Justifying Scale Type for a Latent Variable: Formative or Reflective?

摘要

The study attempted to explore the possibilities to create a procedure at the experimental level to double confirm whether manifest variables scale type is formative or reflective. Now, the criteria of making such a decision are heavily depended on researchers' judgment at the conceptual and operational level. The study created an experimental procedure that seems could double confirm the decisions from the conceptual and operational level judgments. The experimental procedure includes the following tests, Variance Inflation Factor (VIF), Tolerance (TOL), Ridge Regression, Cronbach's alpha, Dillon-Goldstein's rho, and first and second eigenvalue. The procedure considers manifest variables' both multicollinearity and consistency. As the result, the procedure received the same judgment with the carefully established decision making at the concept and operational level.