Psychometric Rigor for the Real World: Item Banking in Developmental Research

Measurement—a critical component of scientific progress—of attributes in developmental psychology provides a number of challenges. I argue that probabilistic models of item responses resolve many of these challenges and deliver desirable features. Such models allow us to systematically investigate whether measures function equivalently across groups of people, make it possible to deliver consistently meaningful scores from different versions of the measurement instrument (which may be better suited to different ages or developmental periods), and can be used to efficiently maximize the value of time spent with the respondent. To further enhance usability of a measure, its functioning can be codified in the form of an item bank. Item banks, databases containing the information required for reuse of the item, can streamline future measurement tasks given that items can be screened for invariance and selected given the developmental stage of the respondents in question. To illustrate these points, I discuss three recently developed item banks—the National Institutes of Health Infant and Toddler Toolbox, LEVANTE, and the D-score system—meant to manage the kinds of challenges discussed herein.