Note: This article is addressed to those interested in writing test items, such as teachers and other educators. However, anyone administering or reviewing tests can benefit from this information.
About Test Items
An item is the basic unit of interaction on a test. What we often call a test question is more properly known as an item, since it may not be worded as an actual question. The student's feedback is also more properly known as a response rather than an answer, but we won't get too particular on that point. Items can be written in various formats, including multiple choice, matching, true/false, short answer, and essay. We will discuss some of these formats in another article.
Since items are the actual points of interaction of students with the test, item quality is probably the most recognizable indicator of the overall quality of the test. High quality test items take time and effort to write, but are essential to a valid test. Items must test skills and knowledge of the subject at hand, not the student's test taking skills.
Seven Characteristics of a Good Test Item
- A good test item is relevant. It should test the learning objective(s) being measured; nothing more and nothing less. This may sound obvious, but when a student who is highly skilled at taking tests scores better on an item than one who is less skilled, even though he has no more knowledge on the subject, this principle is probably being violated.
- A good test item is important. Items must clearly address learning objectives, not trivia. Memorization of obscure facts is much less important than comprehension of the concepts being taught. Trivia, on the other hand, should not be confused with "core" knowledge that is the foundation of a successful education. Examples of "core", nontrivial knowledge include multiplication facts, common formulas, and common geographic names.
- A good test item is comprehensible. Reading difficulty and choice of vocabulary should be as simple as possible relevant to the grade level being tested. This is a corollary of Characteristic #1. If you are not testing reading skills with an item, then do not make reading the item part of the problem. A good author is invisible; that is, you can read his story without being distracted by the style or skills of the storyteller. In the same way, the wording of a good test item should be "invisible". It should be simple, clear, and not a distraction from the concept at hand. In addition, because of this principle, there should be no objection to an item being read verbally to reading impaired students. This, of course, assumes that the item is not intended to evaluate reading skills.
- A good test item is unambiguous. If a word has more than one possible definition, the context in which it is used should leave no reasonable doubt as to which definition is intended. Directions also should contain no ambiguity. If the student is to circle the correct answer, he should not be instructed to mark the correct answer.
- A good test item is straightforward. There should be no trick questions. Tricky items often turn on the meaning of a single word that is not the focus of the item. This is often a flaw in true/false items. Use of the words always and never, and opinions stated as facts are often an unneeded source of confusion to test-takers. If the correct response hinges on a single word, that word should be clearly emphasized. Humor should be used with care as well. The personality of an individual teacher may shine through in the tests he gives his students, but for serious or high-stakes tests, any attempt at humor can be confusing and distracting.
- A good test item is uncontroversial. Items should be supportable facts or qualified opinions, not unqualified opinions. This principle is closely related to Characteristic #5. For selected-response items, there should be an unarguably correct answer. If more than one option could possibly be correct, the directions should call for the best answer, rather than the correct answer.
- A good test item is independent. Items should not provide clues to the answers of other items. Sometimes a series of comprehension items all relate to a single reading passage, or multiple math problems are taken from a single scenario. This approach simplifies item-writing and can be effective, as long as the individual items are still independent of each other. On the other hand, if getting the correct answer on Item #2 depends on getting the correct answer on Item #1, then item #2 tells you absolutely nothing about the skills of the student who missed Item #1. Furthermore, this student is being penalized twice, in effect, for one mistake.