Hi ,
i performed multivariate Cox regression analysis and one of the covariates had a p-value of 0.08. since it is a replication study and i have a hypothesis, I can have a directional hypothesis. Is it legitimate to divide the p-value by 2 to have a significant p-value?
Also, how do I report 95% CI in such a case?
Thanks
First, I think you are confused. Let me recommend the references in the links.:
https://www.amazon.com/Mathematical-Statistics-Applications-Dennis-Wackerly-ebook/dp/B00B7K7N7K/ref=sr_1_2?keywords=mendenhall+mathematical+statistics&qid=1556948006&s=gateway&sr=8-2
and
https://www.amazon.com/Regression-Modeling-Strategies-Applications-Statistics/dp/3319194240/ref=sr_1_fkmrnull_1?keywords=Frank+Harrell+regression&qid=1556948106&s=gateway&sr=8-1-fkmrnull
Now I notice that a) you didn't list your research question and associated hypotheses. b) Also I notice that 0.08/2=0.04 .05. I am thus suspicious that this is an attempt to get a p-value < 0.05 by any means possible. If so you are very close to dishonesty. In statistics We ask a question then collect data which either supports or does not support a chosen answer to our question. You do not seem to be doing that. The above books are about how to get answers to questions that you can TRUST. Is that REALLY your goal? D. Booth
N.B. multivariate means more than one DV.
Dear David Eugene Booth Sir,
Your perception and explanation are excellent!
Regards,
Dr. Md. Alamin
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS