interval

iormuntidenu two
between populations

How to compare the means of two

Statistical Analysis independent populations.

for Business How to compare the means of two

related populations.
Two-Sample
Decisions Tests How to compare the proportions of
two independent populations.
• Class 14
• Instructor: Prof. Yi-Shan Lee
• TA: Ms. Sandy Lee How to compare the variances of
two independent populations.

Reference: “Statistics for Managers Using Microsoft Excel” 8th edition, Pearson, 2017

Two-Sample Tests
Difference Between Two Means
Two-Sample Tests

• Goal: Test hypothesis or form a confidence interval for the

Population Population difference between two population means, μ1 – μ2.
Means, Means, Population Population
Independent Related Proportions Variances i
• The point estimate for the difference is: !! − !"
categorical.ariane
Samples Samples
Examples:
Group 1 vs. Same group Proportion 1 vs. Variance 1 vs.

th 責
Group 2 before vs. after Proportion 2 Variance 2

i
treatment
HM
and
CUHK
mnus.mn 不胂 iz 不不
符it

Difference Between Two Meansoo Hypothesis Tests for

Independent Samples Two Population Means
• Different data sources Two Population Means, Independent Samples
Population means, • Independent
independent samples • Sample selected from one
population has no effect on the Lower-tail test: Upper-tail test: Two-tail test:
sample selected from the other
population. H0: μ1 ³ μ2 H0: μ1 ≤ μ2 H0: μ1 = μ2
or H1: μ1 < μ2 H1: μ1 > μ2 H1: μ1 ≠ μ2
• Use Sp to estimate unknown σ. 23
σ1 and σ2 unknown, i.e., i.e., i.e.,
assumed equal • Use a Pooled-Variance t test.
H0: μ1 – μ2 ³ 0 H0: μ1 – μ2 ≤ 0 H0: μ1 – μ2 = 0
H1: μ1 – μ2 < 0 H1: μ1 – μ2 > 0 H1: μ1 – μ2 ≠ 0
σ1 and σ2 unknown, i Use S1 and S2 to estimate σ1 and σ2.
not assumed equal Use a Separate-variance t test
 and
related before a fter
sharedemograpnamioriies

Hypothesis tests for μ1 – μ2 Hypothesis Tests for µ1 - µ2

Two Population Means, Independent Samples σ1 and σ2 Unknown and Assumed Equal
Lower-tail test: Upper-tail test: Two-tail test:
H0: μ1 – μ2 ³ 0 H0: μ1 – μ2 ≤ 0 H0: μ1 – μ2 = 0 • Assumptions
H1: μ1 – μ2 < 0 H1: μ1 – μ2 > 0 H1: μ1 – μ2 ≠ 0 • Samples are randomly and independently drawn.
• Populations are normally distributed or both sample sizes
a a a/2 a/2 torttat
are at least 30. standardassumptions
• Population variances are unknown but assumed equal.
• Use pooled-variance t test.
-ta ta -ta/2 ta/2
m H
Reject if tSTAT < -ta Reject H0 if tSTAT > ta Reject H0 if tSTAT < -ta/2
0
or tSTAT > ta/2

Hypothesis Tests for µ1 - µ2 Hypothesis Tests for µ1 - µ2

σ1 and σ2 Unknown and Assumed Equal σ1 and σ2 Unknown and Assumed Equal
• The pooled variance is:
dtt
Nmnesamnennn by
correspondingdiotcnnnnngsnmne
nnnenimauot
(n
Sp2 = 1
- 1)S 1
2
+ (n 2 - 1)S 2
2
• The confidence interval for μ1 – μ2 is:
_sumotd the
under
populations ineuananuummon.mn
anmn (n1 - 1) + (n2 - 1)

( X1 - X 2 ) ± tα/2 æ1 ö
tnutwopornnmnane
1
• Theetest
n statistic is:
are Hypothesised
S 2p çç + ÷
Point estimate difference
n n ÷
è 1 ø
( X1 - X 2 ) - ( μ 1 - μ 2 ) 2
t STAT = in
æ1
S 2p çç +
1
è n1 n 2
ö
÷
÷
ø in 譽 品
• tα/2 has d.f. = n1 + n2 – 2
• tSTAT has d.f. = n1 + n2 – 2 y.mn
in spt.tn
灬灬
cnn.in
H

_change a

enǐiiiiio
on_no
cv.tx.d.it
t.ir 6802
t.neje.tn
Pooled-Variance t Test Example Pooled-Variance t Test Example:
• You are a financial analyst for a brokerage firm. Is there a difference in
Confidence Interval for µ1 - µ2
dividend yield between stocks listed on the NYSE & NASDAQ? You collect
the following data: Since we rejected H0 can we be 95% confident that µNYSE > µNASDAQ?
NYSE NASDAQ
Sample size 21 25 95% Confidence Interval for µNYSE - µNASDAQ:
Sample mean
Sample std dev
n.s.sz
n3.27
1.30
2.53
1.16
si 偷偷
• Assuming both populations are approximately normal with equal
variances, is there a difference in mean dividend yield (a = 0.05)?
u nn n.in ii i
cnun54o
mmz
Hi u Ho ic 灬

ijiii
mmto
⼆
5以44
cnn.name
叫 Since 0 is less than the entire interval, we can be 95% confident that
44 哈
灬 灬 灬
nnn2 µNYSE > µNASDAQ.
in
in n tnnn
mii.in in54
no
4numparina

国 不不2
snninent
evidence
that is
thereadinerenuinmeundnnentyien
neienno
戀
Hypothesis Tests for µ1 - µ2
σ1 and σ2 Unknown and Not Assumed Equal
In this class we discussed:
• Comparing the means of two
Assumptions
• Samples are randomly and independently drawn. Class independent populations
• Point estimation
• Populations are normally distributed or both sample sizes
are at least 30.
Summary • Interval estimation
• Hypothesis testing
• Population variances are unknown and cannot be assumed
to be equal.
-> Use the separate-variance t test (not covered in this
course)
 is butenq
populationvariance unknown

apopulationisnorman distributionorsample
ns.so

pooktvananietten

不上了5，5 51⼆ 3
⼜2 33 Szi 4

SF 24Ǘtmx42

35.5 33
ttq
12.5 新店

215

jyreiea
venue
d.fi48 驠
 ǐorf
theAnytest
edsto
⼝ ⼀
些

pvalue Yiiiii

T r

ii
li.in It 2
必
前坊以
⼆ 215I2.6822 ⼈與