This page was last updated on December 19, 2009.
All grading has now been completed. My thanks to everyone who participated in the course, and best wishes for the holiday season!
Final exam (corrected) with solutions
The spreadsheet contains the complete set of scores and grades for those students who gave me a 5-digit number. If you are not on the spreadsheet but you would like to know how your final grade was computed, feel free to email me.
I would like to make one comment about how the grades were computed. The weights given in the final grades were 30% homeworks, 30% midterm and 40%, as announced early in the semester. At the time of the midterm, I indicated that I might downweight the midterm to compensate for the very high variability in scores on that exam. While the scores on the final were indeed much less variable than those on the midterm (SD=12.9, IQR=16 as against 20.6, 36 on the midterm), after experimenting with different weightings I concluded that changing the weighting system would not have changed the final grades. Therefore, I stuck with the original weighting of the three components of the course.
Older announcements:
Review sessions: 4pm Thursday Dec 10; 2pm Sunday Dec 13; both in Hanes 130.
Analysis of Variance Notes (covered December 9, in class)
There will be no class December 7. The last regularly scheduled class is on December 9.
Homework 6 (optional): Chapter 7, questions 2, 4, 5, 12. Due December 14 (in the final exam, or sooner).
New scoring rule for homeworks: The best 5 of the 6 homework assignments will count for the final grade. Therefore, if you have completed the previous homeworks, you can skip this one without penalty, but you also have the opportunity to improve your overall score if you do not have a perfect score on past homeworks and do hand in HW6.
The final exam is a 3-hour in-class exam, in Hanes 130, same conditions as the mid-term (open book), from 4:00-7:00pm, Monday December 14.
There will be a review session at 4pm on Thursday December 10 (TBC).
Older announcements:
Slides on Florida election analysis
HW5, due 11/30/09: Chapter 5, problem 13 (only one problem this time, but it's a long multi-part question).
A query from a member of the class led me to identify a likely error in question 4.3. The right hand side should be multiplied by s (the sample standard deviation).
Homework 4, due November 9: Chapter 4, problems 2, 3, 5, 11
There will be no class November 9 or November 11. (The HW will be handed in to the TA - detailed instructions later.)
Influence diagnostics in R.
Midterm with sketch solutions.
Midterm exam (open book): In class, October 19 2009.
The midterm exam is an open-book exam covering the material in Chapters 2 and 3, but omitting sections that we have not covered in class. You can expect the style of questions to be somewhat algebra-intensive, but not requiring formal proofs. Any theorems or mathematical results that have been stated in the text may be quoted without proof. You are allowed to bring the course text and your own personal notes, and a calculator is also recommended. Please being a blue book to write your answers.
A review session has been arranged for Thursday, October 15, 6:00-7:30 pm, in the regular class room (Hanes 130).
Updates to class material:
SAS Power Calculations Example
Documentation for PROC POWER
Documentation for PROC GLMPOWER
Homework 3, due 10/19/09: Chapter 3, questions 5, 7, 9, 13.
Homework 2, due 10/07/09: Chapter 2, questions 14, 15; chapter 3, question 2, and this problem.
Notes: Questions 2.14 and 2.15 are both "computer questions": you may use R or SAS for this. Note that the data sets are available here and here. Unless you have previous experience of SAS, you may find it easier to use R for these examples, but later in the course there will be problems that specifically require you to use SAS (so it may be a good idea to start learning it now). Both questions do, however, require a significant amount of additional calculation in addition to the R or SAS output.
The two questions from chapter 3 are intended to be easier (but more theoretical in nature). Note that the last problem is the same as question 3 of chapter 3 but I've reworded it here as the original wording of this question was confusing; you should be able to do it (in this form) by directly imitating the argument for the proof of the Gauss-Markov theorem in chapter 2, that I also went through in class.
Homework 1, due 9/21/09: Chapter 2, questions 3, 4, 10, 13.
Note: The last two are intended as "calculator questions" -
can be done by hand - though you may use the computer to assist you
Chapter 1 presentation (class of 08/31/09)
Richard Potter's revised SAS code for the Amherst example
Return to Richard Smith's page