I am not sure what you are plotting - perhaps post an image of the plot, but note that normality is not a requirement for your dependent or independent regression variables. The normality assumption applies to the distribution of the error term in the equation. Of course, since the Likert values are integers, their distribution can't truly be normal, but that may be a minor matter.
------------------------------
Jon Peck
------------------------------
Original Message:
Sent: Thu May 14, 2020 09:16 PM
From: Anne Heath
Subject: Assessing Normality (that is not normal)
I am assessing normality and my histograms have an undesirable shape, not one in which the P-values are most dense near zero. I want to make sure I have not entered the data into SPSS incorrectly. Since my output does not look appropriate, I am questioning my process. I used two surveys that had a (one with a 7 point and another with a 9 point Likert scale) and plan to conduct a multilinear regression (1 DV and 5 IVs). Because my normality check does NOT look normal (histograms and scatterplots), I am not sure what to do.
- Should I use the Wilcoxon or Kruskal–Wallis to assess normality?
- Do I need to enter specific values for Likert data such as 1 (not at all) and 7 (always) when I enter the data? if so what about the unlabeled values (between not at all and sometimes)?
Thank you for any clarification or suggestions!
------------------------------
Anne
------------------------------
#SPSSStatistics