#SmartPLS4 Series 15 - What is a Formative Construct?

Higher-order constructs in SmartPLS can be either reflective or formative, and the choice changes what gets validated and how the measurement model behaves. Reflective measurement treats the latent construct as the cause of its indicators: arrows run from the latent variable to the observed items. Formative measurement flips that logic—indicators define or “build” the latent construct, so arrows run from the indicators to the latent variable. Getting this direction right matters because it determines whether indicators can be treated as interchangeable and whether deleting an item leaves the construct intact.

In reflective models, the latent construct exists independently of the specific indicators. Because the indicators are manifestations of the same underlying theme, they are considered interchangeable: removing one or two indicators may not destroy the construct’s content validity. The construct still “exists” even if measurement changes slightly, since the latent variable is assumed to cause the observed measures. This is why many scale-development traditions in business and methodology literature favor reflective approaches, especially at lower levels and often at higher levels.

Formative models work differently. Indicators are not interchangeable because each one contributes a distinct meaning to the latent construct. In formative measurement, changes in the indicators lead to changes in the construct itself. That also means the indicator set must be comprehensive: leaving out an important component can change what the construct even represents. The transcript illustrates this with the Human Development Index (HDI): HDI is a composite of health, education, and income. Remove income and HDI no longer represents the same concept—so the indicators are essential parts of the construct rather than interchangeable reflections.

The same contrast appears in examples. Social responsibility (CSR) is used as a reflective case: even if the economic dimension is removed, the broader idea of social responsibility remains as a latent construct. By contrast, social class index (SCI) is treated as formative: SCI is formed from educational level, occupational prestige, and income. Remove income and the index can’t be called SCI anymore because the construct depends on that component.

The transcript also ties the reflective-versus-formative distinction to how error is interpreted. Reflective theory assumes latent constructs cause measured variables, and measurement error limits how fully the indicators can be explained. Formative theory assumes indicators cause the construct, and the key limitation is that the indicator list must be complete; otherwise the construct is not properly formed.

Finally, the discussion sets up what comes next: higher-order construct validity. Before assessing higher-order constructs, the reflective/formative distinction must be understood because formative constructs require different validation logic than reflective ones. The session ends by pointing to a subsequent discussion on validating higher-order constructs.