ABSTRACT

Although structural equation model (SEM) is a powerful and widely applied tool particularly in social sciences, few studies have explored how SEM and statistical learning methods can be combined. The purpose of this paper is to explore how gradient component-wise boosting (GCB) can contribute to item selection. We ran 200 regressions with different farmer psychological variables collected to explain variation in an animal welfare indicator (AWI). The most frequently selected variables from the regressions were selected to build a SEM to explain variation in the AWI. The results show that boosting selects relevant items for a SEM.

中文翻译：

---

结构方程建模 (SEM) 中的规范搜索：梯度分量提升如何发挥作用

摘要

尽管结构方程模型 (SEM) 是一种功能强大且应用广泛的工具，尤其是在社会科学领域，但很少有研究探讨如何将 SEM 和统计学习方法结合起来。本文的目的是探索梯度分量提升（GCB）如何有助于项目选择。我们使用收集的不同农民心理变量进行了 200 次回归，以解释动物福利指标 (AWI) 的变化。选择回归中最常选择的变量来构建 SEM 以解释 AWI 的变化。结果表明，提升为 SEM 选择了相关项目。