In our last article, we discussed the development of the Test Blueprint or Table of Specifications which sets the framework for developing the tests. The quality of a test depends largely on how clearly the test meets its purpose and on the accuracy with which the items match a well-designed blueprint.
The next step is developing the test itself. There are different test formats and each is appropriate and an efficient way to assess understanding of a particular learning outcome. An item is the basic building block of any test. The process of constructing the items required for a given test is referred to as Item Writing, and the people who write these items are called item writers. Items can be questions, tasks, or prompts.
For paper-and-pencil or written tests, there are basically four different formats, being multiple choice response, closed constructed response, open-ended short response, and essays or extended response. Multiple-choice items have one indisputably correct option and several plausible but incorrect distractors. Multiple choice items are dichotomously scored as either correct or not correct.
Closed constructed-response items have one correct answer that the learner generates. They may also require learners to match a series of pairs of sentences or diagrams. Open-ended short-response items require learners to generate a response for which several different but correct answers may exist. Essay or extended-response items require learners to develop a lengthy, sometimes complex, response to a prompt. There are many correct ways to respond in an essay or extended response.
Different item formats can be combined in the same test. Decisions about which format to use in a test and what proportion to use them should be based both on the appropriateness of the format to measure a construct and on practical constraints. High-stakes projects also require the expertise of reviewers, psychometricians, editors and item writing is done in iterative processes, to ensure efficiency and cost savings.
The following are some of the guidelines of writing good items for a test:
The item should be set to measure a single construct/concept.
Items should be set independent of each other. The answer of the preceding item should not be used to answer the next item.
Don’t give the answer away to one item in the stem of another item.
Use the simplest words and sentence structure possible for the items.
Avoiding excessive verbiage. Verbosity is an enemy to clarity. Wordy items take longer to read and answer and sometimes are confusing.
Do not teach during item writing nor introduce new material. Avoid unnecessary background information.
Avoid Opinion-Based Items such as “What would you … do”, “ … use”, “ … try”, etc. because the examinee’s answer can never be wrong.
Use caution when asking for the “best” thing, or the “best” way of doing something, unless it is clearly the best amongst the options. Else, qualify the standard for “best”.
Avoid Absolute Modifiers such as ‘’always’’, ‘’never’’, ‘’only’’ and ‘’none’’. The use of absolute modifiers in multiple choice items’ options makes it easy for test-wise learners to eliminate other options, thus increasing the guessing probability.
Write items as briefly as possible without compromising the construct and cognitive demand required. Get to the point in the stem and present clean, clear options for the examinee to choose.
Do not set tricky items. The intention is not to find out how much the learner does not know, but rather how much they know.
Avoid biased items. Potential sources of bias include cultural, gender, or geographical. There should be no chance that some learners could be offended, frightened, or upset by the test material.
Set items to the right difficulty level. Pretesting items helps in determining the item difficulty levels.
Item layout and design help in making the test clear and appealing. Where possible, images should be used to improve clarity and reduce the number of words in an item
Yes, it’s possible!
