All the work of examination development funnels down to a single number, the cut score, a decision on who passes and who fails. Each stage of exam construction contributes to the overall quality of this decision, but one of the most critical is establishing the criterion referenced standard.
Manager, Test Development and Analysis
|Criterion Referenced Standard Setting for Multiple Choice Examinations
The purpose of a criterion-referenced standard is to determine "how much is enough" to be considered capable to practice in the field. According to the Standards for Educational and Psychological Testing, verifying the appropriateness of the cut score is critical for validity. The validity of the pass or fail inference depends on whether the standard for passing makes a valid distinction between acceptable and unacceptable performance (p. 157). The level of performance required for passing should depend on the knowledge and skills necessary for acceptable performance and should not be adjusted to regulate the number of candidates passing the test (p. 162). These standards suggest that an absolute expectation for passing should be established prior to the examination based on the examination content. When a criterion referenced standard is established, theoretically all candidates can pass the examination or all candidates can fail, depending on their ability levels.
For multiple choice examinations, popular methods of establishing a criterion referenced standard are modified Angoff approaches which require expert judgment concerning the difficulty of each item based on an assessment of the content. Theoretically, each item is a criterion. The question usually is 'what is the probability that the minimally capable candidate would answer the item correctly?' Participants consider the difficulty of the content of the item, relevance, and frequency, and make the assessment. The Angoff method is relatively convenient, has a long history of use, can be learned quickly by the subject matter experts, and has been subject to extensive research.
The criterion referenced pass point depends, in part, on the subjective judgments of experts. Even with training, the outcomes of a standard setting exercise may be different when different methods, experts, or items are used in the process. Thus, there is error associated with the criterion referenced standard setting process. Measurement error can increase the probability that able candidates will fail or less able candidates will pass. Therefore, adjusting for the error of measurement surrounding the pass point when setting the criterion standard, is legitimate and provides confidence that only candidates who should pass, will pass. Once the criterion referenced standard is established, test equating methods allow it to be carried forward to subsequent tests, so that candidates are required to meet the same standard to pass regardless of when they take the examination.