|
 |
James W. Pennebaker, Chairman | SEA 4.212 | The University of Texas at Austin | Austin, TX 78712 | 512-471-1157 |
Home | About Us | Contact Us | Search UT | Seay Building Info | Alumni |
Faculty DirectoryFaculty & TA Office HoursBooks by FacultyArea Chairs and FacultyFaculty Labs & Affiliated Organizations
|
Greg Hixon, Ph.D.
|
||||
 |
Email: hixon@psy.utexas.edu
|
I am a magna cum laude graduate of Case Western Reserve University, where in 1986 I received a B.A. with honors in economics. I received my Ph.D. from the University of Texas at Austin in 1991, concentrating in social psychology with a specialization (M.S. equivalent) in statistics.. After serving on the faculty at the University of Connecticut, I went on to work with several governmental agencies, culminating in my service as Director of Quantitative Analysis for the Texas Department of Human Services from 1999-2003 and then as Senior Research Analyst for the Texas Legislative Council from early 2003 until I returned to the University of Texas in 2004. I currently teach four Ph.D.-level classes in statistical analysis at the University of Texas covering everything from basic approaches like analysis of variance and linear regression to advanced techniques like multivariate non-parametric modeling, simulation methods, structural equations, and more. My research focuses on nonlinear dynamic analysis systems, with primary applications in the investment and financial markets. In addition to my duties at the university, I serve as a statistical consultant to individuals, corporations, and government agencies.
I use statistical analysis literally every working day of my life and I have a deep appreciation for the power of properly deployed analysis to extract knowledge from data. Whether your ultimate aim is a career in academic research, government service, private industry, or practically anywhere else, knowledge of modern statistical analysis techniques will serve you well. The more training you have in these techniques, the more informed your understanding of others' research will be and, perhaps more importantly, the better able you will be to extract reliable and valid findings from your own data. I have often said to the students in my classes that the last 20 years or so have belonged to the people who have provided us with such incredible technological advances in our ability to collect and process data, but the next 20 years will belong to the folks who know how to extract knowledge from all of that data. That is what statistical analysis is all about. I truly enjoy teaching, my enthusiasm for the subject matter is probably evident, and I think all of this comes across in my classes.
Thank you for your interest. Via this page you can get some more details in the form of descriptions of my classes and the answers to some frequently asked questions. But don't hesitate to get in touch if I can be of any help.
This course is a comprehensive introduction to regression methods. Among other topics, coverage includes background theory on the least squares criterion, single-predictor and multiple-predictor models including interaction terms, procedures for selecting the best from many candidate predictors, diagnostic procedures to check assumptions, coding schemes for qualitative predictors, polynomial expansions and other transformations to address model deficiencies and/or capture non-linear relationships, and logistic regression for binary outcome variables.
This course offers a concise introduction to a wide variety of statistical topics beyond basic ANOVA and linear regression. Underlying theory is addressed, but the emphasis is on application. Topics in this course include: 1) robust regression techniques such as least absolute deviation and maximum likelihood estimation with non-normal error structures for use with data containing outliers or characterized by non-normality; 2) nonlinear regression; 3) multivariate ANOVA (MANOVA) for the simultaneous assessment of multiple dependent measures on the same experimental units; 4) hierarchical linear modeling (HLM) for designs containing measures at many levels; 5) principal components and factor analysis techniques for assessing the structure of data.
This course covers common experimental designs and associated analysis of variance (ANOVA) procedures. Coverage includes single-factor and multiple-factor designs with interaction terms, repeated measures, blocking designs, ANCOVA, post hoc tests, and the general linear model approach to analysis of variance problems.
This course offers a concise introduction to a wide variety of statistical topics beyond basic ANOVA and linear regression. Underlying theory is addressed, but the emphasis is on application. Topics in this course include: 1) analysis of categorical outcomes via chi-square, loglinear analysis, multinomial logistic regression and ordinal regression; 2) non-parametric regression techniques such as loess and smoothing splines as well as multi-predictor extensions of those techniques via generalized additive models; 3) randomization and permutation methods for generating p values and conducting significance testing when parametric assumptions aren't met; 4) an introduction to structural equation methods and applications, and 5) time series analysis.
Q: What is your general approach to teaching statistics?
A: One thing that has changed over the last decade or so is the availability of classroom technology. I teach in classrooms with workstations at every desk, and I take full advantage of that. My classes are a balance of theory with application. I give you enough mathematical theory that you understand the various approaches, but I also emphasize hands-on experience executing analyses and interpreting the outputs from modern statistical software.
Q: What software package or packages do you use?
