Rule and Custom as Individual Variation of Behavior Distributed Upon a Continuum of Conformity

Floyd Henry Allport


Since conformity is a fact of central importance for sociology and the other social sciences, a knowledge of the facts of conformity phenomena is essential. Such societal concepts as folkways, mores, and fashions are vague and obscure and lead to confusion and inaccuracy. Specific questions essential to clear understanding of conformity are whether uniformity and conformity exist in the behavior of individuals under observation, in what respect people conform, to what degree, what proportion of the individuals conform, the nature of "conformity-producing agents," etc. The J-curve hypothesis of conformity rests upon the following conceptual basis. Instead of regarding conformity in the dichotomous "all-or-none" manner, the approach here suggested is one more consistent with general scientific procedure, namely, to measure degrees of conformity on a continuum. In this context continua are classified under two types, empirical and nonempirical, and the latter is subdivided into a personality continuum and a telic continuum (measuring purpose fulfilment). A "field of conformity" is defined when there is a generally accepted, though not necessarily explicitly stated, rule and purpose in the situation, and when 50 per cent or more of the population fall upon the first step of a telic (conformity) continuum whose variable is degrees of fulfilment of this rule and purpose. The two major formulations of the J-curve hypothesis are: (1) the distribution of degrees of conformity upon their appropriate continuum is in the form of a curve of positive acceleration toward the mode and (2) in any conformity field the distribution of measurable variations of the behavior upon a relevant empirical, or nontelic, continuum is in the form of a steep unimodal, double J-curve, in which the mode is likely to be off center and the slopes are likely to be asymmetrical.


Conformity is a fact of central importance for sociology and the other social sciences. It is not the only fact of importance, for social organization, conflict, and interaction involve reciprocities of behavior as well as likenesses. Nevertheless, without large-scale similarities of thinking and acting, clustering around accepted modes of performance, social organization itself would be impossible. From the standpoint, therefore, both of theory and of practice, a knowledge of the facts of conformity phenomena becomes one of the first essentials of social science.

When we turn to the literature of this subject, however, we are impressed by the vagueness and obscurity of working concepts and by the lack of logical methods and precise techniques. There have come to the writers' attention no systematic empirical investigations

(898) of this important problem. The prevailing approach to conformity is limited almost entirely to such societal concepts as folkways, customs, mores, fashions, conventions, traditions, and institutions. Customs have been defined as "group habits" created by "group action."[1] Custom in general has been defined as "the totality of behavior patterns which are carried by tradition and lodged in the group, as contrasted with the more random personal activities of the individual. "[2] At times the terminology used refers to common aspects of the behavior of individuals rather than to groups; but, even when this occurs, the usage is not carefully scrutinized, so that it is likely to be incomplete, confused, or inaccurate. Folkways and customs have frequently been defined as "uniform" or "common ways of behaving."[3] It has been said by a recent authority that the wearing of "slacks" and a sport jacket at the opera would be to disregard an American folkway.[4] By this token to wear sober civilian clothes or evening dress would be considered as conforming to our folkways. But, if we understand the term "way" in any exact sense, the wearing of sober or evening clothes, in contradistinction to slacks, is not a different kind of "way" at all. It is the same "way," for the "wearing of," or "moving about in," clothes at the opera would be about the same behavior whether in evening dress or in slacks. It is not the behavior but the clothes which are different. The case is similar with concepts derived from notion of folkways. In the election of representatives to Congress in the United States, the election only of persons from the district in which they reside might be called an American stateway, and this might be contrasted with the British stateway of electing also representatives from other places of residence. This is not, however, a difference in "ways" at all, for the voter performs the same kind of behavior (scratching a ballot) in either case. There are, of course, underlying differences of political attitude and ideology, but such uniformities of the people in the given country should then be studied in ideological categories, not in

(899) their ways of overtly behaving or in alleged superindividual "ways" of their "state."

The same obscurity is illustrated by the term "mores." One large division, at least, of the mores is concerned with strongly enforced prohibitions against disapproved acts. When you conform to the mores, in this sense, you do not do anything at all; or you may do anyone of so great a number of things that the term has no positive meaning. To marry one's sister is a breach of the mores, but you may marry any one of thousands of other persons, or no one at all, and not violate the mores. When you break the mores (as prohibitions), you do not fail to carry out an act expected of you; you do an act which is not expected of you. The act which gives definition or meaning to the mores is, therefore, the act which negates them. The mores, in this sense, are not ways in which you may act but only ways in which you may not act. In other words, they are not ways at all. The only alternative to this conclusion is to consider them not as the behavior of the individual in question but as the disapproving attitudes and behavior of others who would prohibit him from doing something. But here again lies confusion. Are we to regard human acts both as the behaviors of punishing breaches of the mores and also as the mores themselves? If so, then the punishment is the law, and the law the punishment; and the whole structure of societal norms collapses, as perhaps, in a realistic approach, it should. Punishment, then, is a measure exercised by some individuals for the purpose of restraining others from behaving in a manner defined by the former as undesirable. Once we get into the problem, a direct, explicit statement of this sort is indeed better as a working conception for the study of conformity than are societal terms such as "custom," "mores," "law," and "institution"—terms whose meaning evaporates when employed with reference to actual observations.

