The Practice of Citizen Satisfaction Measurement: Statistical Analysis and Modeling

Abstract

In the last chapter, we reviewed a number of important steps in a citizen satisfaction measurement project: agency and population-of-measurement selection, the identification of drivers and outcomes to measure within a survey alongside the satisfaction concept, various perspectives on how the satisfaction concept can be operationalized, questionnaire design, interviewing methods, and sample size considerations. In this chapter, we proceed by considering some of the statistical methods that can be used to examine the data collected through a satisfaction survey. While statistical methods for examining survey data are numerous and vary widely, and while researchers often have their own preferred set of methods (and sometimes cling to these methods passionately), here we will suggest that the type of statistical analysis we ultimately recommend—a multiple-item measure, latent variable, structural equation modeling approach (sometimes also called a “path modeling” approach)—is one often-used and well-tested technique for generating valid, reliable satisfaction scores and for dynamically relating satisfaction to its key influencing factors and its outcomes. In turn, this kind of data analysis can offer useful, accurate, and actionable insights into the practical reforms government agencies can make toward improving citizen satisfaction, as well as related outcomes like citizen trust.