Assignment-3-solutions(Chs 7, 8 and 9) Due by Midnight of Sunday, June 23rd, 2013: (Use dropbox 3): Total 125 points True/False

(two points each) Chapter 7 1. A sample size of 1000 is not large enough to conclude that the sampling distribution of p is a normal distribution, if the estimate of the population proportion is .996. TRUE Here, n(1-p) = 1000*0.004 = 4 which is less than 10 or even 5 (the liberal rule suggested in the book) 2. The standard deviation of all possible sample proportions decreases as the sample size decreases. FALSE 3. If the population is normally distributed then the sample mean may or may not be normally distributed for small sample size. FALSE If the population is normally distributed then the sample mean is also normally distributed even for small sample size.(Instructions on Ch 7, property 4)

Chapter 8
4. First a confidence interval is constructed without using the finite population correction factor. Then, for the same identical data, a confidence interval is constructed using the finite population correction factor. The width of the interval without the finite population correction factor is narrower than the confidence interval with the finite population correction factor. FALSE 5. When the population is normally distributed and the population standard deviation is unknown, then for any sample size n, the sampling distribution of is based on the t distribution. TRUE 6. When the level of confidence and sample standard deviation remain the same, a confidence interval for a population mean based on a sample of n=100 will be narrower than a confidence interval for a population mean based on a sample of n= 150. FALSE

7. When the level of confidence and the sample size remain the same, a confidence interval for a population mean will be wider, when the sample standard deviation s is large than when s is small. TRUE 8. When the level of confidence and sample proportion p remain the same, a confidence interval for a population proportion p based on a sample of n=100 will be wider than a confidence interval for p based on a sample of n=400. TRUE 9. The sample mean, the sample proportion and the sample standard deviation are all unbiased estimators of the corresponding population parameters. FALSE The Sample Standard Deviation is not an unbiased estimator (Clearly stated in Instructions on Ch8 page6) Chapter 9 10. The manager of the quality department for a tire manufacturing company wants to know the average tensile strength of rubber used in making a certain brand of radial tire. The population is normally distributed and the population standard deviation is known. She uses a Z test to test the null hypothesis that the mean tensile strength is 800 pounds per square inch. The calculated Z test statistic is a positive value that leads to a p-value of .045 for the test. If the significance level()is .01, the null hypothesis would be rejected. FALSE Reject only when pvalue is less than the alpha value or level of significance, because it is the probability of rejecting a true Null or of Type I error. (book page 357) 11. You cannot make a Type I error when the null hypothesis is true. FALSE Type I error is rejecting a True Null. 12. The power of a statistical test is the probability of rejecting the null hypothesis when it is false. TRUE 13. The level of significance indicates the probability of rejecting a true null hypothesis. TRUE The level of significance is the probability of type I error 14. When conducting a hypothesis test about a single mean, other relevant factors held constant, increasing the level of significance from .05 to .10 will increase the probability of a Type II error. FALSE increasing Type I error reduces the size of Type II error other things remaining the same.
2

Multiple Choices (3.5 points each) Chapter 7 1. A manufacturing company measures the weight of boxes before shipping them to the customers. If the box weights have a population mean and standard deviation of 90 lbs. and 24 lbs. respectively, then based on a sample size of 36 boxes, the probability that the average weight of the boxes will be greater than 94 lbs. is: A. 34.13% B. 84.13% C. 15.87% D. 56.36% E. 16.87% The = 24/36 = 4. Therefore the Z score is (94-90)/4 = 1.0 P(Z 1.0) = 1- P(Z 1) = 1- 0.8413 2. The internal auditing staff of a local manufacturing company performs a sample audit each quarter to estimate the proportion of accounts that are delinquent more than 90 days overdue. The historical records of the company show that over the past 8 years 13 percent of the accounts are delinquent. For this quarter, the auditing staff randomly selected 250 customer accounts. What is the probability that less than 40 accounts will be classified as delinquent? A. 42.07% B. 92.07% C. 7.93% D. 40.15% E. 90.15% Here = 0.13, n = 250, and 1- = 0.87. We have n = 32.5 and n(1- ) = 217.5, both greater than 10, and n (1- ) = 28.275 > 10. So, normal approximation without continuity correction is appropriate. The standard error of p = p = = 0.0213

Next, 40 accounts out of 250 in proportion is 40/250 = 0.16. The question refers to less than 40. Therefore, the question is P(p 0.16)?