Such terms, moreover, if not used with great caution, lead to inaccuracy as well as to confusion. Sumner says that folkways are uniform, universal (in the group), variable, and imperative ways of meeting human needs.[5] In Sumner and Keller's The Science of So-

(900) ciety [6] we read that the removal of the hat when a civilian meets a woman on the street is a folkway. In a recent investigation the writers have found evidence that the practice of the removal of the hat in men's greeting of women is neither uniform, universal, invariable, nor imperative. The hat was "removed," in fact, in only 26 per cent of a fairly good sample of cases, and there were at least four other acts which were done with reference to the hat in this connection, 18 per cent doing nothing whatever about the hat. "Removal" or even "tipping" of the hat, therefore, cannot be a folkway. Clearly, the term "folkway," as applied to such phenomena must be given up, or else its definition must be altered so as to be more in accordance with the facts. Uniformity and variability are facts to be ascertained by investigation, not by definition from some preconceived category. The degree of universality also must be discovered not by naming an act a custom, and therefore reasoning that it is universal, but by observing an adequate sample of the population to see in how many cases it happens in the situation designated.

Sumner also assumed that modes of adaptation which have been found universally useful and have become automatic and "unconscious" form a kind of collective or mass behavior with which the individuals of the society are compelled to conform. There is a failure here to distinguish behavior which is universal because universally useful and behavior which is not only useful but is accepted as the proper or expected form of response and therefore enforced by individuals upon one another or by smaller numbers of individuals upon the mass. By the trick of creating a societal category of similar behaviors we conceal the possibility of conformity-producing agents and the problem of their discovery and study. There results the purely gratuitous notion of a determinism by some assumed cultural force which acts upon the individual. This confusion is latent in the statement that "conventions are rules or standards of conduct or behavior prescribing what is to be done or not to be done by the members of a given group or community."[7] Does "prescribing" here mean that the rule itself or some force over and above the rule makes

(901) individuals do or refrain from the acts they specify; or does it mean that human beings themselves do the enforcing, and the rules merely clarify and render specific that which is to be enforced? Certainly, before we adopt the former unverifiable hypothesis, we should do our best to explore the possibilities of the latter. The latter can be tested only through a technique which will enable us to break down the "custom," "rule," or "convention" into a multi-individual situation and to examine in numbers the behavior of the individuals involved.

Not all writers, of course, use the familiar conformity terms in this loose and uncritical fashion. We are not pleading that they be given up but only that they be more critically defined and used with greater discrimination with reference both to facts and to scientific methods. The genuine use which they have, both for practical and for scientific purposes, will be later discussed. Our problem in the present paper is to develop methods by which more exact definitions of these terms can be formulated and to suggest techniques of observation by which they may be supplemented for the essential work of empirical investigation. Without, therefore, denying the value of such terms as "custom" and "folkway," our task is to turn to the actual situations to which they refer, so that we shall see them as large numbers of people acting under known conditions of their human and nonhuman environment. We shall then set to work to observe what the individuals are doing when they are "conforming" to the "rule" or "custom" and to describe, count, and measure significant aspects of their behavior.

The specific questions essential to a clear understanding of conformity which might be investigated in such a study include the following: (1) Is there a similarity, or uniformity, of behavior of individuals in the field studied? If so, in what respect are their behaviors similar—where does the uniformity lie? (2) Is there conformity? What test can we apply to see whether the field of behavior __ should be called a conformity field? (3) If there is conformity, in at respect do people conform? Where does conformity lie? (4) Are there degrees of conformity or only a dichotomy of conformists and nonconformists? (5) How many and what proportion of the individuals fully conform? (6) If the conformity may be a matter of degree, how many conform in part but not fully? (7) How many do

(902) not conform at all? (8) Are there common biological, psychological, or "energy" tendencies which, as laws of human behavior, influence individuals toward or away from conformity, quite apart from conformity-producing factors of a "social" sort? (9) Is it possible to discover and estimate the part played in the conformity picture by "conformity-producing agents" (individuals, few or many in the entire population, who are interested and exert an influence in making individuals conform)? (10) What is the relation of the picture of conformity in the population to natural psychological differences of individuals and to personality traits which might be expressed differently if conformity were absent or diminished? (i I) What variations of individual behavior and personality traits with respect to the dimension of conformity are possible within that range of individuals who may be said fully to conform? (12) What is the significance of the aspect of behavior upon which individuals conform to the behavior and purposes of these individuals as human beings? What, in other words, does conformity mean to those who conform; and is this an adequate account of their behavior, or are other explanations, such, for example, as society or culture, necessary for complete understanding?

Concerning these fundamental questions scarcely a shred of systematic evidence exists. Barring some investigations of specific phenomena, mainly for practical purposes, practically no attempt has been made to answer them through scientific inquiry or even to raise them in theoretical discussions. Nor can these questions be answered, or clearly asked, through an approach in terms of "custom," "folkways," "mores," and the other rubrics now in familiar use for describing conformity situations. Useful as these terms may be for other purposes, they fail utterly when used as tools of analysis for problems such as these. In the opinion of the present writers the answers to these questions lie through an investigation in terms not of the group or society as a whole but of individuals acting together in a multi-individual situation. This paper is devoted to an attempt to develop a methodology and technique for such investigations and to indicate its merit through reference to its application in a study of a familiar situation to which the name of "rule" or "custom" might be applied.



