A latent variable causes its indicators, so indicators are not the construct itself.

Evidence

Causal relationships within concepts lie at the core of the latent variable approach to measurement. Typically, the unobserved latent variable causes the indicators. I like to call this the disease–symptom model, since diseases cause symptoms. One infers diseases based on symptoms. But one would be wrong to confuse the symptoms with the disease: the disease is not—an ontological claim—the symptoms. — Gary Goertz, Social Science Concepts and Measurement, 65

Categorization

• DEFINE: Names and explains the disease-symptom model as the core of latent variable measurement.
• XREF: Connects directly to construct-validity and reflective-measurement models the reader encounters in psychometrics and social-science methodology.