How to measure diversity when you must.

Abstract

Racial/ethnic diversity has become an increasingly important variable in the social sciences. Research from multiple disciplines consistently demonstrates the tremendous impact of ethnic diversity on individuals and organizations. Investigators use a variety of measures, and their choices can affect the conclusions that can be drawn and limit the ability to compare and generalize results across studies effectively. The current article reviews 3 popular approaches to the measurement of diversity: the simplistic majority-minority approach and 2 multiple categories variants, the generalized variance and the lesser used entropy statistic. We discuss the properties of each approach and reject the majority-minority approach. We provide 5 examples using the generalized variance and entropy statistics and illustrate their versatility and flexibility. We urge investigators to adopt these multicategory measures and to use our discussion to determine which measure of diversity is most appropriate given the nature of one's data set and research question.

Associations Between Adolescents’ Perceptions of Alcohol Norms and Alcohol Behaviors: Incorporating Within-school Variability

“…The index score was constructed using Likert-scaled responses that might lack the specificity to meaningfully differentiate students from one another without the use of multiple items or a reliable scale. 34 Future alcohol research that uses the generalize variance index should first devise more discrete groupings that approximate students’ norm perceptions by either leveraging categorical variables within the setting or using person-centered approaches (e.g., latent-class analysis) to create these groups from a collection of alcohol use norm perceptions already assessed in the population of interest.…”

“…School-level covariates included school size, percent of students receiving free and reduced-priced meals (FARMS), percent of students suspended, and a measure of the ethnic heterogeneity of the school population. 34,35 …”

“…Variability in norm diversity was measured using a generalized variance approach that takes into account the number of attribute groups possible (i.e., level of student-reported norm perception) and the representation of that group across the entire school population to generate an index score. 34,35 Simply, this school-level variability score represents the probability that two randomly chosen students in a specific school will have different perceived alcohol use norms, with scores ranging from 0 (representing complete homogeneity within that school) to 1 (representing complete heterogeneity).…”

“…To better understand the distribution of adolescents’ social norms within schools, intergroup contact 33 and diversity literature 34,35 provide a framework for modeling and predicting the influence of group variability. Specifically, intergroup contact theory suggests that the more homogenous an environment, the more likely individuals are to interact with those who hold similar attitudes and beliefs.…”

Defining and Measuring Transnational Social Structures

“…In this paper we focus on the Clustered Graph methodology , Lerner et al 2007, 2008, and furthermore, we suggest the application of a diversity index (Budescu and Budescu 2012) The Clustered graph consists of representing personal network data according to some relevant study variables, for instance sending country and host country (see Figure 1). The combination of these two variable form four classes or circles connected among them.…”