HLT 3672 Topic 3 DQ 1 hypothesis testing guide

Two of these four prompts start with evaluate and two start with define or describe, and the difference is worth real marks. Define and describe are answered by an accurate statement. Evaluate asks you to weigh something against something else and reach a judgement, so prompt one needs a comparison — one-tailed against two-tailed, parametric against non-parametric, superiority against non-inferiority — with a sentence saying when each is the right choice in a health care setting. Answering an evaluate prompt with a definition is the commonest way to lose marks on a discussion question that otherwise looks complete and covers every bullet. Notice too that prompt one says application to health care specifically, so at least one clinical example should appear: a trial comparing thirty-day readmission rates between two discharge protocols, a screening study with a new cut-off, an infection rate measured before and after a hand hygiene intervention.

The variables prompt looks trivial and hides a trap. The independent variable is the one you manipulate or group by — the treatment arm, the protocol in use, the exposure a patient did or did not have. The dependent variable is what you measure — the outcome. Their role in hypothesis testing is that the null hypothesis is a statement about the dependent variable being unaffected by the independent one, so identifying them correctly is what tells you which test applies. That last clause is the part markers look for, because the level of measurement of the dependent variable, together with the number of groups in the independent variable, is what selects a t-test rather than a chi-square rather than an analysis of variance. Give one worked clinical example rather than an abstract restatement, and state the test your example implies so the reasoning is visible rather than left to be inferred.

The last two prompts are more closely linked than they appear, and answering them together produces a stronger post. The evidence for rejecting the null hypothesis is the test statistic and its p-value compared against a pre-set alpha, and the honest version says what a p-value is not: it is not the probability the null is true, and it is not a measure of effect size. Failing to reject is not proof of no difference, which is why an underpowered study is uninformative rather than negative. The confidence interval then carries exactly the same information with more of it visible — if a 95% interval for a difference excludes zero, the test rejects at 0.05 — but the interval also shows the direction and the plausible magnitude, which is why clinical papers increasingly lead with it. Make that link explicitly and the fourth prompt answers itself.