Provided that my model is well-fit and correctly classifies 77% of the cases with sensitivity, specificity, positive predictive value, and negative predictive value of 65%, 58%, 61%, and 62% respectively. Would this amount of variance weaken my conclusion?
To rephrase my question: I used a binomial regression model for my dependent dichotomous variable. The model meets all the statistical assumptions and is well fit. However, it explains only 41% of the variance in my dependent variable, i.e. r-square=0.41. This piece of info is not a stand-alone while interpreting my results, yet I’ve seen models explaining >85% of the variance using the same dependent variable I’ve used. Could this r square result weaken my model, in spite of having all the assumptions met? Thank you in advance!
Your question is a bit confusing. R2=0.41 IS the amount of variance explained by your model. It seems to me that that the problem is that you were expecting it to be higher on the basis of other studies. It might be fruitful to ask yourself why your R2 is lower.
Martin Willis Thank you for your answer! I, indeed, was expecting a higher value, yet the independent variables I loaded into my model are different from those in other research papers, which is fair enough. So I am embracing my results ;)
My question is: do I need to report this finding as a limitation of my study? Would it weaken my regression results?
Could you explain what you mean by "the independent variables I loaded into my model are different from those in other research paper"? Do you mean they were not the same types of variables or that they produced different coefficients? If it's the latter, I don't think it is a "limitation" or "weakens" your results, but it is something you probably need to discuss. If it's the former, then you can't really compare with other studies.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation