12 Tips for Writing Good Test Questions

Posted by Joan Phaup

Writing effective questions takes time and practice. Whether your goal is to measure knowledge and skills, survey opinions and attitudes or enhance a learning experience, poorly worded questions can adversely affect the quality of the results.

I’ve gleaned the following tips for writing and reviewing questions from Questionmark’s learning resources:

1. Keep stems and statements as short as possible and use clear, concise language.toolbox
2. Use questions whenever possible (What, Who, When, Where, Why and How).
3. Maintain grammatical consistency to avoid cueing.
4. List choices in a logical order.
5. Avoid negatives, especially double negatives.
6. Avoid unnecessary modifiers, especially absolutes (e.g. always, never, etc.).
7. Avoid “All of the above” and use of “None of the above” with caution.
8. Avoid vague pronouns (e.g. it, they).
9. Avoid conflicting alternatives.
10. Avoid syllogistic reasoning choices (e.g. “both a and b are correct”) unless absolutely necessary.
11. Avoid providing cues to correct answer in the stem.
12. Avoid providing clues to the answer of one question in another question.

If you would like more information about writing question and assessments, a good place to start is the Questionmark Learning Cafe.

5 Responses to “12 Tips for Writing Good Test Questions”

  1. sven says:

    Did you hear about SAQ ? It’s a free online tool developed by the University of applied Sciences Western Switzerland. SAQ helps to evaluate the quality of test questions (before submitted to students). Actually the analysis results are only given in French, but you can submit questions written in English and set the right parameters.

    The questions have to be submitted as IMS/QTI files (Questionmark allows to export to this format).

    Here a blog entry about SAQ:

    SAQ can be found here: http://saq.heig-vd.ch

  2. […] received a tweet about tips for writing test questions on the Questionmark blog. Here is the list of tips they […]

  3. Claudia Miller says:

    AT what % of my test group missing a question should I consider it “bad” and need re-writing ?

  4. Joan Phaup Joan Phaup says:

    It depends, since the number of people who miss a question is only part of the issue.

    For instance, if you were doing criterion-referenced testing (CRT) for pilots who must absolutely know how to lower the landing gear, you wouldn’t want to change the question because it is matched perfectly to the job. If people are failing such a question you may want to improve the instruction or performance support system rather than the question.

    If you are trying to be norm-referenced instead of criterion-referenced, you would aim for questions in a mid-range difficulty. Or, if you need to separate at the edge of the cut, you would aim for higher difficulty. It’s important to look not just at how many people miss a question but also at who get the question right and who gets it wrong: What if 80% get a question right but they turn out to be the ones with the lowest scores? And if the people with the highest scores missed the question, this could mean it doesn’t make sense to them. In this case, it’s time to rethink the question.

    Criterion-referenced Test Development by Sharon Shrock and William Coscarelli explains these concepts in full.

    Reporting tools in Questionmark Perception, including the Item Analysis Report, can take the guesswork out of identifying problematic questions.

Leave a Reply