Steven Goldberg’s “Sociology and Common Sense” (March 1991) contains some bits of wisdom, but its central premise is badly flawed. I first encountered the “Common-Sense Sociology Test” as a graduate student in the early 1960’s, and by then it was at least a decade or two old, so its ancestry is considerably older than Goldberg claims. Like Professor Goldberg, I have felt uncomfortable when reading many such quizzes, particularly those that appear in sociology texts. Yes, many of them are questionable or just plain wrong, and the examples he presents offer eloquent testimony to that fact.

One of the main problems with such quizzes is that they fail on the criterion of specificity: the generalizations they make are so broad the student senses that one question is being asked while, in fact, a very different one is addressed. To be effective, such questions must be worded in such a way that they are specific, concrete, and permit only one possible interpretation. In the second edition of my introductory sociology textbook (l988), I included just such a “common-sense” quiz. Are these statements true or false? The South has the highest homicide or murder rate of all regions in the country. (True.) Blacks have a higher suicide rate than whites. (False.) The higher the social class, the higher the likelihood that someone will drink alcohol. (True.) The divorced have a lower likelihood of getting married in the future than someone of the same age who has never been married. (False.) Here, there is no possibility of misinterpretation, as there is in Goldberg’s examples. And the answers to all the questions I asked about, again, unlike some of those Professor Goldberg cites, are solidly established empirically. The last time I gave such a quiz, the class as a whole got the right answer 48 percent of the time. Thus, while I agree with Goldberg’s point that many such quizzes ask poor questions, they can nevertheless be effectively worded and used as instruments to argue against relying on common sense.

However, my objection to Goldberg’s argument runs far deeper than this. His main point—that common sense is accurate with respect to observations but wrong with respect to explanations—is egregiously wrong, as my common-sense quiz shows. The fact is, many—most—undergraduates do not notice a great many things that take place around them; they have an extremely mistaken view of the size and shape of the social world. Their preconceptions lead them to use their observations to make erroneous generalizations about who does what, how, under what circumstances, and—as Goldberg agrees—why. Permit me to submit two exhibits in support of my position, both taken from social psychology, not sociology.

Exhibit A: Eyewitness Testimony. As research on eyewitness testimony has shown—overwhelmingly and convincingly—people are extremely poor observers of what takes place right in front of them. Moreover, individuals listening to eyewitness testimony have a great deal of confidence that what they are being told is accurate and valid. Again and again, we see that eyewitnesses mistakenly identify perpetrators, recall details of events that don’t exist, introduce nonexistent objects, forget crucial details, and are influenced by erroneous post-event cues. Research does not say that eyewitness testimony is sometimes wrong with respect to major details—it is usually wrong with respect to most major details. On the most basic level, contrary to Goldberg’s assertion, people are extremely poor observers; human observation is deeply flawed, unreliable, and often inaccurate.

Exhibit B: Judgmental Heuristics. Much of what people “know” to be true is wrong, fundamentally distorted by invalid inferential strategies, as an entire school of social psychologists has shown. The commonsensical techniques of reasoning of the so-called “common man” and “common woman” fail to yield correct answers when applied to realms other than the mundane, quotidian world. People tend to be far more influenced by what they are able to recall than what does not come readily to mind (the “availability” heuristic), and they are far more influenced by the features of things that resemble one another than those that don’t (the “representativeness” heuristic). Thus, students think that interracial violence is far more common than it is because it is often reported in the media, it is vivid, and it readily sticks in one’s mind; white students usually think that most welfare recipients are black—when, in reality, most are white—because of their profound ignorance of the difference between proportions and absolute numbers.

The fact is, most students—and the hypothetical man and woman on the street—are neither good observers nor accurate with respect to explanations. They often—and under certain circumstances, usually—make erroneous observations. They walk around believing that certain entirely mistaken generalizations are true. Many do not know the difference between an assertion and a fact. Or how to document an assertion or a generalization. Or how ideology may influence the observations that they and others make about the social world. Many cannot reach a reasonable conclusion even after they have been supplied with evidence bearing on a given question. The problem is far more profound than Professor Goldberg suggests: it is most people’s observations and explanations that need training. We do not have to have the observational acuity of an Erving Goffman to know that, for most people, being able to observe events accurately in the external world is a chancy business, one that would be improved by making use of a systematic discipline such as sociology or social psychology.

