An overview of the role of biostatistics in public health, basic statistical concepts, types of data, and the importance of statistical analysis in public health research.
Introduction to Biostatistics
Key concepts, terminology, and tools for applying statistical methods to biological and health‑related data.
| Aspect | Description |
|---|---|
| Definition | The branch of statistics that deals with the collection, analysis, interpretation, and presentation of data arising from biological, medical, and public‑health research. |
| Purpose | • Summarize complex biological data • Quantify uncertainty and variability • Test scientific hypotheses • Inform clinical and policy decisions |
| Typical Applications | Clinical trials, epidemiologic studies, genetics, environmental health, health‑services research, bioinformatics. |
| Data Type | Scale | Examples | Typical Analyses |
|---|---|---|---|
| Nominal (Categorical) | No intrinsic order | Sex, blood type, disease presence (yes/no) | Chi‑square test, logistic regression |
| Ordinal | Ordered categories | Pain rating (none, mild, moderate, severe) | Mann‑Whitney U, ordinal logistic regression |
| Discrete Count | Integer values, often non‑negative | Number of infections, gene mutation counts | Poisson/negative‑binomial regression |
| Continuous | Measured on a real‑number scale | Blood pressure, cholesterol, gene expression levels | t‑test, ANOVA, linear regression, survival analysis |
Note: Many biological variables are right‑skewed (e.g., serum triglycerides). Transformations (log, square‑root) are often used to meet model assumptions.
| Design | Key Features | Primary Statistical Considerations |
|---|---|---|
| Cross‑sectional | Snapshot of a population at one time point | Prevalence estimation, logistic regression |
| Case‑control | Subjects selected by outcome status (cases vs. controls) | Odds ratios, conditional logistic regression |
| Cohort (prospective/retrospective) | Follow subjects over time to observe incidence | Relative risk, Cox proportional hazards model |
| Randomized Controlled Trial (RCT) | Random allocation to intervention vs. control | Intention‑to‑treat analysis, ANCOVA, survival analysis |
| Longitudinal / Repeated Measures | Multiple observations per subject | Mixed‑effects models, generalized estimating equations (GEE) |
| Clustered / Hierarchical | Groups (e.g., clinics, schools) as sampling units | Multilevel models, intraclass correlation coefficient (ICC) |
| Concept | Definition | Why It Matters in Biostatistics |
|---|---|---|
| Population vs. Sample | Population = all possible units; Sample = subset actually observed | Inference relies on sample representing the population |
| Parameter vs. Statistic | Parameter = true value in population; Statistic = estimate from sample | Estimation (e.g., mean, proportion) aims to approximate parameters |
| Bias | Systematic deviation of an estimate from the true value | Study design and measurement methods must minimize bias |
| Variance & Standard Error | Variability of an estimator; SE = SD of sampling distribution | Determines precision of estimates and width of confidence intervals |
| Confidence Interval (CI) | Range of values that, with a specified probability (e.g., 95 %), contains the true parameter | Provides a measure of uncertainty beyond a single point estimate |
| p‑value | Probability of observing data as extreme as (or more extreme than) those observed, assuming the null hypothesis is true | Used |