Social Problems, Vol. 14, No. 3. (Winter, 1967), pp. 239-247.
WHOSE SIDE ARE WE ON?*
HOWARD S. BECKER

Northwestern University
T o have values or not to have values: the question is always with us.
When sociologists undertake to study problems that have relevance to the world we live in, they h d themselves caught in a crossfire. Some urge them not to take sides, to be neutral and do research that is technically correct and value free. Others tell them their work is shallow and useless if it does not express a deep commitment to a value position. This dilemma, which seems so painful to so many, actually does not exist, for one of its horns is imaginary. For it to exist, one would have to assume, as some apparently do, that it is indeed possible to do research that is uncontaminated by personal and political sympathies. I propose to argue that it is not possible and, therefore, that the question is not whether we should take sides, since we inevitably will, but rather whose side we are on.
I will begin by considering the problem of taking sides as it arises in the study of deviance. An inspection of this case will soon reveal to us features that appear in sociological research of all kinds. In the greatest variety of subject matter areas and in work done by all the different methods at our disposal, we cannot avoid taking sides, for reasons firmly based in social structure.
W e may sometimes feel that studies of deviance exhibit too great a sympathy with the peo le studied, a sympathy reflected in t e research carried out. This feeling, I sus e d , is entertained off and on both I! y those of us who do such research and by those of us who, our work lying in other areas,

R

"Presidential address, delivered at the annual meeting of the Society for the Study of Social Problems, Miami Beach, August,
1966.

only read the results. Will the research, we wonder, be distorted by that sympathy? Will it be of use in the construction of scientific theory or in the application of scientific knowledge to the practical problems of society? Or will the bias introduced by taking sides spoil it for those uses?
W e seldom make the feeling explicit. Instead, it appears as a lingering worry for sociological readers, who would like to be sure they can trust what they read, and a troublesome area of self-doubt for those who do the research, who would like to be sure that whatever sympathies they feel are not professionally unseemly and will not, in any case, seriously flaw their work. That the worry affects both readers and researchers indicates that it lies deeper than the superficial differences that divide sociological schools of thought, and that its roots must be sought in characteristics of society that affect us all, whatever our methodological or theoretical persuasion.
If the feeling were made explicit, it would take the form of an accusation that the sympathies of the researcher have biased his work and distorted his findings. Before exploring its structural roots, let us consider what the manifest meaning of the charge might be.
It might mean that we have acquired some sympathy with the group we study sufficient to deter us from publishing those of our results which might prove damaging to them. One can imagine a liberal sociologist who set out to disprove some of the common stereotypes held about a minority group. T o his dismay, his investigation reveals that some of the stereotypes are unfortunately true. In the interests of justice and liberalism, he might well be tempted, and might even succumb to the temptation, to suppress those findings, publishing with scientific

candor the other results which confirmed his beliefs.
But this seems not really to be the heart of the charge, because sociologists who study deviance do not typically hide things about the people they study. They are mostly willing to grant that there is something going on that put the deviants in the position they are in, even if they are not willing to grant that it is what the people they studied were originally accused of.
A more likely meaning of the charge, I think, is this. In the course of our work and for who knows what private reasons, we fall into deep sympathy with the people we are studying, so that while the rest of the society views them as unfit in one or another respect for the deference ordinarily accorded a fellow citizen, we believe that they are at least as good as anyone else, more sinned against than sinning.
Because of this, we do not give a balanced picture. W e focus too much on questions whose answers show that the supposed deviant is morally in the right and the ordinary citizen morally in the wrong. W e neglect to ask those questions whose answers would show that the deviant, after all, has done something pretty rotten and, indeed, pretty much deserves what he gets. In consequence, our overall assessment of the problem being studied is one-sided.
What we produce is a whitewash of the deviant and a condemnation, if only by implication, of those respectable citizens who, we think, have made the deviant what he is.
It is to this version that I devote the rest of my remarks. I will look first, however, not at the truth or falsity of the charge, but rather at the circumstances in which it is typically made and felt. The sociology of knowledge cautions us to distinguish between the truth of a statement and an assessment of the circumstances under which that statement is made; though we trace an argument to its source in the interests of the person who made it, we

