To the Editor:
Walter Berns’s review of The Responsible Electorate [August], the final work of V. O. Key, deserved more space and publicity. Many of your readers have not had the opportunity to hear solid criticisms of behavioral social science, at least from writers as able as Berns. COMMENTARY, as the most able and responsible forum for American liberal opinion, has the continuing obligation to encourage the exploration of the damaging dimensions of the “new orthodoxy” in social science.
Allan A. Spitz
Department of Political Science
Washington State University
To the Editor:
Walter Berns’s review of V. O. Key’s last work is needlessly confusing. Berns seems to feel that he must defend Key’s book as if it contradicted earlier voting studies, but I do not see that it does. . . .
As Berns states, the voting studies conducted by Lazarsfeld, Berelson, and others are concerned with correlations. These correlations may be stated in terms of “laws of behavior.” But the point must be made that correlation simply implies that when X happens, Y happens (or when X is the case, Y is the case). Correlation does not mean cause. The statement that “wealthy Protestant farmers vote Republican” does not tell us what causes that behavior. Key indicates that rational judgment of issues and candidates is the cause. These items of information are obviously not incompatible. Secondly, when behavioral scientists speak of “laws of behavior” they are, in the vast majority of cases, speaking of statistical laws. These laws are primarily based on correlations of man’s social milieu and his behavior, and they are useful in explanations or predictions since they are true in more than 50 per cent of the cases to which the law applies. They are not expected to be true in 100 per cent of the cases, as would be so if man’s rationality were the same as that of dogs or ants. The discovery of laws in no way indicates that man is not a rational creature.
Berns writes that the origins of party loyalty are not sociological or psychological, but political. I fail to see the distinction. Generally speaking, sociology is the study of human behavior in groups. Psychology is the study of individual human behavior. Political behavior can be a subdivision of one of these, but surely not a third category.
It is admitted that errors were made in the earliest voting studies that Berns uses as scapegoats. However, much improvement is to be found in more recent works such as those by Angus Campbell and Bernard Berelson’s latest book (written in conjunction with Gary Steiner), Human Behavior. I do not think contradictions would be found between these and Key’s book. Key’s book is, in fact, a useful contribution to the behavioral sciences.
[Dr.] Susan G. Clark
Fort Rucker, Alabama
To the Editor:
Scholars, sad to say, are as inclined as anyone else to paint things black and white when the true color is gray, perhaps, or checkered. Thus, Walter Berns, in his review of V. O. Key’s book, The Responsible Electorate, tells us that Key’s analysis of American voting behavior is good, and previous studies by social scientists are bad, because Key found voters rational whereas earlier scholars found them non-rational.
Professor Berns’s (and Key’s) main targets are sociologists who explain voting as the outcome of social pressures from voters’ friends and relatives . . . and psychologists who see unconscious personality needs lurking behind the voting decision. . . . Most students of voting now agree that these kinds of explanations are not, by themselves, enough. They leave out a host of specifically “political” influences, ranging from the great campaign issues to the effectiveness of precinct organization. They are based on studies of single elections and cannot explain long-range shifts in party preference.
But must we throw out the baby with the bath? It makes as little sense to suggest that voters are usually rational and are influenced mainly by significant issues as to say that they are seldom rational and are uninfluenced by issues. If, as Professor Berns seems to feel, voters make their decisions only after careful consideration of the facts, and rarely make uninformed choices on the advice of associates, then voting behavior is radically different from most kinds of behavior. Indeed, the most “rational” thing an ignorant voter can do is to follow the advice of someone whose interests are the same as his and who knows more than he does.
Apart from this, how rational are voters? The 1964 Presidential election is often cited, perhaps justly, as a triumph of rationality. Lifelong Republicans switched to Johnson on the basis of genuine campaign issues. At the same time, however, the normally Democratic white voters of five states overwhelmingly chose Goldwater because of one issue, race. Their behavior was impeccably logical, once we grant them the assumption that their wish to keep segregation outweighed all other campaign issues in importance; but my definition of rationality in voting includes a sensible ranking of issues, not merely sound reasoning from whatever premise one starts with.
