Cross-Cultural Multigroup‑SEM Misspecification Monte Carlo Repository (149 Conditions)

This repository (version v2) contains the complete data outputs from a large-scale Monte Carlo simulation study evaluating path-level robustness in multigroup structural equation models (MG-SEM). The dataset supports the manuscript “Path‑Level Diagnostics for Multigroup SEM: A Tutorial and Monte Carlo Evaluation with FWER‑Aware Inference”. A total of 149 unique simulation conditions were generated, systematically crossing different structural parameter values, factor loading strengths, residual variances, and cultural group configurations. Each condition follows a fixed design of N = 1,000 and R = 10,000 replications; a representative subset of R = 200 replications per condition is preserved in cleaned .csv format to facilitate downstream analyses, diagnostic procedures, and figure/table generation. The simulated data include both observed covariates (academic year, GPA, gender) and psychological constructs measured through latent variables: Passive AI Habits (PAH), Attribution Styles (AS), Amotivation (AMO), Cultural Norm Sensitivity (CNS), Help-Seeking Avoidance (HSA), and Help-Seeking Readiness (HSR). Interaction terms (e.g., PAH × AS) and grouping variables indicate cultural origin (West vs. Asia), with allocation scenarios reflecting both 50/50 and 70/30 group splits. All latent constructs follow reflective measurement models with realistic item structures and residual variances. This dataset supports cross-study comparisons under both correctly specified and misspecified models (e.g., reverse, mirror, omit, proxy-pattern), enabling benchmarking of parameter bias, confidence interval (CI) coverage, model selection indices (AIC/BIC), and Type I/II error rates—particularly in culturally moderated structural pathways. For convenience, this repository includes a sample data package (PathC_Baseline_Ctrl10_R10000_Example.zip) that demonstrates the folder structure, file contents, and naming conventions used across all 149 conditions. New in v2: This version includes condition-level folder structure for transparency, and a representative example of full output files for a single simulation condition. Citation notice: This dataset supports a manuscript currently under review. For citation purposes, please reference the forthcoming article once published. Pre-publication inquiries may be directed to the corresponding author.