252solnD3 3/23/05 (Open this document in 'Page Layout' view!) D. COMPARISON OF TWO SAMPLES 1. Two Means, Two Independent Samples, Large Samples. Text 10.1-10.3, 10.7 [10.1 – 10.3, 10.5] (10.1 – 10.3, 10.5) 2. Two Means, Two Independent Samples, Populations Normally Distributed, Population Variances Assumed Equal. Text 10.4, 10.13a, 10.20a, b, e [10.4, 10.15a, 10.13a,b,e]. For the last problem: x1 17 .5571 , s1 1.9333 , x 2 19.8905 , s 2 4.5767 (10.4, 10.14, 10.12a,b,e) 3. Two Means, Two independent Samples, Populations Normally Distributed, Population Variances not Assumed Equal. Optional Text 10.20[10.13c,d] (10.12c,d) See data above. D3, D4 4. Two Means, Paired Samples (If samples are small, populations should be normally distributed). Text 10.26, 10.29[10.36, 10.37], D1, D2 (10.32*(in 252hwkadd.), [10.34] (different numbers), 10.25[10.35], D1, D2) 5. Rank Tests. a. The Wilcoxon-Mann-Whitney Test for Two Independent Samples. Text 12.65[10.48] (10.46) b. Wilcoxon Signed Rank Test for Paired Samples. Text 12.74-12.76[10.57-59] (10.80-82 on CD), Downing & Clark 18-15, 18-9 (in chapter 17 in D&C 3rd edition), D5 6. Proportions. Text 10.32, 10.38, 10.39, 12.32** [12.2, 12.7*, 12.8*] (12.2) 7. Variances. Text 10.40, 10.43-10.48 [10.16, 10.19 - 10.24, 10.25] (10.15, 10.18 - 10.23, 10.24) D6a (below), D6, D7 (A summary problem), D8 (A summary problem) Graded assignment 3 will be posted. Solutions to outline points 6 and 7 are in this document. Summary problems D7 and D8 are in the next document. ------------------------------------------------------------------------------------------------------------------------------- Problems with 2 Proportions. x 45, n1 100 Exercise 10.32 [12.2 in 9th] (12.2 in 8th edition): Assume that 1 . a) At the 5% significance x2 25, n2 50 level is there a significant difference between population proportions? b) Construct a 95% confidence interval for the difference between the two population proportions. Solution: From Table 3 of the Syllabus Supplement: Interval for Confidence Hypotheses Test Ratio Critical Value Interval pcv p0 z 2 p Difference p p 0 p p z 2 sp H 0 : p p0 z between If p0 0 p H 1 : p p0 p p1 p2 proportions p p0 q 0 1 n 1 n If p 0 p 0 p 01 p 02 p1q1 p2 q 2 q 1 p s p n1 1 n2 or p 0 0 p p01q 01 p02 q 02 n1 n2 Or use s p 2 n p n2 p2 p0 1 1 n1 n 2 H 0 : p1 p 2 H 0 : p 0 H : p p 2 0 x1 45, n1 100 .01 0 1 Given or or x2 25, n2 50 H 1 : p1 p 2 H 1 : p 0 H 1 : p1 p 2 0 x 45 x 25 .45, p2 2 .50 . p p1 p2 .45 .50 .05 . a) p1 1 n1 100 n2 50 p0 x1 x2 45 25 70 n p n2 p2 100 .45 50 .50 0.4667 1 1 q0 1 p0 1 .4667 .5333 n1 n2 100 50 150 n1 n2 100 50 p p0q0 1 n1 1 n3 .4667 .5333 1100 150 .0074667 .08641 z 2 z.005 2.576. 1 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) (Only one of the following methods is needed!)Test Ratio: z p p 0 p .05 0 0.579 . Make a .08641 diagram showing a 99% 'accept' region between -2.576 and +2.576. Shade the 'reject regions below -2.576 and above 2.576. Since -0.579 lies in the 'accept' region, do not reject H 0 . or Critical Value: pcv p0 z p 0 2.5760.08641 0.233 . Make a diagram showing a 2 99% 'accept' region between -0.223 and +0.223. Shade the 'reject' regions below -0.223 and above 0.223. Since p .05 lies in the 'accept' region, do not reject H 0 . b) p p z sp , where p p1 p2 .45 .50 .05, 2 q1 1 p1 1 .45 .55, q2 1 p2 1 .50 .50, sp p1q1 p2 q2 .45.55 .50 .50 .002475 .005000 .007475 0.086458 . n1 n2 100 50 So p .05 2.576 0.086458 .05 .223 or -.173 to .273. Note that this interval includes zero and thus supports the null hypothesis. Exercise 10.38 [12.7 in 9th] (Not in 8th edition): Of 56 white workers terminated, 29 claimed bias. Of 407 black workers terminated, 126 claimed bias. a.)At the 5% significance level, is there