549 PUBLIC OPINION AND PUBLIC OPINION POLLING
paste, going to motion picture shows, and reading newspapers. Such actions, which I like to think of as mass actions of individuals in contrast to organized actions of groups, lend themselves readily to the type of sam-pling that we have in current public opinion polling. In fact, it is the existence of such mass actions of individuals which explains, in my judgment, the successful use in consumer research of sampling such as is employed in public opinion polling. What I find question- able, and what this paper criticizes, is the use of such sampling with its implicit imagery and logic in the study of a matter which, like the process of public opinion, functions as a moving organization of interconnected parts. The last item I wish to consider briefly
refers to the interesting and seemingly baf- fling question of how one should or can sam- ple an object matter which is a complicated system of interacting parts, having differen- tial influence in the total operation. Perhaps the question in itself is absurd. At various times I have asked different experts in sampling how one would sample an organic structure. With a single exception these indi- viduals looked a t me askance as if the ques- tion were idiotic. But the problem, I think, remains even though I find it difficult to state. In human society, particularly in mod- ern society, we are confronted with intricate complexes of moving relations which are roughly recognizable as systems, even though loose systems. Such a loose system is too complicated, too encumbered in detail and too fast moving to be described in any one of its given "cycles" of operation adequately and faithfully. Yet unless we merely want to speculate about it we have to dip into it in some manner in order to understand what is happening in the given cycle of operation in which we are interested. Thus, using the pub- lic opinion process in our society as an illus- tration we are able to make a rough charac- terization as to how it functions in the case, let us say, of a national issue. However, if we want to know how it functions in the case of a given national issue, we are a t a loss to make an adequate description because of the complexity and quick movement of the cycle of its operation. So, to know what is going on, particularly to know what is likely to go
on in the latter stages, we have to dip in here and there. The problems of where to dip in, how to dip in, and how far to dip in are what I have in mind in speaking of sampling an organic structure. I suppose, as one of my friends has pointed
out, that the answer to the problem requires the formulation of a model. We have no such model in the instance of public opinion as it operates in our society. My own hunch is that such a model should be constructed. if it can be a t all, by working backwards instkad of by working forward. That is, me ought to begin with those who have to act on public opinion and move backwards along the lines of the various expressions of public opinion that come to their attention, tracing these expres- sions backward through their own various channels and in doing so, noting the chief channels, the key points of importance, and the way in which any given expression has come to develop and pick up an organized backing out of what initially must have been a relatively amorphous condition. Perhaps, such a model, if it could be worked out, would allow the develo~ment of a realistic method of sampling in place of what seems to me to be the highly artificial method of sampling used in current public opinion poll- ing.
DISCUSSION THEODOREM. NEWCOMB University of Michigan
Professor Blumer has long been known as a formidable critic, and I am sure there are many others who shared with me the anticipation of seeing him turn his battery of high-powered guns upon the practitioners of "public opinion." In my judgment, his guns in this instance have misfired. This is not to say that those who study attitudes by sampling methods are beyond criti- cism; even a lesser critic than Professor Blumer could point to many shortcomings on their part. It may truly be said of all of them, probably, that in one way or another they have done what they ought not to have done and that they have not done what they ought to have done, and in some of them there is little health indeed. His target was thus an easy one, and I want to raise the questioil of why it is that he has neverthe- less, in my judgement, missed it. The first reason, I think, is that he was not
quite selective enough in his aim. I wish he had
AMERICAN SOCIOLOGICAL REVIEW
not tried to direct his fire toward all the people who, to use his phrase, use polling as a device, all at once. The most conscientious of them dif- fer so much from those who are less so that I can hardly think of what it is that they have in common. Perhaps it is that they all ask ques- tions, and that they all do some sort of selecting of the individuals whom they question. But I had supposed that it is quite as important to "identify the class" of people one is criticizing as to do so for concepts one is discussing. If professor Blumer had aimed at a sharply defined segment instead of the entire spectrum, I suspect he might have hit it. Some of those who use "pol-ling" devices are, as he says, preoccupied with techniques. A good many of them reveal "an ab- sence of using specific studies to test a general proposition." Some of them, I suppose, do not even know whether their respondents are arch- bishops or intinerant laborers. The shotgun ap- proach is bound to hit somebody, but there are so many whom it does not touch that the shot must be called a miss. No one in this audience, a t least, will feel that the shot could fairly be aimed at him.
Secondly, I wish Professor Blumer's major target had been a real one instead of an illusory one. I just don't know of anyone who is trying to study by "polling" methods what he described as public opinion. Let me restate his very spe- cific definition of what public opinion "in any realistic sense" is: "the pattern of the diverse views and positions on the issue that come to the individuals who have to act in response" to that pattern (Blumer's italics). This is one of many possible "indentifications as a class" of phenomena. There have been many other identi- fications of the class which goes by the same name in various times and in various places. But it is not the kind of "generic object" which American "pollers" have in mind during the 1940)s. In fact, I see nothing in common be-tween Professor Blumer's "generic object" and that of any "pollers" whom I know, except that the words "public opinion" are involved. In short, I think this shot was aimed at the phan- tom of a word and not at any real target at all. I think, further, that when he judges that "pol- ling" is unable "to isolate 'public opinion' as an abstract or generic concept," he really means that he prefers his own formulation of the con- cept to that of others. For this he is scarcely to be blamed; I too prefer my own formulations to those of others. But this is not to say that others do not have them. This shot went wild, I think, because the target just wasn't there to be shot at. Nobody in this audience, again, is
trying to do the things which "they" are criti- cized for doing badly.