Using the standardization process with p = 0.13 (the population proportion) and p = 0.0213, we have P(Z table or MegaStat. 3. If we have a sample size of 100 and the estimate of the population proportion is 0.10, the mean of the sampling distribution of the sample proportion is: A. 0.0009 B. 0.09 C. 0.03 D. 0.90 E. 0.10 Chapter 8 4. The width of a confidence interval will be: A. Wider for 90% confidence than 95% confidence B. Wider for a sample size of 100 than for a sample size of 50 C. Narrower for 95% confidence than 99% confidence D. Wider when the sample standard deviation (s) is small than when s is large 5. When the level of confidence and sample size remain the same, a confidence interval for a population proportion p will be __________ when p(1-p) is smaller than when p(1-p) is larger. A. Narrower B. Wider C. Neither A nor B, they will be the same D. Cannot tell from the information given 6. A confidence interval increases in width as A. The level of confidence increases B. n decreases C. s increases D. All of the above E. None of the above ) = P(Z 1.41) = 0.9207 from the

7. In a manufacturing process a random sample of 9 bolts manufactured has a mean length of 3 inches with a standard deviation of .3 inches. What is the 95% confidence interval for the true mean length of the bolt? A. 2.804 to 3.196 B. 2.308 to 3.692 C. 2.769 to 3.231 D. 2.412 to 3.588 E. 2.814 to 3.186 Since the population standard deviation is not given (only sample std. dev given) and the sample is small, t distribution is appropriate. Using MegaStat (you could use t Table too)

Confidence interval - mean 95% confidence level 3 mean 0.3 std. dev. 9 n 2.306 t (df = 8) 0.231 half-width 3.231 upper confidence limit 2.769 lower confidence limit

8. The internal auditing staff of a local manufacturing company performs a sample audit each quarter to estimate the proportion of accounts that are delinquent (more than 90 days overdue). For this quarter, the auditing staff randomly selected 400 customer accounts and found that 80 of these accounts were delinquent. What is the 95% confidence interval for the proportion of all delinquent customer accounts at this manufacturing company? A. .1608 to .2392 B. .1992 to .2008 C. .1671 to .2329 D. .1485 to .2515 E. .1714 to .2286

Confidence interval - proportion 95% confidence level 0.2 proportion 400 n 1.960 z 0.039 half-width 0.239 upper confidence limit 0.161 lower confidence limit

9. The internal auditing staff of a local manufacturing company performs a sample audit each quarter to estimate the proportion of accounts that are current (between 0 and 60 days after billing). The historical records show that over the past 8 years 70 percent of the accounts have been current. Determine the sample size needed in order to be 99% confident that the sample proportion of the current customer accounts is within .03 of the true proportion of all current accounts for this company. A. 1842 B. 1549 C. 897 D. 632 E. 1267 You can use the formula given in my instructions or book or MegaStat. We always round up the required sample size to the whole number because the formula gives the minimum.

Sample size - proportion 0.03 E, error tolerance 0.7 estimated population proportion 99% confidence level 2.576 z 1548.143 sample size 1549 rounded up

10. A company is interested in estimating the mean number of days of sick leave taken by its employees. The firm's statistician randomly selects 100 personnel files and notes the number of sick days taken by each employee. The sample mean is 12.2 days and the sample standard deviation is 10 days. How many personnel files would the director have to select in order to estimate to within 2 days with a 99% confidence interval? A. 2 B. 13 C. 136 D. 165 E. 166
Sample size - mean 2 E, error tolerance 10 standard deviation 99% confidence level 2.576 z 165.872 sample size 166 rounded up

Chapter 9 11. . If a null hypothesis is rejected at a significance level of .01, it will ______ be rejected at a significance level of .05 A. Always B. Sometimes C. Never
If the calculated test statistic exceeds the 1% level critical value it will definitely exceed the 5% level critical value.

