hello everyone,
Is it possible that a particular variable shows positive relationship with dependent variable in correlation analysis and the same variable shows negative sign of coefficient as independent variable in regression analysis? If it may happen what may be the interpretation for the same?
Dear Sarita,
It is not quite possible as such. The difference between a raw regression coefficient and correlation being just a standardization factor.
However, if you mean multiple regression, then this is possible depending the relationship of this predictor to other predictors in the model. In this case you have a misspecified model and will want to do some more investigation about the relationships between all of the predictors.
I hope this helps.
The correlation coefficient between the dependent and the independent variable cannot have an opposite sign to the coefficient of the independent variable in a linear regression equation. This is because the coefficient is the rate of increase of the variable for a unit increase in the dependent variable if the other independent variables are kept constant. Is your regression nonlinear? Let us see your data and the regression equation.
Thank you Bruce, Ette and Sanjeet for your precious suggestions, it cleared my doubt.one of My friend had this question in her DRC, so we were in search of possible answers.
As far i understood, this case may happen in multiple linear regression where presence of another predictors variable may influence particular IV position or Logistic regression which is non linear
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation