Quantitative Methods and Statistical Analysis
Inference and estimation
Candidates should understand the logic of hypothesis testing: null and alternative hypotheses, type I and type II error, power, and the p-value as the probability of data at least as extreme given the null. Know that a p-value is not the probability that the null is true, and that confidence intervals convey both magnitude and precision. Understand that statistical significance depends on sample size and is not the same as importance.
Choosing a test
Know the common tests and their assumptions: t-tests for continuous outcomes comparing means, chi-square for categorical associations, correlation and regression for relationships, and analysis of variance for multiple groups. Understand when nonparametric alternatives are appropriate. Learn the distinction between crude and adjusted estimates, and the role of multivariable models in controlling confounders. Understand effect measures: risk ratio, odds ratio, risk difference, hazard ratio, and number needed to treat.
Multiplicity and missing data
Understand why multiple comparisons inflate type I error and what correction or pre-specification achieves. Know the consequences of complete-case analysis when data are not missing completely at random, and the rationale for multiple imputation.
Traps
- Interpreting a non-significant result as proof of no effect.
- Confusing odds ratio with risk ratio when the outcome is common.
- Ignoring that adjustment for a variable on the causal pathway introduces bias.
- Treating p less than 0.05 as a threshold for truth rather than a convention.
Worked rule of thumb
Report the effect estimate with its confidence interval and the test used. If the interval is wide and crosses the null, the study is inconclusive rather than negative; if it excludes the null but the effect is trivial, significance does not imply importance.