4) In the dataset there is a total score of an aggression questionnaire (agtot). This measure is broken down into three subscales, which are listed as separate variables: Hostility (aghost), Anger (agang), and Physical Aggression (apa). Run the regression syntax that is in the midterm syntax file under the heading ***QUESTION 4***. Imagine you are advising a high school student on their honors thesis. They come to you with this output printed out, excited that they have a significant result. They are examining how the total aggression score and the subscales predict depressive symptoms (bdi) in your sample. Looking at the data, you tell the student that this finding is not publishable because their output shows that one of the assumptions of multiple regression has been violated. Explain the problem suggested by the output: HINT: Do not use structure/squared structure coefficients for this) (2 points) 5) You decide to run a different model that 'fixes' the problem that you identified in Question #4. Run the model that is in the syntax under the heading ***QUESTION 5***. You are more excited about these results, but the student has a few questions for you: a) You're amazing! But, where in the output are you looking to determine if the full model is significant? And how can you tell that it is a significant result? (1 point) b) What percentage of variance in depression do the hostility, anger, and physical aggression subscales collectively account for? (1 point) c) Which subscale or subscales are uniquely related to depression? (1 point) d) Generally, in multiple regression, what does it mean if a predictor is "uniquely" related to an outcome?