12. If a null hypothesis is not rejected at a significance level of .05, it will ______ be rejected at a significance level of .01 A. Always

B. Sometimes C. Never Because the critical value 1% is even larger. 13. When carrying out a large sample test of H0: =10 vs. Ha: 10 by using a pvalue, we reject H0 at level of significance when the p-value is: A. Greater than /2 B. Greater than C. Less than D. Less than /2 E. Less than Z p is always compared with alpha whether one-tailed or two-tailed. See book page 361 rule in box 14. If you live in California, the decision to buy earthquake insurance is an important one. A survey revealed that only 133 of 337 randomly selected residences in one California county were protected by earthquake insurance. Calculate the appropriate test statistic to test the hypotheses that at least 40% buy the insurance. (rounded) A. 0.20 B. 0.39 C. -0.13 D. -0.20 E. -0.39 Ho: 0 and H1: < 0 one-tailed p= 133/337 = 0.3947 0 = 0.40 from Instruction on Ch 9, page 12, sp = { 0(1- 0)/n}= (.4*.6)/337 = .0267 calculated Z = (.3947- .40)/.0267 = -0.20 (rounded to two decimals) Essay Type (show your work) (Six points each) Chapter 7 1. A PGA (Professional Golf Association) tournament organizer is attempting to determine whether hole (pin) placement has a significant impact on the average number of strokes for the 13th hole on a given golf course. Historically, the pin has been placed in the front right corner of the green and the historical mean number of strokes for the hole has been 4.25, with a standard deviation of 1.5 strokes. On a
9

particular day during the most recent golf tournament, the organizer placed the hole (pin) in the back left corner of the green. 36 golfers played the hole with the new placement on that day. Determine the probability of the sample average number
of strokes exceeding 4.60. Here = 4.25, = 1.5, and n= 36. So, Z = (4.60-4.25)/0.25 = 1.4 The right hand side area or probability is required because the question is ..exceeding 4.60. So we will subtract from 1 the probability of the calculated Z to the left (given in the table or by MegaStat). P( 4.60) = P(Z 1.4) = 1- P(Z 1.4) = 1- 0.9192 = 0.0808 or 8.08% chance.

= 1.5/36 = 0.25. Therefore,

2. The population of lengths of aluminum-coated steel sheets is normally distributed with a mean of 30.05 inches and a standard deviation of 0.32 inches. A sample of 16 metal sheets is randomly selected from a batch. What is the probability that the average length of a sheet is between 30.19 and 30.27 inches long?
The Standard deviation of the sample mean = 0.32/16 = 0.08. Therefore, P{(30.1930.05)/0.08 Z (30.2730.05)/0.08} = P(1.75 Z 2.75) = 0.9970 -0.9599 = 0.0371 or 3.71%.

Chapter 8 3. A small town has a population of 20,000 people. Among these 1,000 regularly visit a popular local bar. A sample of 100 people who visit the bar is surveyed for their annual expenditures in the bar. It is found that on average each person who regularly visits the bar spends about $2400 per year in the bar with a standard deviation of $200. Construct a 99 percent confidence interval around the mean annual expenditure in the bar.
This problem needs finite population correction because the population from which sample is taken is only 1000 not 20,000 (that number was given to trick you) and is Not at least 20 times the sample of 100. The finite population correction factor (given in my Instructions) is (N-n)/(N-1) or 900/999 in this case or 0.901. Its square root is 0.949. So,

= (200/10)*0.949 = 18.98; For 99% confidence,

2.576*18.98 = 48.89. Therefore, the interval is 2400 48.89 or between 2351.11 and 2448.89
10

4. The production manager for the XYZ manufacturing company is concerned that the customer orders are being shipped late. He asked one of his planners to check the timeliness of shipments for 1000 orders. The planner randomly selected 1000 orders and found that 150 orders were shipped late. Construct the 95% confidence interval for the proportion of orders shipped late. (150/1000) 1.96 [(0.15*0.85)/1000] = 0.1279 to 0.1721 5. The weight of a product is measured in pounds. A sample of 50 units is taken from a batch. The sample yielded the following results: = 80 lbs. and s=10 lbs. Calculate a 99 percent confidence interval for the mean.
The t-value for 99% confidence interval with df 49 is 2.68. Therefore, the confidence interval is 80 2.68(10/50) = 76.21 to 83.79. However, if a student followed the practical rule and calculated using Z value as follows: 80 2.576(10/50) = 76.35 to 83.64. I accepted it. You can see that the difference is very small considering the large integral values.

