Hi,

Am I correct in assuming that if a variable was found to be good predictor in univariate analysis, it should also be found to be a good predictor in multivariate analysis as multivariate takes into account multiple variables including interaction terms.

I am going through a research paper in which univariate and multivariate analysis were done on a dataset. In univariate analysis, two independent variables were found to be related to dependent variable. However in multivariate analysis only one independent variable was found to be related to dependent variable. Moreover the only independent variable found to be a good predictor in multivariate analysis was different that the two independent variable found to be good predictors in univariate analysis.

Do these result seem to be ok?

Cheers,
Ambi.