Paper prepared for the session on “History of Public Opinion Research” WAPOR, Berlin, September 19-21, 2007
By Hans L Zetterberg, ValueScope AB, Bromma, Sweden
After seventy years of polling it is well documented that pollsters have at times inspired and facilitated democratic processes and decisions (Lijphart 1999). At other and more recent times, they have often impeded the democratic process by rushing into print and TV with poll results on issues that the public has not yet had a reasonable chance to be informed about, that they have not discussed, particularly not discussed with those who could contribute more to the understanding of the issue, and about which they have not yet formed mature opinions. Such polls contribute to premature closure of debates, and they mislead news editors, political commentators, and legislators (cf. Yankelovich 1996). The urgency and speed that are the life-blood of journalism as well as its built-in limitation of space and time slots have corrupted much polling. Frankly, in the poll-rich countries, much opinion research in mass media is nowadays often mere conversation pieces or entertainment for viewers and readers, not useful guidance for politicians, nor reliable sources for social science. Something seems to have gone wrong in this opinion polling, and we might look for remedies.
A. Lawrence Lowell (1913) considered public opinion as the outcome of talk and arguments about a common issue among ordinary citizens in a situation where most everyone’s view had become known and influenced by everyone else’s view, i.e. a degree of public consensus of a real community or of a functioning group. The New England town meeting was a nearby inspiration, also to a Boston Brahmin as Lowell.
Lowell became president of Harvard University. He influenced his colleagues among political scientists and historians to study “public opinion” so defined. Also journalists began more regularly to have stories about opinions in local communities, the business community, in the military, in the church, the Washington establishment, university campuses, et cetera. In journalism Walter Lippmann (1922) pursued and developed Lowell’s conception of public opinion. Lowell’s writing was a stimulus to scholars of the Chicago School of Sociology – dominant in its field between the World Wars – to incorporate this conception of public opinion in their teaching, research, and theorizing. It fitted well with the conceptions developed by their founding father, Ezra Park.
The pollsters’ use of statistical samples of discrete individuals and demographic units of analysis rather than real communities and groups was the most serious criticism leveled at the young polling enterprise in the middle of the twentieth century. This critique by the Chicago sociologist Herbert Blumer (1948) is still cited. The critics give opinion research in national samples limited relevance for the serious study of the actual social and political processes. The necessary links to the relevant publics in society are simply not offered by a few demographic “background variables” such as age and sex.
At a symposium on polling in 1949, Clyde Hart, Head of the National Opinion Research Center (NORC) at Chicago University echoed Blumer’s view and admitted: “Our samples are always samples of discrete individuals in a population; they are not samples of a public” (Hart 1949, p 28). This seemingly innocuous remark points to the gap between the prior opinion theory and the new opinion research. To stay within the confines of interview surveys and nevertheless cope with this gap is the topic of this paper.
The only way I know for pollsters to manage the gap is by adding auxiliary issue-handling questions to each interview question about an issue. They aim to establish the extent to which a polled opinion about an issue is a product of a discussion in a public or merely an aggregation of idiosyncratic views. Those individuals who belong to publics that have discussed the issue at hand can then be tabulated separately; they represent opinions of publics.
The first efforts along these lines in Chicago occurred in small surveys, not the national polls of NORC.
Chicago-trained sociologist Arnold M. Rose, active at the University of Minnesota, was first to publish about the problem (Rose 1949, revised 1954). I was his research assistant in the academic year 1950-51 and can attest to how he struggled with the practical side of the problem. Here are some of the auxiliary questions he tried in local surveys:
R1. How long ago did you first hear about this issue?
R2. Have you ever discuss this issue with others?
R3. If yes, tell all the settings where you talked about the issue? Have you talked about this issue --
in your own family,
among your friends,
among your neighbors,
at your work place,
in a tavern, pub or restaurant,
in the meeting of an association,
at a church gathering
or in other places.
R4. If yes, did you discuss the issue over the phone, by mail, in personal contacts?
R5. In general, did the others in such talks have the same views about the issue as you, or other views?
R6. Do you think that this issue has been discussed enough as far as you are concerned, or do you welcome more discussion before you definitely make up your mind?
The question here numbered R2 “Have you ever discussed this issue with others?” is a direct outflow of Lowell’s idea that public opinions are a product of interactions in the encounters of citizens. The other questions are theoretically relevant, but do not get at the considered opinions in the big contexts – the business community, opinions in the military, in the church, the Washington establishment – that Lowell had in mind and that Lippmann wrote about.