It will be necessary at the outset to clarify our point of view with respect to the phenomena of conformity. It is important to make a distinction between an administrative classification or "control" viewpoint, upon the one hand, and the disinterested scientific approach, upon the other. If one takes the former, which is essentially the "societal," position, one is likely to think of customs and folkways not as finally graded variations of behavior but as a simple dichotomy. They are regarded as all or none. Either there is a custom of tipping the hat when men greet women acquaintances or there is no such custom. A man either conforms to the custom or he does not. The custom is thought of as an undifferentiated, qualitative norm, operative in a certain cultural area, not as a range of individual behaviors. Such a point of view is justified from the stand-point of culture-trait classification or administrative control. Every judge or jury must decide whether certain behavior is a violation of the law or not. They must think in terms of dichotomies. For those who wish to impose conformity upon other people, or for social scientists who are interested in studying these methods of control, the dichotomous way of thinking is the natural one. Disinterested science, however, is found upon measurement rather than upon dichotomy. Its phenomena are not of the "all-or-none" type but are matters of degree. Categorical definitions are used only as a starting-point to help select a phenomenon for study. In this way, for ex-ample, the familiar notion of custom as an all-or-none affair is useful merely to help us locate our problem. Once we start our task of observation, however, the measurement of degrees along a continuum will inevitably supersede the method of dichotomy.

Let us consider, for illustration, certain well-known "customs" from the societal, dichotomous standpoint and later contrast them with the other approach. A sociologist is likely to say that for a particular culture area there is a custom of turning out to the right in passing another person upon a highway. This is an all-or-none matter: you either turn to the right or you do not. If you turn to the right, you are acting properly; if you turn to the left or go straight ahead, you are violating a custom, if not, indeed, a law. He might

(904) also say that there is a custom of men removing the hat in the house. This again is a dichotomy. "Hat off" is conforming; "hat on," not conforming. When a man is walking with a woman on the street, the custom is for him to allow her to walk on the inside. To compel her to walk on the outside is a violation of custom. In going through a door a man allows a woman to precede him, not to follow him, etc. Similarly with fashions and things used or worn. To wear clothes is to conform to custom; to go nude in public is a breach both of the mores and of the law. Where veils are in fashion, to wear a veil is fashionable and to be without one is unfashionable. To wear shoes for ordinary street dress is approved; to go without shoes is disapproved. A short skirt at a certain time is a la mode; a long skirt will mark its wearer as out of date.

Considered from the "pure" science standpoint these dichotomies, however useful for practical purposes, are misrepresentations of fact. There is no sharp line between "tipping the hat" and "not tipping the hat." There are a number of things of different sorts which men do with, or with reference to, the hat in greeting women; and it is an arbitrary matter where the line between "tipping" and "nontipping" is drawn. The wearing of clothes is also a question of degree. There are all stages between complete dress and complete undress; and both laws and customary standards state different stages, for different times and places, as the dividing-line between conformity and nonconformity. Variations of veils are numerous, so also of foot-gear; and it might be difficult at times to tell whether the objects worn on the feet are shoes or not, or whether or not a veil is being worn. Even in the more crucial cases of turning to the right on a highway the simple dichotomy will not hold. Suppose that there is a one-way highway with occasional wider places for automobiles to turn out in passing. Here the rule would be for the one who, when approaching another, finds a turn-out place on his right to turn out in that direction; but for the other individual the conforming behavior would be to go straight ahead. One who takes the administrative viewpoint will be likely to set up a new dichotomy to cover this situation. Following this procedure consistently, human interactions may be all broken up into an increasing number of dichotomies as the social order becomes more complex. For a descriptive, disinter-

(905) -ested scientist, however, the procedure would be different. He would say that, if you treat the behavior as on a continuum rather than by multiplying dichotomies, one rule and purpose will hold for all occasions; that is, provided you accurately state your continuum. In this instance he would say that the conformity requirement would be neither turning out to the right in passing in certain situations nor going straight ahead in others. He would say that within the limitations of safety and possibility there is one rule, namely, always giving way to the right. This rule and purpose would probably fit both drivers and all occasions. It also fits the two-lane highway better than the old dichotomous statement. The driver who turns out to the right in the usual two-lane highway, is not fully conforming to law or custom unless he turns out far enough to the right, in proportion to his ability, to allow the other car to pass safely and conveniently. This clearly is not a matter of dichotomy but of degree, for it depends upon a ratio of spatial quantities in which there is al-ways some margin. If the driver turns out far enough for the other car to pass, but not conveniently, he is not fully conforming; but neither is he totally nonconforming. He is conforming in a lesser degree. Even the dichotomy of wearing or not wearing a hat in the house may be, on special occasions, a matter of degree. What, for example, are we going to do about skullcaps worn to prevent exposure to drafts, fancy hats worn at the table during parties, and the soldier who keeps on his cap while bearing side arms? Here, again, new dichotomies might be set up for control and standardization; but, if this is our only approach to the problem, the observed individual variations of behavior lose their full significance and the purpose of the "custom" itself is concealed. The dichotomous statement of conformity would here be merely an arbitrary statement, not a description. It is better from the descriptive standpoint to construct continua which would employ the true variable or meaning of the practice. All these cases of unusual hat-wearing would have their proper place upon the continuum and would be understood with reference to their conformity to the rule. Similarly, when dangers or annoyances are more frequent on the inside of the sidewalk than the outside, permitting the woman to take the outside would not be considered an act of nonconformity if we discarded dichotomies and

(906) found the right continuum variable upon which to allocate meaning-fully the man's behavior. And so with all the examples we have cited and many more.

