Hypothesis Testing and

Estimation for Two Population

Parameters
Vicente E. Montaño, DBA
 Hypothesis test and Estimation for two
population means when 1 and 2 are
unknown-Independent Samples
• In most situations in which there is a need
to test two populations standard deviation .
Instead there will be a estimate 1 and 2
using s1 and s2, computed from samples
selected from the two populations. This is
the case of of t-distribution to establish the
critical value. The situation is that samples
are considered to be independent samples.
Independent samples
Samples selected from two or more
populations in such a way that the occurrence
of values in one sample has no influence on
the probability of the occurrence of values in
their other sample (s).
 Retirement Investing
• A major political issue for the past decade has
involved questions about the long-term future of
Social Security system. Many people who have
entered the workforce in the last 20 years believe the
system will not be solvent when they retire and are
thus actively investing their own retirement accounts.
One investment alternative is a tax shelter annuity
(TSA) marketed by life insurance companies. Certain
people depending on occupation, qualify to deposit
part of their paycheck in a TSA and pay no federal
income tax on this money until it is withdrawn.
 Cont …………
• While the money is on deposit, the insurance
companies invest it in their stock or bond portfolio.
If the portfolio performs well, the TSA accounts
grow. A second alternative open to many people is a
plan known as 401 (k), in which employees
contribute a portion of their paycheck to purchase
stocks, bonds, or mutual funds . In some cases
employers match all or part of the employee
contribution. In many 401 (k) systems the employees
can control how their funds are invested.
 Cont …………
• A recent study in North Carolina investigated the
question of whether employees with equivalent
annual incomes covered by two plans [TSA or 401
(k)] differ in their average annual contribution. A
random sample of 15 people was selected from the
population of adults in North Carolina who have
401 (k) plans. The variable of interest is the dollar
amount of money invested in the retirement plan
during the previous year. The samples are
considered to be independent since the amount
invested by one group should have no influence
on the amount invested by the other.
 Cont…..
• Testing whether the mean dollars invested
in TSAs and 401 (k) programs are the same
or different, the hypothesis test will be two-
tailed, using the significant level of 0.05
Hypotheses:
Ho: 1 - 2 = 0 or 1 = 2
Ha: 2 - 2 = 0 or 1 = 2
Where:
1 = mean dollar invested by the TSA eligible
population during the past year.
2 = mean dollar invested by the 401 (k) eligible
population during the past year.
TSA Eligible 401 (k) Eligible

n1 = 15 n2 = 15

x1 = $2,255 x2 = $2,140

s1 = $ 645 s2 = $ 708
 To determine whether the null hypothesis should be
rejected or not the test statistics should be first be
calculated.

x _ x _ ( _  )
( 1 2) 1 2
t= df = n 1 + n 2 - 2
sp 1 + 1
n1 n2

_ _
Where: x1 and x 2 =sample means from population 1 and 2.
1 - 2 = Hypothesized difference between population
means
n 1 and n = sample sizes from the two populations
2
sp = pooled standard deviation
The test statistics is based upon two important
assumptions:
1. Each population has a normal distribution.
2
 2

2. The two populations variances 1 and 2 are
equal.

If standard deviation is equal then the pooled

standard deviation would be computed.
 Pooled standard deviation
(n 1 _ 2
1 ) 1 +( n _ 1 )s 2
s
2 2
sp =
n + n _ 2
2 1

2
s1 = sample variance from population 1
2
s2 = sample variance from population 2
n 1 and n 2 = sample size from population 1 and 2
Are population variance equal?
Before proceeding to test whether the population mean are
equal, a test is needed to determine the validity of the equal
variances assumption. The F-test can be used to determine
this.

2
s1
F=
2
s2
Using  = 0.10 a two tailed test statistics for variance we put the
larger variance in the numerator giving:

2
708
F= = 1.205
2
645

The critical value from the F-distribution table for D1 = (15 – 1

= 14) and D2 = (15 – 1 = 14) degrees of freedom, and the upper
tail equal to /2 = 0.05, is 2.484. Since F = 1.205 < 2.484, based
on the sample data there is insufficient evidence to conclude that
the population variance are different. This means that we can
proceed to test if population means are equal using the t-test
statistics.
Are the population means equal?
(n 1 _ 2
1 ) 1 +( n _ 1 )s 2
s
2 2
sp =
n + n _ 2
2 1

( 15 _ 2
1 ) 645 + ( 15 _ 1 ) 708 2

sp =
+ _ 2
15 15

= 677.23
 x _ x _ ( _  )
( 1 2) 1 2
t=

sp 1 + 1
n1 n2

_ ( 0.0 )
t= ( 2,255 – 2,140)

