In research, we often want to understand how different concepts are connected. Simple statistics can tell us if two things are related, like how hours spent studying might relate to a test score. But what if we want to test a whole theory at once? For instance, how does a student's home environment and their motivation together influence their study habits, and how do those habits then lead to a final grade?

This is where Structural Equation Modeling (SEM) comes in. It's a powerful statistical method that lets us test complex webs of relationships all at the same time. Think of it as a combination of two other techniques: factor analysis (which helps identify underlying concepts) and multiple regression (which examines how several variables predict an outcome).

With SEM, researchers can build a model that represents their theory and then test how well that model fits the actual data they've collected. It’s a way to see if the story you’re telling about how things work is supported by the evidence.

To understand SEM, we first need to grasp its two fundamental components: observed and latent variables.

Latent variables are more abstract. They're concepts we're interested in but can't measure with a single question or tool. You can't put a ruler on "intelligence," "anxiety," or "brand loyalty." Instead, we infer their existence from several related observed variables.

For example, to measure the latent variable "Depression," a psychologist might ask questions about sleep patterns, appetite changes, and feelings of sadness. None of these questions is depression, but together, they paint a picture of the underlying concept. In SEM diagrams, we typically draw observed variables as squares and latent variables as ovals or circles.

An SEM model has two main parts. The measurement model specifies how the latent variables are measured by the observed variables (like the diagram above). The structural model shows the theorized relationships between the latent variables. For example, a structural model might propose that 'Work Stress' directly influences 'Job Satisfaction'.

Structural Equation Modeling didn't appear out of thin air. Its origins trace back to the early 20th century with two key ideas.

First was path analysis, developed by biologist Sewall Wright in the 1920s to study genetic inheritance. Path analysis allowed researchers to map out cause-and-effect pathways between variables. Around the same time, psychologists like Charles Spearman were developing factor analysis to identify underlying factors (latent variables) from a battery of tests, such as different measures of intelligence.

It wasn't until the 1970s that researcher Karl Jöreskog and his colleagues merged these two streams of thought. They created a single, flexible framework that could simultaneously analyze the measurement of latent variables and test the relationships between them. This synthesis became what we now know as SEM.

Because of its flexibility, SEM is used across many different fields to test complex theories.

In each case, SEM allows researchers to move beyond simple, one-to-one relationships and test a more holistic model of how the world works. It provides a way to statistically evaluate an entire theory, not just bits and pieces of it.

Now that we have a foundational understanding of what SEM is, let's test your knowledge.

With this foundation, we can begin to explore the different ways to build the measurement part of a model.