have still not proved it false. Recognizing the point and promising to address it eventually, I shall turn to the typical situations in which the accusation of bias arises.
When do we accuse ourselves and our fellow sociologists of bias? I think an inspection of representative instances would show that the accusation arises, in one important class of cases, when the research gives credence, in any serious way, to the perspective of the subordinate group in some hierarchical relationship. In the case of deviance, the hierarchical relationship is a moral one. The superordinate parties in the relationship are those who represent the forces of approved and official morality; the subordinate parties are those who, it is alleged, have violated that morality.
Though deviance is a typical case, it is by no means the only one. Similar situations, and similar feelings that our work is biased, occur in the study of schools, hospitals, asylums and prisons, in the study of physical as well as mental illness, in the study of both
"normal" and delinquent youth. In these situations, the superordinate parties are usually the official and professional authorities in charge of some important institution, while the subordinates are those who make use of the services of that institution. Thus, the police are the superordinates, drug addicts are the subordinates; professors and administrators, principals and teachers, are the superordinates, while students and pupils are the subordinates ; physicians are the superordinates, their patients the subordinates.
All of these cases represent one of the typical situations in which researchers accuse themselves and are accused of bias. It is a situation in which, while conflict and tension exist in the hierarchy, the conflict has not become openly political. The conflicting segments or ranks are not organized for conflict; no one attempts to alter the shape of the hierarchy. While

Whose Side Are W e On? subordinates may com lain about the treatment they receive rom those above them, they do not propose to move to a position of equality with them, or to reverse positions in the hierarchy.
Thus, no one proposes that addicts should make and enforce laws for policemen, that patients should prescribe for doctors, or that adolescents should give orders to adults. W e can call this the apoliticd case.
In the second case, the accusation of bias is made in a situation that is frankly political. The parties to the hierarchical relationship engage in organized conflict, attempting either to maintain or change existing relations of power and authority. Whereas in the first case subordinates are typically unorganized and thus have, as we shall see, little to fear from a researcher, subordinate parties in a political situation may have much to lose. When the situation is political, the researcher may accuse himself or be accused of bias by someone else when he gives credence to the perspective of either party to the political conflict. I leave the political for later and turn now to the problem of bias in apolitical situation~.~
W e provoke the suspicion that we are biased in favor of the subordinate parties in an apolitical arrangement when we tell the story from their point of view. W e may, for instance, investigate their complaints, even though they are subordinates, about the way things are run just as though one ought to give their complaints as much credence as the statements of responsible officials. W e provoke the charge when we assume, for the pur-

?

1 No situation is necessarily political or apolitical. An apolitical situation can be transformed into a political one by the open rebellion of subordinate ranks, and a political situation can subside into one in which an accommodation has been reached and a new hierarchy been accepted by the participants. The categories, while analytically useful, do not represent a fixed division existing in real life.

241

poses of our research, that subordinates have as much right to be heard as superordinates, that they are as likely to be telling the truth as they see it as superordinates, that what they say about the institution has a right to be investigated and have its truth or falsity established, even though responsible officials assure us that it is unnecessary because the charges are false.
W e can use the notion of a hierarchy of credibility to understand this phenomenon. In any system of ranked groups, participants take it as given that members of the highest group have the right to define the way things really are. In any organization, no matter what the rest of the organization chart shows, the arrows indicating the flow of information point up, thus demonstrating (at least formally) that those at the top have access to a more complete picture of what is going on than anyone else. Members of lower groups will have incomplete information, and their view of reality will be partial and distorted in consequence.
Therefore, from the point of view of a well socialized participant in the system, any tale told by those at the top intrinsically deserves to be regarded as the most credible account obtainable of the organizations' workings. And since, as Sumner pointed out, matters of rank and status are contained in the mores,2 this belief has a moral quality.
W e are, if we are proper members of the group, morally bound to accept the definition imposed on reality by a superordinate group in preference to the definitions espoused by subordinates. (By analogy, the same argument holds for the social c!asses of a community.) Thus, credibility and the right to be heard are differentially distributed through the ranks of the system. As sociologists, we provoke the
2 William Graham Sumner, "Status in the Folkways," FolRway~,New York: New
American Library, 1960, pp. 72-73.

