英文:
Forcing Treatment Group to Remain in Model Despite Statistical Insignificance?
问题
我试图在SAS中运行二元逻辑回归，寻找因变量（FRSTVACREC和CMPLTSRS）、处理组（对照组与治疗组）和协变量（GENDER AGE QAGE RESP RACEETH）之间的交互作用效应。协变量是根据以前的相关矩阵识别的。我正在尝试强制处理组变量TREAT保留在模型中，尽管它在统计上不显著，以查看TREAT与其他协变量对因变量的交互作用效应。以下是我的代码：
proc logistic data=work.hpvboys_pre_analysis_final plots=(oddsratio(cldisplay=serifarrow));
class Treat / param=glm;
model FRSTVACREC(event='1') = Treat GENDER AGE QAGE RESP RACEETH / clparm=wald link=logit lackfit technique=fisher selection=forward slentry=0.05 sle=0.10;
run;
proc logistic data=work.hpvboys_pre_analysis_final plots=(oddsratio(cldisplay=serifarrow));
class Treat / param=glm;
model CMPLTSRS(event='1') = Treat GENDER AGE QAGE RESP RACEETH / clparm=wald link=logit lackfit technique=fisher selection=forward slentry=0.05 sle=0.10;
run;
我如何强制TREAT保留在这个模型中？
一位高级统计学家建议我使用include语句来强制Treat保留，但我不认为语法允许这样做。尝试了前向和后向淘汰选择，但无法弄清如何手动强制其保留。
英文:
I am attempting to run a binary logistic regression in SAS, looking for interaction effects between the dependent variables (FRSTVACREC and CMPLTSRS), treatment groups (control vs treatment), and covariates (GENDER AGE QAGE RESP RACEETH). Covariates were identified using a correlation matrix previously. I am trying to force the treatment group variable TREAT to stay in the model despite it being statistically insignificant to see the interaction effects between TREAT and the rest of the covariates on the dependent variables. Here is my code:
proc logistic data=work.hpvboys_pre_analysis_final plots=(oddsratio(cldisplay=serifarrow));
class Treat / param=glm;
model FRSTVACREC(event='1') = Treat GENDER AGE QAGE RESP RACEETH / clparm=wald link=logit lackfit technique=fisher selection=forward slentry=0.05 sle=0.10;
run;
proc logistic data=work.hpvboys_pre_analysis_final plots=(oddsratio(cldisplay=serifarrow));
class Treat / param=glm;
model CMPLTSRS(event='1') = Treat GENDER AGE QAGE RESP RACEETH / clparm=wald link=logit lackfit technique=fisher selection=forward slentry=0.05 sle=0.10;
run;
How would I force TREAT to remain in this model?
A senior statistician recommended I use the include statement to force Treat to remain, but I do not believe the syntax allows for this. Tried forwards and backwards elimination selection, cannot figure out how to manually force it to stay.
答案1
得分: 2
设置INCLUDE=1并将要包含的变量放在MODEL语句中的第一个位置。然后你会在输出中看到类似以下的内容：
>每个模型顶部将包括以下效应：
>截距 TREAT
英文:
Set INCLUDE=1 and put the variable to include as the first in the MODEL statement. Then you'll see something like the following in your output:
>The following effects will be included in each model at the top:
>Intercept TREAT
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论