| Feature | Statistical Association | Data Validity |
|---|---|---|
| Focus | How variables relate to each other | How reliable the data is |
| Metric | Correlation coefficients, Odds Ratios | Sample size, control variables |
| Conclusion | 'Factor X is linked to Disease Y' | 'The study results are/are not representative' |
Sample Size: Small groups may produce anomalous results that do not reflect the wider population, leading to misleading Odds Ratios.
Confounding Variables: Factors like diet, exercise, or genetics must be controlled to ensure the observed effect is actually due to the risk factor being studied.
Analyze Trends: When presented with a graph, always check if the trends for both variables move in the same direction (positive correlation) or opposite directions (negative correlation).
Question Causality: If an exam question asks if a factor 'causes' a disease based only on a graph, the answer is usually 'no' or 'not necessarily' because correlation alone is insufficient proof.
Evaluate Anomalies: If data contradicts known biological facts (e.g., smoking appearing to reduce disease risk), look for flaws in the study such as a lack of statistical testing or small sample size.
Check the Units: Always verify the axes; for example, cancer rates might be per 100,000 people while smoking rates are percentages.
The 'Lag Time' Error: Students often forget that there is a delay between exposure to a risk factor (like smoking) and the development of a disease (like lung cancer). Trends may not align perfectly in time.
Misinterpreting OR < 1: A value below 1.0 does not always mean a factor is 'healthy'; it may indicate a flawed study or the influence of an unmeasured variable.
Overgeneralization: Assuming that a trend found in one specific demographic (e.g., males in the 1950s) applies universally to all groups today.
