Optimal Scale Length and Single-Item Attitude Measures: Evidence from Simulations and a Two-Wave Experiment

Abstract

Research on the optimal number of survey response options (scale length) has focused on multi-item indices, leaving us with less understanding of how the chosen length of a single-item attitude measure can affect its reliability, validity, and performance in statistical analyses. We employ both Monte Carlo simulations and a two-wave experiment to address these questions. The simulations provide an important basic framework for conceptualizing the issues at play in optimal scale length, while the experiment provides evidence against which the simulation-based expectations can be compared.