Table of Contents Table of Contents
Previous Page  176 / 762 Next Page
Basic version Information
Show Menu
Previous Page 176 / 762 Next Page
Page Background

176

Chapter 4: Measures of Position

Figure 4-26:

Box Plot with the upper inner fence and the

outlier for

Example 4-11

Next we will discuss another test which one can use to test for outliers. This

test is referred to as the

Outer Fence Test

.

Outer Fence Test

Some authors also compute the interval [

- 3

,

+ 3

].

These endpoints are called the

outer fences

of the box plot. Using these end

points, another procedure can be presented using the inner and outer fences

to determine whether a data value can be classified as a “suspected” outlier

or if the value can be classified as an “outright” outlier. This procedure for

testing for an outlier is much more conservative than that of

the Inner

Fence Test

. This is so because the value has to be outside the outer fences

for it to be classified as an outlier.

A data value may be classified as a “suspected” outlier if it falls between the

limits of the inner and outer fences, including the values of the fences.

A data value may be classified as an “outright” outlier if it falls outside the

limits of the outer fences.

Figure 4-27

graphically shows when a data value may be classified as a

suspected or outright outlier.