A: In my own personal work, listed from most-used to least, I use R (open source, free, and extremely powerful), Excel with a variety of specialized add-ons, SPSS, S-Plus, and a handful of specialized packages. For teaching, the classrooms I use currently have SPSS installed on every workstation (that might change to another package in the not-too-distant future) and we use SPSS for my foundational classes (ANOVA and regression). For my advanced classes, we use a combination of packages including SPSS, R, and Excel. Once you truly understand a particular type of analysis, though, you'll find that the specific software package on which you execute it is not terribly important - if you know how to do something via SPSS, it's pretty easy to do it in S-Plus or Stata for example.
Q: Are there any prerequisites for your classes?
A: Not officially. My foundational classes (the ANOVA class and the regression class) start at a pretty basic level, so if you're comfortable with math you could probably handle them with very little prior exposure to stats beyond what you probably got in a basic undergrad stats class (particularly if it was a good one and not so long ago that the basic concepts of statistical inference are still somewhat fresh in your mind or would come back to you relatively quickly). Many students, however, take a semester-long intro graduate-level stats class (covers a little of everything, from the basic concepts of statistical inference to correlation and regression, to t-tests and basic ANOVA) prior to taking my classes. Dr. Cormack in the psychology department has a very good class of that sort, and they are common in many departments throughout the university. If you have any doubts, get in touch and I'll give you an honest assessment of what I think would work best for you. All of that pertains to my foundation classes, ANOVA and regression. For the advanced applied courses, you definitely need to have had a good graduate-level regression class and a decent ANOVA class wouldn't hurt.
Q: Can I take your classes in any order?
A: Other than that you should take a good regression class (and a good ANOVA class wouldn't hurt) prior to taking either of my advanced applied classes, the order doesn't matter. You could take my ANOVA and regression classes in either order without any problem. The same goes for Advanced Applied Analysis and Advanced Applied Analysis II - they cover independent topics and can be taken in either order without any problem. (Advanced Applied Analysis II came into being a few years after the original, which explains the name. It's not an indication that these courses form any sort of a meaningful sequence.)
Q: There are semester-long classes in other departments for some of the topics you cover in your advanced applied classes. For those topics, is your class an effective substitute?
A: No, it isn't. In the two 3-hour sessions I might devote to, say, HLM or to principal components and factor analysis, I can't possibly cover everything that would be covered in a semester-long course. But what I aim to do is to provide enough coverage that you can build on it for yourself as needed, either by pursuing one of those other courses or through self-education. I try to provide the basic mathematical theory necessary to understand the analysis procedure, coverage of the main issues involved in practical application, and experience using software to perform the analysis. With those building blocks you'll at least be a competent user of the technique. Moreover, you'll know enough to assess whether it is a technique that you'll likely use extensively in your own research and if so you can pursue one of those semester-long courses (and have a good head start in it) or have a sufficient foundation that you could pick up any good book on the technique and build on your knowledge via that route.
Q: I'm not a grad student in the psychology department. Can I register for your class?
A: I'm employed by the psychology department and my classes are intended primarily for that audience, so naturally students in this department have priority (and they're the only ones who can register without my permission). If you're not in the psychology department, you'll need my permission to register. So if you're interested, send me an email, tell me which course you're interested in registering for, let me know what department you're in and what stat courses or other math background you've got, and we'll take it from there. My courses seem to be fairly popular so there are no guarantees, but in most semesters there ultimately winds up being space available and if that's the case I'd be glad to have you in my class. Note too that while many of the examples I use are from psychology and other social sciences, not all are -- and in any case the examples generalize easily. I'm happy to say that I have had grad students from engineering, the business school, physical science departments, and other social science departments in my classes. The vast majority have done well and generally say that the experience was of significant benefit to them.
Q: Why are three of your classes called PSY394T?
A: It does seem odd, doesn't it? The answer is that I don't know. I asked one of our friendly administrative staff that question once, got what at the time sounded like a reasonably sensible answer, promptly forgot the substance of that answer, and now I'm too embarrassed to ask again. Fortunately, every course has a unique number that is for all intents and purposes unique. Especially if you're registering for one of my fall semester classes, make sure you get the correct unique number.
Q: Are you the same faculty member whose name shows up as John Hixon or J. Hixon in various places here at UT?
A: Yes, I am. I go by Greg, but that’s my middle name. My first name is John and despite my attempts to ensure that I appear consistently everywhere as Greg, I’ve learned that it’s difficult to succeed entirely in that regard. The last time I checked, I was the only Hixon on the faculty at UT Austin, so if you see a Hixon – particularly in association with a statistics course – you can be reasonably sure that it’s me.
