The Psychometric Performance of the PROMIS Smoking Assessment Toolkit: Comparisons of Real-Data Computer Adaptive Tests, Short Forms, and Mode of Administration.

introductionThe PROMIS Smoking Initiative has developed six item banks for assessment related to cigarette smoking among adult smokers (Nicotine Dependence, Coping Expectancies, Emotional and Sensory Expectancies, Health Expectancies, Psychosocial Expectancies, and Social Motivations). This article evaluates the psychometric performance of the banks when administered via short form (SF), computer adaptive test (CAT), and by mode of administration (computer vs. paper-and-pencil).

methodsData are from two sources: an internet sample (N = 491) of daily and nondaily smokers who completed both SFs and CATs via the web and a community sample (N = 369) that completed either paper-and-pencil or computer administration of the SFs at two time points. First a CAT version of the PROMIS Smoking Assessment Toolkit was evaluated by comparing item administration rates and scores to the SF administration. Next, we considered the effect of computer versus paper-and-pencil administration on scoring and test-retest reliability.

resultsAcross the domains approximately 5.4 to 10.3 items were administered on average for the CAT. SF and CAT item response theory-scores were correlated from 0.82 to 0.92 across the domains. Cronbach's alpha for the four- to eight-item SFs among daily smokers ranged from .80 to .91 and .82 to .91 for paper-and-pencil and computer administrations, respectively. Test-retest reliability of the SFs ranged from 0.79 to 0.89 across mode of administration.

conclusionsResults indicate that the SF and CAT and computer and paper-and-pencil administrations provide highly comparable scores for daily and nondaily smokers, but preference for SF or CAT administration may vary by smoking domain.