比较配对设计的SPSS与SAS实现

配对设计主要研究某种实验处理前后的对比状况，来判定该种处理是否有效。该种方法能够有效地控制非实验因素对结果的影响。

需要注意的是实验前后的结果不是独立的，就是说实验前的状况对实验后的结果是有影响的。统计学上通常会通过差值来消除这种影响，也可以考虑进行两独立样本的检验。这里分别用两种软件实现第一种方法。

例子：考察16个病人服药前后的weights和triglyceride levels,具体数据见附件。

1.SPSS

Analyze-Compare Means-Paired Samples T test,然后选择配对变量

2.SAS（UNIVARIATE语句详解参考：http://543484067.qzone.qq.com/blog/1294717032）

将数据导入SAS后进行如下操作

data y;
   set a;
   tg=tg0-tg4;
   wgt=wgt0-wgt4;
run;
proc univariate data=y normal;
var tg wgt;
run;

下面我们看看两种结果的比较：   

---

Paired Samples Statistics　MeanNStd. DeviationStd. Error MeanPair 1Triglyceride138.441629.047.26Final triglyceride124.381629.4127.353Pair 2Weight198.381633.4728.368Final weight190.311633.5088.377

 

 

Paired Samples Test

 

 

Paired Differences

 

 

 

 

 

Mean

Std. Deviation

Std. Error Mean

Pair 1

Triglyceride - Final triglyceride

14.063

46.875

11.719

Pair 2

Weight - Final weight

8.063

2.886

.722

 

Paired Samples Test

 

 

Paired Differences

 

 

95% Confidence Interval of the Difference

 

 

Lower

Upper

Pair 1

Triglyceride - Final triglyceride

-10.915

39.040

Pair 2

Weight - Final weight

6.525

9.600

 

Paired Samples Test

 

 

 

 

 

 

 

 

t

df

Sig. (2-tailed)

Pair 1

Triglyceride - Final triglyceride

1.200

15

.249

Pair 2

Weight - Final weight

11.175

15

.000

 

SAS结果（部分）：

      变量:   tg

                                                              矩

                                    N                     16               权重总和                16
                                    均值             14.0625           观测总和               225
                                    标准差        46.8749667       方差             2197.2625
                                    偏度          -0.0645782         峰度            -1.1826074
                                    未校平方和         36123         校正平方和      32958.9375
                                    变异系数      333.333096      标准误差均值    11.7187417

                                                         基本统计测度

                                               位置                    变异性

                                         均值     14.06250     标准差         46.87497
                                         中位数   18.00000     方差               2197
                                         众数       .                  极差          152.00000
                                                                        四分位极差     79.50000

                                                        位置检验: Mu0=0

                                          检验      --统计量---    -------P 值-------

                                          学生 t    t  1.200001    Pr > |t|    0.2487
                                          符号      M         1          Pr >= |M|   0.8036
                                          符号秩    S        23        Pr >= |S|   0.2522

 

          变量:   wgt

                                                              矩

                                    N                     16                          权重总和                16
                                    均值              8.0625                      观测总和               129
                                    标准差        2.88602957                 方差            8.32916667
                                    偏度          0.72763584                   峰度             -0.479326
                                    未校平方和           1165                    校正平方和        124.9375
                                    变异系数      35.7957156                标准误差均值    0.72150739

                                                         基本统计测度

                                               位置                    变异性

                                         均值     8.062500      标准差          2.88603
                                         中位数   7.000000      方差            8.32917
                                         众数     6.000000       极差           10.00000
                                                                       四分位极差      4.50000

                                                        位置检验: Mu0=0

                                          检验      --统计量---    -------P 值-------

                                          学生 t    t  11.17452    Pr > |t|    <.0001
                                          符号      M         8          Pr >= |M|   <.0001
                                          符号秩    S        68         Pr >= |S|   <.0001
附：

patidagegendertg0tg1tg2tg3tg4wgt0wgt1145018014810611310019819625601399411975922372333500152185861491182332314461112145136149821791815640156104157799721921764911671388810717116916676301381321461431322222198631160128150118123167167952010712012919517419920010450156103126135922332291161194144114114121179181124911079315614815015815313611145107129861591571511459018614212812210121621315520112107103891482572551660110410311779130151146

---