Mirror Sites:
CSUDH - Habermas - UWP - Archives

California State University, Dominguez Hills
University of Wisconsin, Parkside
Soka University Japan - Transcend Art and Peace
Created: June 28, 2003
Latest Update: June 29, 2003
jeannecurran@habermas.org
takata@uwp.edu

Jeanne's Notes on Introduction to Everyday Statistical Reasoning

Site Copyright: Jeanne Curran and Susan R. Takata and Individual Authors, June 2003.
"Fair use" encouraged.

Timothy J. Lawson's Everyday Statistical Reasoning: Possibilities and Pitfalls, 2002. Wadsworth and Thomas Learning, is one of the texts for our Statistics course this Fall. I chose it particularly because the availability in almost every workplace of some form of computer-assisted statistical analysis means that there is no longer a trenchant need for you to learn to calculate statistical formulae. There is, however, a very real need for you to understand the difference between personal experience and statistics as presented in articles, news reports, and legislation on which you're going to vote. Our course will be focused on the understanding and interpretation of statistics in our lived experiences. This emphasis should help to relieve some of your math anxieties.

First of all, let me share my biases about the book. It came to my attention through Diana Rothberg, a Wadsworth representative in our area. Diana listened to me, really listened, when I complained about the nonsense of teaching students to calculate formulas. And she suggested Lawson's book. I was delighted. It's small, and can't be too expensive, I hope. That matters to me, because I want you to spend the money you can afford on books that will be meaningful to you in your lived experience, not just in the classroom.

I liked that the chapters were short, and I could read quickly through the whole book. That means that you'll be able to read the material fairly easily within your fifteen-minutes parcels of study time whenever you can catch them. Reviewers spoke of the book's "readability," and I agree, up to a point. Lawson does speak fairly straight-forward English, and does a reasonably good job of explaining some things. But he's a psychologist, and he teaches statistics. So he and I part ways very early, when his explanations bog down in three major areas:

• He leans heavily on quantitative analysis. He seems to rely on the fact that you should have learned formulas at some point.

  I don't believe that. I lean much more heavily on qualitative analysis and on critical interpretation. The use of quantitative data, or data which lend themselves to ready counting or measurement are often categorized for convenience in measurement. Some of that is because we, as humans, tend to need to categorize in order to bring structure out of chaos. We are also just at the rather ambiguous end of the positivist period, in which most of us were sure that the quantitative measurement of scientific methodology would answer all the big questions. That hasn't happened. We're discovering that the world is far more complex than we had imagined, and that reducing our studies to easy-to-measure categories has missed important aspects of understanding.

  Qualitative analysis forces us to take more complexity into account: grounded theory, for example. Lawson deals with this to some extent in his chapters on analysis of covariance, or of multiple variables, but he shifts his reasoning back into mathematical language, which I'm afraid will lose most of us. My concern with much of this quantitative emphasis is that we are often not in a position in the social sciences to get data that are worth such intense analysis. Unlike psychologists, who often use their students as subjects, sociologists tend to pay more attention to the context, and the context of classes of sophomores bears little resemblence to the context of lived experience for most of us. This leads me into my second concern:

• He leans heavily on many very small studies, almost all of which are based on the study of college sophomores.

  This book was written to supplement regular statistics texts and courses. Lawson did not intend it as an alternative, but as a supplement. Therefore it's geared to texts that he must use in his psych classes. As a sociologist, that disturbs me. I want to supplement a much broadrer universe of ideas. And I want to use the text as your primary reading. That's because we're in an urban college where few of you are going to go on for a doctorate in sociology or any of the other behavior sciences, and most of you are pleased when you can find 15 quiet minutes for deep concentration.

  Most of you are never going to have to conduct these little studies that would get you promoted in assistant professor jobs. Most of you are going to work in some corporate capacity at a job that will hopefully permit you to help people. So the skills you need to develop are the very skills that Lawson is talking about, the skills of reasoning critically with statistical principles. That involves interpretation. Good qualitative analysis of measures can make quantitative data more interesting, and more likely to be useful. So we're going to focus on interpretation.

  Oddly enough, Lawson's use of all these little studies based on students' behavior are good practice for that kind of interpretation. The only problem is that they made me feel like I was reading Jones and Gerard on Foundations of Social Psychology again. And that's not an easy read. This is the kind of review text that adds up all the little studies published in myriad journals and given as myriad presentations at professional groups to give you an overview of the field. Good reference work, not so readable.

  But the book is short; the chapters are short, unlike the Jones and Gerard text. I think you can manage this aspect of the text. Just remember to look at all these studies as examples that you might have tried on your fellow students, and reason them through for yourself.

• He tries to use statistical terms in a non-mathematical context. It doesn't work.

  This starts in Chapter 5 on Correlation. He introduces a whole section on Z scores, without an adequate explanation of why you would want to use Z scores. In my whole life, the only people I've ever known who used Z scores were teachers. Now that's personal experience, and so I mustn't generalize to the fact that probably no one uses Z scores out there in the world. But they're not the major focus of our course.

  He does this again with regression in the next chapter. I'll supplement these sections of the test. We may use just SPSS, the statistical program, to explain these because I want to spend most of our time on interpretation. At any rate, I don't want you to feel overwhelmed if these sections and the sections on ANOVA and ANCOVA confuse you. They're actually quite simple, and we'll go over them together.

  Also, upon rereading Lawson's ANOVA and ANCOVA sections, I realize that he does a pretty good job of taking the social situatedness into account and he doesn't really use formulas. I think it's his tables that confused me. What he actually does is to walk through the table verbally, and I think the problem is that he needs to redesign the table. The verbal description is pretty good, it's just wordy. That's why mathematicians came up with math in the first place, to make things clearer than one could by verbal description. I'll try to upload my lectures before you read the chapter. I think that might help.

  Lawson concludes with Bayesian probability. That's the first time I've come across that in an intro stat course. But it's a very useful concept, and I think it will make sense to you. Never mind whether you learn how to calculate the probability. Just learn that sometimes the probability depends on something that has already happened. Like Let's Make a Deal and choosing between Doors 1 and 2 is different once you've been shown that the prize is not behind Door 3 than the 50% chance you would have had if you just had those 2 doors to start with. I care less that you can calculate Bayesian probability than that you remember the Let's Make a Deal example and how the con man profits from it in the shell game.

  I think Lawson is kind of unrealistic in assuming that people, even professional people, resort to such recall of their statistical learning in real life. He does cite a study that says you can teach students to think that way. (Ibid., at p. 109.) But I'll bet that if I tried to get you to think statistically about many of these problems, you'd throw your hands up in frustration. What I'd like to teach you is an awareness of the basic rules of analysis, practiced to a level where you just kind of automatically run through them as a check. Trust me, that's a skill you need.

I hope these notes will help you approach the text comfortably. This is a course in which you won't be able to answer most of our questions off the wall. You need to read my lectures, and to read the text. And you'll need to attend lab sections to run SPSS. Understanding statistical interpretations and minimal ability at computer literacy in statistical analysis are useful skills in the real world.