George H. Gallup who (parallel to Elmo Roper) had introduced modern polling in the 1930s was not overly concerned that he did not conform to classical opinion conceptions.
He suggested three auxiliary background questions reflecting his interest in media:
G1. Have you heard about this issue before [this interview]?
G2. Have you noticed the issue being mentioned or discussed in newspapers?
G3. Have you searched for more information about this issue?
The first of these questions is absolutely fundamental in issue polling. If it had been consistently used, the amazingly high number of people who never heard of an issue until the interviewer asks would have classified many poll results as uninformed and saved equally uninformed editorial comments.
In the meetings of the International Gallup Association, Dr Gallup frequently encouraged his affiliates to use the above interview questions. I was a member of this Association 1970-1986 and attended most of its annual meetings. However, his appeals were rarely headed. At mid-century the Norwegian member in the Association, Björn Balstad, had invented “the omnibus survey” that saved the financing of media polling when the newspaper market changed to one-city-one-paper, and polls everywhere except a few metropolitan areas lost their place in the competition between papers. With a fixed calendar schedule and a published pricelist based on the number of interview questions, this efficient research vehicle spread rapidly among pollsters and market researchers. To have “an omnibus” became a criterion for membership in Gallup International Association. Needless to say, in polling on an issue, media wanted to pay only for the publishable question, not for any auxiliary ones.
Daniel Yankelovich became a foremost interpreter of American public opinion in the generation of pollsters that followed the pioneering ones of Gallup and Roper and those trained by them. Yankelovich had studied social psychology at the Sorbonne before he became involved in opinion research. He anchored public opinions more in personality than in social encounters and publics. His thinking about auxiliary questions is presented in a paper from the research house of Yankelovich, Skelly and White (1981) on immaturity and instability of many published opinions and on the resistance of some to accept new opinions. His team developed a “mushiness scale” based on four auxiliary questions:
Y1. “On a scale of 1 to 6, where 1 means that the issue affects you personally very little and 6 means that you really feel deeply involved in this issue, where would you place yourself?”
Y2. “On some issues people feel that they really have all the information that they need in order to form a strong opinion on that issue, while on other issues they would like to get additional information before solidifying their opinion. On a scale of 1 to 6, where 1 means that you feel you definitely need more information on the issue and 6 means that you do not feel you need to have any more information on the issue, where would you place yourself?”
Y3. “On the scale of 1 to 6, where 1 means that you and your friends and only rarely, if ever, discuss the issue and 6 means that you and your friends and family discuss it relatively often, where would you place yourself?”
Y4. “People have told us that on some issues they come to a conclusion and they stick with that position, no matter what. On other issues, however, they may take a position, that they know that they could change their minds pretty easily. On a scale of 1 to 6, where 1 means that you could change your mind very easily on this issue and 6 means that you are likely to stick with your position no matter what, where would you place yourself?”
Answers to question Y1 are often of interest to the general public and fit in most press releases. The third question deals with the same concern as among the Chicago sociologists; it is a compact form of questions R2 and R6 in the above list. The other questions refer in the main to personality variables. Question Y4 is a very important one. It relates to a notion that the American philosopher Ralph Waldo Emerson had presented in his classical essay “Self-Reliance” from 1841. He held that “What I must do is all that concerns me, not what the people think.” Emerson stressed that nothing at last is sacred but the integrity of your own mind, and that you should insist on yourself and never just imitate. The self-reliant people are not arrogant. They have developed a healthy resistance to the tendency to automatically be conformists.
Yankelovich, Skelly and White gave away their scale for free. They did not seem to have found any lasting takers. By and large the practice of pollsters throughout the twentieth century when doing issue polling for media remained deficient: one-shot questions without any information about the level of preparedness and involvement of the respondents.
In Europe, meanwhile, Elisabeth Noelle-Neumann (1980) reformulated the problem of Lowell’s conception of opinions in a theory she called "the spiral of silence." Her theory is based on events in everyday encounters.
The spiral of silence is demonstrated in a series of experiences which follow when a person enters into an encounter with strangers. As always when we deal with encounters we suffer from the phenomenon that the familiar blurs knowledge. Let’s therefore be pedantic in our analysis.
1. Newcomers to a social encounter will notice opinions prevailing among the others in the encounter. People have a kind of "social skin" says Noelle-Neumann. Rather quickly they can scan and sense a "climate of opinion" prevailing in the small talk of any social encounter. Her theorem of the social skin should take account of selective perception, and I would phrase it like this:
· In any social encounters, the participants (a) scan the others for the descriptive language in use, particularly utterances that present opportunities or threats for them, (b) scan the others for the evaluative language in use, particularly opinions about people such as themselves, and (c) scan the others for the prescriptive language in use, particularly for any norms that may apply to themselves.