More questions: If college-educated suburban Jews vote consistently for liberal candidates while college-educated suburban WASPs vote consistently for conservatives, can sociology and psychology tell us nothing about the reasons for the difference? Where is the rationality in Jews’, Italians’, or WASPs’ scanning the ballot to select candidates they may never have heard of, but whose names have the proper ethnic flavor? And if the Democrats in 1964 had nominated a highly competent Negro, or a woman, whose views on all issues were identical to those of Lyndon B. Johnson, how “rational” would the electorate have been, and who would now be in the White House?
When people vote, as when they make other decisions, they are somewhat rational, some of the time. They are not always or entirely so, and on occasion they are capable of the wildest unreason. If social scientists pose their questions about voting in a way that allows only oversimplified extremes as answers, they will never learn much about it.
Richard L. Simpson
Department of Sociology
The University of North Carolina
Chapel Hill, North Carolina
Mr. Berns writes:
Miss Clark says my review is “needlessly confusing” because, she suggests, I felt under some obligation to “defend Key’s book as if it contradicted earlier voting studies.” She denies any contradiction. But, as my review made clear with ample quotations on this very point, Key himself thought his findings contradicted the voting studies, and I think he was right. Of course both Key and I could be wrong. Whether we are wrong depends on whose understanding of these voting studies is correct, Miss Clark’s or Key’s and mine. She seems to argue that Lazarsfeld and Berelson used correlations merely to state “statistical laws” and did not intend to make any statement respecting causation. I, on the other hand, quoted them as writing (in The People’s Choice, p. 27) that “Social characteristics determine political preference” (italics added). Does she deny that they wrote this? 1 offer her more evidence: Berelson and Lazarsfeld (and William Mc-Phee) describe their later book (Voting) as an attempt to provide the “‘total’ picture of the central decision in voting—how and why people come to favor this candidate rather than that one” (p. xii). Does Miss Clark deny that the how and why of a vote is another way of designating the cause of a vote? Indeed, as I have argued in another place, there is something in the nature of the behavioral scientist’s enterprise that requires him to move beyond the discovery of “statistical laws” or the demography of the vote and to attempt to establish scientific laws of voting behavior. Whatever Miss Clark may think about this, leading behavioral scientists agree with me. Witness Samuel Eldersveld (Political Behavior, p. 273): “Until researchers in voting behavior escape from paying homage to the marvels of the numerical symbol and concentrate on developing scientific laws with a relatively high degree of probability demonstrating causation, . . . our discipline can make no progress toward achieving scientific understanding.” The point at issue is not whether a correlation indicates a cause but whether the authors of the voting studies made causal assertions based on correlations, or whether they regarded it as their purpose to find the causes of votes, or, to use the language of the Michigan Survey Research people, the “motivation of voting behavior.”
Miss Clark also denies a distinction between the political, on the one hand, and the sociological and psychological, on the other, and asserts that political behavior is only a subdivision of one or the other or, presumably, both, thereby denying the existence of political science. The tacit assumption behind her position is that there are no forms of individual or group behavior that are qualitatively different from all others and hence require a special discipline for their study. It is not my purpose here to argue that an ideal psychology or an ideal sociology could not comprehend what is specifically political; my only contention is that as these disciplines exist today, they are not able to explain the phenomena of political rationality concerning questions of the common good and justice which Key, proceeding as a political scientist, brings to light. The value of his work is that he preserves our awareness of these facts against the onslaught of certain dogmatisms that deny this capacity without proof.
As for Professor Simpson’s objections, I need only add that neither Key nor I argued that “voters are usually rational and are influenced mainly by significant issues. . . .”
For a full statement of my views on the study of voting by behavioral scientists, I refer readers to my long chapter in Herbert J. Storing (ed.), Essays on the Scientific Study of Politics (Holt, Rinehart &c Winston, New York, 1962).