Common Error 9: Opaque or Incorrect Interaction Term Creation with SEM Method The creation of latent variable interaction terms in SEM presents significant challenges in transparency and accuracy (Becker et al., 2023; Rasoolimanesh et al., 2021). Studies show that a majority of CB-SEM (73.47%) and half of PLS-SEM research fail to adequately document their procedure (Rasoolimanesh et al., 2021). A common error is manually computing latent scores (e.g., by summing/averaging indicators) and then multiplying them—an approach that ignores measurement models and introduces error (Rasoolimanesh et al., 2021; Becker et al., 2022).

SEM offers three established methods: (1) the product indicator approach (Chin et al., 2003), (2) the orthogonalization approach (Little et al., 2006), and (3) the two-stage method (Hair et al., 2022). The two-stage method has superior statistical power and is the only option for formative constructs, while orthogonalization excels in point estimation and mitigates multicollinearity (Henseler & Chin, 2010; Fassott et al., 2016). Method selection should consider measurement type (reflective vs. formative) and research priorities (Hair et al., 2022).

常见问题九：潜变量交互项创建不透明或不正确

在SEM框架下创建潜变量的交互项是方法学难点。Rasoolimanesh等人（2021）发现，高达73.47%的CB-SEM研究和50%的PLS-SEM研究未能充分报告其创建方法。一个常见错误做法是：手动计算潜变量得分（如将指标简单平均）后相乘，这完全忽略了测量模型，引入了严重的测量误差。

问题实质：

不能用“苹果和橘子的均价”去乘“香蕉和葡萄的均价”来代表“水果组合”效应。潜变量交互项必须在其测量模型的框架下严谨构建。

解决建议：

根据模型类型和研究目标，明智选择并明确报告以下三种方法之一：

Ÿ  乘积指标法：将构成潜变量的观测指标两两相乘。适用于反映型模型，但可能产生多重共线性。

Ÿ  正交化方法：先对潜变量得分进行正交化处理再生成交互项。能有效缓解共线性，提升点估计精度。

Ÿ  两阶段法：第一阶段估计潜变量得分，第二阶段将其用于回归分析包含交互项。统计功效最高，且是形成性指标模型的唯一选择。

方法选择需权衡测量类型（反映型vs形成性）、统计功效和估计精度需求（Hair等, 2022）。

Reference

• Becker, J. M., Cheah, J. H., Gholamzade, R., Ringle, C. M.,      & Sarstedt, M. (2023). PLS-SEM's most wanted guidance. International      Journal of Contemporary Hospitality Management, 35(1), 321-346.

• Chin, W. W., Marcolin, B. L., & Newsted, P. R. (2003). A      partial least squares latent variable modeling approach for measuring interaction effects. Information Systems Research, 14(2), 189-217.

• Henseler, J., & Chin, W. W. (2010). A comparison of approaches for the analysis of interaction effects between latent variables using partial least squares path modeling. Structural      Equation Modeling, 17(1), 82-109.

• Little, T. D., Bovaird, J. A., & Widaman, K. F. (2006). On the merits of orthogonalizing powered and product terms. Structural      Equation Modeling, 13(4), 497-519.

• Rasoolimanesh, S. M., Wang, M., Mikulić, J., &      Kunasekaran, P. (2021). A critical review of moderation analysis in tourism and hospitality research toward robust guidelines. International      Journal of Contemporary Hospitality Management, 33(12), 4311-4333.

• Xu, Y., & Shiau, W. L. (2026). Moderation analysis in business and management research: Common issues, solutions, and guidelines  for future research. International Journal of Information      Management, 86, 102995.