大牛，菜鸟请教您啦！！！ 有关CONTRAST 语句

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

分析线性回归和平方回归时 用 contrast语句，后面还要配相应的系数，系数的获得是用这样的语句，
proc iml;
x={2 4 6};
xp=orpol(x,2);
print xp;
run;

由获得的系数再用在contrast 语句后，做线性和平方回归的分析，有谁用过这个呢？
contrast "zuhe linear" zuhe -0.707107        0        0.7071068;
contrast "zuhe quadratic" zuhe 0.4082483        -0.816497        0.4082483;

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

分享0 收藏0 回帖

关键词：CONTRAST contr RAS con Quadratic

where is the contrast statement

大数据之魂 发表于 2012-11-12 20:02
where is the contrast statement

proc GLM;
class         zuhe        youshi        junyundu;
model net =        zuhe junyundu youshi zuhe*junyundu/ss4;
random youshi/q;
lsmeans zuhe junyundu;
means zuhe junyundu/duncan;

contrast "zuhe linear" zuhe -0.707107        0        0.7071068;
contrast "zuhe quadratic" zuhe 0.4082483        -0.816497        0.4082483;

run;

这是我的程序，您能帮我分析下这个可行吗

you are right. since your test variable is equally spaced, i.e., { 2 4 6}, so all contrasts written like the following forms

Linear   -1   0   1   &       Quad      1  -2   1

were correct to test linear &  quad trend.
we know that the constrast does not matter about a factor added, for example {-1 0 1} == {-1.234 0 1.234}, and so on.

Jingju

continue on the last post:
I saw an interaction term  zuhe*junyundu. so i doubt the above code is proper to test linear trend for ' zuhe', but i dont know how to adjust for it.
\jingju

jingju11 发表于 2012-11-17 09:57
continue on the last post:
I saw an interaction term  zuhe*junyundu. so i doubt the above code is p ...

ok, thanks very much, you are a very nice man~~