When you finish an ARRT exam, your score report does not show "X out of Y correct." Instead, you receive a scaled score from 1 to 99 where a score of 75 represents passing the exam. Importantly, this scaled score is NOT a percentage. A score of 75 does not correspond to answering 75% of exam items correctly: in fact, the number of correct answers needed to earn a 75 can vary depending on the exam form taken. ARRT regularly updates its exams, and multiple forms may be used at the same time for any given discipline. Even when forms match the same content specifications, they can be slightly easier or harder. To keep scoring interpretable across these differences, ARRT uses scaled scoring.
Scoring begins in a simple way. Each correct answer earns a candidate one raw point (in other words, items are not weighted). However, a raw score (the total number correct) can mean different things if one form is harder than another. To adjust for this difference across forms, ARRT uses the Rasch model to estimate item difficulty and overall form difficulty. That's why two candidates can both earn a scaled score of 75 on different exam forms even if their raw scores differ.
A = (99.49 - 74.50) / (C - D)
B = 74.50 - (A x D)
Scaled Score = (raw score x A) + B
We call the value A the slope and the value B the intercept. If you have ever heard of y = mx + b, that's exactly what we're doing here: finding the two values (m as the slope, or b as the intercept) that specify a line. In this case, the line connects the raw score metric and the scaled score metric.
In short, scaling helps ensure your score reflects your performance, not which form you happened to receive. Scaled scoring also preserves the meaning of "75" across exam versions and over time. It supports comparable pass/fail decisions even as ARRT exams evolve.
For more examples and additional information on ARRT scaled scores, interested readers are encouraged to explore the ARRT Technical Appendix, which can be found on the ARRT website here: ARRT Annual Exam Report - Technical Appendix 2024.pdf.
