Comparison of latent growth curves: A parameter constancy test

Area of Science:

• Psychometrics
• Longitudinal Data Analysis
• Structural Equation Modeling

Background:

• Latent growth curve (LGC) models are standard for analyzing developmental trajectories.
• Current model selection relies on goodness-of-fit indices, neglecting parameter temporal constancy.
• Parameter stability is vital for reliable forecasting and interpretation of longitudinal data.

Purpose of the Study:

• To introduce a novel parameter constancy test (PCT) for LGC models.
• To address the gap in assessing temporal parameter stability in LGC analysis.
• To improve the reliability and interpretability of longitudinal developmental and learning trajectory models.

Main Methods:

• Developed and applied a parameter constancy test (PCT) for LGC models.
• Utilized structural equation modeling for LGC implementation.
• Compared quadratic and negative exponential models using PCT on real-world data.

Main Results:

• The negative exponential model demonstrated superior parameter constancy compared to the quadratic model, even with fewer data points.
• PCT effectively identifies potential breakpoints and determines the minimum required measurement waves for reliable modeling.
• Inappropriate model selection or instability can lead to misinterpretations, especially when evaluating interventions or extrapolating data.

Conclusions:

• Parameter constancy is as crucial as statistical fit in selecting LGC models.
• PCT integration enhances model consistency, optimizes resource allocation, and prevents erroneous conclusions in longitudinal research.
• The negative exponential model is a robust choice for longitudinal analysis due to its consistent parameter stability.