Scoring/Reporting

Criterion-Reference Testing Licensure exams, such as the ones administered by the National Board of Examiners in Optometry, are criterion-referenced tests. A criterion-referenced test bases a candidate’s score solely on her or his knowledge of the content, without reference to other candidates. In contrast to this, norm-referenced tests provide percentile rankings of test-takers, meaning that one person’s score is in reference to the rest of the test-takers. Some examples of norm-referenced tests include the SAT and intelligence tests. Norm-referenced tests provide percentile rankings to convey a candidate’s score relative to other candidates and are appropriate in certain contexts such as school admission decisions. However, criterion-referenced tests compare candidates’ performance to an absolute standard, and not to the performance of others. Criterion-reference tests are appropriate for licensure because there is no limit to the number of candidates who can pass or fail. Every candidate’s performance is judged independently against the absolute standard, regardless of when the test is taken.

Performance on a criterion-referenced test is judged against a pre-determined standard of performance, and therefore, is scored the same regardless of when or where the exam is completed. Norm-referenced exams utilize a normal curve in setting percentile rankings. Criterion-referenced exams do not utilize a normal curve in scoring.

Establishing the pass-fail cutoff score for an examination is accomplished through a process referred to as standard setting. A standard setting study involves convening a variety of experts and stakeholders in order to account for all aspects of the examination including the format and content of the exam, the candidate population, the necessary level of knowledge in the field, and the implications of various decisions points and scores. Because these are high stakes examinations, NBEO uses only state of the art psychometric practices to set its standards and to maintain the integrity of the scores.

The goal of a criterion-referenced standard setting is to determine how much knowledge is “enough” in order to pass an exam. Additionally, NBEO uses different standard setting methods appropriate to the distinct formats of the Part I ABS, Part II PAM, and Part III CSE to determine each cut score.

Although NBEO establishes overall pass-fail standards, it does not make licensure decisions. Licensure is the legal responsibility of the state boards of optometry, some of which require that candidates attain individual section standards. Therefore, it is possible for candidates to meet the NBEO threshold for passing, but not a state boards’ standards. To avoid confusion, all candidates are advised to contact the administrative offices of the state boards’ to which they seek licensure to fully understand the manner in which NBEO scores are utilized.

For NBEO examinations a total raw score is computed and then converted to a scaled score ranging from 100 to 900. A score of 100 represents the lowest scaled score, while 900 represents the highest possible scaled score. A score of 300 represents the overall pass-fail cutoff score. The use of scaled scores allows for direct comparison of scores from one examination administration to another because the passing standard is always the same -- a scaled score of 300.

Sections and special examinations are reported on a 0 to 99 scale. On this scale, zero represents the lowest possible score, 99 represents the highest possible scores, and 75 represents the pass-fail cutoff score. These scores also are transformed using a single scaling equation.

Section scaled scores cannot be added to reproduce the overall scaled scores. This is because sections have a different number of items and overall point value. A simple sum of these scaled scores would skew the outcome. The scaling of the section from 0 to 99 does not have a summative translation to the scaling from 100 to 900. That is, adding up the scaled scores in sections is not equivalent to the overall scale score due to the process of scaling itself.