Chapter 10

Statistical Inference: One-

Sample Hypothesis Test
I Scientific Hypothesis
A testable supposition that is tentatively adopted
to account for certain facts and to guide in the
investigation of others.
1
II Statistical Hypotheses
Statements about one of more parameters of a
population distribution

A. Null Hypothesis (H0)

Hypothesis that the researcher tests

B. Alternative Hypothesis (H1)

Hypothesis that corresponds to the researcher’s

scientific hunch 2
 C. Examples of Null and Alternative Hypotheses

H 0 : μ ≤μ0 H 0 : μ ≥μ0 H 0 : μ = μ0
H 1 :μ > μ0 H 1 :μ < μ0 H 1 :μ ≠μ0

1.  denotes the unknown mean of a population;

0 denotes a hypothesized value of the population

mean that is to be tested.

3
D. Characteristics of the Null Hypothesis, H0,

and the Alternative Hypothesis, H1

1. H0 contains an exact statement,  = 0, that can

be tested. H1 is an inexact statement:  > 0,

 < 0, or  ≠ 0.

2. The null and alternative hypotheses are mutually

exclusive and exhaustive. If the null hypothesis is
rejected, the alternative is probably true.
4
E. Statistical Hypotheses for An Experiment
1. The dean at Idle-On-In College believes that a
new registration procedure will reduce the time
required for students to register. For the past
several years, the current procedure has required
3.10 hours. The statistical hypotheses are
H 0 : μ ≥3.10
H 1 :μ < 3.10,
where denotes the unknown population mean
for the new procedure and 0 = 3.10 denotes the
mean of the current procedure.
5
3. To test the null hypothesis, the dean conducts a
trial run with a random sample of n = 27 students
and obtains the following data:
X = 2.90 and σ̂ = 0.3013.

4. The new procedure appears to require less time

than the old procedure. But the dean worries
because she knows that sample means vary from
sample to sample. She wonders, “Is it possible that
the population mean of the new procedure is really
equal to or greater than the old procedure?”
6
4. To determine if the null hypothesis is credible or
not credible, the dean computes a t test statistic
for her data and determines the probability of
obtaining a sample mean of 2.90 if the true
population mean is really equal to or greater than
3.10.

5. She discovers that the probability of obtaining a

a sample mean of 2.90 if 0 = 3.10 is .001. She

decides to reject the null hypothesis and concludes

that the new procedure is better.
7
 F. Role of Logic in Hypothesis Testing

S c i e n t i f i c
S t a t i s t i c a l
D e d u c t i v e i n f e r e n c e
h y p o t h e s i s
h y p o t h e s e s

R a n d o m s a m p l i n g

I n d u c t i v e
a n d e s t i m a t i o n o f
i n f e r e n c e

p o p u l a t i o n p a r a m e t e r

S t a t i s t i c a l

t e s t

8
III Two Statistics for Testing H0:  = 0

A. t Test Statistic

X −μ0 X = ∑ X i /n
t=
σ̂ / n μ0 = hypothesized pop.m ean

σ̂ = ∑(X i − X )2 /(n −1)

n = sam ple size

9
 B. Sampling Distribution of the t Statistic

ν = ∞ ( same
) as normal distribution)

ν = 12

f ( t )

ν = 4

Figure 1. Graph of the distribution of t for 4, 12, and ∞

degrees of freedom. When n = ∞, the t distribution is
identical to the z (standard normal) distribution.
10
 1. Concept of degree of freedom, df or 

Degrees of freedom for the t statistic:  = n – 1

3. Variance of the t statistic:

ν
Var(t) =
ν −2

4. A bit of history: 1908, the beginning of a new

era in statistics

5. Contribution of William Sealey Gossett (Student)

11
Table 1. Percentage Points of Student’s t Distribution

Level of Significance for a One-Tailed Test

.05 .025 .01 .005

Degrees of Level of Significance for a Two-Tailed Test

Freedom,  .10 .05 .02 .01

10 1.812 2.228 2.764 3.169

19 1.729 2.093 2.539 2.861
20 1.725 2.086 2.528 2.845
26 1.706 2.056 2.479 2.779
30 1.697 2.042 2.457 2.750
38 1.686 2.024 2.429 2.712
∞ 1.645 1.960 2.326 2.576
12
C. z Test Statistic

X = ∑ X i /n
X −μ0
z=
σ/ n μ0 = hypothesized pop.m ean
σ = know n population
standard deviation
n = sam ple size

1. The z test statistic is rarely used because the

population standard deviation is usually not
known.
13
 2. The sampling distribution of z is the standard
normal distribution.
3. Var(z) = 1

D. Comparison of t and z

X −μ0 Random variable – Constant

t= =
σ̂ / n Random variable

X −μ0 Random variable – Constant

z= =
σ/ n Constant
14
IV Steps in Testing a Null Hypothesis
A. Registration Example at Idle-on-in College
1. Is a new registration procedure better than the
current procedure? For the past several years, the
current procedure has required 3.10 hours.

