Understandable Statistics Seventh Edition By Brase and Brase Prepared by: Lynn Smith Gloucester County College Chapter Nine Part 5 (Section 9.7) Hypothesis Testing Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 1 Independent Sampling Distributions There is no relationship whatsoever between specific values of the two distributions. Example: comparison of the incomes of individuals from two groups. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 Testing the Differences of Means for Large, Independent Samples • Let x1 and x2 have normal distributions with means 1 and 2 and standard deviations 1 and 2 respectively. • Take independent random samples of size n1 and n2 from each distribution. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 3 The n the variablex1 x 2 hasthefollowingcharacte ri stics: 1. A normal distribution. 2. Mean 1 2 3. Standard deviation Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 1 n1 2 2 n2 4 If both n1 and n2 are 30 or larger The Central Limit Theorem can be applied even if the original distributions are not normal. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 5 When testing the difference of means The null hypothesis usually indicates that there is no difference between the means. H0: 1– 2 = 0 or, equivalently H0: 1= 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 6 Left-Tailed Test H0: 1= 2 H1: 1< 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 7 Right-Tailed Test H0: 1= 2 H1: 1 > 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 8 Two-Tailed Test H0: 1= 2 H1: 1 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 9 Use the Critical Values for the Normal Distribution = 0.05 = 0.01 Critical z for Left-Tailed Test 1.645 2.33 Critical z for Right-Tailed Test 1.645 2.33 Critical z for Two-Tailed Test 1.96 2.58 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 10 To convert the sample x1 x 2 value to a z value, use : z ( x1 x 2 ) ( 1 2 ) 2 1 n1 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 2 n2 11 A college professor wishes to determine (at 0.05 level of significance) if there is a difference in the final exam results between the two groups of students. One group studies Calculus in the traditional classroom setting; the other is taught via distance learning. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 12 Sample exam results: x 88 Distance Learning 85 3.3 4.1 n 31 30 Traditional Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 13 We wish to determine if the observed difference in exam results (88 versus 85) is significant. H0: 1= 2 H1: 1 2 Using a two-tailed test, the critical z values are 1.96 for = 0.05. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 14 Compute the sample test statistic x1 x2 88 85 3 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 15 Convert the test statistic to a z value z ( x1 x 2 ) ( 1 2 ) 12 n1 22 n2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 30 3 . 14 3.3 2 4.12 31 30 16 Since 3.14 > 1.96 We reject the null hypothesis - the statement that there was no difference between the two groups. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 17 We conclude that the exam results for the distance learning class are significantly different from those for the traditional group at 5% level of significance. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 18 P Value Approach • To find the P value for the sample statistic of the difference in the means, use the corresponding z value, z = 3.14. • Since we are working with a two-tailed test, add the area to the right of z = 3.14 to the area to the left of z = – 3.14. • P value = 2(0.0008) = 0.0016 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 19 P value = 0.0016 We would reject H0 for any 0.0016. We, therefore reject H0 for = 0.05. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 20 Inferences about the Differences of Two Means of Small Independent Samples Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 21 Assumptions • Independent random samples are drawn from two populations with means 1 and 2. • The parent populations have normal distributions. • The standard deviations for the populations (1 and 2) are equal. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 22 Best Estimate of the Common or Pooled Standard Deviation for Two Populations ( n1 1) s ( n2 1) s s n1 n2 2 2 1 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 2 23 When testing the difference of means The null hypothesis usually indicates that there is no difference between the means. H0: 1– 2 = 0 or, equivalently H0: 1= 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 24 Left-Tailed Test H0: 1= 2 H1: 1< 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 25 Right-Tailed Test H0: 1= 2 H1: 1 > 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 26 Two-Tailed Test H0: 1= 2 H1: 1 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 27 To Test the Hypothesis Use the t Statistic x1 x 2 t with d.f. n1 n 2 2 1 1 s n1 n2 where s the pooled standard deviation Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 28 The Pooled Standard Deviation ( n1 1) s ( n2 1) s s n1 n2 2 2 1 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 2 29 Test H0: 1 = 2 Against H1: 1 > 2 Use a 1% Level of Significance. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 30 Sample Data Population 1 Population 2 x 17.3 16.1 s 1.6 1.5 n 8 9 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 31 Degrees of Freedom d.f. = n1 + n2 – 2 = 8 + 9 – 2 = 15 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 32 For a Right-Tailed Test at 1% Level of Significance and d.f. = 15 Use the column headed by = 0.010 The critical t value is t = 2.602. