Problems message: StataQuest will calculate many of these statistics.

Those using Quick's data files should load stata006 by choosing File and Open. Use the scroll arrow next to the Drives box to locate and double click on a:. Load by double clicking on stata006.

Others should load their page 6 data file.

All StataQuest users should choose Summaries. Choose Means and SDs, highlight CDSales, and choose OK. Study the result. Choose Summaries again. Choose Median/Percentiles, highlight CDSales, and choose OK. Study the result. Quick Notes Statistics has or will explore many of these statistics. StataQuest will not calculate an individual percentile. Compare StataQuest answers with those in Quick on pages PS 12-13 and PS 18-19. Where answers differ, it is because of rounding.

Practice Set 3 answers: 1A) 17 4) Median = 50th percentile = 16 7A) 14 7B) 21 7E) 29 Do other question 7 problems by hand.

Practice Set 4 answers: 1C) 30.6 1D) 5.5 Do other problems by hand.

Saving and Printing Results is easy. Saving output stored in a Stata Results window requires creating a log file by choosing log. Type stata006 in the File name box and choose OK. StataQuest will add the extension .log and stata006.log will be stored on the A drive. This file can be used by choosing Log, highlighting the desired log file, and choosing OK. Choose the default Append to existing file and choose OK. To print an open log file, choose File and Print Log file.

Chapters 5 and 6 problems are very similar to those of chapters 3 and 4. These StataQuest Practice Set Instructions and Answers will explore only the ungrouped practice sets of chapters 3 and 4.

Problem I message: These problems should be done by hand.

Problem II, III, and IV message: The StataQuest calculator will be used to solve these problems.

All StataQuest users should select Calculator, Statistical tables, and Binomial. Set No. of trials to 5, Prob. of success to .6, No. of successes to 3, and choose Run. The answer will appear in the Stata Results window. To determine the entire distribution repeat for successes of 0, 1, 2, 4, and 5.

Problem V message: The StataQuest calculator will be used to solve this problem.

All StataQuest users should select Calculator, Statistical tables, and Poisson. Set lambda (the mean) to 1, the count to 0, and choose Run. To determine the entire distribution repeat for counts of 1-7.

Problem VI message: The StataQuest calculator will be used to solve this problem.

All StataQuest users should select Calculator, Statistical tables, and Poisson. Set lambda (the mean) to 3 (1,500 x .002), the count to 0, and choose Run. Answer the next problem by setting the count to 2. See page PS55 of Quick for the answer to the last problem.

Problem I message: Calculate z by hand. Choose Calculator, Statistical Tables, and Normal. Insert z and choose Run. The probability to z and from z is given.

Problem II message: Determine the appropriate probability. Choose Calculator, Inverse Statistical Tables, and Normal. Insert the probability and choose Run. Use the resulting z to solve the problem.

Problem I message: You can find a confidence interval for the mean of a sampling distribution by using the actual data or the StataQuest calculator.

Those using Quick's data files should load stata068 by choosing File and Open. Use the scroll bar next to Drives to locate and click on a:. Double click on stata068.

Others must create a data file for the variable part size. You may want to review creating a data file on page 1 of Quick Start for StataQuest located at http://businessbookmall.com. Create a data file using 36 rows and 1 column. I named my variable Weight. To name a variable, double click on Var1. When saving my data file, I named my data file stata068 for the software being used and the problem page number.

All Stata users should select Summaries and Confidence intervals. Highlight Weights. Set the Confidence level to 99. Quick's answer differs because of rounding.

Note: This problem can also be done with the StataQuest calculator. Choose Calculator, 1-sample normal test. Fill in the appropriate data using .065 for the population standard deviation, 30.025 for the sample mean, and 30.0 for the Hypothesized mean. Choose Run.

Problem II message: Redo problem I with a 95% confidence interval.

Problem III message: Answer this problem by hand.

Problem I message: You can find a range for the proportion of a sampling distribution by using the actual data or the StataQuest calculator.

Those using Quick's data files should load stata072 by choosing File and Open. Use the scroll bar next to Drives to locate and click on a:. Double click on stata072.

Others must create a data file for page 72 data. Enter a 1 for samples that passed and a 0 for those that failed. To name a variable, double click on Var1. I names this variable Inspect.

All Stata users should select Summaries and Confidence intervals. Highlight Inspect, set the Confidence level to 95, and choose OK.

Note: This problem can also be done with the StataQuest calculator. Choose Calculator, 1-sample test of the proportion. Fill in the required data using .9 for the Prob. of success. Choose Normal approximation and Run. Quick's answer differs because of rounding.

