Linear regression, structural relations, and measurement error.
Linear regression, structural relations, and measurement error.
复制标题
线性回归、结构关系和测量误差。
DOI：
10.1037/h0029777
复制
发表时间：
1970
影响因子：
22.4
通讯作者：
P. Isaac
中科院分区：
文献类型：
--
作者：
P. Isaac
区分线性回归分析适用的情况和涉及线性结构关系的情况。虽然这两种情况都涉及线性参数的估计，但参数在这两种情况下的用途不同，因此，它们的估计值通常会有所不同。回归的假设和预测性质进行审查和对比的假设和性质的线性结构关系。在结构关系中，没有关于2条回归线的问题;只有1条线是感兴趣的。指出了这两种情况下测量误差的影响。给出了一个例子，其中线性回归分析被不适当地使用，并在其中的线性结构关系的分析将是适当的。（PsycINFO数据库记录（c）2016阿帕，保留所有权利）
Distinguishes between situations to which linear regression analysis is applicable and situations which involve linear structural relations. Although both situations involve estimates of linear parameters, the parameters serve a different purpose in the 2 situations, and, consequently, their estimates in general will differ. The assumptions and predictive nature of regression are reviewed and contrasted with the assumptions and nature of linear structural relations. In the structural relation, there is no question about 2 regression lines; only 1 line is of interest. The implications of measurement errors for the 2 situations are pointed out. An example is given in which linear regression analysis was used inappropriately, and in which the analysis for a linear structural relation would have been appropriate.(PsycINFO Database Record (c) 2016 APA, all rights reserved)