charge of bias, in ourselves and others, by refusing to give credence and deference to an established status order, in which knowledge of truth and the right to be heard are not equally distributed. "Everyone knows" that responsible professionals know more about things than laymen, that police are more respectable and their words ought to be taken more seriously than those of the deviants and criminals with whom they deal. By refusing to accept the hierarchy of credibility, we express disrespect for the entire established order.
W e compound our sin and further provoke charges of bias by not giving immediate attention and "equal time" to the apologies and explanations of official authority. If, for instance, we are concerned with studying the way of life inmates in a mental hospital build up for themselves, we will naturally be concerned with the constraints and conditions created by the actions of the administrators and physicians who run the hospital. But, unless we also make the administrators and physicians the object of our study (a possibility I will consider later), we will not inquire into why those conditions and constraints are present.
W e will not give responsible officials a chance to explain themselves and give their reasons for acting as they do, a chance to show why the complaints of inmates are not justified.
It is odd that, when we perceive bias, we usually see it in these circumstances. It is odd because it is easily ascertained that a great many more studies are biased in the direction of the interests of responsible officials than the other way around. W e may accuse an occasional student of medical sociology of having given too much emphasis to the complaints of patients.
But it is not obvious that most medical sociologists look at things from the point of view of the doctors? A few sociologists may be sufficiently biased in favor of youth to grant credibility

to their account of how the adult world treats them. But why do we not accuse other sociologists who study youth of being biased in favor of adults? Most research on youth, after all, is clearly designed to find out why youth are so troublesome for adults, rather than asking the equally interesting sociological question : "Why do adults make so much trouble for youth?" Similarly, we accuse those who take the complaints of mental patients seriously of bias; what about those sociologists who only take seriously the complaints of physicians, families and others about mental patients?
Why this disproportion in the direction of accusations of bias? Why do we more often accuse those who are on the side of subordinates than those who are on the side of superordinates ?
Because, when we make the former accusation, we have, like the well socialized members of our society most of us are, accepted the hierarchy of credibility and taken over the accusation made by responsible officials.
The reason responsible officials make the accusation so frequently is precisely because they are responsible. They have been entrusted with the care and operation of one or another of our important institutions: schools, hospitals, law enforcement, or whatever. They are the ones who, by virtue of their oficial position and the authority that goes with it, are in a position to "do something" when things are not what they should be and, similarly, are the ones who will be held to account if they fail to "do something" or if what they do is, for whatever reason, inadequate.
Because they are responsible in this way, officials usually have to lie. That is a gross way of putting it, but not inaccurate. Officials must lie because things are seldom as they ought to be.
For a great variety of reasons, wellknown to sociologists, institutions are refractory. They do not perform as society would like them to. Hospitals do not cure people; prisons do not re-

Whose Side Are We On? habilitate prisoners; schools do not educate students. Since they are supposed to, officials develo ways both of denying the failure o the institution to perform as it should and explaining those failures which cannot be hidden. An account of an institution's operation from the point of view of subordinates therefore casts doubt on the official line and may possibly expose it as a lie.3
For reasons that are a mirror image of those of officials, subordinates in an apolitical hierarchical relationship have no reason to complain of the bias of sociologica1 research oriented toward the interests of superordinates. Subordinates typically are not organized in such a fashion as to be responsible for the overall operation of an institution.
What happens in a school is credited or debited to the faculty and administrators; they can be identified and held to account. Even though the failure of a school may be the fault of the pupils, they are not so organized that any one of them is responsible for any failure but his own. If he does well, while others all around him flounder, cheat and steal, that is none of his affair, despite the attempt of honor codes to make it so. As long as the sociological report on his school says that every student there but one is a liar and a cheat, all the students will feel complacent, knowing they are the one exception. More likely, they will never hear of the report at all or, if they do, will reason that they will be gone before long, so what difference does it make?
The lack of organization among subor-dinate members of an institutionalized relationship means that, having no responsibility for the group's welfare, they likewise have no complaints if

!

