0
$\begingroup$

In order to do a linear regression, I need to test my data set for normality. I have one dependent variable plus three independent variables. The independent variables are all scale variables, i.e. without any groups/levels (e.g. male/female). How do I test for normality in SPSS? Unfortunately, I can only find procedures that describe what to do if the independent variables have different groups/levels. In such a case, it seems you have to establish normality for all separate combinations of the groups/levels of the independent variables by splitting the file and organizing the output by the different groups/levels. See this instruction video on youtube: https://www.youtube.com/watch?v=IiedOyglLn0. But what if my variables don't have any groups/levels?

$\endgroup$
4
  • $\begingroup$ I cannot make any sense of this because variables with "sub-groups/sub-levels" do not sound like numerical variables at all: what would Normality even mean and why would you want to test it? Could you provide some context and a concrete example of what you're dealing with? $\endgroup$
    – whuber
    Commented Jan 4, 2020 at 21:39
  • $\begingroup$ Thank's for the comment. I've edited my question. $\endgroup$
    – M_S
    Commented Jan 4, 2020 at 21:54
  • $\begingroup$ You do not need to test any variables for normality in linear regression: that's not an assumption any linear regression model makes. $\endgroup$
    – whuber
    Commented Jan 5, 2020 at 4:39
  • $\begingroup$ Okay, thanks. I thought that I have to test whether or not my variables are normally distributed in order to do a linear regression... $\endgroup$
    – M_S
    Commented Jan 5, 2020 at 9:35

0