        —Erich Goode
Professor of Sociology
State University of New York
Stony Brook, NY

Mr. Goldberg Replies:

The purpose of the “Common-Sense Sociology Test” is to demonstrate that which cannot be demonstrated: that many of the observations made by students (and conclusions closely connected to observations) are faulty and that these can be corrected by an acceptance of “opposite” observations made by sociologists, “opposite” observations that are usually concordant with an ideology that is shared by the majority of academic sociologists.

To choose a typical example: the “correct” answer to the true/false question, “The income gap between male and female workers has narrowed in recent years,” is: “[It] has widened rather than narrowed; women hold few high-paying positions and the average working white woman earns less than the average working black man.”

Now, the freshman student, who is unfamiliar with the full meaning of the term “income gap,” clearly thinks that he is being asked whether a woman performing the same job as a man now receives a more-nearly equal income than she formerly would have. The student responds that a woman does, and this would have to be incorrect for the test to have its desired effect of making the student understand that one of his basic beliefs is incorrect. The test cannot do this, of course, because the student is correct. That the same forces that have made this true have also brought large numbers of women into the labor force at entry-level positions (thereby increasing the “income gap”) is interesting, but not in the least surprising to a student to whom is explained what the question and this term literally mean.

A lack of specificity is often the method that enables the test to seem to demonstrate that the student is incorrect. But it is the tendentious and false “answer”—and not a mere and easily corrected lack of specificity—that is the deeper problem. That this is the case is clear from the fact that neither Professor Goode nor anyone else can, by increasing the specificity of the questions on the test, enable the questions to demonstrate successfully to the student that his deeply held beliefs are incorrect. Specificity serves to demonstrate to the student that his beliefs are correct.

My general point is that the observations people repeatedly make over time are astonishingly correct and that the sociologist’s legitimate role is to explain that which is observed, not to pretend that the observation is incorrect.

Professor Goode attempts to refute this in three ways. First, he points out that eyewitness accounts are notoriously inaccurate. This is certainly as correct as it is irrelevant; I wrote not of observations made midst gunfire and mayhem, but ones made repeatedly and over time. Second, in his discussion of judgmental heuristics, Professor Goode gives only the examples of interracial violence and blacks on welfare to make his point. To be analogous with the basic beliefs considered by the test we discuss, it would have to be the case that students are incorrect in their beliefs that more blacks attack whites than vice versa and that disproportionate numbers of blacks are on welfare. The students are not, of course, incorrect. That they are incorrect about the absolute figures relevant here hardly casts doubt on any basic, observationally based beliefs.

Third, Professor Goode offers alternative test questions. The question about social class and drinking strikes me as unsurprising or surprising only in the illegitimate way in which the questions are presented on the Common-Sense Sociology Test. Read literally, the question is sufficiently specific, but I doubt that many students—picturing the cocktail party of the well-to-do—are particularly surprised that as high a percentage of the higher classes drink, if “drink” means “ever taking a drink.” I suspect that most students—who, after all, are freshmen on their first day of college—misread and interpreted the question as “members of the lower class drink more” (i.e., have a higher per capita rate of consumption and alcoholism). A colleague specializing in the study of alcoholism tells me that this question is complicated methodologically, but that a strong case can be made for a “false” answer.

Professor Goode’s other two questions have the virtue of specificity and clarity, but they are not in the least surprising—at least they were not to half of my students nor to half of Professor Goode’s (results that we would expect if the students have no strongly held beliefs about the issues). In contrast, the incorrect “answers” to the test we discussed were surprising to nearly all of my students, as I’m sure they would be to Professor Goode’s.

I would hope that we could make a test of questions that, on the first day of class, only half the students answered correctly. After all, the average student gets only about three-quarters of the questions right on the last day of class. I never implied that we don’t teach them anything, only that what we teach them has to do with the explanation of that which is observed, not with the incorrectness of their correct observations.