A decision was made across the PROMIS network to administer approximately 150 items to each respondent in the calibration sample, anticipating a testing session of about 30 minutes at a response rate of 5 to 6 items per minute. Approximately 50 items were devoted to questions about demographic and social characteristics and medical history, including 10 questions asking about global perceptions of physical, mental, and social health (Hays, Bjorner, Revicki, Spritzer, & Cella, 2009 (link)). Because of the multiple domains being tested, final item banks for calibration testing were limited to 56 items in each PROMIS domain. We attempted to choose the final 56 items each for depression, anxiety, and anger on the basis of content balancing (i.e., having a representative group of symptoms and complaints associated with each of these constructs) and balancing with regard to likelihood of endorsement (i.e., having items that represented a lower-to-moderate range of severity as well as items reflecting higher severity).