677.23 1
_ + 1_
15 15

= 0.465
This indicate that the difference in sample means is 0.465
standard errors above hypothesized difference of 0. Next
determine the critical value from the t-distribution table
(Excel’s TINV function) with degrees of freedom equal to n1
+ n2 – 2 = 15 + 15 – 2 = 28 and alpha = 0.05 for the two tail
test. The appropriate value is 2.048.
Since t = 0.465  2.048, the null hypothesis should be
rejected.
What if the population variance
are not equal?
_ _ 
( x1 x2 ) _ (  1 2 )
t=
2 2
sp s1 s2
_ + _
n1 n2
 Degree of Freedom for t-test
statistics with unequal population
variance
2 2
( s1 / n 1 + s2 / n ) 2
2

2 2
( s1 / n 1 s2 / n ) 2
2

n _1
1 n2 _1
Television Viewers Income Case
• The managers believe the mean household income
for cable subscribers is less than for those who do
not subscribe. Thus, by advertising on the network
station , the business could reach a higher income
audience. However, has no actual data to support
their contention. In the spirit of friendly
cooperation, the network managers joined forces
with the local cable provider, Future Vision, to
survey a total of 548 households in the market
area.
Hypothesis Testing-Two Means
- Future-Vision
Issue:
The managers believe the mean household
income for cable subscribers is less than for those
who do not subscribe.
Objective:
Use Excel to help test a hypothesis about the
mean income of two groups. Data file is Future-
Vision.xls
Hypothesis Testing – Future Vision

Open File Future-Vision.xls

File contains sample

data for 548 viewers –
last row is 549
Hypothesis Testing – Future Vision

Click on PHStat, then

Data Preparation and
finally on Unstack Data
Hypothesis Testing – Future Vision

Click on Tools – then Data

Analysis – then F-Test Two-
Sample for Variances
Hypothesis Testing – Future Vision

Define ranges for Variable 1 (No)

and Variable 2 (Yes) and Select
the Alpha value (.05)
Hypothesis Testing – Future Vision

The Excel output shows we

should reject the hypothesis of
equal variances. The p-value is
less than alpha.
Hypothesis Testing – Future Vision

Since we reject the

hypothesis of equal
variances, return to
the unstacked data,
and Click on Tools,
then Data Analysis
and finally, t-test
Two-Sample
Assuming Unequal
Variances
Hypothesis Testing – Future Vision

Enter data range for Variable 1

and Variable 2 and Hypothesized
Mean Difference
Hypothesis Testing – Future Vision

Since the p-value is

greater than alpha,
do not reject the
hypothesis.
 Estimating the Difference
Between Two Population Means
When 1 and 2 Are Unknown-
Independent Samples
Common situation the standard deviation are not known
and must be estimated from sample data. There are two
types of estimates, point estimates and confidence interval
estimates. In attempting to estimate the difference between
two population means  1 -  2 , the point_estimate
_ is
the difference between the two sample means, x1 x -2 .
However, point estimates are subject to sampling error, a
confidence interval estimates can be developed.
 Point Estimate  Critical value (standard error)
Confidence Interval Estimate for  1 -  2
standard deviation unknown and  2  2.
= 2
1

_ _
( x1 _ 1 + 1
x 2)  t/2 s p
n1 n2
Where:
(n 1 _ 2
1 ) 1 +( n _ 1 )s 2
s
2 2
sp =
n + n _ 2
2 1

t/2 = Critical value from the t-distribution for

the desired confidence level and degrees
of freedom equal to n 1 + n 2 - 2
 Confidence Interval Estimate for  1 -  2
standard deviation unknown and  2   2 .
2 1

_ _ s1
2 2
s2
( x1 _ x 2)  t/2 s p +
n1 n2

t/2 = Critical value from the t-distribution for

the desired confidence level and degrees
of freedom equal to ( s 2 n + s 2 n 2
1 / 1 2 / 2)

2 2
( s1 / n 1 s2 / n ) 2
2

n _1
1 n _12
Confidence Interval Estimate for  1 -  2 Large
sample nsizes
1
( nand
2
> 30.)

_ _ 2 2
( x1 _ x 2)  z/2 s p s1 s2
+
n1 n2

z/2 = critical value from the standard normal

distribution for the desired confidence level.
 Interval Estimation
- Insurance Crash Tests

Issue:
A automotive insurance company wants to
determine the difference in mean vehicle damage for
a import luxury car and a comparably priced U.S.
model
Objective:
Use Excel to help construct a confidence
interval estimate for the difference in population
mean damage time. Data file is Crash.xls
Interval Estimation – Insurance Crash Tests

Open File Crash.xls

File contains sample

data for 9 cars of each
type.
Interval Estimation – Insurance Crash Tests

Click on Tools – then Data

Analysis – then Descriptive
Statistics
 Interval Estimation – Insurance Crash Tests

Define Input Range,

Output Location
and check
Summary Statistics
option.
Interval Estimation – Insurance Crash Tests

The Standard Deviation or

Sample Variance values
are used to test the
hypothesis of equal
variances.
Interval Estimation – Insurance Crash Tests

Substituting the Mean

values and the
Standard Deviation
values into Equation
9-6 gives the
Confidence Interval
$563.94 + $894.40
Please answer the QUICK LUBE CASE

Click Here Please