Another of his shots went astray, in my judg- ment, because it was aimed at two targets blurred together in his finder, and so both were missed. "The inherent deficiency of public opin- ion polling," he says, "is contained in its sam-pling procedure." And yet his comments under this heading have largely to do with the kind of questions asked by those who use sampling pro- cedures. I wish he had shown some inherent connection between sampling and failure to obtain adequate information.
All investigators necessarily sample. Some of them are systematic enough about it so that they know what universe their sample repre-sents, and within what margin of probable er-ror, and some are not. As Professor Blumer himself says, "we have to dip in here and there." Careful investigators, since they cannot ask everything of everyone, plan selectively as to what they will ask of whom, so that the one will be relevant to the other. I t is quite as necessary for Professor Blumer to sample when he studies his version of public opinion as it is for anyone else. Since you sample, I sample, Professor Blumer samples and we all sample, it behooves all of us to do three things: to know what universe our population represents with what degree of certainty; to make sure that our questions are relevant to that universe; and to get all the relevant information that we can.
I know of no reasons why relevant and ade- quate information cannot be obtained from in- dividuals systematically selected by sampling methods quite as well as from unsystematically selected samples. Professor Blumer would like to know "whether individuals in the sample represent that portion of a structured society that is participating in the formation of public opinion on a given issue." Very well. There are many avenues of investigation open to him, all of which will involve sampling. He may, e.g., hypothesize that certain sorts of individuals, or members of certain sorts of groups, do "partici- pate in the formation of public opinion on a given issue," while others do not. Let him then sample both universes, obtain the relevant infor- mation, and either confirm or discard his hy- pothesis. Or, alternatively, let him sample the total adult population, obtain the relevant in-formation-as to individual characteristics, group memberships, activities which bring pres- sure upon policy makers, etc., et~.-and again confirm or reject the hypothesis. Or he may draw up a hypothesis about the u7ays in which leaders, policy makers and representatives "as-
551 PUBLIC OPINION AND PUBLIC OPINION POLLING
sess the public opinion as it comes to their attention." If so, let him draw a sample of such individuals, obtain the relevant information from them, and test his hypothesis. Very few investigators of "public opinion," I
agree, have put such hypotheses to the test. But I fail to see the slightest incompatibility be- tween sampling (whether professional or ama-teur) and the obtaining of relevant information. If Professor Blumer is not "able to determine [that there is any] way in current public opinion polling" to obtain the relevant informa- tion, I think he has simply not turned his abili- ties in this direction. I will not say that he prefers not to see any way in which this can be done, but I do think he shows a curious reluc- tance to try.
Society, I agree, is in many ways like a biological organism. I t functions, as Professor Blumer says, not as an aggregation of inter-changeable parts but as a dynamic configuration of member parts related by gradients, nodes of influence, etc. We must study not only the parts but also their modes of interdependence. But these interdependent processes and functions do not occur independently of human beings. We know about them only by observing people, not just as discrete individuals but as members of interrelated groups. We must observe them in order to answer our questions, but we cannot observe all of them. Both our record-taking and our sampling will be better if they are sys-tematic than if they are not. But systematic sampling and systematic record-taking do not preclude the possibility of choosing relevant groups to sample nor of obtaining information relevant to functioning in a dynamic society. The two targets-sampling and relevant infor-mation-are so unrelated that they cannot serve simultaneously as a single target. Finally, I think Professor Blumer's shots
went wild because his guns were badly mounted. He has chosen as a foundation the method of first "isolating the generic object" following which propositions "could" be formed and, presumably, be tested. I think that he has put the cinders on top of the concrete, and that you don't get a firm foundation that way. I wish he had done it the other way around. The issue on which we disagree is one about which scien- tists have never been completely agreed. I h a p pen to believe that Professor Blumer's stand is one which delays scientific progress. But it is a reasonable stand, and even his mistreatments of fact about the limitations of "polling" do not weaken it. The issue in its broadest form is that of just
how theory and research most effectively con-tribute to each other. In the more specific form in which he has posed it, it is the issue of how concepts and verified facts mutually support each other. At the one extreme in regard to this issue are those who believe that you must first think out a master theory and then find ways to test it. At the other extreme are those who believe in gathering mountains of verified facts which somehow will eventually fit themselves into a master theory. I do not accuse Professor Blumer of belonging with the first group of ex-tremists, but I think his position as outlined in this paper is too close to that extreme to result either in good theory or in good research. None of our older-brother sciences has devel-
oped any body of useful theory except by set- ting itself manageable problems-i.e., hy-potheses limited enough to be tested by sys-tematic observation. Physicists never succeeded in "isolating generic objects" of importance to them (if, indeed, they have yet done so) except as they set themselves limited problems, e.g., concerning falling bodies under specified con-ditions. Geneticists did not first sit down and dream up the concept of the gene before en-tering their laboratories. The history of science is studded with concepts which are wrong or half wrong, useless or half useless. This is no argument against concepts-quite the reverse. I t is an argument in behalf of constant refine- ment of concepts in the crucible of laboratory and field investigation. Concepts which are "iso- lated" a priori are of unknown usefulness until they are put to work, until i t is shown that more fruitful hypotheses are successfully tested with them than without them. To assert that one kind of concept is more right or more "rea- listic" than another, before it has been put to this kind of test, is to remain at the level of dogmatic assertion. Let me develop this point one short step
further. Concepts may be developed either a t the phenotypical or at the genotypical level. That is, "a class of empirical items" may be "identified" either on the basis of apparent similarities or on the basis of necessarily inter- dependent factors. For example, schizophrenic insanity may be genotypically like "normal introversion" though phenotypically more like manic-depressive insanity. Sadism and masoch- ism, psychiatrists tell us, are genotypically similar, though phenotypically at opposite...