evidence that white workers are more likely to claim bias than black workers? b) Find and interpret the p-value in a). Solution: H 0 : p 0 H : p p 2 H : p p 2 0 x 29, n1 56 Given 1 or 0 1 or so .05 0 1 x 2 126 , n 2 407 H1 : p 0 H 1 : p1 p 2 H 1 : p1 p 2 0 Population 1 contains white workers and population 2 x x 29 126 p1 1 .5179 , p 2 2 .3096 . p p1 p2 .5179 .3096 .2083 n1 56 n 2 407 a) p0 x1 x2 29 126 155 n p n p 56 .5179 407 .3096 0.3348 1 1 2 2 n1 n2 56 407 463 n1 n2 56 407 p p 0 q 0 1 n1 1 n3 .3348 .6652 156 1 407 .22270 .017857 (Only one of the following methods is needed!)Test Ratio: z .002457 .0052393 .06726 p p 0 p .2083 0 3.097 . This is a .06726 one-tail test! Because the alternate hypothesis says that the difference between the proportions is above zero, this is a right-tail test. Make a diagram with a mean at zero showing a 95% 'accept' region below z z .05 1.645 . Shade the 'reject region above 1.645. Since 3.097 lies in the 'reject' region, reject H 0 . or Critical Value: Since the alternate hypothesis says p 0 , we need a critical value for the difference between the two proportions that is above zero, so the two-tail formula, pcv p0 z p becomes 2 pcv p0 z p 0 1.6450.06726 .1106. . Make a diagram showing a 95% 'accept' region between 0.1106. Shade the 'reject' regions above 0.1106. Since p .2083 is above .1106 and lies in the 'reject' region, do not reject H 0 . or Confidence Interval: The two sided formula is p p z s p . 2 Since the alternate hypothesis is p 0 , a one-sided interval will be p p z s p where 2 s p p1 q1 p 2 q 2 .5179 .4821 .3096 .6904 .0044586 .0004616 .004920 0.07014 . So n1 n2 56 463 the confidence interval will say p 0.2083 1.645 0.07014 0.2083 .1154 .0929 . The null hypothesis says p 0 , which is contradicted by p .0929 . Reject H 0 . 2 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) b) Remember that the alternate hypothesis is p 0 Extreme, in this case means high, since high values of p will discourage our belief that p 0 . pvalue Pz 3.097 .5 .4990 .0010 The Instructor’s Solution Manual says (a) There is sufficient evidence to conclude that white workers are more likely to claim bias than black workers. (b) Using Excel, the p-value is 0.00098. The probability of obtaining a difference in two sample proportions as large as 0.20828 or larger is 0.00098 when the null hypothesis is true. Exercise 10.39 [12.8 in 9th] (Not in 8th edition): 500 African Americans and 500 whites (all with incomes above $50000) were surveyed with the result that 74% of the African-Americans and 84% of the whites owned stocks. a. At the 5% significance level, is there a significant difference between the proportion of African Americans and the proportion of whites who own stocks? b. Find and interpret the p-value in a. c. Set up a 95% confidence interval for the difference between the two proportions. H : p p 2 H : p 0 H : p p 2 0 p .74, n1 500 Solution: Given 1 or 0 1 or 0 .05 0 1 p 2 .84, n 2 500 H 1 : p1 p 2 H 1 : p 0 H 1 : p1 p 2 0 p p1 p2 .74 .84 .10 (Population 1 = African American, Population 2 = Whites) a) p0 n1 p1 n2 p2 500 .74 500 .84 .79 q0 1 p0 1 .79 .21 n1 n2 500 500 p p 0 q 0 1 n1 1 n3 .79 .21 1500 1500 .1659 .004 (Only one of the following methods is needed!)Test Ratio: z .02576 p p 0 p z 2 z.015 1.960. .10 0 3.882 . Make a .02576 diagram showing a 95% 'accept' region between -1.960 and +1.960. Shade the 'reject regions below -1.960 and above 1.960. Since -3.882 lies in the lower 'reject' region, reject H 0 . or Critical Value: pcv p0 z p 0 1.9600.02576 0.0505 . Make a diagram showing a 2 95% 'accept' region between -0.0505 and +0.0505. Shade the 'reject' regions below -0.0505 and above 0.0505. Since p .10 lies in the lower 'reject' region, reject H 0 . The Instructor’s Solution Manual says that there is sufficient evidence to conclude that a significant difference exists in the proportion of African Americans and whites who invest in stocks. b) Since this is a 2-sided test, p value 2Pz 3.882 2.5 .4999 .0002 . The Instructor’s Solution Manual should say that the probability of obtaining a test statistic as large in absolute value as 3.8819 or larger is 0.0002 when the null hypothesis is true, but it apparently got this one wrong. c) p p z sp , where p p1 p 2 .10, q1 1 p1 1 .74 .26, q 2 1 p 2 1 .84 .16, 2 s p p1 q1 p 2 q 2 .74 .26 .84 .16 .0003848 .0002688 .0006536 0.02557 . n1 n2 500 500 So p .10 1.960 0.02557 .10 .050 or -.150 to .050. Note that this interval does not include zero and thus contradicts the null hypothesis. 3 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) Exercise 12.32 (New): A random sample of 200 coffee drinkers is taken and their preferences between Brand A and Brand B are taken before and after an advertising campaign. After Before A B The results are as follows: . a) Is there evidence at the 5% significance level that the A 101 9 22 68 B proportion of coffee drinkers who prefer brand A has risen over the campaign? b) Compute a p-value and interpret it. Solution: a) The 2005 version of the outline says the following. In the McNemar Test we compare two proportions taken from the same sample, which is equivalent to paired samples. Assume that two different question 2 question 1 yes no questions are asked of the same group with the following responses. So, for x yes 11 x12 x no 21 x 22 example x 21 is the number of people who answered no to question 1 and yes to question 1. x11 x12 x 21 x 22 n , p1 z x12 x 21 x12 x 21 H 0 : p1 p 2 x11 x12 x x 21 and p 2 11 . If we wish to test , let n n H 1 : p1 p 2 . (The test is valid only if x12 x 21 10 .) Here we have a left-sided test because for us to believe that the proportion that prefer A to rise, we need to find that x11 x 21 x11 x12 , but this implies x 21 x12 , which means that the numerator of z will be H 0 : p1 p 2 negative. Our hypotheses are . Using our formula for z , we compute H 1 : p1 p 2 z x12 x 21 x12 x 21 9 22 9 22 13 2 2.3349 . For a left – sided test our rejection region is below 31 z .045 1.645 . Since -2.3349 is in the rejection region, we reject the null hypothesis and conclude that the proportion has risen. b) The p-value is Pz 2.34 .5 .4904 .0096 , which says that if the null hypothesis is true the probability of getting results as extreme or more extreme than what we actually got is 96 one-hundredths of 1%. 4 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) Problems with 2 Variances. Exercise 10.40 [10.16 in 9th] (10.15 in 8th edition): Find FU and FL the upper and lower critical values of the F ratio for the following two-tailed tests: (a) .10, n1 16, n2 21 , (b) .05, n1 16, n 2 21 , (c) .02, n1 16, n2 21 , (d) .01, n1 16, n 2 21 , (e) As gets smaller (or as the confidence level gets larger) what happens to the size of the ‘accept’ and ‘reject’ regions? Solution: For a 2-tail F test, the upper bound is FU Fn1 1,n2 1 and the lower bound is 2 n1 1,n2 1 FL F1 1 2 n2 1,n1 1 F 15,20 FL F.95 15, 20 and . For .10, n1 16, n2 21 , for example, FU F.05 2 1 20,15 F.05 From the Instructor’s Solution Manual (edited): These values come from table E.5 in the text. Since my table does not have a df1 15 column, you would have to average the values in the df1 14 and df1 16 columns to get these numbers. 