dySEM: An R Package for Dyadic Structural Equation Modeling with Latent Variables

dySEM is an R package (R Core Team, 2024) that was created to make it easier for users to deploy the widely popular lavaan package (Rosseel, 2012) to fit structural equation models (SEMs) using latent variables to data sets with dependent observations (Little, 2013) collected from interdependent dyads (e.g., romantic partners, pairs of friends, parents, a parent and a child, etc.) (Kenny, Kashy, & Cook, 2006).Initially, the package was exclusively designed to facilitate the testing of dyadic invariance (Sakaluk, Fisher, & Kilshaw, 2021)-the psychometric equivalence of survey responses across members of dyads.However, the package has since been expanded to include functions that streamline the process of specifying, fitting, and reporting on a variety of models for dyadic data that are amenable to specification in SEM with latent variables (e.g., Kim & Kim, 2022;Sakaluk et al., 2025).Currently, dySEM facilitates the deployment of several kinds of dyadic SEMs, including:• "uni-construct" dyadic models: models in which both members of the dyad are measured on the same, singular construct (e.g., both partners complete a measure of relationship satisfaction) (Sakaluk, 2021;Sakaluk et al., 2025); • "bi-construct" dyadic models: models in which each member of the dyad is measured on two different constructs, with one being used to predict the other (e.g., the popular Actor-Partner Interdependence Model [APIM], and the less popular Common Fate Model [CFM]) (Kenny, 1996(Kenny, , 2018;;Kenny et al., 2006;Kim & Kim, 2022;Ledermann & Kenny, 2012) • "multi-construct" dyadic models: models in which each member of the dyad is measured on multiple constructs, such as when both partners are asked to complete a "multidimensional" survey measure (e.g., a measure of relationship quality with subscales for satisfaction, commitment, intimacy, etc.)