Using Structural Equation Modeling to Assess Functional Connectivity in the Brain: Power and Sample Size...

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Developmental Neuroscience

Background:

• Functional brain connectivity research often employs structural equation modeling (SEM).
• Determining optimal sample sizes for SEM in pediatric neuroimaging is crucial for reliable findings.
• Previous research has not extensively evaluated sample size effects on SEM in this specific context.

Purpose of the Study:

• To assess the impact of varying sample sizes on the statistical power and model fit of SEM applied to functional brain connectivity.
• To establish recommended sample sizes for SEM analyses in pediatric brain imaging studies.

Main Methods:

• Simulated functional brain connectivity data from 51 typical readers (aged 7.5-12.5 years) using magnetoencephalography (MEG).
• An autoregressive model with 5 latent variables (brain regions) and 3 indicators each was used.
• Sample sizes ranged from 20 to 1,000 participants, with 1,000 replications per size, analyzed for convergence, fit indices (RMSEA, D-Fit), and path stability.

Main Results:

• Sample sizes of 70–80 participants were found adequate for modeling relationships with acceptable fit.
• A sample size of 50 participants demonstrated satisfactory model fit.
• Model convergence and structural path stability were influenced by sample size, with larger samples generally yielding more robust results.

Conclusions:

• Structural equation modeling is a viable method for investigating complex brain activation interdependencies in pediatric populations.
• A minimum of 50–80 participants is recommended for reliable SEM analyses of functional brain connectivity in children.
• These findings provide essential guidance for researchers designing neuroimaging studies in pediatric populations.