3 I have stated a portion of this a r e ment more briefly in "Problems of Publication of Field Studies," in Arthur Vidich,
Joseph Bensman, and Maurice Stein (Eds.),
Reflections on Community Studies, New
York: John WiIey and Sons, 1964, pp. 267284.

243

someone maligns it. The sociologist who favors officialdom will be spared the accusation of bias.
And thus we see why we accuse ourselves of bias only when we take the side of the subordinate. It is because, in a situation that is not openly political, with the major issues defined as arguable, we join responsible officials and the man in the street in an unthinking acceptance of the hierarchy of credibility. W e assume with them that the man at the top knows best.
W e do not realize that there are sides to be taken and that we are taking one of them.
The same reasoning allows us to understand why the researcher has the same worry about the effect of his sympathies on his work as his uninvolved colleague. The hierarchy of credibility is a feature of society whose existence we cannot deny, even if we disagree with its injunction to believe the man at the top. When we acquire sufficient sympathy with subordinates to see things from their perspective, we know that we are flying in the face of what "everyone knows." The knowledge gives us pause and causes us to share, however briefly, the doubt of our colleagues.
When a situation has been defined politically, the second type of case I want to discuss, matters are quite different. Subordinates have some degree of organization and, with that, spokesmen, their equivalent of responsible officials. Spokesmen, while they cannot actually be held responsible for what members of their group do, make assertions on their behalf and are held responsible for the truth of those assertions. The group engages in political activity designed to change existing hierarchical relationships and the credibility of its spokesmen directly affects its political fortunes. Credibility is not the only influence, but the group can ill-afford having the definition of reality proposed by its spokesmen discredited, for the immediate conse-

quence will be some loss of political power. Superordinate groups have their spokesmen too, and they are confronted with the same problem: to make statements about reality that are politically effective without being easily discredited. The political fortunes of the superordinate group-its ability to hold the status changes demanded by lower groups to a minimum-do not depend as much on credibility, for the group has other kinds of power available as well.
When we do research in a political situation we are in double jeopardy, for the spokesmen of both involved groups will be sensitive to the implications of our work. Since they openly conflicting definitions o reality, our statement of our problem is in itself likely to call into question and make problematic, at least for the purposes of our research, one or the other definition. And our results will do the same. The hierarchy of credibility operates in a different way in the political situation than it does in the apolitical one. In the political situation, it is precisely one of the things at issue.
Since the political struggle calls into question the legitimacy of the existing rank system, it necessarily calls into question at the same time the legitimacy of the associated judgments of credibility. Judgments of who has a right to define the nature of reality that are taken for granted in an apolitical situation become matters of argument. Oddly enough, we are, I think, less likely to accuse ourselves and one another of bias in a political than in an apolitical situation, for at least two reasons. First, because the hierarchy of credibility has been openly called into question, we are aware that there are at least two sides to the story and so do not think it unseemly to investigate the situation from one or another of the contending points of view. W e

propose

know, for instance, that we must grasp the perspectives of both the resident of Watts and of the Los Angeles policeman if we are to understand what went on in that outbreak.
Second, it is no secret that most sociologists are politically liberal to one degree or another. Our political preferences dictate the side we will be on and, since those preferences are shared by most of our colleagues, few are ready to throw the first stone or are even aware that stone-throwing is a possibility. W e usually take the side of the underdog; we are for Negroes and against Fascists. W e do not think anyone biased who does research designed to prove that the former are not as bad as people think or that the latter are worse. In fact, in these circumstances we are quite willing to regard the question of bias as a matter to be dealt with by the use of technical safeguards. W e are thus apt to take sides with equal innocence and lack of thought, though for different reasons, in both apolitical and political situations. In the first, we adopt the commonsense view which awards unquestioned credibility to the responsible official.
(This is not to deny that a few of us, because something in our experience has alerted them to the possibility, may question the conventional hierarchy of credibility in the special area of our expertise.) In the second case, we take our politics so for granted that it supplants convention in dictating whose side we will be on. (I do not deny, either, that some few sociologists may deviate politically from their liberal colleagues, either to the right or the left, and thus be more liable to question that convention.)
In any event, even if our colleagues do not accuse us of bias in research in a political situation, the interested parties will. Whether they are foreign politicians who object to studies of how the stability of their government may be maintained in the interest of

