Achieving Efficiency Without Losing Accuracy: Strategies for Scale Reduction with an Application to Risk Attitudes and Racial Resentment

Objectives Researchers often employ lengthy survey instruments to tap underlying phenomena of interest. However, concerns about the cost of fielding longer surveys and respondent fatigue can lead scholars to look for abbreviated, yet accurate, variations of longer, validated scales. In this article, we provide a template to aid in scale reduction. Methods The template we develop walks researchers through a procedure for using existing data to consider all possible subscales along several reliability and validity criteria. We apply our method to two commonly used scales: the seven†item Risk Attitudes Scale and the six†item Racial Resentment Scale. Results After applying the template, we find a four†item Risk Attitudes Scale that maintains nearly identical reliability and validity as the full scale and a three†item Racial Resentment Subscale that outperforms the two†item Subscale currently used in a major congressional survey. Conclusions Our general template should be of use to a broad range of scholars seeking to achieve efficiency without losing accuracy when reducing lengthy scales. The code to implement our procedures is available as an R package, ScaleReduce.