Chapter 9 6. The average waiting time per customer at a fast food restaurant has been 7.5 minutes. The customer waiting time has a normal distribution. The manager claims that the use of a new cashier system will decrease the average customer waiting time in the store. Based on a random sample of 16 customer transactions the mean waiting time is 6.25 minutes and the standard deviation is 2 minutes per customer. Test the managers claim at 5% and 1% significance level tests.
Here n = 16, hypothesized is 7.5, sample mean = 6.25 and sample std dev s = 2. The standard error of is s/16 = 0.5. Since population is unknown and sample is small we use ttest. The calculated t-value is (6.25- 7.5/.5) = -2.5. The test is left-tailed as shown below. Ho: 7.5 (managers claim is wrong) H1 : < 7.5 (managers claim is valid) The 5% and 1% t-values for 15 df one-tailed are 1.753 and 2.602. Since the absolute value of calculated t is between these critical values, we reject the Null at 5% but not at 1%. Also using MegaStat:

11

Hypothesis Test: Mean vs. Hypothesized Value

7.5000 6.2500 2.0000 0.5000 16 15 -2.50 .0123 hypothesized value mean input std. dev. std. error n df t p-value (one-tailed, lower)

The p-value shows that the Null hypotheses will be rejected at 5% but not at 1%. The managers claim can have 95 % confidence. Students who used Z-distribution got the conclusion wrong for 1% level and lost 2 points for using the wrong distribution.

7. In an early study, researchers at an Ivy University found that 33% of the freshmen had received at least one A in their first semester. Administrators are concerned that grade inflation has caused this percentage to increase. In a more recent study, of a random sample of 500 freshmen, 175 had at least one A in their first semester Calculate the appropriate test statistic to test the hypotheses related to the concern and test at 5% and 1%. Ho: 0.33 ( No Grade Inflation) H1: > 0.33 (Grade Inflation) The sample proportion p = 175/500 =.35 and hypothesized = 0.33 (from earlier research). Std error of p or p = (.33)(.67)/500 = 0.021. Therefore, Z= (.35.33)/.021= 0.9524. One-tailed critical Z for 5% is 1.645 and for 1% is 2.33. So, cannot reject the Null at even at 5% level.
Hypothesis test for proportion vs hypothesized value
Observed 0.35 175/500 175. 500 Hypothesized 0.33 165/500 165. 500 0.021 0.95 .1708 p (as decimal) p (as fraction) X n std. error z p-value (one-tailed, upper)

12

8. A microwave manufacturing company has just switched to a new automated production system. Unfortunately, the new machinery has been frequently failing and requiring repairs and service. The company has been able to provide its customers with a completion time of 6 days or less. To analyze whether the completion time has increased, the production manager took a sample of 36 jobs and found that the sample mean completion time was 6.45 days with a sample standard deviation of 1.5 days. At significance levels of 5% and 10%, test whether the completion time has increased. Indicate which test you are performing; show the hypotheses, the test statistic and the critical values and mention whether onetailed or two-tailed. H0: 6 and H1: > 6 t = (6.45 6)/(1.5/36) = 1.8 The critical t-values for 35 df and one-tailed 10% and 5% are: 1.306 and 1.690. The calculated value exceeds both. So the Null hypothesis is rejected at both significance levels. Based on the evidence of this sample, the completion time has increased.
Some of you have done two-tailed test instead of one-tailed which is obviously wrong. These are serious conceptual errors, which in real life applications may lead to very wrong conclusions. I took only two points for this mistake. Some of you have used Z-test perhaps following the so called practical rule given in the book to use Z-test for 30 or more observations. In my Instructions I discussed this and recommended using t-test regardless of sample size if population variance is unknown. However, I ignored this part in this question because it has some basis in the practical rule. Nevertheless, I still recommend that t-test should be used whenever population variance is not known, regardless of sample size (unless it is extremely large such as 200 or 300, where the difference between Z and t is quite negligible). Finding tvalues used to be a big deal long time ago, but not anymore. So why go for approximation unnecessarily? Some of you still need to learn the simple rule for rejection which I asked in many forms in the Assignments and gave the answer too. It is so simple. If p-value is less than the significance level Reject the Null, otherwise do not reject.

13