Virginia Commonwealth University

Department of Sociology
Department of Statistical Sciences and Operations Research

An e-mail address is required of all students. Send me an email message right now even if you think I know your address.If you did not take 508 this fall, include a summary of your previous work in statistics, research and computer use as well as your initial reactions to the textbook.

Additional information and handouts will be posted at the VCU Blackboard Site

 PREREQUISITE:

• Simple random sampling
• The use of the normal probability distribution
• The distribution of  means and  proportions from simple reandom samples
• Confidence intervals and P-values for means and proportions
• Crosstabulation and the chi-square test of independence

STATISTICAL METHODS FOR THE SOCIAL SCIENCES, by Alan Agresti and Barbara Finlay, 3rd edition, 1997. Prentice-Hall. At the VCU Bookstore. This semester we will cover Chapters 7-14, with a nod or two toward 15 and 16. If you have not used this text before, review Chapters 1-6 immediately, with special attention to 5 and 6. Chapter 8 is somewhat of an anomaly, since it deals with categorical variables and everything else involves numerical variables until Chapter 15. The material in Chapter 8 on crosstabulation and chi-square tests was discussed in STAT/SOCY 508. I will deal with this chapter in an indirect manner, citing specific sections as appropriate in conjunction with other topics.

COMPUTING (SPSS):

The class will use the Windows version of SPSS. The current version is 12.0. VCU has a site license that allows us to distribute SPSS for Windows to all students, faculty and staff.  For a nominal fee of $15 you can get a CD-ROM that contains the same version as that in the classroom at the VCU ONLINE stores in the Student Commons and the MCV Campus Bookstore. This is a really good deal! If you have access to a personal computer that is powerful enough to handle the program you should take advantage of it immediately.  SPSS says that minimum requirements are Microsoft® Windows® 95, 98 or NT® 4.0, 68MB available hard drive space, and 64MB RAM. Note: Version 12 has a few nice improvements over Version 11.5, but for the purposes of this class the earlier version will be fine. Version 12 may not be available in the Hibbs computer labs until midway in the semester.

If you do not have a PC that you can install SPSS on, you can use a VCU computer.  Most computers in the public access computer  labs at VCU have SPSS installed, as do all the computers in the open lab in Hibbs. Your home school or department may also have a computer lab with SPSS available to graduate students. Check it out!

Prentice-Hall publishes a lot of SPSS books. The most widely known is probably the  Guide to Data Analysis by M. Norusis.   It first appeared some 20 years ago, and is a handy reference text to SPSS usage and traditional statistical methods. It is outrageously expensive, however, and a used copy of earlier versions of the book will serve almost as well. You do not really need books like this since the SPSS package includes very detailed help pages.

RECOMMENDED READING:

• Statistics as Principled Argument. Robert P. Abelson, 1995.
• Tricks of the Trade: How to think about your research while you're doing it.  Howard S. Becker, 1998.
• Medical Uses of Statistics, 2nd ed., John C. Bailar and Frederick Mosteller, 1992
• Statistics in Britain 1865-1930: The social construction of knowledge.  Donald A. Mackenzie, 1981.
•  The Cartoon Guide to Statistics, by Larry Gonick and Woollcott Smith (HarperPerennial 1993) is more accurate statistically and definitely funnier than many introductory textbooks!
• Visual Revelations: Graphical Tales of Fate and Deception from Napoleon Bonaparte to Ross Perot. Howard Wainer, 1997, 2nd ed. 2000.  There is a copy of the 1997 edition in Cabell Library.

VIDEO, CD-ROM, MULTIMEDIA, and INTERNET RESOURCES:

It seems as though new resources for statistics are being created every day and distributed freely on the Internet. One useful site is at The Claremont Colleges' web interface for statistical education. This site includes tutorials as well as links to other resources such as the Chance Project at Dartmouth College (where I first studied probability theory and stochastic processes). I particularly like the interactive graphics that demonstrate how the Central Limit Theorem works: try it now.

At the University of Minnesota   an NSF sponsored project has created a website "to develop and disseminate materials that help students learn core concepts underlying statistical inference".

If you find other interesting websites write a review for me and I'll post it on the Blackboard Course site.

