Information General about STA 4322/5325 Mathematical Statistics Spring 2004
Lectures: TR 12:301:45 in Room 108 OSB Instructor: Fred W. Huer, Room 209C OSB, Phone [phone]. Oce Hours: TR 11:2012:20 (or by appointment) e-mail: [e-mail] Announcements posted at: http://stat.fsu.edu/huffer Old Exams and Quizzes posted at: http://stat.fsu.edu/huffer/mordor/4322 This address must be typed completely; you cannot get there by clicking on a series of links. The posted les are pdf les and require Acrobat Reader to read and print them. This syllabus and some homework solutions will also be posted in mordor. Required Text: Mathematical Statistics and Data Analysis (Second edition), John A. Rice, Duxbury Press, 1995. (There are many other textbooks covering similar material. One commonly used book is Introduction to Mathematical Statistics by Hogg and Craig. Another is Mathematical Statistics with Applications by Wackerly, Mendenhall, and Scheaer.) Prerequisites: STA 4442 (Intro to Prob) and MAC 3313 (Calc III). Goal of Course: To introduce the theory and concepts which underly and justify commonly used statistical methods. Tentative Syllabus: We will cover some background probability material in chapters 3 and 4. Then we will cover chapter 6 and parts of chapters 8 and 9. A tentative list of topics is given below. Review of basic probability. Joint distributions. Distributions of sums and quotients. Expectations of functions of random variables. Moment generating functions. TEST #1 (date to be determined later) Distributions derived from the Normal distribution: 2 , t and F distributions. The distribution of the sample mean and the sample variance. Parameter estimation: method of moments and maximum likelihood. Pearsons chi-squared goodness-of-t test. Comparison of estimators using mean squared error and bias. Asymptotic distributions via the -method. 1
The Fisher information for parametric families of distributions. Asymptotic distributions for maximum likelihood estimates. The Cramer-Rao lower bound. Sucient statistics and the factorization criterion. The Rao-Blackwell theorem. Introduction to Hypothesis Testing. The Neyman-Pearson Lemma. TEST #2 will be given sometime in the period April 615 (date to be determined later). There will be a quiz on the last day of class (Thursday, April 22). Course Grades: Your course grade will be based on quizzes (given roughly every other week) and two exams. The percentage weights attached to these items are: Quiz Average Score on Exam # 1 Score on Exam # 2 40% 30% 30%
A course total will be computed using the weights above. This will be used to assign grades according to the following scale (which reports the lowest percentage score in each of the grade ranges): A A B+ B B C+ C C D 90.00 85.00 80.00 75.00 70.00 65.00 60.00 55.00 50.00 Your course total is computed in two ways. The raw course total is computed as Total = .40Q + .30T1 + .30T2 where Q = Quiz Average, T1 = Test #1 score, T2 = Test #2 score, with all values being raw scores expressed on a percentage scale. The curved course total is computed in a similar way except I rst curve each of the scores Q, T1 , T2 so that they have a of median 70 and an interquartile range of 20 (unless Note 3 applies). Then I compute the weighted average .40Q + .30T1 + .30T2 of these curved values. Finally, I curve this quantity to get the curved course total. The curved course totals (for those people who complete the course) will have a median of 70 and an interquartile range of 20 (unless Note 3 applies). Use the grade cutos given above to compute a letter grade for both the raw course total and the curved course total. Your nal letter grade is the better of these two grades.
Note 1: Lower grade cutos may be used (at my discretion) if the above grade cutos lead to what I consider an unreasonable distribution of grades. Note 2: All the medians and interquartile ranges used in curving T1 and T2 are computed using only the scores of those people who actually take the exam which is being curved. The curving of the quiz average is done using only those people who take at least 75% of the
2
quizzes. The curving of the nal course total is done using only those people who take both tests and at least 75% of the quizzes. Note 3: If the interquartile range of any quantity is less than 20 to start with, then it is not adjusted. That is, when curving, the median is always adjusted to be 70, but the interquartile range is only adjusted to be 20 if it is greater than 20 to start with.
Quizzes: Quizzes will be given roughly every other week and will be announced in advance. Quizzes will cover recent homework and lecture material. They will not be cumulative. Quiz problems will typically be modied versions of problems assigned as homework or modied versions of examples presented in lecture. They could also be short answer or multiple choice questions testing recollection of denitions, formulas or concepts covered in lecture. Schedule for rst few quizzes (tentative): January 20 (Tuesday) May include short question(s) on this handout. So, read it over! February 3 (Tuesday) February 17 (Tuesday) Your quiz average, which counts for 40% of your grade, is computed after dropping your lowest quiz score. If you miss a quiz, you receive a zero for that quiz (and this zero counts as one of your lowest scores). If you miss a quiz for a legitimate reason (medical problem or family emergency) and you bring adequate documentation of this reason, you will receive an excused absence. (The documentation should be brought in as soon as possible after the missed quiz.) If you have more than one excused absence, you will be given make-up quizzes for the additional excused absences. (Nothing special will be done if you have only one excused absence; this will still count as a zero. Non-excused absences will always count as a zero.) First Assignment: Reading (Review of Expected Values): pages 116120 (4.1.1 and 4.1.2 up through Example A) Reading: 3.13.4 Exercises from 3.8: 1(a), 3, 6, 7, 8(a,b), 9(a), 10(a,b), 12(a,b), 13, 14(a), 15(ad), 16, 17(a,b), 18(ac), 19 Homework exercises are not handed in. However, quiz problems will strongly resemble homework exercises (or lecture examples). Solutions to all (or most) of the exercises will be handed out to the class (in advance of the quiz covering that material).
3