s**m 发帖数: 340 | 1 the studied population has 114 patients. Univariate cox model of my index
results hazard ratio of 2.6 (p<0.001). I put the index into a multivariate
cox model with 5 other variables, and results a hazard ratio>4 (P<0.02).
However, a collaborator criticized that the univariate results does not have
enough statistical power to justify the use of multivariate model. Does
that make sense? I am not an expert in statistics. I just use epi-info to do
the analysis, and it does not give me this statistical power value. Do I
need this statistical power stuff?
Thanks a lot for any suggestion! | w*******9 发帖数: 1433 | 2 You may need some rough diagnostic analysis to justify the model. Try mating
ales residual or defiance residuals. |
|