2. To evaluate the new procedure, the dean conducts

a trial run with a random sample of n = 27
students and obtains the following data:

X = 2.90 and σ̂ = 0.3013.

15
3. Is the new procedure, X = 2.90,really better than
the current procedure, μ0 = 3.10,

4. Because of sampling variability, it is unlikely that

the sample mean is equal to the population mean
of the new procedure.

5. It seems reasonable to reject the null hypothesis if

the observed sample mean, X = 2.90,is very
improbable given that 0 is equal to 3.10.

16
 X = 2.62

n = 27

μ =? X = 2.98

n = 27

= 3.24
X N = 12,660

n = 27

= 2.90
X
X = 2.86

n = 27
n = 27

X = 3.21

n = 27

Figure 2. Illustration of sampling variability of sample means

for six random samples from a population
17
6. A sample mean that would be observed 5 or fewer
times in a hundred replications of an experiment
(Probability = 5/100 = .05) is considered very
improbable.

7. A sample mean with probability 5/100 = .05 is

reasonable grounds for rejecting a null hypothesis.
This criterion is called a significance level and
denoted by  = .05

18
 B. Five-Steps in Testing a Null Hypothesis for
the Registration Example
Step 1. State the statistical
hypotheses: H 0 : μ ≥3.10
H 1 :μ < 3.10
Step 2. Specify the test
statistic: t statistic because she wants
to test  ≥ 3.10,  is unknown,
the sample is random, and the
population distribution of X is
probably approximately normal
19
Step 3. Specify the sample
size: n = 27
and the sampling
distribution: t distribution with  = n – 1 = 26,
because  is unknown and must
be estimated, and the population
distribution of X is probably
approximately normal
Step 4. Specify the
significance level:  = .05
20
Step 5. Obtain a random
sample of size n,
compute t, and
make a decision

Decision Rule: Reject the null hypothesis if t falls in the lower

5% of the sampling distribution of t; otherwise, do not
reject the null hypothesis. If the null hypothesis is
rejected, conclude that the new registration procedure
reduces the time to register; if the null hypothesis is not
rejected, do not draw this conclusion.
21
C. Selecting a Level of Significance, 
1. Common conventions in specifying 

D. Specifying the Critical Region for Rejecting

H0

1. The location of the critical region is determined by

the alternative hypothesis.
2. The size of the critical region is determined by the
level of significance, .
3. The critical region identifies values of X that
are very unlikely if the null hypothesis is true.
22
 Critical region

for α = .05
f ( t )

–3 –2 –1 0 1 2 3

Critical

value = –1.706

2.926 2.984 3.042 3.100 3.158 3.216 3.274

Reject
H Don't reject H
0
0

Figure 3. Critical region for rejecting the null hypothesis

23
V Computation of t Statistic
Registration
( X i − X )2
Student Time, Xi

1 2.9 0.00
2 2.7 0.04
3 2.4 0.25
. . .
27 2.5 0.16
∑ X i = 78.3 ∑( X i − X )2 = 2.36

∑ X i 78.3 ∑ X i − X )2 2.36
X= = = 2.90 σ̂ = = = 0.3013
n 27 n −1 27 −1
24
 X −μ0 2.90−3.10
−0.20
t= = = = −3.449
σ̂ / n 0.3013/ 27 0.0580

ν = n −1= 27 −1= 26

−t.05,26 = −1.706

A. Reject H 0 : μ ≥3.10 because

t = −3.449 < t.05,26 = −1.706

25
 Critical region

for α = .05
f ( t )

–3 –2 –1 0 1 2 3

Critical

value = –1.706

2.926 2.984 3.042 3.100 3.158 3.216 3.274

Reject
H Don't reject H
0
0

Figure 4. X = 2.90 and t= −3.449 fallin the  =

critical region
26
VI Some Experimental Design Considerations

A. John Henry Effect

B. Importance of Control Groups

VII One- and Two-Tailed Tests

A. One-Tailed Test
1. Critical region is in either the upper or
lower tail of the sampling distribution.
27
 Critical region
f ( t )

