# Non-Normal data in multiple linear regression [duplicate]

My study involve 73 subject. I want to carry out a multiple linear regression in order to form a predicting formula. However my dependent variable is not normally distributed according to a normality test. It is close to normal distirubtion in the p-value of normality test and it looks like normal distributed data in histograms.

In this case should I continue using multiple linear regression or Should I transform the variable/using non-parametric regression method?

If I can continue using multiple linear regression, is there any statistical journal that I can cite for explaining why I can do that? Do you have some key words that I can google for?

• Which normality test are you using? Oct 26, 2016 at 13:53
• First recognize that it's in no way a requirement of linear regression that the dependent variable should follow a normal distribution - see e.g. What if residuals are normally distributed, but y is not? or Normality of dependent variable = normality of residuals?. Oct 26, 2016 at 13:58
• Tp Ferdi: Shapiro–Wilk test Oct 26, 2016 at 14:19
• To Scortchi: So it is ok to run the linear regression am I right? Thanks you! Oct 26, 2016 at 14:20
• Yes - then examine the residuals to check assumptions. Oct 27, 2016 at 8:49