To sum up the matter, we may say that for purposes of scientific descriptive purposes the dichotomy should, wherever possible, be re-placed by the use of a continuum. Nothing is lost in this, for every case of true conformity will still fall upon the full conformity step of that continuum. And much will be gained, for we shall be able, on the other hand, to observe and record as distinct a large number of other behaviors which really are distinct, and which in a system of dichotomies would be concealed by being lumped all together either as conformity or as nonconformity. In this way the true and full significance of "custom" as a human practice will be seen, whereas it might otherwise be left to conjecture or actually misrepresented. Human behavior and traits, unlike administrative practices, cannot be reduced to dichotomies. Individual differences range widely over a continuum falling in a normal probability distribution. In order intelligently to relate a conformity situation to psychological realities, we must therefore express conformity itself in terms of degree.

Proceeding upon the assumption that continua and scales of measurement are useful tools for the measurement of uniformities in behavior, an earlier study was conducted in which they were employed as methods. With their aid a hypothesis was formulated which was supported by the data then at hand, as well as by subsequent findings. This theory, which is known as the "J-curve hypothesis of conforming behavior,"[8] has advanced our knowledge of the use of continua for conformity studies and has given us a generalization concerning the nature of conformity which we shall employ as a methodological tool for the study of "custom situations."

As a preliminary to understanding this hypothesis, the meaning of certain continua which have been employed must be explained. These we may call, in brief, the empirical, the personality, and the telic continua. The empirical continuum is one which is already

(907) familiar in the measurement of physical quantities. Its units are in terms of such variables as time, space, weight, velocity, etc. We may cite as examples the velocity with which a motor car crosses a boulevard intersection, the extent to which automobiles cross over white lines marking prescribed traffic lanes, and the time of arrival of workers in a factory or of students in a classroom. Types of uniformity other than those of obedience to rules and laws may also be measured upon such a continuum: for example, the width of trousers or skirts, the length of time after receiving an invitation to a dinner at which the invitation is answered, the angle of the neck and head in bowing during a prayer, the position of the hand and arm saluting the flag, and the degree of energy put into a handshake. All these are examples of the application of empirical continua to measurable aspects of approved behavior.

Valuable as such forms of measurement are, the study of behavior, however, would be seriously limited if we had no other variables upon which to measure it. It is important that we supplement these methods of measuring physical quantities with some way of stating degrees of qualities or meanings. The classic attitude-measurement studies of Professor L. L. Thurstone give us psychophysical methods which can be adapted to the construction of scales for this more subtle type of measurement. We have modified these methods, originally desired to measure degree of affect, so as to employ them for constructing continua in terms of degrees of logical meaning, traits of personality, and extent of fulfilment of purpose. Continua of this general sort we shall call nonempirical. In using such a continuum, statements of acts of behavior are allocated upon it according to the degree which they manifest some trend or trait of personality or of the fulfilment of some purpose, with the same logic that we employ in the allocation of acts on an empirical continuum in terms of their variables of time or space. There are many types of nonempirical behavior continua. We shall be interested here, however, mainly in types. These are (a) those measuring trends or traits of personality and (b) those measuring purpose fulfilment. We shall refer to the former as a personality (or personality trait) continuum and to the latter as a telic continuum.

To illustrate personality continua, let us return to the illustration

(908) of the handshake. Instead of employing a physical continuum of energy, we now select some trait of personality upon which to measure instances of this act. Suppose we choose as our variable the degree of the trait "warm-heartedness to cold-heartedness," or "affectionate to distant." By observing a considerable number of acts not in special aspects but as wholes, we may recognize different type variations or manners of shaking hands, such, for example, as the limp hand, the sudden hard grip, the extending of two fingers, etc. These types of handshake, as found among people in general, may be rated by the psychophysical technique so as to determine the degree of the trait of warm-heartedness or affection-giving which each expresses. A scale can thus be constructed upon this trait variable in which these and other type acts may serve as "landmarks" with known scale values. By reference to such a scale the instances of handshake in our manifold of observations can be measured as to their manifestation of the personality trait in question. Similarly, traits such as that from extreme caution to extreme boldness might be measured on a personality continuum in terms of the types of acts one might do on approaching an intersection (stopping, slowing down and looking, going on without speed change, etc.). Such measurements, of course, should be taken when there are no institutional controls, such as stop signs or traffic lights, which might hamper the free expression of personality characteristics. Again, we might construct a continuum of the trait "considerateness to inconsiderateness," the step values of which could be given in terms of types of acts which an individual might do in acknowledging hospitality he had received. Degree of piety might be shown on a scale in the manner in which one bowed or knelt at religious devotions. It is admitted that appearances are sometimes deceiving and that individuals might perform these acts for reasons other than the expressions of the traits indicated. Nevertheless, we are dealing here not with a thorough personality study of an individual but with a cross-section of the acts of many individuals. If enough act items are used, and if the continuum is constructed to represent degrees of the trait as manifested "on the average" by acts of this type, differences due to extraneous reasons will tend to cancel one another if a sufficiently large population is measured. Our scale will then be adequate for the

(909) estimates of personality traits relevant to the conformity situations we are studying.

