Hello,
I have done a multiple logistic regression to compare low dose versus high dose ATG and it effect on developing PTLD
Inthe logistic regression model, the odds ratio was 1.9 which I understand that high dose ATG is associated with 90% increase in risk of developing PTLD.
However, the P value was 0.2 and confidence interval was crossing 1
Can I debate that odds ratio 1.9 is statistically significant despite the P value of 0.2?
Looking forward to hearing form you
In my opinion, you should not count that variable as a significant one. The reason why are you getting this high odds ratio but insignificant p-value is because maybe your referencing factor/level had a comparatively smaller population than the other factors/levels.
The p-value of 0.2 tells you that, assuming the dose of ATG has absolutely no impact on the risk, getting odds ratios more extreme than 1.9 from data sets like yours is 20%. By all common standards this is really far away from being statistically significant. Your data is simply not conclusive (by common standards).
It's ok if you want to believe that these data give you some hint on a positive dose-effect to plan further studies, but it is clearly not enough to put some "official claim" that your data shows a positive dose-effect (e.g. in a publication).
You need to say more about effective sample size , eg gives us a cross tab of the raw 2 by 2 data. Atg by ptld as that is all the information you have on the relation.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication