Calibration sets serve as tools to prevent scorer drift from the standards exemplified in the scoring guide and anchor and practice papers. Calibration takes place after any sustained break in scoring (e.g., lunch break) or at the discretion of the trainer. The trainer selects sets of approximately 10 to 20 papers (from the training trend set or current-year responses.) These sets of 10 to 20 papers are called calibration sets. The trainer also may use a calibration set to focus the team’s attention on specific item-level scoring issues that have been discovered in backreading. When a calibration set is used, each scorer scores the same set of papers; the trainer reviews the results and, if necessary, reviews the scoring standards with the team.