0% found this document useful (0 votes)
2 views9 pages

Practice 4

The document contains practice problems for a statistics course, focusing on hypothesis testing, power analysis, and sample size determination. It includes problems related to automobile acceleration times, geometric distributions, and Poisson distributions, providing detailed calculations and rejection regions for various statistical tests. Additionally, it explores the implications of significance levels and power in hypothesis testing scenarios.

Uploaded by

wedxwe
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
2 views9 pages

Practice 4

The document contains practice problems for a statistics course, focusing on hypothesis testing, power analysis, and sample size determination. It includes problems related to automobile acceleration times, geometric distributions, and Poisson distributions, providing detailed calculations and rejection regions for various statistical tests. Additionally, it explores the implications of significance levels and power in hypothesis testing scenarios.

Uploaded by

wedxwe
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

STAT 410

Practice Problems

1. An advertisement for a particular brand of automobile states that it accelerates from


0 to 60 mph in an average of 5.0 seconds. Makers of a competing automobile feel
that the true average number of seconds it takes to reach 60 mph from zero is above
5.0. Suppose the population standard deviation is believed to be 0.43 seconds. We
wish to test H 0 : µ ≤ 5.0 vs. H 1 : µ > 5.0.

a) Fifty automobiles were tested (each automobile was tested a single time). The sample
mean time to reach 60 mph from zero was 5.09 seconds. Find the p-value of this test.

b) For which values of the sample mean X should we reject H 0 , if a 5% level of


significance will be used, and 50 automobiles will be tested (each automobile to
be tested a single time)?

c) Find the power of this test if the true average time is 5.15 seconds, a 5% level of
significance will be used, and 50 automobiles will be tested (each automobile to
be tested a single time).

d) What is the minimum sample size required if we want to have the power of at least
0.90 at µ = 5.15 for the test with a 5% level of significance.

2. 4.5.4 ( 7th edition ) 5.5.4 ( 6th edition )

3. 4.5.9 ( 7th edition ) 5.5.9 ( 6th edition )

4. 4.5.12 ( 7th edition ) 5.5.12 ( 6th edition )


5. Let X 1 , X 2 , X 3 , X 4 be a random sample of size n=4 from a Geometric ( p )
distribution ( the number of independent trials until the first “success” ). That is,

P ( X 1 = k ) = ( 1 – p ) k – 1 p, k = 1, 2, 3, … .
We are interested in testing H 0 : p = 0.30 vs. H 1 : p = 0.40.

a) Use the likelihood ratio to find the best rejection region in terms of ∑ in=1 x i .

b) Find the best rejection region with the significance level closest to 0.07.

c) If the rejection region from part (b) is used, find the power of the test.

Hint: If X 1 , X 2 , … , X n are independent Geometric ( p ), then Y = ∑ in=1 X i


has Negative Binomial distribution ( the number of independent trials until
the n th “success” ). Then

 y −1 
P( Y = y ) =   ⋅ p n ⋅ ( 1 − p ) y − n , y = n, n + 1, n + 2, … .
 n −1 

EXCEL: = NEGBINOMDIST ( y – n , n , p ) gives P ( Y = y )


1. An advertisement for a particular brand of automobile states that it accelerates from
0 to 60 mph in an average of 5.0 seconds. Makers of a competing automobile feel
that the true average number of seconds it takes to reach 60 mph from zero is above
5.0. Suppose the population standard deviation is believed to be 0.43 seconds. We
wish to test H 0 : µ ≤ 5.0 vs. H 1 : µ > 5.0.

a) Fifty automobiles were tested (each automobile was tested a single time). The sample
mean time to reach 60 mph from zero was 5.09 seconds. Find the p-value of this test.

X − µ0 5.09 − 5.0
Test Statistic: σ is known. Z= = = 1.48.
σ 0.43
n 50

Right – tailed test. p-value = ( area of the right tail ) = P ( Z > 1.48 ) = 0.0694.

b) For which values of the sample mean X should we reject H 0 , if a 5% level of


significance will be used, and 50 automobiles will be tested (each automobile to
be tested a single time)?

σ = 0.43. n = 50. α = 0.05.


X − µ0 X − 5 .0
σ is known. Test Statistic: Z = = .
σ 0.43
n 50

Rejection Region: Right - tailed. Reject H 0 if Z > z α = 1.645.


0.43
X > 5.0 + 1.645 ⋅ = 5.1.
50

c) Find the power of this test if the true average time is 5.15 seconds, a 5% level of
significance will be used, and 50 automobiles will be tested (each automobile to
be tested a single time).

Power = P ( Reject H 0 | H 0 is NOT true ) = P ( X > 5.1 | µ = 5.15 )

 
 5.1 − 5.15 
= P Z >
0.43  = P ( Z > – 0.82 ) = 0.7939.
 
