Signal-to-Noise Ratio in Estimating and Testing the Mediation Effect: Structural Equation Modeling versus Path...

Area of Science:

• Social and behavioral sciences
• Quantitative psychology
• Statistical modeling

Background:

• Mediation analysis is crucial for understanding causal processes.
• Previous comparisons of SEM and PAWC for mediation analysis may lack validity due to issues with latent variable scaling.
• Measurement errors in composite scores can bias parameter estimates in path analysis.

Purpose of the Study:

• To compare the statistical efficiency and power of path analysis with weighted composites (PAWC) versus structural equation modeling (SEM) for mediation analysis.
• To evaluate parameter estimates using signal-to-noise ratio (SNR) to overcome metric-dependent comparisons.
• To identify conditions influencing the performance of PAWC and SEM in mediation analysis.

Main Methods:

• Comparison of PAWC and SEM using signal-to-noise ratio (SNR) for parameter estimates.
• Analysis of indirect effect estimation and testing in mediation models.
• Investigation of the impact of predictor-mediator relationship strength and prediction error on SNR.

Main Results:

• PAWC consistently yields greater SNR than SEM for estimating and testing the indirect effect, even with measurement errors.
• Path analysis via factor scores and mediation analysis with equally weighted composites (EWCs) also tend to yield greater SNRs than SEM.
• The advantage of PAWC over SEM is more pronounced with a strong predictor-mediator relationship, while prediction error in the mediator negatively impacts PAWC.

Conclusions:

• PAWC is statistically more efficient and powerful than SEM for mediation analysis in empirical research.
• SNR is a suitable metric for comparing mediation analysis methods independent of latent variable metrics.
• Understanding the conditions affecting SNR is vital for selecting the appropriate mediation analysis methodology.