texva
Start free

Blog ·

The twelve question types a THPT paper can use, and what each one measures

The twelve types do not measure the same thing. What each one is actually good at, the cognitive level it fits, and the mistake each one invites.

Choosing a question type gets treated as a formatting decision — multiple choice to mark quickly, written answers for the hard part. But the type decides what you can measure, and a blueprint that looks balanced on paper still measures crookedly if every cell uses the same one.

The selection types

  • Single-answer multiple choice — good for recall and understanding, weak for application because elimination works. The usual mistake: filler distractors, which turn the question into a test of elimination.
  • Multi-statement true/false — measures application within one context and discriminates better because marks scale with correct statements. The usual mistake: statements that depend on each other, so one error costs two marks.
  • Matching — measures relationships between concepts, which plain multiple choice rarely reaches. The two columns must be unequal in length, or the last pair is a gift.
  • Multiple response — useful when the correct answer genuinely has several parts, but the partial-credit rule has to be stated before the paper goes out.

The constructed types

  • Short numeric answer — measures real calculation because there is nothing to guess from. Settle the units and the rounding while writing; an answer of 3.75 against a key that says 3.8 costs every correct student a mark.
  • Fill in the blank — measures precise terminology and recalled formulas. Only use it where exactly one answer is right; otherwise you are marking phrasing.
  • Short written answer — measures step-by-step reasoning, and the only type that shows you where a student went wrong. It needs a per-step scheme, not one number for the whole question.
  • Extended writing — measures synthesis and presentation. The most expensive to mark, so one or two per paper at most, and always with a rubric or two markers will disagree.

The types that hang off data or a figure

  • Reading a graph or variation table — measures interpreting a representation, separately from computing. The figure must be vector; an enlarged screenshot has students misreading the axis.
  • Contextualised questions — measure high-level application. The context has to be load-bearing; if the question still solves without it, the context is decoration that only slows readers down.
  • Question sets sharing one stimulus — save space and reach depth, but never shuffle their order when generating paper codes, and never let a later item depend on an earlier answer.
  • Questions requiring the student to draw — measure genuine geometric thinking, but only when the answer sheet leaves real room to draw.

Templates for this

More posts

Start free — no card needed