Structural Equation Modeling (SEM) is a modeling approach that makes use of sets of associations that are designed to reflect cause-effect relationships. SEM is a flexible analytic strategy that can incorporate unmeasured latent variables. The process involves modeling the unobservable trait using traits that seem related and are observable, to input the now estimated unobserved variables into a model. The estimated relationships will tell the researcher whether a factor was actually connected and to what degree.

We analyzed the complete HINTS 4 data by implementing Structural Equations Modeling to find the best structure and set of features that underlie and explain the four sub scales of well-being, physical, social/family, emotional, and physical/functional. This research question is modeled using several important factors: Depression rating, Support System, Self-Belief/Independence, Lifestyle (these can include food habits and intentions, exercise, unhealthy tendencies, etc.), information they seek out, who they seek the information for, do they share information or talk to family and friends. We used the lavaan package in R to carry all relevant model fitting computations. We will discover new connections between old variables that could be unexpected and useful in future studies or in practice. We will also determine how the new variables in the subcategories of well-being interact with the old variables as well as how they interact with each other and to what degree of an effect all of these components will have on each other and the latent variables.

Our results reveal new insights into better understanding the complex pathways governing well-being.



Author: Kyle Anderson

Coauthor(s): Cyril Rakovski

Status: Work In Progress