Problem II, III, and IV message: These problems dealing with determining the appropriate sample size should be done by hand.

Problem I message: Data for this 1-sample test of a mean was first presented on page 68. StataQuest assumes a small sample and calls the calculated z for this test statistic a t value.
Ignore this for now. We study the small sample t distribution in chapter 16.

Those using Quick's data files should load stata068. Others should load their page 68 data file.

All StataQuest users should choose Statistics, Parametric tests, and 1-sample t test. Highlight weight and set the Hypoth. mean to 30 mg. Choose OK. The t-value of 2.32 is not beyond 2.33 (it is less than 2.33) and H₀ is accepted. Parts are not too heavy.

Note: I use the term beyond to indicate further from H₀, the hypothesized population parameter.

Problem II message: If a 1-tail test is accepted, a 2-tail test must be accepted. Here. 2.32 < 2.58.

Problem III message: Redo problem II with the new level of significance. Reject H₀ because 2.32 > 1.96. Parts are not too heavy.

Note: These problems can be done with the StataQuest calculator. Choose Calculator and 1-sample normal test. Fill in the required data using .065 for the population standard deviation. Choose Run.

Problem I message: This problem concerns the difference between 2 independent means.

Those using Quick's data files should load stata090.

Others should create a data file with 70 rows and 2 columns (variables). In the first data variable column entitled Days, enter the data for Supplier A and then the data for Supplier B. In the second group variable column entitled Supplier, enter a 0 next to Supplier A's data and a 1 next to supplier B's data.

All StataQuest users choose Statistics, Parametric tests, and 2-sample t test. Check 1 data variable, 1 group variable under Options, and choose OK. Highlight Days for the Data variable and Supplier for the Group variable. Check unequal variances. The test statistic t (z) of -1.32 is not beyond -1.96 (it is greater than -1.96), the critical value of t (z), accept H₀. Delivery times are the same.

Note: This problem can also be done with the StataQuest calculator. Choose Calculator and 2-sample normal test. Assume the sample standard deviation equals the population standard deviation.

Problem II message: StataQuest calculates z and p simultaneously. Redo the chapter 13 test for p.

Those using StataQuest's data files should load stata068.

Others should load their page 68 data file.

All StataQuest users should choose Statistics, Parametric tests, and 1-sample t test. Highlight Weight and set the Hypothesized mean to 30 mg. StataQuest gives a two-tail p-value using t of .0261. H₀ is accepted because .0261/2 = .013 > .01. Parts are not too heavy. Quick Notes calculates z = .0103 and also accepts H₀.

Problem III message: Answer this problem by hand. See page PS 91 for the answer.

Problem I message: This 1-sample test of the proportion.

StataQuest users should load stata096.

All StataQuest users should choose Statistics, Parametric tests, and 1-sample test of proportion. Check Normal approximation, highlight passed, and set Probability of success to .86 and choose OK. The p value of .4150 is for two-tails. P for one-tail is .2075. It is > .01 and H₀ is accepted. The proportion of parts passing inspection has not increased.

Problem II message: This is a 2-sample proportions test similar to those explored in chapter 14.

Those using Quick's data files should load stata96.

All StataQuest users should choose Statistics, Parametric tests, and 2-sample test of proportion. Check 1 data variable, 1 group variable under Options and choose OK. Highlight Defects for the Data variable and Shift for the Group variable. P of .03 is for two-tails. Because .03 > .01, accept H₀. Shift defects are the same.

Problem I message: This is an analysis of 2 independent means. It is similar to problems done earlier.

Those using Quick's data files should load stata100.

All StataQuest users should choose Statistics, Parametric tests, and 2-sample t test. Check 1 data variable, 1 group variable under Options, and choose OK. Highlight SickDays for the Data variable and Group for the Group variable. Choose Unequal variances. Choose OK. Reject H₀ because p = .0005 < .01. Sickdays are not the same.

Note: This problem can also be done with the StataQuest calculator with a 2-sample t test.

Problem II message: Here, a paired difference test is used to determine if 2 variables are dependent.

Those using Quick's data files should load stata100.

Others should add 2 variables to their page 100 data file, one for the efficiency rating before training and the other for the efficiency rating after training. Be sure to match each employee with their before and after efficiency rating.

All StataQuest users should choose Statistics, Parametric tests, and Paired t test. Highlight Before for Data variable #1 and After for Data variables #2. Choose OK. Reject H₀ because p = .0028/2 = .0014 < .01. Training increased efficiency.