1 = 0.429 2.33 1 (b) .05, n1 16, n 2 21 FU = 2.57, FL = = 0.362 2.76 1 (c) .02, n1 16, n2 21 FU = 3.09, FL = = 0.297 3.37 1 (d) .01, n1 16, n 2 21 FU = 3.50, FL = = 0.258 3.88 (e) As gets smaller, the rejection region gets narrower and the acceptance region gets wider. FL gets smaller and FU gets larger. (a) .10, n1 16, n2 21 FU = 2.20, FL = Exercise 10.43-10.46 [10.19-22 in 9th] (10.18-21 in 8th edition): Assume that you have the following information: n1 25, n2 25, s12 133.7 and s 22 162.9 . a) What is the value of the F ratio used for testing equality of variances? b) How many degrees of freedom do you have in the numerator and the denominator of the f test? c) What are the critical values FU and FL if you do a test for equality of variances? d) What is your statistical decision? Solution: a) Given: H 0 : 12 22 , H1 : 12 22 , n1 25, n2 25, s12 133.7 and s 22 162.9. Fcalc s12 s 22 133 .7 0.826 . b) Since n1 25 and n2 25, the degrees of freedom are df1 n1 1 161 .9 24, 24 . 25 1 24 and df 2 n 2 1 24 . c) Part of the F table appears below. We find F.025 14 df 2 24 * 0.100 0.050 0.025 0.010 1.80 2.13 2.47 2.93 16 * 1.77 2.09 2.41 2.85 20 24 30 * 1.73 2.03 2.33 2.74 * 1.70 1.98 2.27 2.66 * 1.67 1.94 2.21 2.58 df1 40 50 * 1.64 1.89 2.15 2.49 * 1.62 1.86 2.11 2.44 24, 24 2.27 and FU F n1 1,n2 1 F.025 FL F1n1 1,n2 1 2 2 75 100 * 1.59 1.82 2.05 2.37 * 1.58 1.80 2.02 2.33 1 1 200 500 * 1.56 1.77 1.98 2.27 * 1.54 1.75 1.95 2.24 10000 * 1.53 1.73 1.94 2.21 1 . d) The F 24,24 2.27 0.441 F n2 1,n1 1 .025 2 Instructor’s Solution Manual says that, since Fcalc = 0.826 is between the critical bounds of FU = 2.27 and FL = 0.441, do not reject H0. There is not enough evidence to conclude that the two population variances are different. 5 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) Exercise 10.47 [10.19 in 9th] (10.18 in 8th edition): Assume that you have the following information: n1 16 , n 2 13, s12 47.3 and s 22 36.4 , but the populations are very skewed to the right, should you use the F test to test for equal population variances? Solution: Given: From the Instructor’s Solution Manual , In testing the equality of two population variances, the F-test statistic is very sensitive to the assumption of normality for each population. If the populations are very right-skewed, the F-test should not be used. It really should be noted that the Levene test (to be covered in section F) is a more appropriate test in this case. However, the entire data sets are needed for the Levene test. Exercise 10.48 [10.24 in 9th] (10.13? in 8th edition): Again assume that you have the following information: n1 16 , n 2 13, s12 47.3 and s 22 36.4 and use a 5% significance level. a) Is there evidence of a difference between 12 and 22 ? b) Test 12 22 . c) Test 12 22 . Solution: a) H 0 : 12 22 or H 0 : Fcalc s12 s 22 12 22 1. and H1 : 12 22 or H 1 : 47 .3 1.2995 . Since n1 16 and 36 .4 n 2 13, 12 22 1. the degrees of freedom are df1 n1 1 15,12 3.18 and 16 1 15 and df 2 n2 1 12 . For our upper and lower critical values, we find F.025 15,12 F.975 1 1 . The first value comes from the table in the text since values of df1 14 12,15 2.96 0.338 F.025 and df1 16 appear in Table 2, but not df1 15 . This is not really a problem, since if we find 14,12 3.21 F.025 and 16,12 3.15 F.025 it is not hard to guess that 15,12 F.025 14,12 F 16,12 F.025 .025 2 3.21 3.15 3.18 . The following comes from the Instructor’s Solution Manual. 