全部版块 我的主页
› 论坛 › 数据科学与人工智能 › 数据分析与数据科学 › SAS专版
One side test of coefficients in a regression
edwardzxf
2018 7
2014-02-19
How do I perform a one sided test of the coefficients in a linear regression? for example:

proc reg data = data;
        model  y=x1 x2 x3;
        test x1+x2 =0;
run;


However, test x1+x2 =0 is two sides test, I want to test x1+x2>0, then how? Many thanks.



二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

全部回复
bridge1989
2014-2-19 20:31:37
I have looked up some relatives books. the test and mtest in reg process must be an equation.
But,your need can be meet by a little change:
     every statistical test wiil give out a statisstical value, then the P will be ensured based on relative statistical limitation table. So one-tailed or two-tailed is based on which table you choose.
      
二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

jingju11
2014-2-20 11:22:02
You could NOT do that in SAS so far. when you do the one-sided test, you are at risk of delivering a valid test. If I were you, I may do bootstrapping for the p value.
JingJu
二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

edwardzxf
2014-2-20 13:35:24
bridge1989 发表于 2014-2-19 20:31
I have looked up some relatives books. the test and mtest in reg process must be an equation.
But,y ...
I think you are right, the pvalue in a one side test is a half of that in a two-side test.
二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

edwardzxf
2014-2-20 13:36:33
jingju11 发表于 2014-2-20 11:22
You could NOT do that in SAS so far. when you do the one-sided test, you are at risk of delivering a ...
It's a little bit pity as it's easy to do that in STATA.
二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

bridge1989
2014-2-20 14:06:28
jingju11 发表于 2014-2-20 11:22
You could NOT do that in SAS so far. when you do the one-sided test, you are at risk of delivering a ...
高级ＳＡＳ人才啊　我关注了你的博客
二维码

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

点击查看更多内容…
相关推荐
栏目导航
热门文章
推荐文章

说点什么

分享

微信QQ空间QQ微博
扫码加好友，拉您进群
各岗位、行业、专业交流群