We come now to the telic continuum. In the same way that an act can be allocated to a continuum through its physical properties, or through its expression of a trait, it can also be allocated with reference to its fulfilment of a purpose. Here we turn our attention from physical units, or qualities inherent in the organism, to the general "satisfactoriness" or adequacy of the adjustment which is made through the act itself. Does it carry out what the individual is trying to do? If not fully, does it carry out that purpose in some degree? If so, in what degree? The left extreme of the telic continuum is generally taken to indicate complete fulfilment of the purpose, while lesser degrees of fulfilment are represented in the steps to the right of that step. Telic continua may be subclassified as individual or societal. In either case, of course, the purpose is that of individuals. In the former case, however, it is a purpose upon which the individual is acting alone, or spontaneously, and without reference to the expectancies or like behavior of others; while in the latter it is a purpose or behavior similar to that which many others also are carrying out, and which is associated with the common acceptance of a norm or rule. Since the "societal-telic" type of continuum is that with which we shall here be chiefly concerned, we shall draw our illustrations from that type. To take an example, we may ask with reference to the motorist coming to an intersection at which a red traffic light is showing not what was his speed of driving or the evidence he is showing of a trait of boldness or caution but the degree to which he is carrying out the rule and purpose implied in traffic laws. Is he behaving in such a way that, if his behavior were made universal, it would fully comply with the purpose of facilitating the safety and flow of traffic? If the motorist stops, he carries out fully the rule and the purpose. If he almost stops, but not quite, he is in some degree carrying out the rule and purpose though not so fully. If he slackens his speed only a little, he is still to some degree carrying out the rule and its purpose, though not so much as in the preceding instance. If he drives through without change of speed, he is not fulfilling the rule or purpose in any degree whatsoever. To take another example, the time of arrival of workers of a factory might be

(910) recorded as units upon a purely empirical continuum of time. They might also be recorded, if the conditions were such as to reveal it, upon a continuum of the trait of promptness to dilatoriness. Or they might be plotted upon a telic continuum which would be in terms of the rule and purpose of the factory situation. On this latter continuum all workers who came at or before the time set by the manager for the beginning of work would fall in upon first step of the continuum as being "on time." Those who came later than that moment would be grouped together, in convenient time classifications, as exhibiting successive, recognizable degrees of "lateness." The steps, though measurable by time, do not now mean time but degree of compliance with the factory rule and purpose. Or, again, it might be determined, though the task would be more difficult, what mode or modes of performance constitute the socially "approved" manner of shaking hands. A person performing this act (or acts) would fall upon the left-hand step of complete fulfilment. Acts of other types but still having some "handshaking meaning" would be placed, at known scale values, as steps to the right of this first step. Those who performed such acts would be carrying out the social expectation and purpose in lesser degree. Promptness and degrees of delay in answering the dinner invitation could likewise be scaled as steps of complete and partial carrying-out of the commonly expected performance in matters of this kind.

The right-hand extreme of the telic continuum should be defined as carefully as the left. It represents the particular act, or range of empirical measurements of those acts, which is the smallest degree of fulfilment of the rule or purpose that is still recognizable as some degree of fulfilment. The motorist who slows down only slightly is still on the telic continuum; but, if he fails to slow down at all, he is not. The worker who comes to the factory at the end of the forenoon may be said completely to disregard the rule and purpose of the opening hour. The continuum ends before that point of time. A letter answering an invitation which does not arrive before the occasion to which the individual was invited is not on the continuum. If it arrives on the day before, it might, according to circumstances, be considered as, for example, on the extreme right-hand step. The

(911) limits of the telic continuum are therefore finite and much more definite than those of the other continua.

It will be apparent that, of these various types of continua, the societal telic is pre-eminently the one upon which conformity should be measured. By its very nature it defines the true variable upon which people may be said, in varying degrees, to conform. It affords us a precise definition of the full conformity position and permits us to treat variations of conformity in quantitative terms. It gives us a means of distinguishing between behavioral similarities which are simple uniformities and those which represent true conformity situations. We shall refer to this type of continuum hereafter as the continuum of conformity.

Although these three types of continua are distinct, they are, by the nature of the case, related. Not all empirical or personality continua are likely to have a relationship to a particular telic continuum; but certain variables which might be selected for such continua do bear a clear relationship to degree of fulfilment on the telic scale. For example, the empirical time of arrival of factory workers and degree of promptness as a trait are probably both related, according to some constant function, with "on-timeness" or degree of "lateness." We should not, however, expect weight, stature, or the traits of neatness or sympathy to be so related. Nontelic (empirical or nonempirical) continua whose intervals bear a discoverable relation-ship to a telic continuum may be said to be relevant to that telic continuum and to possess relevant behavior variables. One of the objectives of conformity study is the discovery of behavior variables relevant to the telic continuum of conformity in a particular situation.

Having clarified our definitions and our measuring devices, we are now ready to delimit the field of observation with which the J-curve hypothesis deals. Before attacking the problem of conformity distributions, we must locate a conformity situation to study. And to do this we must know what we mean by conformity. When we consider all cases in which the behaviors of a large number of people are similar, we shall see that those instances fall into two groups: First, there is simple uniformity. This is the mere fact that people, by their very natures, act more or less alike in a given situation. For ex-

(912) -ample, nearly everyone turns up his coat collar when walking out-doors in a temperature of 20 degrees below zero. In addition to simple uniformity, however, there is another class of similar behaviors. This class comprises the situations in which not only the behavior is similar but the particular form, degree, or quality of the behavior upon which the individuals agree is accepted explicitly or tacitly by them as the approved standard which they ought, and are generally expected, to obey. These are situations of true conformity. Conformity, in other words, is a situation of uniformity with the recognition of an accepted norm or rule as to the behavior involved. It is at this point that the sociological definitions are of aid. They enable us by preliminary observation, and through our own behavior and introspection, to select times, occasions, and populations in which conformity situations may be reasonably expected to appear. The "rule," "folkway," or "custom" of "so and so" is a convenient, in fact an indispensable, concept with which to start.[9]

