📸 SEM in AMOS — combining measurement and structural models
What Is SEM?
Structural Equation Modeling (SEM) simultaneously estimates a measurement model (CFA: how latent variables are measured) and a structural model (causal paths between latent variables). It handles multiple dependent variables, mediating paths, and measurement error — all in a single framework.
Two-Step Approach
Anderson & Gerbing's recommended approach: (1) First establish an acceptable measurement model via CFA. (2) Then add structural paths. This separates measurement quality from structural relationships.
Running SEM in AMOS
Step 1: Validate each construct's measurement model separately via CFA first.
Step 2: In the full AMOS diagram, connect latent variables with single-headed arrows (causal paths).
Step 3: For indirect effects: Analysis Properties → Bootstrap → 5000 samples → Bias-corrected confidence intervals.
Step 4: Evaluate fit with CFI, RMSEA, SRMR.
📸 SEM standardized path coefficients and indirect effects
APA Reporting
SEM demonstrated acceptable model fit (CFI=.942, RMSEA=.063). The structural model revealed a significant indirect effect of stress on performance through job satisfaction (β=-.162, p=.001, bootstrap 95% CI [-.248, -.072]), supporting partial mediation.