Some concepts are measured directly while others require fallible symptom indicators

Evidence

In the classic uses in education and psychology it is pretty clear that the test questions are not ontology. Correct answers on IQ tests are an effect of intelligence not intelligence itself. Votes in Congress are a reflection of ideology not the ideology of the individual. In contrast, for the concept of consumption poverty one is trying to measure actual consumption, not indicators of consumption. In the case of human well-being, infant mortality might be a good indicator of the quality of healthcare in a country. — Gary Goertz, Social Science Concepts and Measurement, 328

Categorization

• DEFINE: Sharpens the indicator/ontology distinction with parallel concrete cases: IQ tests, roll-call votes, consumption, infant mortality.
• SEED: Article angle: how the proxy-vs-direct measurement distinction quietly determines what most quantitative social science can even claim to know.