logistic regression: modelling a yes/no outcome15 min
  • Prediction models: discrimination, calibration and ROC curves14 min
  • Poisson regression: counts, rates and person-time14 min
8

Survival analysis

Time-to-event data: when people drop out of view before the event, simple averages and proportions mislead. These tools do not.

  • Time-to-event data, censoring and the Kaplan-Meier curve15 min
  • Comparing survival curves: the log-rank test and hazard ratios14 min
  • The Cox proportional hazards model14 min
9

Sample size and power

How many participants do you need? The question every protocol, ethics committee and funder will ask first.

  • Why sample size matters: the four ingredients of power12 min
  • Calculating sample size for means, proportions and events15 min
  • Clusters, non-inferiority, attrition and writing the section14 min
10

Statistics in clinical trials and evidence synthesis

Analysis populations, missing data, interim looks, and combining trials: where biostatistics meets regulation.

  • Intention-to-treat, per-protocol and missing data14 min
  • Interim analyses, data monitoring and multiplicity in trials14 min
  • Meta-analysis, heterogeneity and reporting standards15 min