求助: 趋势检验

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

我不是学统计的，哪位能帮忙用SAS处理一下数据，做卡方趋势检验，不胜感激。 糖尿病前期查体数据
             阳性数   查体总数
2006年
1153
36968
2007
1555
40396
2008
1881
40970
2009
2225
42438

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1 回帖

关键词：不胜感激 糖尿病 统计 不胜感激 糖尿病

What's the English name of 卡方趋势检验 ?

是不是要这个?

data yifanhe;
input year positive total;
datalines;
2006 1153 36968
2007 1555 40396
2008 1881 40970
2009 2225 42438
; run;
proc freq data=yifanhe;
table year*positive year*total positive*total / chisq;
run;

一帆HE 发表于 2010-3-29 20:02
我不是学统计的，哪位能帮忙用SAS处理一下数据，做卡方趋势检验，不胜感激。 糖尿病前期查体数据
             阳性数   查体总数
2006年
1153
36968
2007
1555
40396
2008
1881
40970
2009
2225
42438

***I googled that 卡方趋势检验 is  Chi-Square trend test.  **;
data diabeScreen(drop=total);
input #1 year #2 positive #3 total;
negative=total-positive;
datalines;
2006
1153
36968
2007
1555
40396
2008
1881
40970
2009
2225
42438
;
run;


data diabets;
input year diabet $ count@@;
datalines;
2006 No 35815 2006 Yes  1153
2007 No 38841 2007 Yes  1555
2008 No 39089 2008 Yes  1881
2009 No 40213 2009 Yes  2225
;

proc freq data=diabets;
      weight count;
      tables diabet*year / trend chisq measures cl;
   run;

果然是高手，是chisq  trend,能不能把结果Z、P告诉我呀，我可是外行呀。谢谢。

5# 一帆HE

See your 短消息 for Z and P.

一帆HE 发表于 2010-3-30 21:44
果然是高手，是chisq  trend,能不能把结果Z、P告诉我呀，我可是外行呀。谢谢。

Some words about the test above:

The following is my personal views about the results from SAS:
In terms of the measures of association by measures options, as well as Pearson and Spearman correlation coefficients, the positive association is weak in the absolute values, i.e., very small values(<0.1); on the other hand, since the asymptotic 95% confidence limits do not contain zero, that implies an evidence of a strong association. This contradiction can be attributed to the extremely large sample sizes, which apparently inflate the strength of the association between the time and diabetes rate(incidence) in your study. Moreover, that probably also indicates the limitation in the study. I mean, the comparison may be not very meaningful at all. For example, the incidence rate for diabetes is increasing because the treatment for diabetes takes time  as we know thus the number of new cases may always or very often grow faster than that of recovered cases…. Overall, you need to be prudent on your conclusion. For me, it may be more important for the increment of new cases every year.
Statistics is not everything, which depends on your own judegement.
JingJu

Statistics is not everything, which depends on your own judegement.