01 · Population, sample, unit
Write three separate nouns: the people the question concerns, the records observed, and what one row represents. Then ask whether one row is a person or a visit. Continue with HEALTH.1.
You do not need to know statistical tests or programming to begin this course. Use this optional review if you want to practise three foundations first: what one row represents, what is being compared, and what a result can and cannot support. If those ideas already feel familiar, continue directly to HEALTH.1.
Read this route from left to right. First decide whom the question concerns and what is being compared. Then confirm what was measured and whether observations come from different people or repeated measurements of the same person. Only after those choices does an analysis method have a clear meaning.
A spreadsheet contains blood-pressure values at discharge and 30 days later. If both rows belong to the same person, the measurements are linked; if they belong to different people, they are separate. A calculation performed before checking that relationship can answer the wrong question even when the software runs without error. Design and measurement determine what the numbers mean. The discussion of units follows the public explanation by Altman and Bland.
For each item, choose “ready” if the sentence already makes sense or “review” if you want the short explanation below. There is no score or pass mark. The activity works without a server and sends no response anywhere.
If you marked an item for review, use the matching activity below. You may also start HEALTH.1 now and return here when the idea appears in context.
Follow the list generated above or browse the eight short activities. Work through only what helps, then return to the question and explain the idea in your own words.
Write three separate nouns: the people the question concerns, the records observed, and what one row represents. Then ask whether one row is a person or a visit. Continue with HEALTH.1.
Name the outcome, the exposure or intervention, and the reference comparison. Do not choose a test until the contrast and time horizon are explicit.
Classify each field as a label, ordered category, count, measurement, date, or identifier. HEALTH.2 turns those labels into a data dictionary.
Two different participants are independent; two visits from one participant are repeated; a deliberately matched control pair is paired. The identifier—not the row count—reveals the structure.
For values 2, 3, 3, 4, 18, compare the mean with the median and identify the unusual value. For every percentage, write its numerator and denominator. HEALTH.3 expands this into the descriptive profile.
If 6 of 24 records meet a rule, calculate 6 ÷ 24 = 0.25 = 25%. Then express the contrast as 6:18 and the percentage-point difference against a stated reference.
Read “effect 2.1; 95% CI 0.4 to 3.8” as one sample estimate plus a range produced by a declared interval method—not as a guarantee that every value is equally plausible. HEALTH.4 develops the interval contract.
Programming is optional. If you choose technical practice, first open the AURORA-30 orientation; it introduces the fictional practice cohort, local files, and guided execution route separately from this course-start orientation.
One completed example: adults starting a blood-pressure program. One row represents one person. The outcome is systolic pressure at day 30, and the comparison is program minus usual care. Each person contributes one day-30 measurement. Notice that the comparison and dependence are stated before choosing a method.
Without software, write five lines for any health-research question: population, unit, outcome, comparison, and dependence. If one line is vague, use the matching review activity. You may begin HEALTH.1 before or after this exercise.