Scaling back survey scales - How short is too short?

To understand childrens development, one must examine an array of constructs. Yet the time and budget constraints of large-scale survey research create a dilemma: how to cut scales to the fewest possible items while still retaining their predictive properties. In this article, the authors compare the predictive validity of several shortened versions of the Behavior Problems Index and the Home Observation for Measurement of the Environment-Short Form with their full scales within the National Longitudinal Survey of Youth-1979 cohort. They use the scales to predict delinquency, reading recognition scores, parent-child activities, and smoking behavior of children at ages 13 or 14 from data gathered 2,4, and 6 years prior Analyses leave the authors cautiously optimistic that short scales, especially scales composed of items gathered at different time points and repeated regularly, may enjoy substantial predictive power However, two-item scales may be too short, and psychometric study on additional scales is warranted.