Whose Side Are We On? the United States (as in the Camelot aflair)4 or domestic civil rights leaders who object to an analysis of race problems that centers on the alleged deficiencies of the Negro family (as in the reception given to the Moynihan
Report) ,6 interested parties are quick to make accusations of bias and distortion. They base the accusation not on failures of technique or method, but on conceptual defects. They accuse the sociologist not of getting false data but of not getting all the data relevant to the problem. They accuse him, in other words, of seeing things from the perspective of only one party to the conflict. But the accusation is likely to be made by interested parties and not by sociologists themselves.
What I have said so far is all sociology of knowledge, suggesting by whom, in what situations and for what reasons sociologists will be accused of bias and distortion. I have not yet addressed the question of the truth of the accusations, of whether our findings are distorted by our sympathy for those we study. I have implied a partial answer, namely, that there is no position from which sociological research can be done that is not biased in one or another way.
W e must always look at the matter from someone's point of view. The scientist who proposes to understand society must, as Mead long ago pointed out, get into the situation enough to have a perspective on it. And it is likely that his perspective will be greatly affected by whatever positions are taken by any or all of the other participants in that varied situation.
Even if his participation is limited to reading in the field, he will necessarily read the arguments of partisans of one
4 See Irving Louis Horowitz, "The Life and Death of Project Camelot," Transaction, 3 (Nov./Dec., 1965), pp. 3-7, 44-47.
5 See Lee Rainwater and William L.
YanceyA "Black Families and the White
House, ibid., 3 (July/August, 1966, pp.
6-11, 48-53).

245

or another side to a relationship and will thus be affected, at least, by having suggested to him what the relevant arguments and issues are. A student of medical sociology may decide that he will take neither the perspective of the patient nor the perspective of the physician, but he will necessarily take a perspective that impinges on the many questions that arise between physicians and patients; no matter what perspective he takes, his work either will take into account the attitude of subordinates, or it will not. If he fails to consider the questions they raise, he will be working on the side of the officials. If he does raise those questions seriously and does find, as he may, that there is some merit in them, he will then expose himself to the outrage of the officials and of all those sociologists who award them the top spot in the hierarchy of credibility.
Almost all the topics that sociologists study, at least those that have some relation to the real world around us, are seen by society as morality plays and we shall find ourselves, willy-nilly, taking part in those plays on one side or the other.
There is another possibility. W e may, in some cases, take the point of view of some third party not directly implicated in the hierarchy we are investigating. Thus, a Marxist might feel that it is not worth distinguishing between Democrats and Republicans, or between big business and big labor, in each case both groups being equally inimical to the interests of the workers.
This would indeed make us neutral with respect to the two groups at hand, but would only mean that we had enlarged the scope of the political conflict to include a party not ordinarily brought in whose view the sociologist was taking.
W e can never avoid taking sides.
So we are left with the question of whether taking sides means that some distortion is introduced into our work so great as to make it useless. Or, less

drastically, whether some distortion is introduced that must be taken into account before the results of our work can be used. I do not refer here to feeling that the icture given by the research is not "~alanced," the indignation aroused by having a conventionally discredited definition of reality given priority or equality with what "everyone knows," for it is clear that we cannot avoid that. That is the problem of officials, spokesmen and interested parties, not ours. Our problem is to make sure that, whatever point of view we take, our research meets the standards of good scientific work, that our unavoidable sympathies do not render our results invalid.
W e might distort our findings, because of our sympathy with one of the parties in the relationship we are studying, by misusing the tools and techniques of our discipline. W e might introduce loaded questions into a questionnaire, or act in some way in a field situation such that people would be constrained to tell us only the kind of thing we are already in sympathy with. All of our research techniques are hedged about with precautionary measures designed to guard against these errors. Similarly, though more abstractly, every one of our theories presumably contains a set of directives which exhaustively covers the field we are to study, specifying all the things we are to look at and take into account in our research. By using our theories and techniques impartially, we ought to be able to study all the things that need to be studied in such a way as to get all the facts we require, even though some of the questions that will be raised and some of the facts that will be produced run counter to our biases. But the question may be precisely this. Given all our techniques of theoretical and technical control, how can we be sure that we will apply them impartially and across the board as they need to be applied? Our textbooks in

