datacamp-logo
Session Ready
Exercise

Construct validity & model fit

We check for construct validity only after checking for CFA model strength (as measured by CFI, TFI and RMSEA) for the same reasons we checked for internal reliability after checking for EFA model strength: a poor-fitting model could actually indicate greater construct validity than a relatively better-fitting model. Let's illustrate with an example. In earlier lessons, we explored whether the ten-item c_sat scale, measuring customer satisfaction, loaded better to a two- or three-item solution. Let's compare their respective CFA models and measures of construct validity.

The three- and two-factor models are loaded as c_sat_cfa_model_3 and c_sat_cfa_model_2, respectively.

lavaan and semTools have been loaded into your environment.

Instructions 1/3
XP
  • 1
  • 2
  • 3
  • Fit the three-factor model to the c_sat data.
  • Inspect the CFI, TLI and RMSEA of the model.