Writing Test Questions
Below are some suggestions faculty may wish to use when constructing various types of summative assessments. In addition, the following handout regarding multiple choice questions may be useful.
- Mix item types.
- Create items to reflect course goals and activities (i.e., If the emphasis of the class is knowledge acquisition, do not test performance).
- Design the test to reflect the emphases the information and activities have received in class.
- Distribute criteria in advance.
- Make sure students understand the structure of the test, criteria.
- If using commercial or testing bank items, make sure they reflect the goals and activities of the course.
- Evaluate each test and save valid and reliable items to create your own test bank.
- Avoid writing items that link to each other: the answer to one question should NOT depend upon the answer to another.
- Start with easy items and progress to more difficult ones.
- Have someone else proofread the exam.
- Pilot test the exam.
- Anticipate special needs students may have (i.e., special testing environments, readers, etc.).
- Explain procedures in advance.
- During the exam, write last minute changes or answers to questions on the board.
- To keep students on track, write remaining time on the board rather than calling it out.
Constructing True/False Items
- Make items simple and clear.
- Use a large number to minimize guessing.
- Reword items that test key information and compare answers to minimize guessing and determine if the knowledge has been acquired.
- Avoid taking items directly from the text.
- Avoid complex statements that test reading skill.
- Use an automatic grading format (circle T or F, ScanTron sheets, etc.).
- Make positive statements and especially avoid double negatives.
- Don't try to trick the students.
- Make sure the statements are entirely true or entirely false. If a question may be open to interpretation, provide a space for clarification.
- Be aware of how students interpret vocabulary.
- False = all, always, never, every, none, only
- True = generally, sometimes, usually, maybe, often.
- Use precise items: (i.e., 50% of the time) rather than several, seldom, frequently, etc.
- Use more false items since they tend to discriminate more than true ones.
Constructing Matching Items
- Keep like items in the same set (i.e., events and dates belong together, dates and places do not).
- Give more information in the stem (left hand column) and keep matches short.
- Make directions clear: can responses be used more than once? Where does the answer go? Can students infer relationships or are they based on real world logic?
- Each stem should have only one response (but responses can be used more than once in some cases).
- Don't let grammatical cues such as tense signal the correct response to a stem.
- Arrange response column in a logical order (alphabetical, etc.) so students can find them quickly and easily.
- Keep a set together on a single page to avoid wasting time in flipping back and forth.
- Keep sets short.
- Give more responses than stems to minimize guessing.
Constructing Completion Items
- Use original questions.
- Make questions clear through the use of concise clues and familiar vocabulary.
- Provide grading criteria in advance so students will understand what you are looking for in each item.
- Establish a grading strategy (i.e., use a computer form, fill in a grade sheet).
- Provide enough context so only one word fits.
- Avoid grammatical cues when testing information recall.
- If items may be ambiguous, add an explanation area.
Constructing Open-Ended and Essay Items
- Use comprehensive or global questions.
- Provide clear instructions.
- Give students grading criteria in advance (i.e., points for accuracy, completeness, creativity, interpretation, repetition, etc.).