Biostatistics and Health Data Analysis
Biostatistics and Health Data Analysis
This domain tests data types, descriptive and inferential statistics, sampling, hypothesis testing, regression, and diagnostic test interpretation. The exam favors correct interpretation over calculation alone.
Core rules and concepts
- Numerical data can be continuous or discrete; categorical data are nominal or ordinal. Choose medians and interquartile ranges for skewed data, means and standard deviations for roughly symmetric data.
- Sampling error decreases with sample size; bias does not automatically. Randomization addresses confounding in trials; blinding reduces measurement and performance bias.
- A 95% confidence interval estimates a range of plausible values for the parameter under repeated sampling. A p-value is the probability of data as extreme or more extreme if the null hypothesis is true; it is not the probability that the null is true.
- Type I error is a false positive; type II error is a false negative. Power is 1 minus type II error. Chi-square tests compare proportions; t-tests compare means; regression estimates adjusted associations.
Worked example
In 1,000 screened people, prevalence is 10%, sensitivity 90%, specificity 95%. True positives = 0.90 × 100 = 90. False positives = 5% of 900 = 45. Positive predictive value = 90/(90 + 45) = 66.7%. Even with good sensitivity and specificity, two-thirds of positives are true positives because prevalence is modest. If prevalence fell, PPV would fall further.
Traps that cost marks
Do not say a p-value is the probability the hypothesis is true. Do not interpret a confidence interval as a 95% probability that the fixed parameter lies inside it. Do not confuse statistical significance with clinical importance, or sensitivity with predictive value.