multicollinearity and interaction I am a bit confused between multicollinearity and interaction. Would anyone mind explaining it to me and the difference between the two and how they affect the variables?
I understand that multicollinearity is when two of the variables correlated but how does that affect the results?
I understand interaction is when the variables cross on the graph and thus interact but how does that affect the results?
Just confused
 A: You are correct in relation to your understanding, but I will detail a little more what would be multicollinearity and interaction.
Multicollinearity: within the context of regression, multicollinearity is when there is a correlation between two or more continuous independent variables. When the objective of your analysis consists only of prediction, there is no problem with multicollinearity, but when the objective involves the interpretation of parameters and inference, the model can indicate that these correlated variables are not important when they are actually important and with that you can come to the wrong conclusions!
Interaction: the term interaction is used mainly in ANOVA, this website here has a very intuitive example of what interaction would be in practice and not just as the crossing of lines in the graph. In the analysis, you decide whether it makes sense to place the interaction based on your knowledge of the problem. If you choose to put it, the interpretations of the parameters will change and you will spend more degrees of freedom to estimate the parameters more, so you may lose some power if there are many interactions.
