Problems and Methods in Statistical Ethics
Albert P. Brogan
The traditional treatment of ethics has been almost entirely theoretical. Discontent with this purely theoretical treatment has led to the development of descriptive or comparative ethics. In the hands of such men as Westermarck, Hobhouse, Tufts, and the French school of sociological moralists, comparative ethics has been primarily historical. By using the genetic or evolutionary approach to moral problems, these writers have made important contributions to our knowledge. Another attempt to make ethics more empirical has come from the schools of ethics which may be called utilitarian, telic, or consequential. These thinkers emphasize the importance of the future in our moral judgments. This approach to ethics has been fruitful also. But neither knowledge of the unverifiable past nor knowledge of the more or less inaccessible future can be substitutes for detailed knowledge of the living present in all its variety and processes of change. So one of bur main needs at the present time is the development of studies in contemporary morality, as well as in other aspects of our valuations.
There have been two approaches to contemporary morality in recent writings. Some men (F. C. Sharp, G. C. Cox, C. F. Taeusch) have studied cases of actual moral problems or codes. This work has led to a survey of what men consider right and what they consider wrong. For reasons to be indicated shortly, this work has been useful, but incomplete. The other approach has been by psychologists (G. G. Fernald, F. L. Wells) who were trying to get tests for defective or delinquent persons. This work has been important, but in a limited way.
In order to secure significant additions to our knowledge, it is necessary to have an adequate combination of hypothetical theory and factual verification. Merely to stare blankly at the facts will not teach us much. So we must ask what theories are pertinent to our study of comparative ethics.
One of the most fruitful of modern discoveries is the logic of relations. From the time of Aristotle until our own day, almost all logic has been concerned with class concepts and their manipulation in such relatively trivial forms of thought as the syllogism. Under the influence of our American thinker, Charles Peirce, there has been developed an elaborate logic of relations by
( 175) such men as Schroeder, Whitehead, and Russell. With this logic as a tool of analysis, much of our traditional thinking needs to be revised. Roughly speaking, we may say that most important matters involve relational facts rather than class concepts. In accordance with this theory, the present writer has at-tempted to work out a relational, or, as it has been called, melioristic, theory of value. Without going into details, it may be said that this theory asserts that all facts of value involve what may be called the relation of better or worse. To the relation of betterness, all other value characteristics are subordinate, whether the moral qualities of right and wrong, the general qualities of good and bad, or what not.
If this relational or melioristic theory of value can be accepted as either completely or partially true, then it follows that our study of comparative ethics must deal with more complex problems than the traditional ones, and must use more complex tools of investigation. Fortunately we have one such tool for inductive investigation in the recent development of the statistics of correlation. With the help of correlational statistics, we can deal adequately with the relational nature of value and with similar complex relational problems.
With the relational theory of value as an hypothesis and with correlational statistics as a method of inductive investigation, the present writer has published three studies in comparative ethics, or what may be called statistical ethics (the International Journal of Ethics, January, I923; April, I924; January, 1925). Although several additional investigations are nearly completed, the present short discussion can deal only with the general nature of the problems and methods which have been involved in this work.
There are many moral topics that might be studied. My previous work has centered around a list of "sixteen worst practices." It is important to note that the list of practices was given by popular opinion, in this case by many hundred students at the start of a course in ethics. Some very serious forms of misconduct, such as murder, are not included because they ate not practices among the groups involved. Other practices, such as smoking or dancing, are included because they are considered objectionable by a small minority among the groups. No good practices are included because it is difficult to get a satisfactory list of such topics. I am now experimenting with a positive list of good traits of character. It is to be noted that classes are necessary as topics because it is desirable to compare the comparative value and the comparative frequency of the topics. For this purpose individual actions would not do, though it may be useful later to study such topics. It may be advisable later to study classes of actions which have been logically and accurately defined. But popular opinion could hardly do this, so our first studies must deal with the topics as they function in popular opinion. The topics, as given by popular opinion, may seem somewhat vague or ambiguous. But they function with surprising definiteness in all our results.