 50 
d) What is the minimum sample size required if we want to have the power of at least
0.90 at µ = 5.15 for the test with a 5% level of significance.

0.43
Rejection Region: Reject H 0 if X > 5.0 + 1.645 ⋅ .
n

0.43
Power ( 5.15 ) = P ( Reject H 0 | µ = 5.15 ) = P ( X > 5.0 + 1.645 ⋅ | µ = 5.15 )
n
 0.43 
 5.0 + 1.645 ⋅ − 5.15 
= P Z >
n  = P  Z > − 0.15 n + 1.645  .
 0.43   0.43 
   
n
 

P ( Z > – 1.282 ) = 0.90.

− 0.15 n − 0.15 n
⇒ + 1.645 ≤ −1.282 . ⇒ ≤ −2.927 .
0.43 0.43

⇒ n ≥ 8.390733 . ⇒ n ≥ 70.4044. n ≥ 71.

2. 4.5.4 ( 7th edition ) 5.5.4 ( 6th edition )

H 0 : p = 1/2 H 1 : p = 1/4

Significance level = α = P ( Reject H 0 | H 0 is true ) = P ( X ≤ 3 | p = 1 2 ) /


3 k 10 − k
1 1
= ∑ 10 k    
C = 0.171875.
k =0 2 2

Power = P ( Reject H 0 | H 0 is not true ) = P ( X ≤ 3 | p = 1 4 ) /


3 k 10 − k
1 3
= ∑ 10 k
C     = 0.775875.
k =0 4 4
3. 4.5.9 ( 7th edition ) 5.5.9 ( 6th edition )

H 0 : θ = 1/2 H 1 : θ < 1/2

Reject H 0 if X 1 + … + X 12 ≤ 2

X 1 + … + X 12 has Poisson distribution with mean 12 θ.

Power: γ ( θ ) = P ( Poisson ( 12 θ ) ≤ 2 ) = ∑
2 (12 θ ) k e −12 θ
k =0 k!

1
γ ( ) = P ( Poisson ( 6 ) ≤ 2 ) = 0.062 = significance level = α
2
1 1
γ ( ) = P ( Poisson ( 4 ) ≤ 2 ) = 0.238 γ ( ) = P ( Poisson ( 3 ) ≤ 2 ) = 0.423
3 4
1 1
γ ( ) = P ( Poisson ( 2 ) ≤ 2 ) = 0.677 γ( ) = P ( Poisson ( 1 ) ≤ 2 ) = 0.920
6 12
4. 4.5.12 ( 7th edition ) 5.5.12 ( 6th edition )

H 0 : µ = 0.5 H 1 : µ > 0.5

Reject H 0 if X 1 + … + X 8 ≥ 8

X 1 + … + X 8 has Poisson distribution with mean 8 µ.

a) significance level = α = P ( Reject H 0 | H 0 is true ) = P ( Poisson ( 4 ) ≥ 8 )

= 1 – P ( Poisson ( 4 ) ≤ 7 ) = 1 – 0.949 = 0.051

∞ (8 µ ) k e − 8 µ 7 (8 µ ) k e − 8 µ
b) γ ( µ ) = P ( Poisson ( 8 µ ) ≥ 8 ) = ∑ = 1− ∑ .
k =8 k! k =0 k!

c) γ ( 0.75 ) = P ( Poisson ( 6 ) ≥ 8 ) = 1 – 0.744 = 0.256


γ ( 1.00 ) = P ( Poisson ( 8 ) ≥ 8 ) = 1 – 0.453 = 0.547
γ ( 1.25 ) = P ( Poisson ( 10 ) ≥ 8 ) = 1 – 0.220 = 0.780
5. Let X 1 , X 2 , X 3 , X 4 be a random sample of size n=4 from a Geometric ( p )
distribution ( the number of independent trials until the first “success” ). That is,

P ( X 1 = k ) = ( 1 – p ) k – 1 p, k = 1, 2, 3, … .
We are interested in testing H 0 : p = 0.30 vs. H 1 : p = 0.40.

a) Use the likelihood ratio to find the best rejection region in terms of ∑ in=1 x i .
b) Find the best rejection region with the significance level closest to 0.07.

c) If the rejection region from part (b) is used, find the power of the test.

