This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: STAT 500 — HW 3 — Solution Grading: one point per part of questions 1-3, 5 points for question 4. 1. Cloud seeding experiment (a) eu = cloud, ou = cloud. No, don’t expect a problem with non-independence. Note: If the eu and ou were different things, you would expect a problem. (b) No, the data are not consistent with the assumption of normally distributed errors. Note: You could use a variety of graphical assessments, e.g box plots, QQ plot, or a formal test. All that I’ve looked at give the same answer. Your explantion should include what you did and what in that suggested non-normality. For example, a QQ plot of the residuals or two QQ plots of the residuals for each group show clear departures from a straight line. (c) It seems that the assumption of equal variances is not reasonable. Both residual or abs(residual) plot show the same pattern: the seeded group of clouds is more variable. Note: other analyses are reasonable so long as the explanation was reasonable. (d) A two-sample t-test of abs(residual) gives p=0.017. There is evidence of unequal variances. (e) ˆ β = log 651- log 278 log 442- log 164 = 0 . 86. This suggests a log transformation. Note: you might suggest a transformation of Y . 14 . This would make sense only if prior knowledge of this type of data indicates that this makes sense....
View Full Document
- Fall '08