In particular Noelle-Neumann uses the evaluative discourse (b) in encounters as a key to the following process of opinion change.
If the prevailing opinions in an encounter are alien to a newcomer's own
views, a normal reaction will be to withdraw from the encounter. There is
no need to withdraw from the conversation if the topic is innocent, like the
weather. But if it rages around partisan and alien political, religious, moral,
or artistic views some may want to withdraw.
This voluntary exclusion is an option on which Noelle-Neumann does not delve for long. Her favorite example of encounters stems from the old era of long trips on European railroads in which six to eight persons are seated face-to-face in compartments. It is difficult to withdraw from such compartments, and they do not facilitate staying silent like the seating arrangement in a bus or airplane. In the face-to-face confinement, however, an interesting process begins. Those who have never been on the old-fashioned trains will recognize the processes from dining parties that include strangers, or, perhaps also from academic seminars.
3. If the prevailing views in the confined encounter seem solid and rigid, then most newcomers begin to lean to the views expressed in the encounter. At least temporarily, the newcomers express some understanding of the views of the others, and may actually assimilate them as their own. A heated argument is thus avoided and the conversation remains polite.
4. If the prevailing opinions in the conversations in the encounter seem uncertain or in a state of fluctuation, the newcomers try to say something acceptable and reasonable that welcomes them among the others. They avoid the risk being excluded from the mainstream of the conversation and they avoid upsetting the exchange. The power of inclusion and exclusion is driven by a “fear of social isolation,” the key human motivation in Noelle-Neumann’s theory.
5. If newcomers feel that their own views are loosing ground in the conversation (or in society at large) they express them with increasing hesitation. Eventually they may turn silent. This is "the spiral of silence" that brings unsupported opinions to a premature death in encounters. It is easy to be a traitor to a losing cause.
6. If the newcomers find that their own views are in general ascendancy within the encounter they express them more freely and with more conviction. It is easier to argue in line with an ever more supportive climate of opinion than against the trend in the climate of opinion. It is easy to be a convert to a winning cause, the so called “bandwagon effect.”
7. People vary in their subjugation to these processes and in their mastery in using these processes. Noelle-Neumann (1983) measured "the strength of personality" to separate the leaders from the followers in the encounters forming opinion.
To identify opinion climate questions Noelle-Neumann uses mostly people’s evaluative scanning. The standard auxiliary question follows this paradigm:
N1. “What view do you think most people have about X, favorable or unfavorable?”
To identify opinion leaders Noelle-Neumann developed a scale with 10 self-rating questions and three background variables measuring “strength of personality”. Among these items are:
N2.1. I usually count on being successful in everything I do.
N2.2. I am rarely unsure about how I should behave.
N2.3. I like to assume responsibility.
N2.4. I like to take the lead when a group does things together.
N2.5. I enjoy convincing others of my opinions.
N2.6. I often notice that I serve as a model for others.
N2.7. I am good at getting what I want.
N2.8. I am often a step ahead of others.
N2.9. I own many things others envy me for.
N2.10. I often give others advice and suggestions.
The three background items are:
N2.A. Holder of a leading position in a profession/being superior.
N2.B. Participation in political party/trade union citizen’s action group in leisure time.
N2.C. Office held in a club or organization.
Translation by Gabriel Weimann (1994)
The scale can be included in any opinion survey, but this is rarely done outside the Institut für Demoskopie Allensbach. The scale covers effort to make others accept someone’s opinion. It is not the same as the Emerson-Yankelovich measurement Y4 that covers efforts to resist other people’s opinions. A better label for this scale might have been “Strength of Persuasive Personality.” To distinguish it from the Emerson- Yankelovich measurement of “Strength of Resistant Personality.” To fully pursue the psychological aspects of opinion formation one would have to use and develop both approaches.
The 21st century has seen a new departure in the conception of public opinion. It is seen as a spontaneous order. This brings it closer to Lowell’s original conception.
Notions of spontaneous orders are old. The Roman Empire did not like them. When Western Rome had collapsed, the theory of spontaneous orders developed at the University of Salamanca in Spain. The term and the phenomenon it stands for has in modern times been an important focus in the economic theory of Friedrich Hayek and his followers.
