Table 2 A continuum exists between MCI and MEI measurement models.
Causal Indicator Model | Model Characteristic | Effect Indicator Model |
|---|---|---|
Item content assesses any cause of the measurement construct that is relevant to a (sub)group of respondents | Content Relevance | Item content assesses the effects of the measurement construct that is relevant to all or most respondents |
Item content tends to be specific (high fidelity) | Content Specificity | Content may be either specific or general |
Item ratings exhibit statistical independence and contribute unique predictive power | Association between Item Ratings | Item ratings are highly correlated, canceling random measurement error |
Item ratings are skewed due to differential content relevance across respondents | Item Score Distributions | Item ratings are normally distributed due to common relevance of item content |
Multivariate regression, cluster, and discriminant analyses against a criterion estimate (of the latent construct) | Construct Validity Statistics | Factor and IRT analyses of item covariance or response probability patterns |