I found that a chi-square or odds ratio was not significant for a 2x2 table. Later when the independent variable was put together with other variables in binomial logistic regression in R, then the same relationship came to be significant. Is it true or my calculation was mistaken?
We haven't seen your models or your data but it's certainly possible if one model is better than the other. D.Booth
The point is quite clear.
No miscalculations occurred.
The reason for the difference goes back to your words:
with other variables.
yes, it can be. sometime, some independent variable(s) did not show significant association with outcome variable (i.e. dependent variable) in univariate analysis but in multivariate, it become significant. that means, the variable(s) under consideration show independent association with outcome variable but with dependent on other covariates (i.e. other independent variables those included in the multivariate analysis).
David Eugene Booth , Abolfazl Ghoodjani Mahendra Pal Singh Negi
Thank you so much all of you. I can not upload the raw data now but I get an idea that I am going in right direction.
So much thanks again for your valuable insights.
it is not clear to me how a variable in binary logistic regression which is insignificant turns significant at multivariate logistic regression, i thought the opposite is true.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size