Page 292 - Statistics II for Dummies

276 Part V: Nonparametric Statistics: Rebels without a Distribution

No need to fret if conditions aren’t met

Many of the techniques that you typically use to analyze data, including
many shown in this book, have one very strong condition on the data that
must be met in order to use them: The populations from which your data are
collected typically require a normal distribution. Methods requiring a certain
type of distribution (such as a normal distribution) in order to use them are
called parametric methods.
The following are ways to help you decide whether a population has a normal
distribution, based on your sample:
✓ You can graph the data using a histogram, and see whether it appears
to have a bell shape (a mound of data in the middle, trailing down on
each side).
To make a histogram in Minitab, enter your data into a column. Go to
Graph>Histogram, and click OK. Click on your variable in the left-hand
box, and it appears in the Graph Variables box. Click OK, and check out
your histogram.
✓ You can make a normal probability plot, which compares your data to
that of a normal distribution, using an x-y graph (similar to the ones
used when you graph a straight line). If the data do follow a normal
distribution, your normal probability plot will show a straight line. If
the data don’t follow a normal distribution, the normal probability plot
won’t show a straight line; it may show a curve off to one side or the
other, for example.
To make a normal probability plot in Minitab, enter your data in a
column. Go to Graph>Probability Plot, and click OK. Click on your
variable in the left-hand column, and it appears in the Graph Variables
column. Click OK, and you see your normal probability plot.

When you find that the normal distribution condition is clearly not met,
that’s where nonparametric methods come in. Nonparametric methods are
those data-analysis techniques that don’t require the data to have a specific
distribution. Nonparametric procedures may require one of the following two
conditions (and these are only in certain situations):

✓ The data come from a symmetric distribution (which looks the same on
each side when you cut it down the middle).
✓ The data from two populations come from the same type of distribution
(they have the same general shape).

24_466469-ch16.indd 276 7/24/09 9:53:54 AM