Determining Linearity, Normality, and Homoscedasticity for Multiple Regression

#1
Hi All,

I am completing a multiple regression and was hoping to get some advice on whether you believed this scatterplot and Q-Q plot of residuals between the dependent variable and independent variables demonstrates linearity, normality, and homoscedasticity.

To me, I believe the graph seems somewhat linear and homoscedastic, but perhaps non-normal?

It would be great to get your opinions on this and what I could possibly do (using SPSS) to remedy any violations of assumptions.

PS. I have always struggled with statistics, so if you can, please be as simple as possible.

Thank you!!
 

hlsmith

Omega Contributor
#9
Can you construct a histogram as well and report the context (y = x + x2 +x3,...,xk; and tell us what the variables are). Is this for a class?
 
#10
Apologies for the delay, I have been trying to get my head around everything that I need to know!!

I have attached a histogram of the standardised residuals for the regression model.

This is for my thesis (eek), and what I am looking at is how well 8 IVs predict anxiety score, as well as looking at each individual IV. As it is psychology, the field is not as stringent with regards to assumptions as most statisticians are. However, if you could provide me with your opinion on whether the graphs suggest linearity, homoscedasticity, and normality, I would be so so grateful!!