※ 本文轉寄自 tomin.bbs@tomin.twbbs.org

---

http://libraryonline.leedsmet.ac.uk/lco/publications/pdf/it/qg-24.pdf
SPSS 14 Guide

Checking Normality of a distribution

A Normal distribution is a statistical distribution in which data are
represented graphically by a symmetrical bell-shaped curve, with the highest
frequency in the middle and smaller frequencies towards the edges.

There are several ways in which you can use SPSS to assess the normality of
a distribution:

‧ The simplest method of assessing normality is by producing a histogram.
The most important things to look at are the symmetry and the peak of the
histogram. A normal distribution should be represented by a bell-shaped
curve.

‧ Another method of assessing the normality of a distribution is by producing
the Normal probability plot, P-P or Q-Q plot. For a Normal distribution, the
probability plot should show a linear relationship.

‧ It is also possible to use Kolmogorov-Smirnov test if your sample size is
greater than 50 or Shapiro-Wilk test if sample size is smaller than 50. What
you need to check on the table is the Sig. value. The convention is that a
Sig. value greater than 0.05 indicates normality of the distribution.

--


--