In paper-and-pencil testing, multiple-choice questions are typically presented in a test booklet and test takers record their final answers on a separate scannable answer sheet. Test takers are often permitted to use the space in their test booklets for scratchwork, which commonly consists of notes, diagrams, underlining of text, or other types of markings. Scratchwork and the relationship it may have with test-taker ability on a large-scale, high-stakes multiple-choice test is the focus of this research. If there is a relationship between test-taker ability, as measured by the number of correct responses, and the amount of scratchwork on the test booklet, this information can potentially be used as secondary evidence to support questioning a test taker’s score if cheating is suspected. As an added benefit, this exploration can also inform the development of a test-taker interface while transitioning from the paper-and-pencil format to other digital formats of test delivery. Three hundred randomly selected test booklets from a single test administration were used. The 25th, 50th, and 75th percentiles of the score distribution were chosen to draw a random sample of about 100 test-taker registration numbers at each of these score levels, and the relationship between the number of correct responses and the amount of scratchwork was explored.

Keywords: scratchwork, multiple-choice, test security