there are so many things in order to imply causality. So with regression analysis you can not say anything about this. Furthermore Rquared or adjusted R squared (with more than one explanatory variable) we defne the proportion of variation in the response that is explained by the regression model.

Two misconceptions there are :
1. If R squared is very large,  the regression model is useful for predicting new observations.
2. If R squared is small, there are no meaningful relationships exist in the data

看样子论坛只能是搜集地方的，自己好好看看吧！