Problem I message: This problem concerns the 30 milligram parts.

Those using Quick's data files should load stata104.

Others should create a data file with four variables. In the first column (variable) entitled Sample enter the numbers 1 to 12. Columns 2-4, which I labeled m1, m2, and m3, should contain the 3 cases for each sample.

All StataQuest users should choose Graphs and Quality control. Choose X-bar chart and highlight m1, m2, and m3. Choose Draw. For a range chart, repeat the above process choosing Range (R) chart. See page PS 104 of Quick Notes Statistics for the answer.

Problem II message: This problem is similar to problem I. See page PS 104 for the answer.

Those using Quick's data files should load stata105.

Others should create a data file with three variables. In the first column (variable) entitled Sample, enter the numbers 1 to 11. Column 2, entitled defects, should contain the number of defects for each sample. Column 3 should contain the sample size of 50.

All StataQuest users should choose Graphs, Quality Control, and Fraction defective (P) chart. Using the scroll bar, load Defects into Variable for number defective, Sample into Identification variable, and Tested into Sample-size variable. Choose Draw. See page PS 105 for the answer.

Problem I message: This problem concerns the difference between 2 variances.

Those using Quick's data files should load stata110.

Others should create a data file with 30 rows and 2 columns (variables). In the first data variable column entitled Variance, enter the data for sample 1 and then the data for sample 2. The second column is used for the group variable. I named it Sample and entered a 1 next to Sample one's data and a 2 next to sample two's data.

All StataQuest users should choose Statistics, Parametric tests, and 2-sample test of variance. Check 1 data variable, 1 group variable under Options, and choose OK. Highlight Variance as the Data variable and Sample as the Group variable. Accept H₀ because the test statistic of F = 1.17 < 3.82. Variance did not increase.

Note: This problem can also be done with the StataQuest calculator.

Problem II message: This is a one-way analysis of variance (ANOVA).

Those using Quick's data files should load stata111.

Others should create a 2 variables data set for the page 110 data. Label the data variable Weight and the group variable Dept.

All StataQuest users should select Statistics, ANOVA, and One-way. Check 1 data variable, 1 group variable under Options, and choose OK. Move Weight into the Data variable and Dept into Group variable. Choose OK. Reject H₀ because p = .0021 < .05. These populations have different means.

Problem I message: Do this nonparametric test using Chi-Square by hand. See page PS 122 for the answer.

Problem II message: This problem emphasizes the concept of statistical independence.

Those using Quick's data files should load stata123.

Others should create a 2 variable data file. The first variable entitled SResult contains 24 ones for making a sale to young people, 16 zeros for not making a sale to young people, 12 ones for making a sale to older people, and 8 zeros for not making a sale to older people. The second variable named age should contain 40 zeros for younger people and 20 ones for older people.

All StataQuest users should select Summaries, Tables, and Two-way. Copy SResult into Row variable, Age into Column variable, and choose OK. Chi square is zero and H₀ is accepted. Making a sale and age are independent. Because the buying percentage of each part of the population (60%) exactly equals the buying percentage of the entire population (60%), chi square is zero.

Problem I message: Do this run test for randomness by hand. See page PS 128 for the answer.

Problem II message: This is a sign test for the difference between two medians.

Those using Quick's data files should load stata128. Others should create a data file with 2 variables. The first should contain current median defects and the second should be last year's median of 5 defects.

All StataQuest users should select Statistics, Nonparametric tests, and Sign test. Load Current into Data variable #1 and Previous into Data variable #2. Choose OK. The probability that n is greater-than or equal to 5 is .1094. This is greater than .05, and H₀ is accepted. Median defects has not increased.

Problem III This is a two sample medians test.

Those using Quick's data files should load stata129.

Others should create a data file with 2 variables. The first should contain the sick days taken by graduates and then non-graduates. The second should contain eleven 0's for graduates and twelve 1's for non-graduates.

All StataQuest users should select Statistics, Nonparametric tests, and Mann-Whitney. Load SickDays into the Data variable box, Group into the Group variable box, and choose OK. P of .0012 is less than .01 and H₀ is rejected. There is a difference in median sickdays taken by graduates and non-graduates.

Problem I message: This is a paired sample (difference) test of the median.

Those using Quick's data files should load stata133. Others should create a data file with 2 variables. The first should contain employee efficiency ratings before training and the second should contain their rating after training. Be sure to match ratings with employees.