The next step in the understanding of the J-curve hypothesis is to know what we mean by a "field of conformity." This definition, however, must be couched in terms more precise and capable of quantitative determination than "custom" and "folkway." A conformity field exists when there is a generally accepted, though not necessarily explicitly stated, rule and purpose in the situation, and when 50 per cent or more of the population fall upon the first step of a telic (conformity) continuum whose variable is degrees of fulfilment of this rule and purpose. An example would be the traffic situation, where an observation of the cases shows that 50 per cent or more of motorists completely stop at the intersection when the signal light shows red. Another instance would be among factory workers where, on a telic continuum of "on time" and "degrees of lateness," 50 per cent or more are on time. In these cases there is an accepted rule and purpose, and the majority of cases fall on the step of complete fulfilment. These cases fulfil the conditions of a "field of conformity," for purposes of the definition in terms of which the J-curve hypothesis is to be stated.[10]


With our definitions of the different continua and the field of conformity in mind, we can now state the first part of the J-curve hypothesis. We quote from the article cited with some changes of wording to adapt the hypothesis to our present purposes.

If, in any field of conformity (see definition given above) we apply a scale whose steps are variations of behavior which represent successive recognizable degrees of fulfilment of the "accepted common purpose," ranging from the prescribed or "proper" act, which most completely fulfils the purpose (on the left) to that which gives it the least recognizable amount of fulfilment (upon the right), the following will occur: (a) more instances will fall upon the step at the extreme left than upon any other; (b) the successive steps from left to right will have a respectively diminishing number of instances; and (c) the decline in the number of instances will decrease as we proceed by successive steps from left to right.

To state the hypothesis in another form : Degrees of conformity in a conformity field are distributed upon their appropriate telic (or conformity) continuum in decreasingly diminishing proportions as one proceeds from the fully conforming act or degree of attribute of the act toward the least conforming act or degree of attribute of the act.

Or, more simply: The distribution of degrees of conformity upon their appropriate continuum is in the form of a curve of positive acceleration toward the mode.

The form of this distribution is illustrated by actual data (from one out of many situations studied) plotted in a column diagram, with suggested curve, in Figure 1a. [11] We have called this form of distribution, for simplicity, the single J-curve (its shape being approximately that of a reversed letter J). The conformity continuum is

( 914)


(915) based upon the variable "on-time"—"lateness." The conformity distribution represents in its left-hand step the proportion of factory workers supposed to be at work at 7:25 A.M. who came on time during the period of the count, and in the successive steps to the right the proportions coming in successive half-hours of lateness. The mode has 78.6 per cent (well above 50 per cent) so that we have here a true conformity field. The proportions plotted on the remaining steps fulfil exactly the requirements of the J-curve hypothesis, namely, that the distribution should be in the single J-form of continuous positive acceleration from the right extreme toward the mode. Such distributions have been found to occur in practically all situations so far investigated in which the conditions of the conformity field are satisfied. The number of different kinds of conformity situations thus far examined is perhaps not yet great enough for us to say that this hypothesis is a law. Nevertheless, its occurrence is so regular that we have ventured to use it as a methodological tool for the defining and identification of conformity situations. The logic of this procedure will be presently described.

The continuum which we have used for the first part of the hypothesis is the telic or conformity continuum. The question now arises: What will be the shape of the distribution upon the nontelic, or in this case, the empirical, continuum? In other words, how were the cases distributed upon the original continuum of time (intervals of 10 minutes) in terms of which they were originally collected? The question may be stated: How will the individuals in a conformity field (i.e., where there is a rule, a purpose, and 50 per cent or more upon the mode of complete conformity) be distributed upon a relevant empirical continuum? This picture is presented in Figure 1b. This figure must be regarded as basic and prior in the investigation to Figure 1a, since the latter employs (merely by regrouping) the data originally gathered upon the former. The cases which fell on the conformity continuum (Fig. 1a) on the first step "on time," are shown on the empirical continuum (Fig. lb) as spread out, at the times they actually occurred, to the left of the point of the factory rule time for beginning work, that is, 7:25 A.M. (see dotted line x, Fig. lb). They are ahead of the time of the factory rule; but they cannot be any more "on time," that is, any more conforming to the

(916) rule of the time to begin work, than those who fall exactly at 7:25. Hence they are all grouped together in the first step on the conformity continuum. Incomplete conformity admits of degree, but there are no degrees of conformity within full conformity itself. It will be noted that the large mode in Figure b falls on the interval 6: 30–6:40. Six-thirty is the time when the factory doors are opened. Though work is not expected to begin until 7: 25, time can be gained by getting tools in readiness and work laid out ahead of time to begin promptly at that moment. Since the pay of the workers is on a piece-work basis, it is therefore economically important to them to get to the factory as early as possible. This is the explanation for the large mode at 6:30-6:40, the time when the factory doors are unlocked. If observation were made not from the clock punchcards but by standing outside the doors before they open and noting the time of still earlier arrivals, some cases would probably be found to fall still farther to the left. These would somewhat lessen the pro-portion falling on the present mode and would probably decline rapidly in frequency as we go toward the left extreme of the time range.[12] We have suggested this probable left portion of the distribution by the left descending limb of the heavy, dashed-line curve and by the dotted area it incloses.

