Composite Variables

Abstract

Background Use of composite variables is a common practice, but knowledge about what researchers should consider when creating composite variables is lacking. Objective The purpose of this paper was to present methods used to create composite variables with attention to advantages and disadvantages. Methods Methods of simple averaging, weighted averaging, and meaningful grouping to create composite variables are described briefly, and the context in which one method might be more suitable than the others is discussed. Study examples and comparisons of statistical power among these methods as well as Bonferroni correction are described. Discussion Each approach to creating composite variables has advantages and disadvantages that researchers should weigh carefully. With normally distributed data, composite variables provide the greatest increases in power when the original variables (that make up the composite variable) have similar associations with the outside outcome variable.

The Brain Health Champion study: Health coaching changes behaviors in patients with cognitive impairment

“…For correlation analyses, changes in the three primary behavior domains and QOL were calculated as posttreatment minus pretreatment for each participant, and correlations between changes in the three primary outcomes and QOL were analyzed using Spearman's rank correlation coefficients. A brain healthy behavior composite variable was calculated using the normed average change in each behavior domain (z‐scored), with each behavior change domain weighted equally [42,43]. The relationship between the normalized composite score and normalized (z‐scored) change in QOL was examined by linear regression using maximum likelihood estimation.…”

Multilevel Interventions Targeting Obesity: Research Recommendations for Vulnerable Populations

“…Although a global evaluation has value, composite variables can result in loss of information and be difficult to interpret. 57,58 Investigators need to balance these factors and be careful to insure proper interpretation.…”

Longitudinal Evaluation of the Role of Academic and Social Impairment and Parent-Adolescent Conflict in the Development of Depression in Adolescents with ADHD

“…The use of composite variables, while having advantages, can present challenges for interpretation. Research has shown that combining raters may, in some cases, diminish any one raters’ point of view, however it can make models more parsimonious (Song, et al, 2013). As part of the composite variables, this study chose to incorporate teacher ratings which has its own set of challenges in the middle school setting given that teachers may not have the same operational definitions of behaviors being rated and may be subject to biases (e.g., Kofler, et al, 2010; Youngstrom, Loeber, & Stouthamer-Loeber, 2000).…”

“…Specifically, consistent with past research in this area (Ostrander & Herman, 2006), scores for each measure were standardized and then summed to create the composite variables which ensures that the outcome variable (depression) is not differentially influenced by source variance. Furthermore, composite variables are a clinically meaningful way to capture multifaceted concepts such as academic and social impairment which may be perceived differently by parents, teachers, and the adolescent themselves (Grace & Bollen, 2008) and statistically efficient and appropriate given normality of data and associations between the variables being combined (Song, Lin, Ward, & Fine, 2013). The composite score for the conflict variable had an α of .92 and item to total correlations ranged from .31 to .60 ( M = .47, SD = .10).…”