Research: Finding, Appraising, and Using Evidence
What this domain tests
Candidates must show they can locate, appraise, and apply evidence, not merely cite it. Know the hierarchy of evidence-systematic reviews and meta-analyses above randomized trials, above quasi-experimental designs, above single-group pre-post studies, above expert opinion and anecdote-and the difference between internal validity (did the design support a causal claim?) and external validity (will the finding hold in my population and setting?). Distinguish reliability (consistency of a measure) from validity (does it measure the intended construct?).
Understand sampling logic: probability versus convenience samples, sample size and statistical power, response bias, and attrition. Know major designs well enough to name their weaknesses-randomized controlled trial, quasi-experimental comparison group, interrupted time series, single-group pre-post, cross-sectional survey, focus group, key-informant interview. Recognize ethical requirements: informed consent, voluntary participation, confidentiality, review-board oversight, and honest reporting of null findings.
Traps
Questions often dress correlation as causation, praise a large sample as proof of validity, or describe a reliable instrument as a valid one. Watch for answers that generalize a single-site pilot to an entire region, and for evidence-based programs adopted without checking whether the original trial population resembles the new one. Another trap treats a literature review as a summary of everything published rather than a critical appraisal against a focused question.
Rule of thumb
Before accepting any study, ask in order: What was compared with what? Who was in the sample? How large and precise was the effect? An answer choice that skips these is usually wrong.
Worked example
A program reports that participants attending more sessions improved more, but attendance was self-selected. The defensible reading is association with likely selection bias-not proof that more sessions caused improvement. A randomized or adjusted comparison is needed.