It will be noted that, if we overlook a local rise in the plotted distribution between 6:50 and 7:20, the curve of the data (suggested by the heavy dashed line of Fig. b) resembles two asymmetrical J-curves of positive acceleration placed back to back. Since this general condition was found in most of the situations studied, the second portion of the J-curve hypothesis is stated as follows:

In any conformity field the distribution of measurable variations of the behavior upon a relevant empirical, or nontelic, continuum is in the form of a steep [13] unimodal, double-J–curve (i.e., a curve having positive

(917) acceleration of both slopes), in which the mode is likely to be off center and the slopes are likely to be asymmetrical.[14]

Experience since the J-curve hypothesis was first published has shown that, while the single-J portion of the theory stands as formulated, the hypothesis of the double-J distribution upon the nontelic continuum must be more carefully studied. Although it is obtained in many instances, it is believed that a better formulation involving more basic principles can be stated. For the present we must point out that differences in absolute and relative strength of the conformity-producing, biological, and inertia factors may alter the form of the distribution. The part played also by personality factors has not been adequately stated; and the forms which the distribution would take if plotted upon a nontelic, personality trait continuum, under varying strengths of the energic and conformity-producing factors, are still undescribed.

We enter now upon a more speculative part of the hypothesis, but one which takes us into the heart of the conformity problem. This is the explanation of how the shape of the distribution on the empirical continuum results from various factors at work. We can best try to construct a picture of what happens if we begin with a theoretical distribution upon a personality continuum. Degrees of personality traits, as we have noted, have usually been found to be distributed in the form of the normal frequency probability curve. If, therefore, we take a trait relevant to punctuality, such as promptness to dilatoriness, and if we borrow the empirical units of time of arrival as roughly expressing degrees of that trait, a hypothetical normal curve of the personality distribution of the factory population upon this time continuum can be drawn. To do this, we must assume that there are no disciplinary or economic factors making the workers come at any particular time. This hypothetical personality distribution is shown by the faint dashed line (p) in Figure 1b. The mode would come somewhere around 7:30–7:40 A.M., and the two extremes

(918) would trail out, roughly from 6:00-9:20, respectively. This curve, though purely hypothetical, may be taken as an approximate picture of what would happen if no influences other than personality differences determined the time of coming to work.

Actually, however, the workers are subject to other influences; and these are common and uniform, rather than differentiating, influences. Each of the workers is obliged to be at work at the set time or suffer loss of pay and perhaps of his position. More than this, the workers are influenced to be at the factory considerably earlier than this starting rule in order to get their work and materials ready so as to start without delay; for they receive their pay on the piece-work basis. The conformity-producing agents in the picture are the factory officials who set and enforce these conditions. We may think of their influence (indicated in Fig. b by the arrow c.-p.a.) as pushing the mode of the distribution over to the left. They push it, in fact, to the point of the opening of the factory doors which is at 6: 30. A skewed distribution results, and the mode becomes steeper. The mode, moreover, is pushed so far to the left that the negative acceleration, characteristic of normal curves as they near the mode, disappears, and positive acceleration, for the most part, takes its place. This accounts for the J-shaped character of the right half of the curve.

Still other influences, however, must be taken into account. These are common influences which compete with that of the conformity producing agents. To get up early enough to arrive at work by this hour requires an expenditure of effort necessary to overcome possible drowsiness and the tendency to stay longer in bed. It requires the expenditure of the energy necessary to attend to dressing, eating, and household duties quickly, and to take a streetcar on time or walk rapidly enough to arrive at the time set for beginning work. A law of all behavior is that an act will be done with the least expenditure of energy possible. We may therefore imagine a factor of biological or physiological inertia (arrow for i in Fig. b). This influence acts in a direction opposite to that of the conformity-producing agents, tending to push the mode toward the right, thereby steepening the curve on the left side, and producing on this side also the J-shaped form. The point of equilibrium between these two influences establishes

(919) the point of the mode and the double-J character of the curve is now accounted for. This theoretical double-J distribution upon the empirical continuum is indicated by the heavy dashed line in Figure b.

We must not forget, however, that personality differences, though skewed in their effects by these influences, are probably not obliterated. Promptness and dilatoriness can be exhibited only with reference to some standard. If the time of starting to work were 9:00 A.M., that point, other things being equal, would probably be-come the mode of the personality distribution, and promptness and dilatoriness would be judged accordingly. If the incentive given by the conformity-producing agents push the "effective" time for arrival at the factory up to 6:30 A.M., the mode will be at this point, and personality differences may be expected again to array them-selves on each side of the mode at this point. This shifting of the standard by which personality differences are revealed and measured through changes in the situation we may call the principle of institutional or cultural relativity. It will be noted, however, that the personality differences cannot remain in their "normal" distribution under these unusual conditions, for strong biasing factors enter to upset what is otherwise a probability distribution. The mode will be higher, the curve skewed, and its slopes steepened. The individual differences will not range out so far upon the left, for they are op-posed by the factor of inertia. Ten minutes difference in time is a more acid test of promptness at 6:20 in the morning than it is at 9:00. On the right side of the curve also the personality differences will be "drawn in" by the strong incentive provided by the conformity-producing agents. Ten minutes of delay is likely to be more strongly avoided, and therefore less of an indicator of an individual's "natural" degree of procrastination, when it means the loss of pay than when there is no such stake involved. For these reasons the distribution as a whole becomes narrowed and positively accelerated upon both sides. Personality differences, however, though distorted in this way, are probably still present and are active in determining the outwardly sloping limbs of the curve. Those who lie near the mode anyway will be drawn into it by the conformity-producing factors. Those lying farther out will not be drawn into the mode but will form the "trailing-out" proportions of the double J.

