0

Can anyone explain the paradox of observing a non-significant Pearson correlation although that independent variable should be included in the Regression model?

Rachel
  • 1
  • Welcome to CV. Since you’re new here, you may want to take our [tour], which has information for new users. Why do you think there is a paradox? – T.E.G. Apr 09 '18 at 21:55
  • Thanks for your reply. I would just like to clarify if one predictor is non-significant in the Pearson correlation but significant in the Regression Model, what does this mean? Aren't these result contradicting? – Rachel Apr 09 '18 at 22:01
  • It might be better if you provide more information. Is there only one predictor in linear regression as well? And in fact, is this a linear regression model? – T.E.G. Apr 09 '18 at 22:03
  • Its a multiple regression, one DV and 5 predictors. The multiple regression shows that one of the predictor is significant in this model but non-signficiant for pearsons. What are the implications of these results? – Rachel Apr 09 '18 at 22:06
  • Yes, my bad, it is in the title. You have other predictors in multiple regression. And the coefficients are partial coefficients, i.e., you control for the effects of other variables. That is why the results are different. – T.E.G. Apr 09 '18 at 22:10
  • 1
    Among many other reasons (example multicollinearity), one common reason is if variance of the predictor variable is too high compared to the variance of dependent variable. – uday Apr 09 '18 at 22:40
  • Maybe look at this: https://stats.stackexchange.com/questions/32896/why-dont-the-results-of-testing-h-0-beta-0-and-h-0-rm-corx-y-0?rq=1 – rannoudanames Apr 10 '18 at 03:17
  • I provided an answer to another question that will also address your question: https://stats.stackexchange.com/questions/338823/multiple-regression-standardized-coef-changed-direction-even-with-low-vif/338861#338861 . One of the correlations I reported had a p-value of .51, but once in the regression equation with another control variable, the p-value dropped to .06. Likely due to suppression, as explained in the link. – Bryan Apr 10 '18 at 03:58

0 Answers0