2 Decision: Since Fcalc = 1.2995 is between the critical bounds of FU = 3.18 and FL = 0.3378, do not reject H0. There is not enough evidence to conclude that the two population variances are different. The Minitab instructions and output for this test on the Normality assumption follow: MTB > VarTest 16 47.3 13 36.4. Test for Equal Variances 95% Bonferroni confidence intervals for standard deviations Sample N Lower StDev Upper 1 16 4.87589 6.87750 11.4026 2 13 4.13630 6.03324 10.7890 6 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) The confidence intervals for the two variances are Bonferroni intervals, which mean that they are simultaneously valid at the 5% level. The test statistic is the one calculated above and rounded. The p-value given to the null hypothesis is above our significance level so that we cannot reject the null hypothesis. b) The statement ( 12 22 ) to be evaluated in b) is a strict inequality and thus must be an alternative hypothesis, so that we have H 0 : 12 22 or H 0 : 12 22 1 and H1 : 12 22 or H 1 : alternative hypothesis is true, we would expect the ratio, Fcalc s12 12 22 1 . If the to be above one. Remember that s 22 df1 15 and df 2 12 . Since this is a one-sided test with a significance level of 5%, we compare our 15,12 2.62 . (We can get this from the textbook table or by averaging the values calculated F ratio, with F.05 14,12 and F 16,12 .) The following comes from the Instructor’s Solution Manual. of F.05 .05 Decision: Since Fcalc = 1.2995 is less than the critical bound of FU = 2.62, do not reject H0. There is not enough evidence to conclude that the variance for population 1 is greater than the variance for population 2. c) The statement ( 12 22 ) to be evaluated in c) is a strict inequality and thus must be an alternative hypothesis, so that we have H 0 : 12 22 or H 1 : 12 22 1 and H1 : 12 22 or H 1 : 12 22 1. The Instructor’s Solution Manual implies that if the alternative hypothesis is true, we would expect the ratio, s2 Fcalc 12 to be below one. Remember that df1 15 and df 2 12 . Since this is a one-sided test with a s2 significance 15,12 F.95 level 1 of 5%, we compare our calculated F ratio, Fcalc s12 s 22 1.2995 with 1 . The following comes from the Instructor’s Solution Manual. 12,15 2.48 0.403 F.05 Decision: Since Fcalc = 1.2995 is greater than the critical bound of FL = 0.403, do not reject H0. There is not enough evidence to conclude that the variance for population 1 is less than the variance for population 2. An alternative way to approach this problem is to note that we can say that our hypotheses are H 0 : 12 22 , H1 : 12 22 or H 1 : ratio, Fcalc s 22 s12 22 12 1. If the alternative hypothesis is true, we would expect the to be above one. Remember that df1 15 and df 2 12 . Since this is a one-sided test with a significance level of 5%, we compare our calculated F ratio, Fcalc s 22 s12 1 0.7695 with 1.2995 15,12 2.48 . Since our calculated F ratio is not above the 5% critical value, we cannot reject the null F.05 hypothesis. 7 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) Problem D6a: Given: H 0 : 12 22 , H 1 : 12 22 , n1 16 , n 2 21, s12 100 and s 22 400 . Test the hypotheses using only right tail tests. .05 . (This is my preferred way of doing problems like this because the F table is set up to make right tail tests most convenient.) s 2 100 Solution: The ratio Fcalc 12 0.250 has 15 and 20 degrees of freedom. We found that s 2 400 15,20 2.57 , and we see that our calculated F-ratio does not exceed it. The ratio F.025 Fcalc s 22 s12 400 4.000 has 100 20,15 2.76, and this second calculated F-ratio does exceed it. Since 20 and 15 degrees of freedom. We found that F.025 one calculated F-ratio is above its bound, we reject the null hypothesis. To get the p-value, compute 2 PF 4.00 on the computer as follows below. Welcome to Minitab, press F1 for help. Results for: notmuch.MTW MTB > %fareaa Executing from file: fareaa.MAC Graphic display of F curve areas Finds and displays areas to the left or right of a given value or between two values. (This macro uses C100-C116 and K100-K120) Enter the degrees of freedom.DF2 must be above 4. DATA> 20 DATA> 15 Do you want the area to the left of a value? (Y or N) n Do you want the area to the right of a value? (Y or N) y Enter the value for which you want the area to the right. DATA> 4 ...working... F Curve Area F Curve with numerator DF of 20 and Denominator DF of 15 The Area to the Right of 4 is 0.0043 0.9 0.8 0.7 Density 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 1 2 3 4 Data A xis 5 6 7 8 Data Display mode 0.890625 Data Display std dev 0.631988 So we can say that 2PF 4.00 2.0043 .0086 . 