(920) We have indicated these effects by the arrows (p) along the heavily dashed line in Figure1b. There exists some evidence in the data itself to support this interpretation.[15]

There must finally be mentioned a possible fourth type of factor represented in the empirical distribution. This factor comprises cases which occur at a certain point by simple chance. Unforeseen house-hold exigencies may delay any of the workers at any time. Street-cars may be delayed in following their schedules by unpredictable weather conditions. Household clocks or watches may be wrong without the knowledge of their owners. Since these factors are unrelated to any of the other influences determining the distribution, and since they may affect all parts of the population, we may think of them as tending to distribute the arrivals equally through-out the time range and, therefore, of lowering and flattening the distribution. This effect is indicated by the arrows (ch) in Figure b. The effect of the operation of this factor in the present distributions is probably comparatively small.

All these explanations can be summed up under the familiar statistical observation that normal probability curves are based upon a large number of unrelated, equally potent, and independently varying factors. Nonnormal or skewed distributions, such as those of our J-curves, are produced by the presence in the situation of a few related, dependently varying, and particularly potent factors acting in a given direction or directions. It would be inaccurate, however, to say that the skewed distributions are normal distributions into which these related and potent factors have subsequently entered. The normal curve does not really become skewed into a J-curve. The J-curve phenomenon is something which exists ab initio. The normal frequency curve in our explanation is introduced as a present factor in the distribution and not as a historical stage in the process of its formation. It is not maintained that these four factors are the only types of influence at work in producing conformity distributions out of the materials of human behavior. They are convenient broad classifications, however, which will probably be found applicable in varying degrees to all situations to which the

(919) concept of custom, folkways, and other conformity terms have been applied.

The asymmetrical, skewed, and highly modal character of conformity distributions cannot be adequately described as a measurable departure of the normal frequency curve. It must be regarded in its own right. It does not indicate something wrong or abnormal about data which would otherwise be normally distributed. It rep-resents the nature of the conformity pattern itself. This fact re-quires a reversal of some of our habits of statistical thinking in which we have based all our indices upon the normal probability curve. New criteria of distribution shape rather than central tendency are now required. The science of conformity is not the science of normal probability but of teleonomic or "voluntaristic" factors, which are the very antithesis of probability. It is not the science of the mean and standard deviation but of the position and steepness of the mode, the asymmetry of the distribution, and the shape of the curve.

Through the use of the continua and scales described above, as well as through the distribution of the observed data upon them, it is hoped that we shall be able to show whether conformity exists, in what way individuals are conforming, how much they conform, what individual variations of conformity exist, who are the individuals who wish the mass of people to conform, what conformity means to them, what influence they have, and, finally, what conformity means to the people who practice it, that is, what they are trying to do through this behavior in terms of their own self-expression, adjustments, and relationships.[16]



  1. Cf. R. E. Park and E. W. Burgess, Introduction to the Science of Sociology (Chicago, 1921), p.799.
  2. Edward Sapir, "Custom," Encyclopaedia of the Social Sciences.
  3. Cf. Daniel Katz and R. L. Schanck, Social Psychology (New York, 1938), p. 15.
  4. R. T. la Piere and P. R. Farnsworth, Social Psychology (New York, 1936), p. 59.
  5. W. G. Sumner, Folkways (Boston, 1906), pp. 2-8.
  6. W. G. Sumner and A. G. Keller, The Science of Society (New Haven, 1927-28), P. 33.

  7. Morris Ginsberg, "Conventions, Social," Encyclopaedia of the Social Sciences.
  8. Floyd H. Allport, "The J-Curve Hypothesis of Conforming Behavior," Journal of Social Psychology, V (May, 1934), 141-83; see also M. C. Dickens and R. S. Solomon, "The J-Curve Hypothesis: Certain Aspects Clarified," Sociometry, I (January—April, 1938),277—91.
  9. Cf. Floyd H. Allport, Institutional Behavior (Chapel Hill, N.C., 1933), pp. 13-15.
  10. In the article previously cited the phrase "field of conforming behavior" was used instead of "conformity situation." This terminology is misleading and should be changed. It suggests that there may be something inherently conforming in the nature of the behavior itself—that the primary, or one of the primary, motives is to conform. Many cases in the field may be those of individuals whose behavior is merely uniform with that of the others, with no feeling of conformity to a standard. In a large number of others who are aware of the standard, the motive is more probably desire to avoid the ill effects of not conforming than a desire for conformity per se.
  11. These data were collected by Dr. Milton C. Dickens. A full account of Dr. Dickens' investigation and its results are included in an unpublished doctoral thesis entitled, "The Prediction of J-Curves of Positive Acceleration in Fields of Conforming Behavior," on file in the Syracuse University Library.
  12. This interpretation is not based on conjecture alone but upon observations actually made by Dr. Dickens, and upon distributions found in other situations where there was not a barrier in the physical situation at the left of the mode.
  13. As shown by subsequent investigation, the term "steep" should be here omitted.
  14. The objection has been raised that the double-J—curve on empirical continua in conformity situations is merely a normal frequency curve the form of which has been concealed by the length of the intervals chosen, the combining of steps, or the manner or absence of smoothing. Unpublished data in hand prove that this contention is not valid. Logical considerations also render the objection untenable.
  15. Dickens, op. cit.
  16. This article as originally submitted contained an application of the method here described in the form of an investigation conducted by Dr. Richard Solomon upon the lengths of conversations in church, library reading-room, and fraternity living-room. Through the data presented, the continua used, and the distributions obtained in this study, it was shown in detail how the foregoing questions regarding conformity situations could be answered. Considerations of space, however, compelled the editors to omit this part of the article. The omitted portion, presented as a method and illustration of the study of social conformity, will appear under joint authorship with Dr. Solomon in the Journal of Abnormal and Social Psychology for October, 1939.

Valid HTML 4.01 Strict Valid CSS2