methodology are no help here. They tell us how to guard against error, but they do not tell us how to make sure that we will use all the safeguards available to us. W e can, for a start, try to avoid sentimentality. W e are sentimental when we refuse, for whatever reason, to investigate some matter that should properly be regarded as problematic. W e are sentimental, especially, when our reason is that we would prefer not to know what is going on, if to know would be to violate some sympathy whose existence we may not even be aware of. Whatever side we are on, we must use our techniques impartially enough that a belief to which we are especially sympathetic could be proved untrue. W e must always inspect our work carefully enough to know whether our techniques and theories are open enough to allow that possibility.
Let us consider, finally, what might seem a simple solution to the problems posed. If the difficulty is that we gain sympathy with underdogs by studying them, is it not also true that the superordinates in a hierarchical relationship usually have their own superordinates with whom they must contend? Is it not true that we might study those superordinates or subordinates, presenting their point of view on their relations with their superiors and thus gaining a deeper sympathy with them and avoiding the bias of one-sided identification with those below them?
This is appealing, but deceptively so.
For it only means that we will get into the same trouble with a new set of officials. It is true, for instance, that the administrators of a prison are not free to do as they wish, not free to be responsive of the desires of inmates, for instance. If one talks to such an official, he will commonly tell us, in private, that of course the subordinates in the relationship have some right on their side, but that they fail to understand that his desire to do better is frustrated

W h o s e Side Are We On?

247

by his superiors or by the regulations conditions are the same elsewhere. I they have established. Thus, if a prison refer to a more sociological disclaimer administrator is angered because we in which we say, for instance, that we take the complaints of his inmates have studied the prison through the seriously, we may feel that we can get eyes of the inmates and not through around that and get a more balanced the eyes of the guards or other inpicture by interviewing him and his volved parties. W e warn people, thus, associates. If we do, we may then write that our study tells us only how things a report which his superiors will re- look from that vantage point-what spond to with cries of "bias." They, kinds of objects guards are in the does not atin their turn, will say that we have not prisoners' world-and presented a balanced picture, because tempt to explain why guards do what we have not looked at their side of it. they do or to absolve the guards of
And we may worry that what they say what may seem, from the prisoners' side, morally unacceptable behavior. is true.
The point is obvious. By pursuing This will not protect us from accusathis seemingly simple solution, we tions of bias, however, for the guards arrive at a problem of infinite regress. will still be outraged by the unbalanced
For everyone has someone standing picture. If we implicitly accept the above him who prevents him from conventional hierarchy of credibility, doing things just as he likes. If we we will feel the sting in that accusation. question the superiors of the prison
It is something of a solution to say administrator, a state department of that over the years each "one-sided" corrections or prisons, they will com- study will provoke further studies plain of the governor and the legisla- that gradually enlarge our grasp of all ture. And if we go to the governor the relevant facets of an institution's and the legislature, they will complain operation. But that is a long-term soluof lobbyists, party machines, the public tion, and not much help to the inand the newspapers. There is no end dividual researcher who has to contend to it and we can never have a "bal- with the anger of officials who feel he anced picture" until we have studied has done them wrong, the criticism of all of society simultaneously. I do not those of his colleagues who think he is propose to hold my breath until that presenting a one-sided view, and his own worries. happy day.
W e can, 1 think, satisfy the demands
What do we do in the meantime? of our science by always making clear I suppose the answers are more or less the limits of what we have studied, obvious. W e take sides as our personal marking the boundaries beyond which and political commitments dictate, use our findings cannot be safely applied. our theoretical and technical resources
Not just the conventional disclaimer, in which we warn that we have only to avoid the distortions that might studied a prison in New York or Cali- introduce into our work, limit our confornia and the findings may not hold clusions carefully, recognize the hierarchy of credibility for what is is, and in the other forty-nine states-which is not a useful procedure anyway, since field as best we can the accusations the findings may very well hold if the and doubts that will surely be our fate.