Question Validation
Validation provides documentation and rating of question quality
After a panel of experts has approved a set of questions for use on the exam, the
validation process allows the experts to rate the questions on scales such as importance,
criticality, and frequency of use for competent performance on the job. It is important
to quantify and document the quality of the questions for several reasons.
First, question validation provides concrete documentation of the relevance of all questions placed on the exam. This becomes key evidence to help defend against challenges
about the exam content. In essence, it provides documentation that a panel of subject
matter experts convened, discussed the relevance of every item, and agreed that
they should be placed on the exam.
Another use of the validation process is in the assembly of the examination. If the experts developed more questions for a blueprint area than necessary, it is
important to quantify the quality of a question to allow selection of only the best
questions. For this reason, the ratings of the panelists are combined using a statistical
formula and translated into a 5-star scale. This way, the questions with five stars
can be selected in place of those only receiving one or two stars.
In fact, ExamDeveloper has the capability of automatically building a draft
exam using item quality and other criteria such as blueprint, cognitive complexity
level, item difficulty, and category type. Once the manager specifies the number
of versions to be built and the thresholds for the criteria, the draft exam
versions are immediately available for viewing by the expert panel. No other exam development software comes close to the capability contained in ExamDeveloper.
Manually building an exam using multiple criteria is a complex, time consuming,
and often error prone task. Using ExamDeveloper to build exams will save your organization
dozens of hours and prevent costly mistakes that occur when trying to trying to
track everything on paper or spreadsheet.
Next Step
Too many is better than not enough
People frequently ask how many questions each user needs to write in order to have
enough for the exam.
A common rule of thumb is to assign 150-200% of the number of questions that the final
(completed) exam versions will contain. For example, if final exam will contain
150-questions, it is advisable to assign 150-200 questions over all areas of the
blueprint.
A more complicated example is the task of assembling two versions of a 100-item
exam that will share 50% of the content between the two versions. In this case,
it would be advisable to assign between 225-300 questions (1.5 or 2 times 150 questions,
50 shared questions + 50 unique to each version).
This information is explained fully in a self-paced tutorial found within the application.
Of course, the Exam Design consultants would be happy to assist with this or any
other question you have about the design or maintenance of your examination program.