Question Formats and Item Types
This domain covers the mechanics of item formats and what each format can and cannot measure. The foundational distinction is selected-response (the candidate chooses from provided options: single-best-answer, multiple-response, matching, hot-spot) versus constructed-response (the candidate produces an answer: short answer, essay, performance task). Selected-response items are efficient and objectively scorable; constructed-response items can assess synthesis and reasoning but require trained raters and rubrics, which introduces inter-rater reliability concerns.
Within selected-response, know the vocabulary: the stem poses the problem; the options are the choices; distractors are the incorrect options; the key is the correct answer. A single-best-answer item has exactly one defensible key. A multiple-response item ("select all that apply") has two or more keys and demands a scoring rule - often all-or-nothing versus partial credit. Know that technology-enhanced items (drag-and-drop, hot-spot, simulation) expand what can be measured but increase cognitive load and require careful usability review.
Traps: (1) Assuming "select all that apply" is scored partially - the item must state its scoring rule; absent a rule, don't infer one. (2) Confusing a distractor with a plausible but wrong option - a distractor that no one would choose measures nothing. (3) Believing constructed-response is inherently "better"; it is better only when the construct requires production. (4) Forgetting that item exposure and guessing behave differently across formats.
Rule of thumb: match format to construct. If the objective is recall of a definition, use single-best-answer; if the objective is to apply a procedure across steps, consider a simulation or constructed response. Worked example: an item asks which format best measures "ability to prioritize competing tasks." Recall items can't measure prioritization; a scenario-based selected-response or simulation can.