Abstract This paper deals with two generalizations of the classical linear structural errors-in-variables model (SEIVM) with univariate observations. In these two models, the data may no longer come from a single linear SEIVM, due to that a change in the slope in the first model and a change in the intercept in the second model may occur after observing the first k ∗ data pairs, where k ∗ is assumed to be unknown, while the explanatory variables as well as vectors of the measurement errors are independent and identically distributed. Nonparametric asymptotic tests are developed to detect a possible change in each of these SEIVM’s, assuming the existence of four moments for the explanatory and error variables. It appears that the SEIVM with a possible change in the slope of this paper has not been considered in the literature, while the SEIVM with a possible change in the intercept has not been studied before nonparametrically.

中文翻译：

---

线性结构变量模型中斜率和截距变化的非参数检测

摘要 本文讨论了具有单变量观测值的经典线性结构变量误差模型 (SEIVM) 的两种推广。在这两个模型中，数据可能不再来自单个线性 SEIVM，因为在观察前 k ∗ 数据对后，第一个模型中的斜率可能发生变化，而第二个模型中的截距可能发生变化，其中 k ∗ 假定为未知，而解释变量以及测量误差的向量是独立同分布的。假设解释变量和误差变量存在四个矩，非参数渐近检验被开发用于检测每个 SEIVM 中的可能变化。文献中似乎没有考虑到可能改变本文斜率的 SEIVM，