GRADING:

I will recommend problems from each chapter of A&F. I recommend that you turn in some of them in to get feedback from me about your writing as well as about your understanding of the material. These will not be graded, and group discussion is encouraged. Short, primarily numerical, answers to many odd-numbered problems are included in the text. These, however, should not be considered adequate answers to the questions! Usually more explanation and attention to how the answers are derived and justified is necessary. Understanding the meaning of numbers and communicating that meaning to others is a primary goal of this course.

I also recommend that you keep a journal. The journal should contain your reaction to the text and to outside readings, as well as to the lectures and exercises. You will get more out of the class if  you are alert to how its topics show up in your life, both professional and personal. It turns out to be very useful to write about statistical ideas and problems, not simply to ask a question and hear a reply. You should  review and revise journal entries every now and then.
 Back to Top

 Back to Top

Comments:

A remarkable presentation of the roots of modern statistics.

Although MacKenzie writes as a sociologist of science, his evaluation of the work of Francis Galton, Karl Pearson and R.A. Fisher provides an excellent summary of their contributions to the development of mathematical statistics as we know it today. His sociological thesis is that the primary concepts of statistics were "invented", not "discovered", and that the social position of  these men strongly influenced the content of their works. While this not a radical position today (at least in sociological circles) it was, in 1987, relatively unfamiliar to many statisticians.

All three of the major figures studied were connected to the eugenics movement, and MacKenzie examines the relationship of eugenics and biometry to their work in mathematical statistics. He shows how Fisher's Genetical Theory of Natural Selection evidences the eugenics goals which are usually associated with the older Pearson. While it is a bit trickier to connect Fisher's books on Scientific Inference and Statistical Methods to eugenics, MacKenzie is quite convincing.

I regret not having read this book when it was first published, since it clarified several issues that I have been struggling with as a teacher of statistics. I looked it up as a result of a reference in an article by John Aldrich in the journal Statistical Science (Vol. 10, No. 4, 364-376) on Karl Pearson and G. U. Yule's views of spurious correlation. Statistical Science, by the way, often contains interesting articles on the history of statistics. Issues over 5 years old are available online through VCU Libraries' contract with JSTOR.

Abelson on Statistics as Principled Argument .

Abelson is a mathematical psychologist and statistician who, after 40 years of teaching statistics at Yale, has written a book for his  students to read after they have taken the "normal" statistical curriculum. His alternative title is "Lots of things you ought to know about statistics but are too stupefied to ask."He reflects upon the way traditional statistics instruction (in psychology, but also in many other disciplines) fails to communicate the ways in which statistical methodology should serve rhetorical and narrative functions. Assigning it to you while you are taking this course is too subversive a thing to do, so I won't, but I will secretly quote some of his insights now and then. Some of you will enjoy browsing in it.

Becker on Social Research

Since Abelson had been one of the first mathematical social psychologists I read in the early ‘60s, I was ready to be impressed by his book, and I was. My latest award for most interesting reflections by a fellow geezer goes to Howard S. Becker, a sociologist whose 40 year career parallels Abelson’s in curious ways. Becker, known as a qualitative researcher, has written on subjects as apparently diverse as medical education, marijuana use, and the art world. In this book he explains what it means to have a sociological perspective on the world, and why social research is possible. Along the way he talks about such statistically related topics as random sampling (and its alternatives), models, causal inference and quotes admiringly several authorities not usually associated with qualitative social research such as John Tukey and Paul Lazarsfeld.
 

Bailar and Mosteller on Medical Statistics.

This book was built around a series of articles that was published in the New England Journal of Medicine in the early '80s. While the chapters are directed at medical researchers who want to publish in NEJM and similar journals, many of them are good reading for educational researchers and other social scientists. I especially recommend Chapter 20, "Writing about Numbers", by Mosteller; and Chapter 10, "P Values". Chapters 2 and 16 are also useful. There is a copy of the second edition in Tompkins-McCaw Library, and Amazon.com lists it for $65. But it is out of print. You might be able to find a used copy somewhere.

Incidentally, Mosteller has just co-edited a book on experimental studies in education: Evidence Matters : Randomized Trials in Education Research . I've only read Chapter 1, which is freely available at The Brroking Institution website. Looks good, especially if you are specializing in educational policy.
 

Handouts for the Class

• Notes on Models for Data
• R-square and Standardization in Regression
• Notes on Subgroup Regression
• Pride in America (ANOVA and Regression)
•  

If you have comments or questions, email me at nhenry@vcu.edu

Updated December 2, 2003