This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Solution to Homework 9 (STAT515, Fall 2011) 9.10. ( 28 points, 4 points/part ) Given information: n 1 = n 2 = 100, x 1 = 70, x 2 = 50, 2 1 = 100, 2 2 = 64. a. X 1 X 2 = s 2 1 n 1 + 2 2 n 2 = r 100 100 + 64 100 1 . 2806 . b. (You may sketch a normal pdf curve that centers at 5. You may also mark some points on the horizontal axis that correspond to one (two, three) standard deviation(s) below/above 5. The normality is a result of Central Limit Theorem since both sample sizes are large.) c. Note that the standard error of X 1 X 2 from part a is very small. With the observed sample mean difference x 1 x 2 = 20, the corresponding zscore (to be computed in part e ) is very extreme (higher than ten!). This suggests that the observed data strongly contradict with the null hypothesis. d. The rejection region for a twosided test with = 0 . 05 based on large samples is ( , z . 025 ) ( z . 025 , + ) = ( , 1 . 96) (1 . 96 , + ). e. The test statistic is z = 70 50 5 1 . 2806 11 . 71 . Because the test statistic falls in the rejection region (in part d ), we reject H at significance level 0.05. That is, the observed data provide sufficient evidence that the difference between two population means in not 5. f. A 95% confidence interval for 1 2 is 20 1 . 96 1 . 2806 (17 . 49 , 22 . 51) . Based on this data set, one can be 95% confident that the first population mean is higher than the second population mean by the amount that lies between 17.49 and 22.51. g. I think that the confidence interval provides more information about 1 2 because now we have a plausible range of the true value of 1 2 . The test of hypothesis simply suggests that 1 2 is not 5, and does not provide further indication of what 1 2 is likely to be....
View Full
Document
 Fall '10
 Zhao

Click to edit the document details