This preview shows pages 1–3. 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: Solutions to Homework #9, 36220 Due 16 November 2005 8.54 We are supposed to test whether “the true average percentage of organic matter in such soil is something other than 3%”. So our null hypothesis H is that μ = 3. We need a twosided test, since we want to tell whether the true percentage is different than 3%, whether the difference is positive or negative. So H 1 is that μ 6 = 3. Because σ is not known, and n is comparatively small, only 30, we can’t use a z (standard normal) test. We can however use a t test, because we’re given that the individual measurements are normal. t = 2 . 481 3 . 295 = 1 . 76 Since n = 30, there are 30 1 = 29 degrees of freedom. To compute the pvalue, we need the probability that a tdistribution with 29 degrees of freedom would give a value as far from the origin as we got — Pr (  T  ≥  t  ). This is p = Pr( T ≤  t  ) + Pr( T ≥  t  ) = 2Pr( T ≥  t  ) since the tdistribution is symmetric. Turning to appendix A.8, we see that, with 29 degrees of freedom, Pr( T ≥ 1 . 7) = 0 . 05, and Pr ( T ≥ 1 . 8) = 0 . 041. Linear extrapolation between these values gives Pr ( T ≥ 1 . 76) = 0 . 0446. (Using statistical software gives 0 . 04456319.) That is, according to the null hypothesis, we should expect to see results this far out along one of the tails 4 . 46 percent of the time. Since we’re using a twosided test, we need to double this probability, p = 0 . 0892. This is less than 0 . 1, so we’d reject the null at the 10% level. At that level, we do have evidence that the mean concentration of organic matter in the soil isn’t 3%. At the stricter α = 0 . 05 level, we don’t have enough evidence to reject the null that μ = 3. 8.58 The SAS includes a twosided pvalue, namely 0 . 0711. We want to perform a twosided test — there will be a problem with our ciruits if the cuton voltage of the diodes is either too high or too low — so this suits us just fine. Because 1 . 0711 > . 01 and > . 05, we conclude that, at that significance level , cuton voltages do not differ systematically from 0 . 60V; with those values of α , we do not need to adjust anything. But 0 . 0711 < . 1, so the deviation from the null is significant at the less reliable 10% level. If that is our chosen value of α , we must adjust, because the average cuton voltage is too high (the calculated test statistic t is positive)....
View
Full
Document
This homework help was uploaded on 01/31/2008 for the course STAT 220 taught by Professor Shalizi during the Fall '05 term at Carnegie Mellon.
 Fall '05
 Shalizi
 Statistics

Click to edit the document details