All StataQuest users should choose Statistics, Nonparametric tests, and Sign test. Load Before into Data variable #1, After into Data variable #2, and choose OK. The probability that n is greater-than or equal to 5 is .0625. Accept H₀ because .0625>.01. Efficiency did not increase. This sample is too small to ever reject H₀.

Problem II message: This is a Kruskal-Wallis test of several medians.

Those using Quick's data files should load stata133. Others should add 2 variables to their page 133 data file; one for weights listed by departments and a second for departments 1-3.

All StataQuest users should select Statistics, Nonparametric tests, and Kruskal-Wallis. Load Weight into the Data variable box and Dept. into the Group variable box. Choose OK. H₀ is rejected because p = .0056 < .01. These medians are not equal.

Problem I message: This is a correlation study.

Those using Quick's data files should load stata148. Others should create a data file with 2 variables. The first should contain employee ages and the second employee sales.

All StataQuest users should select Statistics, Correlation, and Pearson (regular). Load Sales and Age into the Data variable box. Choose OK. The coefficient of correlation is -.9076. Do other problems by hand.

Problem I message: This is a regression study.

All StataQuest users should load stata148. Select Statistics and Simple regression. Load Sales into the Dependent variable box and Age into the Independent variable box. Choose OK. a = 55.94 and b = -1.07. Do other problems by hand.

Other answers on page QQ 8.

3 is .0011, 4 is .0000, and 5 is .0000 4A) See page QQ 56. 4B) .1353 4C) .2707

4D) .1804+.0902+.0361+.0120+.0034+.0009+.0002 = .3233 4E) .6767

2B) t = - 2.5, reject H₀, weekly earnings changed

3) p = .1185/2 = .059 > .05, accept H₀ , tires last > or = to 70,000 miles

4) p = .14/2 = .07>.05, accept H₀ , loan length did not increase 5) See page QQ 92 and 93.

3) p = .0148 < .05, reject H₀ , defined rentals differ at these stores

3) p = .3692/2 = .1846 > .05, accept H₀, women did not do better than men

3) z = -.855 which is not beyond -1.96, accept H₀, median delivery times are the same

2) Chi-square = 11.795 > 9.21 and p = .0027 < .10, reject H₀, median grades are not equal

5) A bin width of 4 results in a Histogram similar to the one on page T 36-37. Do other graphs by hand.

6) Mean = 7, SD = 4.34, Variance = 18.86, Median = 6 (see 50th percentile), and

Skewness (not Pearson's) = 1.68. Answer other questions by hand.

11A) .7351 11B) .0328 11C) See page ST 1. 11D) .0305

12A) .0916 12B) See page T 80-81 which was drawn from page ST 2.

13 and 14) Do by hand.

15) Using the StataQuest calculator, the range is 6.914 to 8.286.

16A) .5934 to .8066 16B) Do by hand.

2) t = -1.77 and p = .0771/2 = .0386 < .05, reject the Null Hypothesis, the mean is not greater than or equal to 80%. Too bad, no chocolate flavored shaving cream.

3) Chi square of 8.4 and p = .015 < .05, reject the Null Hypothesis, supplier and defects are dependent

4) t = -4.24 and p = .0240/2 = .0120 < .05, reject the Null Hypothesis, sales performance improved

5) t = 2.97 and p = .0039 < .05, reject the Null Hypothesis, mean service time at these stores is not the same

6) t = - 4.0 and p = .0012/2 = .0006 < .01, reject the Null Hypothesis, assembly time has gone down

7) F = 1.58 and p = .5044 < .10, accept the Null Hypothesis, shopping time variability has not increased

8) This problem should be done by hand.

9A) F = 5.77 and p = .0244 < .05, reject the Null Hypothesis, the mean time to assemble

computer parts is not equal for these three departments 9B) Do by hand.

10) z = 1.545 and p = .1224 > .01, accept the Null Hypothesis, median assembly times are equal

11) Chi square = 11.23 and p = .0036 < .01, reject the Null Hypothesis, median assembly times for these three methods are not equal.

12) The data set for this problem must have its own file. Therefore, this problem must be done by hand.

13) p = 21.88 > .10, accept the Null Hypothesis, brands are viewed equally by consumers

3A) r = .8852 3B) r squared = .7836 3C) Do by hand.

4) Do by hand.

5) The StataQuest fitted model graph is similar to the graph on page T 160.

6A) a = 13.51 and b = 6.54 6C) The StataQuest fitted model graph is similar to the graph on

page T 160. Do other questions by hand.