Using Lavaan to Estimate Cross-Sectional Structural Equation Models Using R

Social work has a longstanding interest in examining constructs that are inherently latent. However, social work researchers and those in related areas have struggled with analyzing data from studies in which relationships among latent constructs are of focus. In recent years, statistical models and computer software have been developed to better analyze these types of data. Such models are typically referred to as structural equation models (SEM). SEM models allow investigators to test meaningful hypotheses regarding the construct validity of an instrument used to operationalize a construct and relationships between latent and observed variables. The purpose of this workshop is to provide attendees with an overview of common SEM techniques that can be used when analyzing social work data. This presentation has two main objectives.

To meet the first objective, we will introduce the practical concepts driving SEM. Three techniques will be demonstrated, though others will be discussed as applicable. We will begin by examining the construct validity of a widely used scale through confirmatory factor analysis (CFA). We will then identify directly observed predictors of constructs examined in the CFA estimating a multiple-indicators and multiple-constructs (MIMIC) model. Finally, we will estimate an SEM in which we examine the relationships between several latent constructs. In doing this we will discuss the rationale for SEM, the circumstances under which these types of models are useful, and the types of research questions they can answer. We will use a publically available data set to illustrate all SEM concepts (e.g., estimation assumptions, model identification, goodness of model fit, chi-square tests of model comparison, interpretation of coefficients, multi-group analyses), and we will show how much of what participants know about factor analysis and linear regression is applicable (e.g., intercepts, slopes, factor loadings, model fit). After the symposium, we will provide participants with materials that can assist them in performing SEM with their own data (e.g., bibliography of SEM texts and articles, datasets, and other SEM resources).

To meet the second objective, we will demonstrate the ease with which SEMs can be estimated using R, a free and open-source computer language for performing statistical analyses and producing high-quality graphs. R provides several advantages that can benefit social work researchers, such as its packages and their adaptable resources, the flexibility it provides users when executing statistical functions, its ability to simply integrate statistical results in publishable documents (e.g. LaTex and Word documents), and its easy-to-use graphical interface. This presentation will include a brief introduction to downloading R and its available components, but will then move on to demonstrate SEM functions available in R using the “lavaan” package.  Scripts, datasets and additional resources referred to in the workshop will be provided to participants. Social work researchers can easily access R and the “lavaan” package with no cost and use it to improve their research efforts. This workshop will provide the foundation necessary for attendees to get started with R through the introduction of structural equation functions that have relevance to outcomes of interest in social work.