With this list of sixteen worst practices, rankings were secured concerning both the comparative worseness of the practices and the comparative frequency of occurrence of them. At first the comparative worseness was secured for human beings in general, regardless of sex. This gave very definite results, but it was clear that a further question must be raised concerning the comparative value when the practices were done by men and when they were done by women. In the studies of comparative frequency, the sexes were separated for all questions. For all three problems, "general worseness," "double standards," and "frequency," rankings were secured from the following groups: different academic classes in the University of Texas, men and women separately, similar groups at other American universities, and, more recently, from different occupational groups, high-school students, and a few other groups. The studies so far finished have covered American universities with fair, though not complete, adequacy, and have merely sampled other groups. No facts have been secured as yet from groups outside of the United States.
It should be noticed that all the studies which I have made have been concerned with discovering the central tendencies of various groups. Of course attention has been paid to the average deviations or other measures of dispersion, but no attempt has yet been made to give adequate studies of individuals as individuals. We have made a map of what may be called "moral mediocrity." But eccentricity may be more important than mediocrity. Only we must first have the central tendencies from which to measure. Let me emphasize the point that no assertion or suggestion has been made that the central tendencies are or should be normative in any sense. Some judgments may be more reasonable than the central tendencies of these groups; some may be less reasonable. But here we have larger and more philosophical problems.
What are the specific problems upon which light may be thrown by the study of the sixteen practices, when they are ranked according to both comparative value and comparative frequency, and when they are studied with the methods of correlational statistics? Some eight problems may be indicated briefly.
1. If the relational theory of value is true, the question occurs whether popular opinion exhibits the relational structure to fit this theory. The statistical studies have shown a very definite relational structure in all the results. So I think it may be said that the relational or melioristic theory of value has had significant, though of course incomplete, verification.
2. We have had many guesses concerning the alleged differences among human valuations. But the facts are that the most frequent coefficient of correlation is plus .98 among large groups of university students. This is an extraordinarily and almost incredibly high degree of uniformity. Naturally the correlations will be lower when individuals are compared separately, but clearly we need facts rather than ignorant dogmas.
3. Some thinkers would say that the valuations are uniform because they are standardized by tradition. Also they would say that there is no tradition concerning frequency, and therefore there could be no uniformity there. But the facts are otherwise. The coefficients of correlation for frequency usually run from plus .95 to plus .99. Again the uniformity is surprisingly high. For reasons that cannot be discussed here, I think that these frequency estimates are fairly reliable as evidence concerning actual behavior, certainly more reliable than any other evidence that I know.
4. We have had many theories concerning the relation between moral standards and actual behavior. The first accumulation of facts indicates that there is a correlation of negative .56 to .58 between comparative worseness andgreater frequency of practice.
5. Our figures give very definite information concerning the differences between the morality of men and the morality of women. Concerning the general value scale, there is no significant difference, as the correlation between the men and the women is plus .98. But concerning the so-called "double standard," there are interesting differences. Idleness is the only practice which is considered worse for men than for women. But there are many practices which are worse for women to do: smoking, swearing, drinking, vulgar talk, sex irregularity, and gambling. Concerning the actual behavior of the two sexes, there is what is usually called a chance correlation, that is, the coefficient of correlation is plus .03. Whether or not there ought to be a code of double standards, there undeniably is a fact of double behavior.
6. There is space merely to indicate that we can study the influence upon our moral topics from the differences in age, in geographical location, in occupation, and in other ways. There is an underlying uniformity, but there are many interesting differences in the moral results for study and explanation.
7. Of peculiar ethical interest is the problem concerning the criteria which popular opinion uses in making the rankings according to comparative value. In this connection we can study, by the method of correlation, all of the traditional hypotheses, such as hedonism or utilitarianism. This method may trans-form ethical theory into a combination of the formulation of hypotheses and the factual verification of them.
8. Finally, it may be mentioned that these studies give to the general problem of social description an extension into the realm of ethics and other aspects of value.
I shall not spend any time arguing whether this work in statistical ethics is
a part of philosophy or ethics, sociology or psychology. It is obviously on the
borderline of several subjects according to our traditional classifications. Two
things are necessary: first, to enliven our factual descriptions with fruitful
and significant theories or hypotheses; second, to insist that all our theories
and hypotheses must be submitted to factual verification.