STAT217

Worksheet #2 Solutions

 

Question 1

a)      The first thing we have to do is figure out the rejection region in terms of xbar. We reject the null hypothesis if Z > 1.645 or if xbar > (1.645)(5)/sqrt(64) + 25 = 26.028125.

Power = P(reject Ho | Ha true)
Power = P(xbar > 26.028125 | m = 26)
Power = P(Z > (26.028125 - 26)/(5/sqrt(64))) = P(Z > 0.045)

If we use the Z table, we round 0.045 to 0.05.

P(Z > 0.05) = 0.5 – 0.0199 = 0.4801

b)      If the power is 90%, this means the Z value used to calculate the power is –1.282 since P(Z > -1.282) = 0.9. From part a, we reject Ho if xbar > 26.021825.

-1.282 = (26.021825 - m)/(5/sqrt(64))

m = 26.021825 + 1.282(5)/sqrt(64) = 26.823075 which we round to 28.6231.

c)      If the power is at least 95%, we need Z < -1.645 since P(Z > -1.645) = 0.95.

That means (xbar - 26)/(5/sqrt(n)) < -1.645 or xbar < 26 - 8.225/sqrt(n). Note that we get 8.225 by multiplying 1.645 by 5.
But, we also have to take into consideration that the level of significance is 5%. We reject the null hypothesis if Z > 1.645 or xbar > (8.225/sqrt(n)) + 25
This means that 25 + 8.225/sqrt(n) < xbar < 26 - 8.225/sqrt(n). Since it is n that we are interested in solving for, we want to solve 25 + 8.225/sqrt(n) < 26 - 8.225/sqrt(n). Solving for n, n > 271.

 

Question 2

a)      Ho: m Ł 1200

Ha: m > 1200

Reject Ho if p-value < 1%. Do not reject Ho if p-value > 10%.

Z = (1250 – 1200)/(135/sqrt(50)) = 2.62

P-value = P(Z > 2.62) = 0.5 – 0.4956 = 0.0044.

P-value < 1%

Reject Ho

Conclude coffee shops spend significantly more than $1200 per week on average in buying whole beans.

b)      Reject Ho if Z > 2.326

Reject Ho if (xbar – 1200)/(135/sqrt(50)) > 2.326

Reject Ho if xbar > 1200 + 2.326(135)/sqrt(50)

Reject Ho if xbar > 1244.4077

Power = P(reject Ho | Ha true)

Power = P(xbar > 1244.4077 | m = 1250)

Power = P(Z > (1244.4077 – 1250)/ (135/sqrt(50))

Power = P(Z > -0.29) = 0.5 + 0.1141 = 0.6141

c)      Reject Ho if Z > 1.282

Reject Ho if (xbar – 1200)/(135/sqrt(50)) > 1.282

Reject Ho if xbar > 1200 + 1.282(135)/sqrt(50)

Reject Ho if xbar > 1224.4758

Power = P(reject Ho | Ha true)

Power = P(xbar > 1224.4758 | m = 1250)

Power = P(Z > (1224.4758 – 1250)/ (135/sqrt(50))

Power = P(Z > -1.34) = 0.5 + 0.4099 = 0.9099

The power increases from 61.41% to 90.99% which is a difference of 29.58%.

 

Question 3

a)      Ho: p > 0.02

Ha: p < 0.02
Reject Ho if Z < -1.645
Z = (0.012 - 0.02)/sqrt(0.02(0.98)/500) = -1.28
Do not reject Ho.

Conclude that the percentage of people who watch the show is not significantly below 2%.

b)      P-value = P(Z < -1.28) = 0.5 – 0.3997 = 0.1003. Since the p-value is marginally above 10%, this provides strong support for Ho. Thus, we do not reject Ho.

c)      Reject Ho if Z < -1.645

Reject Ho if (phat – 0.02)/ sqrt(0.02(0.98)/500) < -1.645

Reject Ho if phat < 0.02 - 1.645sqrt(0.02(0.98)/500)

Reject Ho if phat < 0.0097
Power = P(phat < 0.0097 | p = 0.012)
= P(Z < (0.0097 - 0.012)/sqrt(0.012(0.988)/500))
= P(Z < -0.47) = 0.5 – 0.1808 = 0.3192

d)     Reject Ho if Z < -1.282

Reject Ho if (phat – 0.02)/ sqrt(0.02(0.98)/500) < -1.282

Reject Ho if phat < 0.02 - 1.282sqrt(0.02(0.98)/500)

Reject Ho if phat < 0.012
Power = P(phat < 0.012 | p = 0.012)
= P(Z < (0.012 - 0.012)/sqrt(0.012(0.988)/500))
= P(Z < 0) = 0.5

e)      If the power is 90%, we need Z = 1.282 since P(Z < 1.282) = 0.9

We reject Ho if phat < 0.012 from part d.

(0.012 – p)/sqrt(p(1-p)/500) = 1.282

0.012 – p = 1.282 sqrt(p(1-p)/500)

Squaring both sides:

0.000144 – 0.024p + p2 = 1.643524p(1-p)/500

Multiplying both sides by 500:

0.072 – 12p + 500p2 = 1.643524p – 1.643524p2

Bringing everything over to the left side:

501.643524p2 – 13.643524p + 0.072 = 0

Solving for the lower root using the quadratic equation, we find p = 0.007165 = 0.7165%. Note that we cannot use the upper root since the left side of the initial equation would be less than zero.

 

Question 4

a)      Ho: m = 25

Ha: m š 25

Reject Ho if t < -2.365 or t > 2.365 (7 degrees of freedom)

t = (20.25 – 25)/(4.7132/sqrt(8)) = -2.8505

Reject Ho

Conclude that the average number of hours per week that semi-retired people work is

signficantly different than 25 hours.

b)      P-value = 2P(t > 2.8505) based on 7 degrees of freedom. The 2.5% critical value is 2.365 and the 1% critical value is 2.998. Therefore, P(t > 2.8505) is between 1% and 2.5% leading to the p-value being between 2% and 5%. Since the p-value is greater than 2%, it would be greater than a level of significance of 1.5% leading to not rejecting Ho. We would conclude that the average number of hours per week that semi-retired people work is not significantly from 25 hours.

c)      Lower limit = 20.25 – 2.365(4.7132)/sqrt(8) = 20.25 – 3.94 = 16.31

Upper limit = 20.25 + 3.94 = 24.19.

With 95% confidence, semi-retired people work between 16.3 and 24.2 hours per week. We would reject Ho at a 5% level of significance since the hypothesis mean of 25 hours falls outside the 95% confidence interval.

 

Question 5

Ho: median Ł 52

Ha: median > 52

There is one value of 52 in the data. We remove the value which decreases the sample size to 6.

Reject Ho if T- Ł 2.

We subtract the median of 52 from the remaining values:

Value

43

59

68

68

70

40

Median

52

52

52

52

52

52

d = Value - Median

-9

7

16

16

18

-12

We then compute the test statistic:

Observation

1

2

3

4

5

6

|d|

7

9

12

16

16

18

Sign

+

-

-

+

+

+

Rank

1

2

3

4.5

4.5

6

T- = 2 + 3 = 5

Do not reject Ho.

Conclude the median is not significantly greater than 52 and that the new teaching method does not significantly increase the grade on basic arithmetic.

 

Question 6

a)      Ho: s = 1200

Ha: s š 1200

Reject Ho if C2 < 6.262 or C2 > 27.488

C2 = 15(1050)2/12002 = 11.4844

Do not reject Ho.

Conclude the wage spread of small factories is not significantly different than that of large factories.

b)      Lower limit of s2 = 15(1050)2/27.4884 = 601,617.4095

Upper limit of s2 = 15(1050)2/6.2621 = 2,640,887.242

We compute the square roots of the two limits to find the confidence interval limits for s.

Lower limit of s = sqrt(601,617.4095) = 775.64

Upper limit of s = sqrt(2,640,887.242) = 1,625.08

If we wanted to interpret this interval, we would say that with 95% confidence, the standard deviation of the monthly wages of small factories ranges from $775.64 to $1,625.08. If we used this confidence interval to test the hypothesis in part a, we would not reject Ho at a 5% level of significance since the hypothesis standard deviation of $1200 falls inside the 95% confidence interval.

c)      Since the sample standard deviation is less than the hypothesis standard deviation, the p-value 2P(C2 < 11.4844) based on 15 degrees of freedom.  The 90% critical value is 8.547. Since P(X2 > 8.547) = 90%, this means P(X2 < 8.547) = 10%. Thus, P(C2 < 11.4844) is greater than 10% leading to the p-value being greater than 20%. Since the p-value is greater than 10%, under the general rule of thumb, this provides strong support for Ho and we do not reject Ho.

 

From the textbook

Section 7-3

Question 27 (page 397)

Reject Ho if (xbar – 25)/(2.4/sqrt(100)) > 1.645 or if xbar > 25.3948

Power = P(xbar > 25.3948 | ľ = 25.2) = P(Z > (25.3948 – 25.2)/(2.4/sqrt(100))) = P(Z > 0.81) = 0.5 – 0.291 = 0.209.

 

Question 28

If the alternative mean is 25.5, Power = P(Z > (25.3948 – 25.5)/(2.4/sqrt(100))) = P(Z > -0.44) = 0.5 + 0.17 = 0.67

If we increase the level of significance to 10%, then reject Ho if (xbar – 25)/(2.4/sqrt(100)) > 1.282 or if xbar > 25.30768

Then power = P(xbar > 25.30768 | ľ = 25.2) = P(Z > (25.30768 – 25.2)/(2.4/sqrt(100))) = P(Z > 0.45) = 0.5 – 0.1736 = 0.3264. The first solution increases the power more.

 

Question 29

For the power to be 95%, the Z value used in the power calculations is -1.645 = (25.3948 – ľA)/(2.4/sqrt(100)) from which we get ľA = 25.3948 + 1.645(2.4)/sqrt(100) = 25.7896.

 

Section 7-4

Question 22 (page 408)

Ho: m = 3.39

Ha: m š 3.39

Reject Ho if t < -2.132 or t > 2.132

T = (3.675 – 3.39)/(0.6573/sqrt(16)) = 1.7344

Do not reject Ho and conclude the vitamin supplement does not appear to have an effect on birth weight.

P-value = 2P(t > 1.7344) based on 15 degrees of freedom. The 5% critical value is 1.753 and the 10% critical value is 1.341. Then, P(t > 1.7344) is between 5% and 10% leading to the p-value being between 10% and 20%. Since the p-value is greater than 10%, it is also greater than the 5% level of significance leading to not reject Ho.

Lower limit = 3.675 – 2.132(0.6573)/sqrt(16) = 3.675 – 0.3503 = 3.3247

Upper limit = 3.675 + 0.3503 = 4.0253

3.3247 < ľ < 4.0253

Since the hypothesis mean of 3.39 lies inside the 95% confidence interval, we do not reject Ho at a 5% level of significance.

 

Section 7-5

Question 17 (page 418)

Ho: p ≥ 0.5

Ha: p < 0.5

Reject Ho if Z < -1.282

Phat = 32/71 = 0.4507; Z = (0.4507 – 0.5)/sqrt(0.5(0.5)/71) = -0.83

Do not reject Ho. The results to not suggest that the nicotine patch therapy is not effective.

P-value = P(Z < -0.83) = 0.5 – 0.2967 = 0.2033 = 20.33%. Since the p-value is greater than the 10% level of significance, we do not reject Ho.

 

Section 7-6

Question 3 (page 426)

Ho: s = 43.7

Ha: s≠ 43.7

Reject Ho if c2 < 57.153 or c2 > 106.629

c2 = 80(52.32)/43.72 = 114.5857

Reject Ho and conclude the errors have a standard deviation that is significantly different than 43.7 feet. Since the new production method appears to have more variation than the old method, it appears to be worse.

P-value = 2P(c2 > 114.5857). The 1% critical value is 112.329 and the 0.1% critical value is 116.321. So, P(c2 > 114.5857) is between 0.1% and 1% leading to the p-value being between 0.2% and 2%. Since the p-value is less than 2%, it is less than the 5% level of significance leading to rejecting Ho.

Lower limit for s2 = 80(52.32)/106.629 = 2,052.1922

Upper limit for s2 = 80(52.32)/57.153 = 3,828.7264

Taking the square roots of the limits, we have 45.3 < s < 61.9. Since the hypothesis standard deviation of 43.7 falls outside the 95% confidence interval, we reject Ho at a 5% level of significance.

 

Section 12-3

Question 9 (page 697)

Ho: median ≤ 2.5

Ha: median > 2.5

Subtracting 1.77 from each value gives differences of -0.03, 0, 0.13, 0.03, 0.11, 0.1, 0.02, 0.04 and 0.12. Subtracting d=0 from the differences gives n=8. Reject Ho if T- ≤ 6.

Observation

1

2

3

4

5

6

7

8

|d|

0.02

0.03

0.03

0.04

0.1

0.11

0.12

0.13

Sign

+

-

+

+

+

+

+

+

Rank

1

2.5

2.5

4

5

6

7

8

T- = 2.5. Reject Ho and conclude the median weight is greater than 1.77 kg.