Hint: If X 1 , X 2 , … , X n are independent Geometric ( p ), then Y = ∑ in=1 X i


has Negative Binomial distribution ( the number of independent trials until
the n th “success” ). Then

 y −1 
P( Y = y ) =   ⋅ p n ⋅ ( 1 − p ) y − n , y = n, n + 1, n + 2, … .
 n −1 

EXCEL: = NEGBINOMDIST ( y – n , n , p ) gives P ( Y = y )

n
∏  0.70 ⋅ 0.30 
xi − 1
L ( p = 0.30 ; x 1 , x 2 ,..., x n ) i =1 
a) λ ( x 1 , x 2 ,..., x n ) = =
L ( p = 0.40 ; x 1 , x 2 ,..., x n ) n 
⋅ 0.40 
xi − 1
∏  0.60 
i =1
n n
∑ xi 0.60 ⋅ 0.30 
n ∑ xi n
 0.70  i =1
⋅   7 
⋅ 
18 
=   =   i =1  .
 0.60   0.70 ⋅ 0.40   6   28 

n  7   18 
λ ( x 1 , x 2 ,..., x n ) ≤ k ⇔ ∑ x i ⋅ ln  + n ⋅ ln  ≤ ln k
i =1  6   28 

n
⇔ ∑ x i ≤ c.
i =1
n
Rejection Region: Reject H 0 if ∑ X i ≤ c. ( left tail )
i =1

Intuition: If p is larger, the first “success” will tend to occur sooner.


b) Want 0.07 = α = P ( Reject H 0 | H 0 is true ) = P ( Y ≤ c | p = 0.30 ) = CDF @ c.

A B C D
1 y PMF @ y CDF @ y
2 4 =NEGBINOMDIST(A2–4,4,0.30) =B2
3 =A2+1 =NEGBINOMDIST(A3–4,4,0.30) =C2+B3
4 =A3+1 =NEGBINOMDIST(A4–4,4,0.30) =C3+B4
5 =A4+1 =NEGBINOMDIST(A5–4,4,0.30) =C4+B5
6 =A5+1 =NEGBINOMDIST(A6–4,4,0.30) =C5+B6
7 =A6+1 =NEGBINOMDIST(A7–4,4,0.30) =C6+B7
… … … …

A B C D
1 y PMF @ y CDF @ y
2 4 0.0081 0.0081
3 5 0.02268 0.03078
⇒ 4 6 0.03969 0.07047
5 7 0.055566 0.126036
6 8 0.068068 0.194104
7 9 0.076237 0.270341
… … … …

CDF @ 6 = P ( Y ≤ 6 | p = 0.30 ) ≈ 0.07.

OR

 4 − 1
P ( Y = 4 | p = 0.30 ) =   (0.30 ) 4 (0.70 ) 4 − 4 = 0.0081,
 4 − 1

 5 − 1
P ( Y = 5 | p = 0.30 ) =   (0.30 ) 4 (0.70 ) 5 − 4 = 0.02268,
 4 − 1

 6 − 1
P ( Y = 6 | p = 0.30 ) =   (0.30 ) 4 (0.70 ) 6 − 4 = 0.03969.
 4 − 1 

P ( Y ≤ 6 | p = 0.30 ) = 0.0081 + 0.02268 + 0.03969 = 0.07047 ≈ 0.07 = α.

n
Rejection Region: Reject H 0 if Y = ∑ X i ≤ 6.
i =1
c) Power = P ( Reject H 0 | H 0 is not true ) = P ( Y ≤ 6 | p = 0.40 ) = CDF @ 6.

A B C D
1 y PMF @ y CDF @ y
2 4 =NEGBINOMDIST(A2–4,4,0.40) =B2
3 =A2+1 =NEGBINOMDIST(A3–4,4,0.40) =C2+B3
4 =A3+1 =NEGBINOMDIST(A4–4,4,0.40) =C3+B4
5 =A4+1 =NEGBINOMDIST(A5–4,4,0.40) =C4+B5
6 =A5+1 =NEGBINOMDIST(A6–4,4,0.40) =C5+B6
7 =A6+1 =NEGBINOMDIST(A7–4,4,0.40) =C6+B7
… … … …

A B C D
1 y PMF @ y CDF @ y
2 4 0.0256 0.0256
3 5 0.06144 0.08704
⇒ 4 6 0.09216 0.1792
5 7 0.110592 0.289792
6 8 0.116122 0.405914
7 9 0.111477 0.51739
… … … …

Power = P ( Y ≤ 6 | p = 0.40 ) = CDF @ 6 = 0.1792.

OR

 4 − 1
P ( Y = 4 | p = 0.40 ) =   (0.40 ) 4 (0.60 ) 4 − 4 = 0.0256,
 4 − 1

 5 − 1
P ( Y = 5 | p = 0.40 ) =   (0.40 ) 4 (0.60 ) 5 − 4 = 0.06144,
 4 − 1 

 6 − 1
P ( Y = 6 | p = 0.40 ) =   (0.40 ) 4 (0.60 ) 6 − 4 = 0.09216.
 4 − 1

Power = P ( Y ≤ 6 | p = 0.40 ) = 0.0256 + 0.06144 + 0.09216 = 0.1792.

You might also like