4 The critical values forthe t-test are found by locatingthe upper andlower limits of

interval. The result defines the regions of rejection. This requires determining the the
24 degrees of freedom (n = 25, df =n- 1), the -value is 2.064. The critical yalueconoftdet.ncFore
values are
5
Lower limit = u - S, = 20 -
2.06425)
=20 - 2.064(1)
= 17.936
5
Upper limit = pt s, = 20 + 2.064|
= 20 + 2.064(1)
= 22.064

Finally,the researcher makes the statistical decision by determining whether the samal.
between the critical limits. For the pizza store sample, X = 22. The sample mean is mean falk
in the region of rejection. Even though the sample result is only slightly less than the not incuded
at the upper limit, the null hypothesis cannot be rejected. In other words, the pizzacritical value
ager's assumption appears to be correct. store man-
As with the Z-test, there is an alternative way to test a hypothesis with the
by using the formula t-statistic. This is
obs
S
obs
=

22201
We can see that the observed t-value is less than the critical
=2
t-value of 2.064 at the 0,05 level wha
there are 25 - 1 = 24 degrees of freedom. As a result, the
p-value is greater than 0.05 and the
hypothesis is not supported. We cannot conclude with 95 percent confidence that the mean is
not 20.

The Chi-Square Test for Goodness of Fit

chi-square (x²) test A chi-square (x*) test is one of the most basic tests for
One of the basic tests for statisti
appropriate for testing hypotheses about frequencies arranged statistical significance and is particulariy
calsignificance that is particularly
appropriate for testing hypoth Univariate tests involving nominal or ordinal variables are in a frequency or contingency table.
the y² test is associated with examined with a y. More generally,
eses about frequencies arranged
in afrequency or contingency
goodness-of-fit
matrix (table) of numbers matches or fits (GOF), GOF can be thought of as how well some
table.
another
between a table of observed frequency counts andmatrix of the same size. Most often, the test
dency) for those counts. another table of expected values (central te
goodness-of-fit (GOF) Consider the following hypothesis that relates back to
A general term representing how
well some computed table or H,: Papa John's Pizza stores are more the chapter vignette:
matrix of values matches some center.
likely to be located in astand-alone location than in a
population or predetermined
SlopPS
table or matrix of the same size. Acompetitor may be
interested in this
ing plan. Aresearcher for hypothesis as part of the competitor analysis in a market-
the competitor gathers a random sample of 100 Papa John Slo
California (where the competitor is andand
the locations are
checked by having
located). The sample is selected from phone directories
an
frequency table. assistant drive to each location. The following
are recorded in a observations

Location
One-Way Frequency Table
Stand-Alone 60 stores
Shopping Center 40 stores
Total 100 stores
Chapter21: Unival| alysis

observed values (O) can be 553

These
complete a x?test.
The x² value will compared to the
expected
reflect the likelihood values for this
distribution reeflected by the
isthatthe expected and observed expected
valuesvalues.
are theThe
that the
higher the value ofobserved
same.
distribution
values come
(E) to
the x' test, the less from a
In
statisticalterms, a x test
determines whether the difference likely it
distribution and the corresponding
expected frequency distributionbctween an observed
Computing a x test is fairly
straightforward and easy. Students who due to sampling frequency
is
have
little trouble
1understanding future significance tests since the master this calculationvariation.
lies these tests as well
basic logic of the x' test should
The steps in computing a X* test are as follows: under
or data and tally the
1. observed
2. Compute the expected values for frequencies
for the categorical
each value of variable.
the categorical variable.
3.
eate the x² value, using the observed
frequencies. frequencies rom the sample and the expected
A Find the degrees of freedom for the test.
t Make the statistical
decision by
predetermined significance levelcomparing p-value associated with the calculated y² against the
(acceptable Type Ierror rate).
These steps can be illustrated with the pizza
store location example.
The data for the location variable
ing frequency table. (stand-alone or shopping center) are provided in the preced
The next step asks, "What are the expected
frequencies for the
another way of asking the central tendency for each category. location variable?" This is
Since the sample size is 100,
finding the expected values is easy. If no pattern exists in the locations, they
tributed randomly across the two categories. We would expect that half (50) ofshould be dis
the locations
would be stand-alone and half (50) would be in a shopping center. This is
another way of
saying that the expected probability of being one type of location is 50 percent. The expected
values also can be placed in a frequency table:

Location Expected Frequencies

Stand-Alone 100/2 = 50stores

Shopping Center 100/2 = 50stores

Total 100 stores

ne actual y² value is computed using the following formula:

x²=-E)?
E,
Where
X*= chi-square statistic
observed frequency in the ith cell
E, = expected frequency in the ith cell

Sum the squared differences:

(o - E)', (9- E)
E,
E,
4:
chi-square value equals
Thus, we determine that the
(60- 50)²
(40 - 50)?
50
50

= 4
 RESE ARCHSNA P H O T

computed as shown
Interested in Retirement? It Often Depends on Your The y' value can be
Age below:
Chi-square tests are used often in business research. Consider
a business that sponsors a program to educate employees on
retirement issues. They need to plan the number and types of
activities that should be the focus of the training and develop
ment seminars. One question is whether or not an equal number
of younger versus older employees will come to the sessions. Expected Observed O-E (0- EY (0- EYIE
They decide to observe the relative frequencies of younger ver -22 484
sus older employees based upon the number of sign-ups they Younger 100 78 48A
122 22 484 484
receive in the first week since the program was announced, with Older 100
a cut-off set at 200. The results are shown in the bar chart below: 968
120
The HR managers want to be sure a difference exists before inyect.
100 ing resources into activities designed for younger or older employ.
80 ees only. Therefore, the acceptable levelof Type l error is set at 0.01
Rather than referring to acritical value table, the p-value associated
60 Younger with ay²value and the associated degrees of freedom can be found
D
Older Internet
40 on any one of several statistical calculators found on the
In this case, the researcher uses the calculator found at http://facuty
20 vassar.edu/lowry/tabs.html#csq. By simply plugging in the observed
0
value of 96.8 and the number of degrees of freedom as indicated. 1
Expected Observed in this case, the calculator returns a p-value. In this case, the p-value
OFLIRT/JUPITER
IMAGES returned is less than 0.0001. Therefore, since the p-value is less than
the acceptable levelof risk, the researcher reaches the conclusion
that the older workers are much more likely to attend the retirement
seminar. They can design the seminar to meet the needs associated
with the number and type of attendees.

Alternatively, the calculation can be followed in tabular form:

Location:
(0, E)
(0,E
Stand-Alone 60 50 10 100/50 = 2.0

Shopping Center 40 50
-10 100/50 = 2.0

Total 100 100 X?=40

Like many other probability distributions, the x² distribution is not a single probability
but a family of curves. These curves vary slightly with the degrees of cunc
degrees of freedom can be computed as
freedom. In this case, i

df= k- 1
where
k= number of cells associated with
column or row data.
Thus, the degrees of freedom equal 1 (df =2-1= 1).
Now the computed y' value needs to be compared with the critical chi-square values associ-
ated with the 0.05 probability level with 1 degree of freedom. In Table A.4 of the appetl-
dix the critical x' value is 3.84. Since the calculated y² is larger than the tabular chi-square,
the conclusion is that the observed values do not equal the expected values. Therefore, the
hypothesis is supported. More Papa John's restaurants are located in stand-alone locations.
We discuss the chi-square test further in Chapter 20 as it is also frequently Used to analyzecon-
tingency tables.
554