Universitat Rovira i Virgili, Tarragona, Spain.
Behav Res Methods. 2010 Feb;42(1):29-35. doi: 10.3758/BRM.42.1.29.
When multiple regression is used in explanation-oriented designs, it is very important to determine both the usefulness of the predictor variables and their relative importance. Standardized regression coefficients are routinely provided by commercial programs. However, they generally function rather poorly as indicators of relative importance, especially in the presence of substantially correlated predictors. We provide two user-friendly SPSS programs that implement currently recommended techniques and recent developments for assessing the relevance of the predictors. The programs also allow the user to take into account the effects of measurement error. The first program, MIMR-Corr.sps, uses a correlation matrix as input, whereas the second program, MIMR-Raw.sps, uses the raw data and computes bootstrap confidence intervals of different statistics. The SPSS syntax, a short manual, and data files related to this article are available as supplemental materials from http://brm.psychonomic-journals.org/content/supplemental.
当多回归用于解释性设计时，确定预测变量的有用性及其相对重要性非常重要。商业程序通常会提供标准化回归系数。然而，它们通常作为相对重要性的指标效果很差，特别是在存在大量相关预测变量的情况下。我们提供了两个用户友好的 SPSS 程序，这些程序实现了当前推荐的用于评估预测变量相关性的技术和最新发展。这些程序还允许用户考虑测量误差的影响。第一个程序 MIMR-Corr.sps 使用相关矩阵作为输入，而第二个程序 MIMR-Raw.sps 使用原始数据并计算不同统计量的引导置信区间。与本文相关的 SPSS 语法、简短手册和数据文件可作为补充材料从 http://brm.psychonomic-journals.org/content/supplemental 获得。
