Statistics / psychometrics · BEGINNER LESSON
Latent variable
A latent variable is a variable that is not directly observed but is inferred from observed variables. In statistical models, it is used to summarize patterns in measurements or responses.
What it is
A latent variable is a quantity in a statistical model that cannot be seen or measured directly. Instead, it is inferred from observed, or manifest, variables, which are the measurable scores, ratings, or responses collected from people, items, or other units. Latent variables may be continuous or discrete, and the type of latent variable helps determine which model is appropriate.
[1] Latent variable modeling - PMC - NIH [2] Recent Integrations of Latent Variable Network Modeling With ... [3] Psychometrics in Psychological Research: Role Model or ...
How it works
The key idea is that observed data are linked to the latent variable through a measurement model, which specifies how the measured variables depend on the unobserved one. In factor analysis, for example, several observed symptoms or test items are treated as indicators of a common factor. The model also separates unique variation and measurement error from the shared pattern, so the latent variable can represent the common part of the data more cleanly.
Why researchers use it
Latent variables help reduce complex, noisy data into a simpler description. They are used in factor analysis, item response theory, latent class analysis, and structural equation modeling. In psychology and education, they can represent abstract constructs such as depression, intelligence, or extraversion, while in network models they can describe hidden positions that affect whether nodes connect.
[1] Latent variable modeling - PMC - NIH [2] Recent Integrations of Latent Variable Network Modeling With ... [3] Psychometrics in Psychological Research: Role Model or ...
Limits and caution
A latent variable is not a directly observed object, so its meaning depends on the model and the data used to estimate it. Good model fit can support a proposed interpretation, but it does not by itself prove that the underlying attribute truly exists in the world. The model must also be specified carefully, including both the measurement model and any relationships among multiple latent variables.
[1] Latent variable modeling - PMC - NIH [3] Psychometrics in Psychological Research: Role Model or ...
SEE THE IDEA
How a latent variable model links data to an unobserved quantity
Measured scores or responses, such as item answers or symptom ratings
Rules that connect the observed variables to the hidden variable
The unobserved factor or trait inferred from the shared pattern
Variation in each observed variable not explained by the latent variable
Observed variables feed into a measurement model, which estimates the latent variable while separating shared signal from error.
The latent variable is inferred from measured data, not observed directly.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.