The Application of Structural Equation Modeling (SEM) in LIS Studies: Theory making and testing

Researchers have to conduct multivariate investigation in most of LIS studies, especially attitudinal and behavioral related fields. Most of these variables are unobservable and immeasurable (known as conceptual constructs or latent variables). Researchers estimate these latent variables via observable factors or indices. Most of them aim to investigate causal relationship between latent variables through observable variables. Conventionally and typically, statistical methods like variance, regression, path, and factor analysis are used in such studies. SEM is the last statistical tool which not only prepares the previous kinds of analysis, but also provides estimate errors in both observable and unobservable variables, theory/model modification, and choosing the best between compared different models. SEM is used both in Confirmatory and Exploratory approaches, in social sciences to test theoretical models. Through analyzing 178 research articles we aimed to: 1) Introduce SEM to LIS researchers; and 2) Clarify SEM use state of the art in LIS studies and draw a close look of fields, topics, variables and theories which have been studied and analyzed via SEM method.