The following is a Normal Probability plot of the residuals in the linear regres
ID: 3218637 • Letter: T
Question
The following is a Normal Probability plot of the residuals in the linear regression model of BAC when given alcohol: Are there any issues with this normal probability plot? Explain. A researcher was not satisfied with residual plot on the previous page. They tried a transformation on the BAC variable in the dataset The tried taking y' = Squareroot y and performing a least squares regression on alcohol consumed (x) and y* instead y. The residual plot and normal probability plot of the new regression model is given below. Did the transformation improve the plots or make them worse? Explain.Explanation / Answer
1)
the pp plot is not S shaped and hence assumption of normality of residuals is not followed.
it is recommned to go for some transformations in order to satisfy assumptions of linear regression
2)
this transformation, y* =sqrt(y) doesnt make much difference. as here also, the pp plot is not S shaped and hence assumption of normality of residuals is not followed. however points are ransomly distributed in residual plot, implying variance is constant, and hence assumtpion of homogenity of error variance is satisfied.
it is recommned to go for some other transformations (logy, exp(y), y(t-1) in order to satisfy all assumptions of linear regression analysis.
Related Questions
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.