Estimating statistical power for structural equation models in developmental cognitive science: A tutorial in R :...

Area of Science:

• Developmental Psychology
• Quantitative Psychology

Background:

• Structural Equation Modeling (SEM) is crucial for understanding psychological constructs.
• Estimating statistical power and sample size for SEM, particularly for non-nested models, is challenging.
• Existing rules of thumb for SEM sample size are often inadequate.

Purpose of the Study:

• To introduce a Monte Carlo simulation approach for estimating a priori statistical power in SEM.
• To provide a practical guide for applying this method to non-nested SEM comparisons.
• To address the need for robust power analysis in developmental science research.

Main Methods:

• Utilized Monte Carlo simulations to estimate statistical power.
• Developed a step-by-step procedure for power analysis in SEM model selection.
• Applied the method to a developmental memory research example.

Main Results:

• Demonstrated the feasibility of Monte Carlo simulations for SEM power estimation.
• Provided a reliable method for determining sample size for non-nested SEM comparisons.
• Highlighted the limitations of generic heuristics for SEM power analysis.

Conclusions:

• Monte Carlo simulations offer a powerful tool for a priori statistical power estimation in SEM.
• This approach enhances the accuracy of sample size determination for complex model comparisons.
• The proposed method is valuable for researchers in developmental science and other fields utilizing SEM.