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 33 Compute the Test Statistic x1 x2 x1 x2 17.3 16.1 1.2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 34 Pooled Standard Deviation ( n1 1) s ( n2 1) s s n1 n2 2 2 1 2 2 (8 1)1.6 (9 1)1.5 1.55 892 2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 2 35 Convert the Test Statistic x1 x 2 to a t Value x1 x2 17.3 16.1 1.2 x1 x 2 1 .2 t 1.593 1 1 1 1 s 1.55 n1 n2 8 9 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 36 Critical t Value = 2.602 2.602 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 37 Test Statistic 1.593 1.593 2.602 Do not reject H0. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 38 P Value Approach To test H0: 1 = 2 Against H1: 1 > 2 t value = 1.593, d.f. = 15 • For a right-tailed test, use the column headed by . • For d.f. = 15, t = 1.593 falls between t = 1.517 and t = 1.753. • We conclude that 0.050 < P < 0.075. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 39 Conclusion: 0.050 < P < 0.075 We reject H0 only for P. We could not reject H0 for = 0.01. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 40 Tests for Differences of Proportions • We will draw independent random samples of size n1 and n2 from two binomial distributions. • For large values of n1 and n2,, the distribution of sample differences in proportions of successes will be approximately normally distributed. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 41 Symbols Used r1 and r2 = number of successes from the first and second samples p1 and p2 = probability of success on each trial from the first and second binomial distributions Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 42 For large values of n1 and n2: The distribution of sample differences r1 r2 pˆ 1 pˆ 2 is approximately normal with n1 n2 mean p1 p2 and standard deviation p1 q1 n1 p2 q2 n2 where q1 1 p1 and q 2 1 p2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 43 The Null Hypothesis p1 = p2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 44 The proportions p1 and p2 are unknown and must be estimated Best (pooled)estimate " p hat" r1 r2 pˆ n1 n2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 45 Critical Values for Tests Involving a Difference of Two Proportions (Large Samples) = 0.05 = 0.01 Critical z for Left-Tailed Test 1.645 2.33 Critical z for Right-Tailed Test 1.645 2.33 Critical z for Two-Tailed Test 1.96 2.58 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 46 “Large” Samples All of the following are larger than five: n1 pˆ , n1qˆ , n2 pˆ , n2 qˆ Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 47 Sample Test Statistic z Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . pˆ 1 pˆ 2 pˆ qˆ pˆ qˆ n1 n2 48 A college is attempting to determine whether a reminder phone call encourages students to participate in early registration. A group of 1200 students was divided into two groups. One received a reminder phone call; the other did not. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 49 The data: Reminder No Reminder n 600 600 Registered early 475 452 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 50 Use a 5% level of significance to test the claim that the reminder phone calls increased participation in early registration. • Let p1 = proportion from the first group (reminder) who registered early. • Let p2 = proportion from the second group (no reminder) who registered early. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 51 The null hypothesis is that there is no difference in proportions. H0: p1 = p2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 52 The alternate hypothesis is that the reminders improve participation rate. H1: p1 > p2 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 53 H1: p1 > p2 • Use a right-tailed test. • For 5% level of significance, the critical z value = 1.645. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 54 For the Reminder group, r1 = 475, n1 = 600. r1 475 pˆ 1 0.792 n1 600 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 55 For the No Reminder group, r1 = 452, n1 = 600. r2 452 pˆ 2 0.753 n2 600 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 56 The Sample Statistic pˆ 1 pˆ 2 0.792 0.753 0.039 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 57 The pooled estimates of proportion r1 r2 475 452 pˆ 0.7725 n1 n2 600 600 qˆ 1 pˆ 0.2275 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 58 Convert the test statistic to a z value: z pˆ 1 pˆ 2 pˆ qˆ pˆ qˆ n1 n2 0.039 1.61 .7725 (.2275) .7725 (.2275) 600 600 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 59 Critical z Value = 1.645 1.645 Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 60 Test statistic = z = 1.61 1.61 1.645 Do not reject the null hypothesis. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 61 Conclusion At 5% level of significance, we cannot reject the claim that the proportions (of students who register early) are equal. We conclude that the reminder phone calls do not make a significant difference in the number of students who participate in early registration. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 62 P Value Approach • For a right-tailed test, the value of P is the area in the right tail of the distribution, larger than z = 1.61. • From Table 5a, we find the P value. • P = 1.000 – 0.9463= 0.0537 • We would reject H0 only for 0.0537. • We cannot reject H0 for = 0.05. Copyright (C) 2002 Houghton Mifflin Company. All rights reserved . 63