This semester, I am teaching a graduate course on structural equation modeling (SEM), an advanced statistical application. This is my third time teaching the course at Texas Tech University and my fourth time overall, as I taught it in 1996 at the University at Buffalo (State University of New York), while I was a researcher at the Research Institute on Addictions.

In developing my lectures, I draw heavily from notes I took myself while taking SEM at Michigan in the Winter 1988 semester from Frank Andrews and Laura Klem. I also took multivariate analysis from Frank and Laura in Fall 1985. I save a lot of stuff, so I have syllabi, notebooks, an extensive set of handouts, and my graded assignments from these two courses.

The classes met twice a week. The way Frank and Laura co-taught their courses was that Frank would lecture one day on the substantive aspects of the relevant statistical technique, and then Laura would teach us how to implement the technique on the computer.

Frank was the clearest, most enthusiastic lecturer one would ever want to hear. In addition to the lecture itself, Frank would hand out extensive bibliographies on the topic, as well as sheets with examples of the technique that had been worked out.

I can't help but think that teaching computer applications of statistical programs must have been so much harder and more time-consuming in the 80s than is currently the case. Back then, data sets were stored on magnetic tape and implementing the commands just to access the data you wanted to work on was a not insubstantial task. Further, we would have to hand type our statistical commands, which usually were quite cryptic. In more recent years, students, faculty members, and other data analysts have had access to much more user-friendly programs that give you an actual spread sheet of your data right on your screen and allow you to select commands from menus. Also, data sets can be transmitted via the Internet and stored/carried on disk.

Despite the seemingly more cumbersome nature of computer data analyses 15-20 years ago compared to now, Laura managed to pull it off. Further, she would use her encyclopedic knowledge of the statistical/data analytic literature in grading our papers. On one paper I just pulled out, for example, she directed me to "See Kenny page 143..."

Frank Andrews died in 1992. To honor Frank, the UM's Survey Research Center created a fellowship to help people attend the Summer Institute. A brief summary of Frank's career is available at this link.

Laura Klem is still around at UM, as seen in this webpage, as a Senior Research Associate with the Center for Statistical Consultation and Research. In fact, she will be teaching a short course on applied structural equation modeling this upcoming May 17 - 20, 2004.

The aforementioned course announcement website notes that, "Enrollees will receive substantial handouts." I couldn't imagine it any other way!