Everyone knows that we bargain when we interact in a marketplace. Hayek, however, focused on something different: the fact we gain information about how others evaluate goods and services in a marketplace of free trade. These informed prices constitute a spontaneous order. It contains more knowledge than a monopoly can have and certainly more knowledge than any government board for price control ever possesses. Likewise, voters may exchange political views while meeting for coffee; the issues may be a candidate’s competence or a clause in the party platform. Such conversations may add pieces of information to each participant. The resulting public opinion is a spontaneous order shaped by more knowledge than any single participant originally had.
Networks in different parts of society develop similar spontaneous orders – wikipedias in science, collective graffiti in the arts, prayers in religion. A prayer such as “Délivre-nous de la rage des Normans” (For the rage of the Northmen save us, O Lord) probably developed as individual appeals in the wake of invasions by brutes from the north; later they became ritualized in church services. When networks change their composition or when they meet changing conditions these spontaneous orders also change.
The rightmost column of the Table below presents some spontaneous orders and their contexts in a modern society.
Type of Freedom
Freedom of conscience
From Hans L Zetterberg, The Many-splendored Society (forthcoming).
Public opinion is the spontaneous order of the body politic. A theory of public opinion is thus a special case of the general theory of spontaneous orders.
There is a rapid advance of knowledge in the latter field. Yochai Benkler (2006) has written on how we develop and share new ideas by “common-based peer production” and how this process transforms markets and freedom. James Surowicki (2005) reviews the new vistas of collective intelligence under the old title of “the wisdom of crowds.”
Using this approach in opinion research requires use of various measures of qualities of the interaction in networks and assessments of their freedom of expressions. But it also requires at a minimum one auxiliary question in addition to the key question about an issue.
Z1. In discussing this issue have you met anyone who had a relevant point of view that you had not thought much of before?
We do not have enough studies using this type of question to evaluate its potential. Much inspiration to the view of informed opinion as a spontaneous order can be derived from the success of the Wikipedia as a spontaneous order of current scientific knowledge.
The innovation and testing of auxiliary questions in issue polling are an accomplished and published achievement of research. Existing auxiliary questions have not received any systematic implementation in polling for media. Serious pollsters for media should make it a rule to include the appropriate auxiliary questions and develop standards of opinion quality on their basis. I would suggest that the editors include the answers to Y1, at least for their own elucidation as editors. I would at present recommend five other auxiliary questions per issue studied for media publication, namely, the mandatory G1, R2, and, whenever a deeper understanding is desired, the N1, N2.5 and (a simplified) Y4.
A financing problem remains: Will any news media or other client pay for five (or more) auxiliary questions per issue studied, when only a percentage figure from the one question about the issue itself is newsworthy? The pricing structure of polling for the media may have to change so that clients pay a reasonable flat fee per issue studied, not per interview question.
The present practice of presenting an opinion on a political issue using one single question means that pollsters have no good answer to critics who claim that their polling is not a reliable tool of democracy, but rather a sham driven by commercial shortcuts, journalistic needs for entertaining content, and/or academic incompetence.
At last, let me admit that in this paper I have thrown stones sitting in a glass house. During my years as a full-time pollster (1970-1986) I failed many times to live up to the proper use of auxiliary questions.
Benkler, Yochai, (2006). The Wealth of Networks: How Social Production Transforms Markets and Freedom, Yale University Press, New Haven.
Emerson, Ralph Waldo, 1841/1993. Self-Reliance and Other Essays, Dover, New York.
Lijphart, Arend, (1999). Patterns of Democracy. Government Forms and Performance in Thirty-Six Countries, Yale University Press, New Haven
Lippmann, Walter, (1922). Public Opinion. Harcourt, Brace, New York, NY.
Lowell, A. Lawrence, (1913). Public Opinion and Popular Government. Longmans Green, New York.
Noelle-Neumann, Elisabeth, (1980). Die Schweigespirale. Öffentliche Meinung – unsere soziale Haut, Langen-Müller, München. English translation 1993 as The Spiral of Silence: Public Opinion, Our Social Skin, 2nd ed. University of Chicago Press, Chicago, ILL.
--“--, (1983) Persönlichkeitsstärke, SPIEGEL-Verlag, Hamburg.
Surowicki, James, (2004). The Wisdom of Crowds: Why the Many are Smarter than the Few and How Collective Wisdom Shapes Business, Economies, Societies, and Nations, Doubleday, New York.
Weimann, Gabriel and Elihu Katz, (1994). The Influentials. People Who Influence People, State University of New York Press, Albany. NY,
Yankelovich, Daniel, (1991). Coming to Public Judgment: Making Democracy Work in a Complex World, Syracuse University Press, Syracuse, NY.
–”–, (1996). "A New Direction for Survey Research," International Journal of Public Opinion Research, vol. 8, pp. 1-9.