| The Basics | |||||||||||||||||||||||||||||||
| Instructor: | Brian Habing, Assistant Professor office: 420G LeConte e-mail: habing@stat.sc.edu office phone: 777-3578 home phone: 739-2686 (9am to 10pm only) | ||||||||||||||||||||||||||||||
| Course Website: | http://www.stat.sc.edu/~habing/courses/518F99.html | ||||||||||||||||||||||||||||||
| Office Hours: |
Whenever the door is open, by appointment, and Monday/Wednesday 3:00 - 4:30 Wednesday/Friday 10:00 - 12:00 | ||||||||||||||||||||||||||||||
| Prerequisite: | A grade of C or higher in Statistics 515 (Statistical Methods I) or equivalent | ||||||||||||||||||||||||||||||
| Credit: | 3 hours (undergraduate or graduate). | ||||||||||||||||||||||||||||||
| Description: | Application of nonparametric statistical methods rather than mathematical development. Levels of measurement, comparisons of two independent populations, comparisons of two dependent populations, test of fit, nonparametric analysis of variance, and correlation. | ||||||||||||||||||||||||||||||
| Purpose of Course: | To introduce the principles and applications of commonly used nonparametric methods. To compare these methods to their parametric counterparts through simulation studies. To introduce the basic methods for analyzing contingency tables. | ||||||||||||||||||||||||||||||
| Text: | Practical Nonparametric Statistics, (3rd Edition) by W.J. Conover, Wiley, 1999. | ||||||||||||||||||||||||||||||
| The Details | |||||||||||||||||||||||||||||||
| Computers: | This course will use the software packages SAS and S-Plus. NO PREVIOUS KNOWLEDGE OF S-Plus OR SAS IS ASSUMED. If you do not currently have access to this package, you will be given accounts on the department of statistics UNIX system and the college of science and math PC network. Computers are available in 124 and 303A LeConte. | ||||||||||||||||||||||||||||||
| Homework: |
Homework is due at the beginning of the class period it was
assigned for. Late homework is not accepted. Homework will be assigned at least three class periods in advance in class, and will also be posted on the class website later in the day on which it was assigned. If the homework is on a handout, that handout will be available in class and during office hours. There will be 8 homework assignments, the 6 highest scores will count toward the final grade. The homework will be scored out of 8 points. THE HOMEWORK WILL NOT BE GRADED UNLESS:
You MAY work on the homework assignments with other students, but each student must write it up individually. (i.e., No photocopies of other students work.) Any questions about the grading of a problem on the homework must be made by the class period following the one in which it was returned. | ||||||||||||||||||||||||||||||
| Project: | An individual final project will be due by 4:30 pm on Tuesday, November 30th. The project will be a combination parametric data analysis, nonparametric data analysis, and simulation study comparing the two. The full details on the project will be announced on September 28th (the class period following the first exam.) | ||||||||||||||||||||||||||||||
| Exams and Topics Covered: | There are three exams, two
hour exams and the
final exam. The topics covered in the exams will generally follow the chapters
of the text listed below. However,
the exams may also cover material which was solely
presented in class, and that is not contained in the text. The final exam
is not cumulative and counts the same as the other exams.
The first exam is in class on Thursday, September 23rd. It will focus on the subjects related to chapters 1 through 3 of the text, including: probability and counting rules, discrete and continuous random variables, properties of estimators and hypothesis tests, the binomial and quantile test, the sign test, and McNemar's test. The second exam is in class on Thursday, November 4th. It will focus on the material related to chapters 5 and 6 of the text, including: the Mann-Whitney test, the Kruskal-Wallis test, the squared rank test, measures of rank correlation, nonparametric linear regression, the Wilcoxon signed ranks test, the Friedman test, and the Kolmogorov-Smirnov-test. The final exam will be held at 9:00am on Thursday, December 9th. It will focus on the material in chapter 4, including: chi-squared tests, the Mantel-Haenszel Test, Cochran's test for related observations, measures of dependence, and loglinear models. Make up exams will be given only in extreme circumstances, and only when accompanied by appropriate documentation. Incidence of cheating and academic dishonesty will be punished to the full extent allowed by university regulations. Any questions concerning a grade received on an exam must be raised by the next class period after it was returned. While it will generally not be necessary to "curve" or rescale the grades on the hour exams or the final, this may be done at the instructors discretion by setting 100% to be 100%, the lowest A to be 90%, the lowest D to be 60%, and 0% to be 0%, and then linearly interpolating to calculate the scores and rounding up to the next half point. In all cases the rescaled scores shall not be lower than the original scores, and the scaling shall be announced when the graded exams are returned. Any student scoring under 59 on either of the first two exams will be given a chance to raise their score. A five problem, five points per problem, no partial credit, exam will be given the following week. No score shall be raised higher than 59. | ||||||||||||||||||||||||||||||
| Grades: | The grade is determined by the scores on the
homework, project, and the three examinations as follows:
with the letter grade determined by the percentage of points obtained out of the maximum possible 448, rounding up.
There is no "extra credit". | ||||||||||||||||||||||||||||||
| Comments and Complaints: | While there are end of semester evaluation forms, they come far too late to resolve any difficulties experienced in the class. All complaints should be raised by either speaking with me directly, or by anonymously leaving a message in my mailbox in 216 LeConte. |