8 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) Exercise 10.49 [10.25 in 9th] (10.24 in 8th edition): A professor claims that there is much more variability in exam scores of students taking the introductory accounting course as a requirement than for students taking the same course as part of their accounting major. A random sample of scores of 13 non accounting majors and a second sample of scores of accounting majors is taken from the recorded scores for one exam. The results are n1 13, n 2 10, s12 210.2 and s 22 36.5. a) Test the professor’s claim at the .05 level. b) Interpret the p-value. c) What assumptions are necessary to justify the use of as F test? Solution: Note that the professor’s conjecture can be written as 1 2 2 2 and must be an alternative hypothesis. a) The Instructor’s Solution Manual gives the following hypotheses: H0: 1 2 2 2 (Where Populations: 1 = nonaccounting students, 2 = accounting students) - The variance in final exam scores of nonaccounting students is less than or equal to the variance in final exam scores of accounting students. H1: 1 2 2 2 - The variance in final exam scores of nonaccounting students is greater than the variance in final exam scores of accounting students. We put the variance that should be larger according to the alternative hypothesis on top to get the test ratio Fcalc s1 2 2 210 .2 12,9 3.07 . Since our = 5.759. This ratio has 12 and 9 degrees of freedom. F.05 36 .5 s2 calculated F-ratio exceeds the table value, we reject the null hypothesis. 12,9 5.11 , we can say that the p-value is above .01. If we check table E.5 b) Since 5.759 is larger than F.01 in the text, we find that the p-value is above .005. The Instructor’s Solution Manual says that using Excel, the p value = 0.0066. c) This test assumes that the parent populations are both Normally distributed. Problem D6: We have the following data for returns on two stocks: Stock A 7, 8, -5, 9, 11 nA = 5 Stock B 6, 7, 0, 4, 9, 15 nB = 6 a. Find a 95% interval for A2 B2 b. Test the following at a 95% level: H 0 : A2 B2 H 1 : A2 B2 Solution: The formulas for this problem are given in “Confidence limits and Hypothesis Testing for Variances” in the Syllabus Supplement and summarized on the formula pages. a) From the data above we can compute s A2 40.00 and s B2 25.36 . The formula given is s 22 s12 22 s 22 ( n1 1, n2 1) F . If we let Stock A be x 2 , and Stock B be x1 , then we can state that F n2 1,n1 1 12 s12 2 1 2 DF1 n B 1 5 , DF1 n A 1 4 and confidence interval formula and get ( 4,5) F.025 7.39 , so 1.577 s A2 s B2 s 22 s12 s A2 s B2 40 .00 1.577 . Substitute these values in the 25 .36 A2 s A2 (5, 4) ( 5, 4 ) 9.36 and F.025 . From the F table, F.025 F4,5 B2 s B2 1 2 A2 2 1 2 1.577 9.36 , which becomes 0.213 A2 14 .76 . If we wish an 7.39 B B interval for the standard deviations, we can take the square roots to get 0.462 A B 3.842 . 9 252solnD3 3/23/05 (Open this document in 'Page Layout' view!) b) We are testing H 0 : A2 B2 supplement, test F DF1 , DF2 s12 s 22 H1 : A2 B2 or H 0 : A2 2 1 H1 : A2 1 . According to the syllabus 2 B B where DF1 n1 1 , DF2 n 2 1 and s12 is the larger of the two variances according to the alternate hypothesis. Since this is a 1-sided test with .05 , s A2 s B2 40 .00 4,5 5.19 . Since the ratio is smaller than the F, we do not 1.577 is thus compared to F.05 25 .36 reject H 0 . 10