α = .05

–3 –2 –1 0 1 3
2

t
.05, 26
= 1.706

Don't reject
H 0 Reject H 0

Figure 5. Critical region for H0:  ≤ 3.10

28
 B. Two Tailed Test

Critical region Critical region

f ( t )

α /2 = .025 α /2 = .025

–3 –2 –1 0 1 2 3

– t = – =
.05/2, 26
2.056 t 2.056
.05/2, 26

Reject Don't reject H Reject

0

H H
0 0

Figure 6. For H0:  = 3.10, the critical region is divided

between the upper and lower tails of the t distribution.
29
C. Denoting Critical Values of t

1. Upper and lower two-tailed critical values are

denoted by, respectively, t/2,  and –t/2,  ,

where  is the significance level and  is the

degrees of freedom.

2. t,  denotes the upper one-tailed critical value

3. –t,  denotes the lower one-tailed critical value

30
 D. Merits of One- and Two-Tailed Tests

VIII Type I and Type II Errors

A. Type I Error, corresponds to , (Rejecting a

True Null Hypothesis)

B. Type II Error, corresponds to , (Failing to

Reject a False Null Hypothesis)

31
C. Correct Acceptance and Correct Rejection

D. Power, 1 – , for the Registration Example

1. Value of the sample mean that cuts off the lower
.05 region of the t sampling distribution

X .05 = μ0 + t.05,26σ̂ / n

= 3.10+ (−1.706)/(0.3013)/ 27 = 3.001

32
 2. Minimum reduction in registration time that
would
warrant changing to the new procedure
μ0 −μ ′ = 3.10−2.95= nine m inutes

where  = 2.95 corresponds to the new

registration time.

3. The size of the region corresponding to a Type II

error can be determined by computing a t statistic
for the difference X .05 −μ ′ = 3.001−2.950

33
4. t statistic for computing the probability of a Type II
error, , and power, 1 – 

X .05 −μ ′
t=
σ̂ / n

3.001−2.9500.051
= = = 0.880
0.3013/ 27 0.058

5. Area above t = 0.880 is .19 (Prob. of a Type II error)

34
 Sampling distribution

under H
0

1 – α = .95

α = .05

t
f ( t )
μ = 3.1
Sampling distribution 0

under μ ' = 3.001

H when = 2.95 X
1 .05

^
1– = .81
β

β = .19

'
μ = 2.95

Reject H Don’t reject

H
0

35
0
 Table 2. Probabilities Associated with the Decision Process

True Situation

μ = 3.10 μ ' = 2.95

Correct acceptance Type II error

μ ≤ 3.10
^

1 – α = .95 β = .19

Researcher’s

Decision

Type I error Correct rejection

μ > 3.10
^

α = .05 1 – β = .81

36
E. Factors that determine the probability of a
Type II error,  and power, 1 – 

1. Level of significance, 

2. Size of the sample, n

3. Size of the population standard deviation, 

4. Magnitude of the difference between  and .

37
 5. Simple way to increase power: increase the
sample size.

IX Determining the Required Sample Size (n)

A. Cohen’s Effect Size, d

d = |μ −μ0 |/σ

 d = 0.2 is a small effect

 d = 0.5 is a medium effect

 d = 0.8 is a large effect
38
Table 3. Approximate n Required for Testing Hypotheses
One-Sample Test

One-Sided Two-Sided
Hypothesis, Hypothesis,
Effect 1 −β 1 −β
Size,
d 

0.2 .05 156 215 272 198 264 326

.01 253 328 396 294 374 447
0.5 .05 27 36 45 34 44 54
.01 43 55 66 51 63 75
0.8 .05 12 15 19 15 19 22
.01 19 24 28 22 27 32
39
B. Statistical Significance Versus Practical
Significance

C. Reporting p Values

1. Obtaining p values from Excel

Select INSERT from the Excel menu

Select Function
Select the TDIST function

40
2. TDIST function

TDIST(x,deg_freedom,tails)

 Replace x with the t value

 Replace deg_freedom with the df for t

 Replace tails with 1 for one-tail and 2 for two-

tail

TDIST(3.449,26,1)

The p value is .0009652, or < .001

41