Reply
Sat 4 May, 2019 01:20 pm
Dissertation Results. X and Y non-correlation, but then X significantly predictive in regression model. Why?
I just finished my dissertation results and trying to interpret them and very confused. Basically, X and Y are not correlated but then when I put X in my regression model with a few other variables X significantly predicts Y. I've been explained this statistically and somewhat understand it --- that when you control for other things the relationship changes, but theoretically it makes absolutely no sense. Can someone give me an example using variables showing how they are not related when looking at them in a vacuum but then becomes a significant predictor in another instance? Having a hard time writing discussion because of this. Thanks.
@disserthelp,
It all depends on the magnitude of the impact. Let's say X and Z impact Y. X impacts it slightly, Z impacts it tremendously. If you plot Y vs X without taking Z into account, it looks like X is not significant. You don't need to explain this. Publish the results of the regression model and note the size of the contribution from each independent variable.