文档介绍：我们要讨论的是?SPSS 建立方程?检验方程总体和各自变量的统计学意义?自变量的筛选?Case 的剔除(强影响点) ?共线性诊断多元统计方法??线性回归线性回归?? Logistic Logistic 回归回归??注:完全可以用于单因素分析,从这一点也可以看出统计学的方注:完全可以用于单因素分析,从这一点也可以看出统计学的方法是相通的。法是相通的。?? SPSS SPSS 无法进行无法进行条件条件 Logistic Logistic 回归回归,就像,就像 sas sas 没有专门针对配对资没有专门针对配对资料的语句一样,任何统计软件均有局限性。料的语句一样,任何统计软件均有局限性。建立方程的计算量很大,一般需软件完成。前人手工计算这两种回归(借助科学计算器),精神可嘉 80 年代,一个 Logistic 回归即使用计算机,计算的时间也以天计。调整或控制??多变量的控制或调整法就是同时记录可多变量的控制或调整法就是同时记录可能有影响的因素,把他们与可能的危险能有影响的因素,把他们与可能的危险因素一起作分析因素一起作分析??单因素分析有统计学意义,多因素分析单因素分析有统计学意义,多因素分析没有意义的解释没有意义的解释, ,举例举例?? In univariate analysis, the family histories In univariate analysis, the family histories of SAH and ICH were positively associated of SAH and ICH were positively associated with each of the subtypes of stroke. whereas with each of the subtypes of stroke. whereas after adjustment for potential risk factors, after adjustment for potential risk factors, family history of ICH family history of ICH no longer no longer showed showed a significant association with haematoma. a significant association with haematoma. 单因素分析有意义,多因素没有意义的解释?? Family history of intracerebral haematoma Family history of intracerebral haematoma was not an independent risk factor for was not an independent risk factor for haematoma, haematoma, but it might be a good predictor, but it might be a good predictor, which indirectly which indirectly influences the pathogenesis influences the pathogenesis of intracerebral haematoma via certain of intracerebral haematoma via certain ponents such as hypertension, ponents such as hypertension, and even lifestyle factors such as alcohol and even lifestyle factors such as alcohol consumption. consumption. ?? Is family history an independent risk Is family history an independent risk factor for stroke? factor for stroke? ?? J Neurol Neurosurg Psychiatry. 1997 J Neurol Neurosurg Psychiatry. 1997 Jan;62(1):66-70. Jan;62(1):66-70. 多元线性回归多元线性回归(简单步骤) (简单步骤) 指标多,难以理解。在计算自动化时代,这是回归的难点,我们不懂如何判断我们的模型的好坏回归的实际应用??在影响因素分析