Richard Gonzalez
Fall, 2003
3243 East Hall; 647-6785
e-mail: gonzo@umich.edu 
The class website:  www.lsa.umich.edu/psych/courses/psyc613

1   Psychology 613

"Go on, Mr. Pratt," says Mrs. Sampson. "Them ideas is so original and soothing. I think statistics are just as lovely as they can be."
O. Henry, The Handbook of Hymen

1.1  Course Description

This is the first course in a two-semester sequence on data analysis. I present the "general linear model" with particular emphasis on exploratory data analysis, contrast analysis, residual analysis, and Euclidean distance. The topics covered over the two semesters include analysis of variance, regression, categorical data analysis, principal components analysis, multidimensional scaling, cluster analysis, multivariate ANOVA, canonical correlation, and structural equations modelling.
The presentation of the material will be intuitive. The goal of the course is to give you tools that you can use in your research. Issues of experimental design will be highlighted throughout the course. Whenever possible I will convey an appreciation for the "art" of data analysis.

1.2  Intellectual Bias of Instructor

The goal of science as science is not prediction and control but understanding. Prediction is merely the test of understanding and control the practical reward. (Macleod, 1947, p. 209)
The purpose of statistics in psychological research is typically misunderstood. Researchers sometimes judge an experiment by the p-value alone. A p-value, important in getting past a journal editor, is not the primary goal of an experiment. Data analysis is an exploratory process where the end product is a description of what happened in the experiment or study. In this course we will learn simple procedures to uncover what the "data are telling us" and techniques to test research hypotheses. The bottom line: understanding comes from how you describe your data-the p-value merely serves to punctuate a sentence.
I believe statistical training should continue throughout one's research career-a two semester graduate sequence is merely the beginning. I teach from the perspective of the general linear model. This provides a broad framework from which the more advanced techniques become simple extensions. Thus, we will work hard to setup the proper foundation in this course.

1.3  Use of a Statistics Package

The computation of a multiple regression or an analysis of variance can be ugly if done by hand. I will not explain the computational details, except when computation aids intuitive understanding.
All problem sets assigned in this course will require the use of a statistics package. I don't care which package you use to complete the problem sets. I will use SPSS throughout the course to standardize the examples. You should use the package that is most often used in your research lab.

1.4  Texts

The required texts for Psychology 613 are:
Designing Experiments and Analyzing Data by Maxwell and Delaney
Applied Linear Statistical Models by Neter, Kutner, Nachtsheim, & Wasserman
The book by Maxwell and Delaney (MD) is a survey of ANOVA designs; the book by Neter, Wasserman and Kutner (NKNW) is a survey of regression, ANOVA, and the connection between the two. Both are excellent reference books. However, I'm am willing to let students use other books as well (such as Keppel, Kirk, Hays, etc.). Feel free to consult with me. For purposes of standardization, I will key my lecture notes to relevant chapters in the MD and NKNW books, and use their notation.
I have also ordered a few other books, all of which are optional. The optional books I ordered include: the SPSS manual, the APA publication guide for learning how to write results sections, and some additional books dealing with advanced topics. As the term progresses, some or all of these books may become appealing to you.
My lecture notes will be available as a course pack at the beginning of the term. These lecture notes consist of about 500 single-spaced pages of conceptual motivation, examples, intuitions, graphs, and annotated output.
Texts for Psych 614 will be reviewed next semester.

1.5  Problem Sets and Exams

I have very high expectations. My job is to train you to be professional researchers. Data analysis is a necessary component of a successful research project. It is a skill, and like any other skill (e.g., learning to play the piano, learning to play tennis) it requires a lot of work to achieve mastery at a "professional" level (to carry this analogy further, playing the piano at the level of a studio musician). So, this course is necessarily time consuming. There isn't any way around that. I put in a lot of time/effort into this course, and I expect that you do the same.
There will be seven or eight problem sets spaced approximately two weeks apart. Due dates will be announced in class. Anticipate spending about 8-12 hours per problem set. More details will be given in class. I have tried to trim away all busy work from the problem sets; they focus on building skills and highlighting subtle distinctions in the concepts we will cover.
There will be one midterm and a final exam. The exams will not involve the use of the computer, but sections of the exams will require interpreting computer output. Both the midterm and the final will be two and a half hour exams. The midterm will occur sometime during the middle of the term. The final will cover material presented after the midterm (i.e., the final is not comprehensive, though there will be some conceptual overlap). Both exams will be open book.

1.6  Grading

The final grade will be a weighted average of the problem sets (50%), midterm (25%), and final (25%). Note the relatively high emphasis on the problem sets.

1.7  GSIs

The GSIs will hold office hours (tba) and, as needed, review sessions such as elementary SPSS operations.

1.8  Prerequisites

An introductory statistics course is a must. You should review the material on hypothesis testing, measures of location (aka central tendency), measures of spread (aka variability), and exploratory data analysis (such as boxplots). If you have not had an introductory course in statistics, or are very rusty, then we should talk during the first two days of the term about what you can do to prepare. The pace in this course will be very quick so falling behind even a couple of days will be stressful.
Use of a statistical package such as SPSS is necessary in order to complete the problem sets. SPSS is available through ITD subscription as well as other locations throughout campus (e.g., B250 East Hall, Angel Hall). SPSS is installed on all machines in East Hall that run Windows NT, and is installed on most Macs. You are free to select any statistics package you wish.

1.9  Class E-mail List & Web Site

We will make heavy use of the class e-mail group (psy613@umich.edu) to post announcements, hints, typo corrections, etc. 
The class website: 
http://www.lsa.umich.edu/psych/courses/psyc613

2  Course Outline

Here is an outline of the material that will be covered in this course. In the past students who claim solid knowledge of ANOVA and/or regression still report learning quite a bit from the course. I present material in a way that will give you new intuitions about what ANOVA and regression do and how these techniques can be used effectively in research settings. I'll be happy to discuss the appropriateness of the course with you.
Topics Readings
I. Quick Review of Concepts from Introductory Statistics
      -central tendency, variability, hypothesis testing,
       t tests, etc. MD chs 1/2; NKNW ch 1/app
II.Comparing Groups on One Factor
      -one way ANOVA (fixed & random); MD ch 3
       examining assumptions
      -multiple comparisons MD chs 4, 5, & 6
III.Comparing Groups on Multiple Factors
      -randomized block designs
      -latin square designs
      -factorial designs, k-way ANOVA MD chs 7 & 8
IV. Advanced ANOVA Designs
      -random & nested effects MD ch 10
      -repeated measures MD chs 11, 12, 13 & 14
V.Correlation and Regression
      -straight-line regression and transformations NKNW chs 2, 3, & 5
      -examining residuals and diagnostics NKNW ch 4
VI.General Linear Model
      -multiple regression NKNW chs 6, 7 & 8
      -reformulation of ANOVA NKNW chs 10 & 14
      -polynomial regression NKNW ch 9
      -analysis of covariance NKNW ch 23; MD ch 9
      -time series analyses NKNW ch 13
VII.Analysis of Categorical Data
      -binomial and contingency tables handout
      -logistic regression NKNW ch 10


File translated from TEX by TTH, version 3.33.
On 19 Aug 2003, 22:52.