PHD Discussions Logo

Ask, Learn and Accelerate in your PhD Research

Question Icon Post Your Answer

Question Icon

10 months ago in Conceptualization By Arjun

What are modification indices, and what are the substantive (theoretical) versus statistical risks of using them to modify your model?

What are modification indices, and what are the theoretical (substantive) and statistical risks associated with using them to adjust your model?

All Answers (1 Answers In All)

By Karkuvel Answered 7 months ago

Modification indices (MIs) estimate how much the model's chi-square would decrease if a fixed parameter (like a covariance) were freely estimated. The statistical risk is capitalizing on chance, leading to a model tailored to sample-specific noise. The substantive risk is creating a theoretically meaningless model by adding paths with no logical justification. MIs should only guide post-hoc revisions if any suggested change has a strong, defensible theoretical rationale.  

Your Answer