EXST 7014 - Experimental Statistics II
Spring 1998
Course Description: EXST 7014 Experimental Statistics II (4)
F,S Prereq.: EXST 7004 or equivalent. 3 hrs. lecture; 2 hrs. lab.
Credit will be given for only one of the following: EXST 7013, 7014, 7015.
Multiple classification analysis of variance and covariance, individual
degrees of freedom, factorial arrangement of treatments, and multiple regression;
emphasis on science/laboratory research problems.
Course Prerequisites: EXST 7004 or equivalent.
Course Objectives:
General objectives include the following:
-
Provide the basic tools and building blocks for performing data analysis
using analysis of variance and multiple regression methods
-
Provide a foundation for further study in statistics including both theory
and methods
-
Stimulate each student's general interest in and an appreciation for statistics
and its application
Specifically, upon completion of the course, students should be able to:
-
Perform simple and multiple regression analysis, interpret basic and standard
regression output, and perform basic model diagnostics
-
Conduct basic analyses and diagnostics of and interpret the results from
analyses of completely randomized designs, randomized complete block designs,
latin-square designs, split-plot designs, and repeated measures designs,
and write the expected mean squares for the balanced designs described
above
-
Perform and interpret the results of a basic analysis of covariance and
multi-source regression
-
Perform basic ANOVA-like and regression analysis of count data
-
Understand descriptions of statistical analyses in subject-area journals
when ANOVA, multiple regression, analysis of covariance, or simple discrete
data models are applied
-
Consult with a statistician on designing experiments and analyzing data
Instructor: Dr. Barry Moser, 149A Ag. Administration Bldg, 388-8376,
barry@stat.lsu.edu.
Instructor Office Hours: TBA and by appointment (schedule with
e-mail when possible).
Lab Instructor: Brad Tiffee, Room 71A Ag. Administration Bldg,
388-8395, btiffee@stat.lsu.edu.
Lab Instructor Office Hours: TBA.
Required Text: Ott, R. L. 1993. An Introduction to Statistical
Methods and Data Analysis, 4th ed. Duxbury Press, Belmont, CA.
Assignments and Grading: Grades will be based upon 3 regular
exams, 2 laboratory practica (midterm and final), a weekly lab grade based
upon homeworks, and a final exam. The final exam will be COMPREHENSIVE.
Projects or outside work for additional credit are not permitted. You must
take the exams and practica and turn in homework when scheduled unless
you have received prior approval for alternative arrangements from the
course or lab instructor, as appropriate. A grade of "incomplete" can
not be given for unsatisfactory work or for failure to take exams or turn
in homework when scheduled. Each exam, practicum, and the overall homework
lab grade will be reported on a 100% basis. The Final Course Grade will
then be computed by applying the percentages below to each of the items
and summing them for the total score. Final letter grades will be assigned
using the scale below. The instructor reserves the right to modify this
scale by extending the lower limits on the minimum score required for a
grade.
Instrument Date Percentage Grade Total Score
---------------- ------- ---------- ----- -----------
Lab assignments Weekly 10% A >=90
Exam 1 Feb 19 15% B 80-89
Practicum 1 Mar 9 15% C 70-79
Exam 2 Mar 19 15% D 60-69
Exam 3 Apr 30 15% F <60
Practicum 2 May 4 15%
Final May 16 15%
---------------- ------- ----------
Total 100%
The Final Exam will be held on Saturday, May 16, 1998, from 7:30a-9:30a,
in our regular classroom.
Topics: The topics and coverage will follow the
text book closely beginning with chapter 8. Text book chapters and/or sections
are indicated within square brackets.
-
I. Review and Preparation
-
ON YOUR OWN: Basic probability; Z, t, chi-square, and F tests (1-sample
and 2-samples as appropriate) [1-7]
-
Linear combinations of random variables, expected values, variance, covariance,
and correlation
-
II. Goodness-of-fit and Contingency Tables
-
Goodness-of-fit measures (Pearson, Neyman, Likelihood Ratio Chisquares)
[8.1-8.3, 8.5-8.6]
-
2-way contingency tables (tests of homogeneity and independence), measures
of association [8.7-8.8]
-
III. Regression
-
Simple linear regression and correlation (review) [9, 10.1-10.5]
-
Matrix notation and some linear algebra [11.7]
-
The General Linear Model, curvilinear and polynomial regression, multiple
regression analysis [11.1-11.5]
-
Variable selection, regression diagnostics [12.1-12.4]
-
Multisource regression and dummy variables, general linear hypothesis test
[11.1, 11.6, 12.5]
-
Logistic regression on binary responses
-
IV. Experimental Design
-
Completely randomized design, factorial treatment arrangements [13, 15.5]
-
Contrasts and multiple comparisons procedures [14]
-
Randomized complete block design, blocks as random effects, other blocked
designs: latin square design and extensions [15.1-15.4, 15.6-15.7, 16.1-16.3]
-
Random and fixed effects and mixed models, nested errors: experimental,
sampling and sub-sampling errors and expected mean squares [17]
-
Repeated measures and split-plot designs, introduction to repeated measures
designs where randomization is restricted in space or time (covariance
structures) [17.6, 18.1-18.3]
-
Analysis of covariance and adjusted means [15.8, 19]
-
V. Analysis of Categorical (Count) Data (time permitting)
-
Logistic analysis with categorical predictors [8.7, 12.6]
-
Conditional independence and three-way contingency tables with the log-linear
model