international journal of cultural studies 1999 lewis 199 221

http://ics.sagepub.com/Studies

International Journal of Cultural

http://ics.sagepub.com/content/2/2/199The online version of this article can be found at:

DOI: 10.1177/136787799900200203

1999 2: 199International Journal of Cultural StudiesJustin Lewis

The opinion poll as a cultural form

Published by:

http://www.sagepublications.com

can be found at:International Journal of Cultural StudiesAdditional services and information for

http://ics.sagepub.com/cgi/alertsEmail Alerts:

http://ics.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://ics.sagepub.com/content/2/2/199.refs.htmlCitations:

What is This?

- Aug 1, 1999Version of Record >>

at UNIVERSIDAD DE CHILE on April 14, 2014ics.sagepub.comDownloaded from at UNIVERSIDAD DE CHILE on April 14, 2014ics.sagepub.comDownloaded from

http://ics.sagepub.com/


http://ics.sagepub.com/content/2/2/199

http://ics.sagepub.com/content/2/2/199



http://ics.sagepub.com/cgi/alerts

http://ics.sagepub.com/cgi/alerts

http://ics.sagepub.com/subscriptions

http://ics.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.sagepub.com/journalsPermissions.nav

http://ics.sagepub.com/content/2/2/199.refs.html

http://ics.sagepub.com/content/2/2/199.refs.html

http://ics.sagepub.com/content/2/2/199.full.pdf

http://ics.sagepub.com/content/2/2/199.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml

http://online.sagepub.com/site/sphelp/vorhelp.xhtml





A R T I C L E

INTERNATIONALjournal of

CULTURAL studies

Copyright © 1999 SAGE PublicationsLondon, Thousand Oaks,

CA and New DelhiVolume 2(2): 199–221

[1367-8779(199908)2:2; 199–221; 008931]

The opinion poll as a cultural form

● Justin Lewis

University of Massachusetts

A B S T R A C T ● The article examines public opinion polls from a culturalstudies perspective. Polls are discussed as a cultural form, a system ofrepresentation that attempts to signify the public. The article explores thedifferences between the products of polling agencies and the interpretation ofpolls in the mainstream media. It is argued that for all its attempts at closure, theopinion poll discourse is a fairly ‘open’ text, signifying a far wider range ofpolitical ideas than is found in mainstream political discourse. Once theserepresentations of the public are re-presented in mainstream media, thediscourse narrows: responses that might be articulated within a progressivepolitical framework are generally ignored or repressed. The article examines thisform of closure and concludes by proposing that a pro-corporate, center-righthegemony in the field of public opinion is a narrow, strategic project rather thana broad ideological victory. ●

K E Y W O R D S ● cultural form ● cultural studies ● opinion polls

Cultural studies has tended to focus on particular moments in the struggleover meaning – not so much to generalize from the particular but to lookfor the general in the particular. My intention here is to take the questionof hegemony and resistance to a broader level through an analysis of publicopinion. This is not an attempt to ‘settle’ the matter or to declare a winnerin the struggle between dominance and resistance, merely an effort to moveanalysis to a site that has been largely ignored by cultural studies. I am not,in so doing, trying to shift cultural studies away from a qualitative focus on

03 Lewis (to) D 28/6/99 12:21 pm Page 199

at UNIVERSIDAD DE CHILE on April 14, 2014ics.sagepub.comDownloaded from

http://www.sagepub.co.uk/

http://www.sagepub.co.uk/journals/details/j0214.html



200

specificity, but it is difficult to discuss forms of political power without, atsome point, implicating pluralities or majorities within civil society (Lewis,1997). As Andy Ruddock argues, ‘the very notion of culture depends on therelative coherence of meaning within a given society’, and that while differ-ence is integral to any sophisticated analysis of media and audiences, ‘so toois the recognition that the possibilities for difference are not boundless’(Ruddock, 1998: 122).

In what follows I shall argue that we might begin to explore the opinionpoll not purely in terms of its veracity but as a cultural form. We shouldtherefore consider not only polls themselves, but the way they are repre-sented. Opinion research, I shall try to show, are a site in which hegemonyis secured through the double signification of media frameworks, wherebyrepresentations of the public (the poll) are represented in media discourse.

Critical approaches to the opinion poll

It is not only the misanthropic who regard the routine expressions of publicopinion that now litter public discourse with a certain contempt. Theopinion poll is, for many, the epitome of empiricist social science. Just asacademics from various disciplines become more adept at picking themapart, the more abundant they become. Opinion polls are, in this meta-phorical sense, the sores of our political culture, and the modern body politicis ridden with them.

Criticism of opinion polling is nothing new. In 1948 Herbert Blumerargued persuasively that polls do not so much represent public opinion asmanufacture it, and that public opinion was something more than a columnof percentages derived from pre-coded questionnaires. The instrument ofmeasurement, he observed, was being mistaken for the thing itself (Blumer,1948). Later, Habermas argued that polls were not an expression of demo-cratic will but a substitute for it, since they curtailed the discursive con-ditions necessary for the development of ideas in the public realm(Habermas, 1989).

The spread of post-structuralism in European universities broadened thegulf between the academy and the pollster. In the late 1970s the French soci-ologist Pierre Bourdieu (who, unlike some of his contemporaries, was notaverse to the use of quantitative data) issued one of the more comprehen-sive dismissals. ‘Public opinion’, Bourdieu argued, ‘does not exist’ – not, atleast, in the pseudo-scientific form of the opinion survey (Bourdieu, 1979).Following Bourdieu, post-structuralist critics have seen polls as the mythicconstructions of a modernist age, the products of a series of epistemologi-cal and methodological blunders whose factual appearance signifies a smugrationality.

Scepticism is not confined to post-structuralism: many of those political

INTERNATIONAL journal of CULTURAL studies 2(2)

03 Lewis (to) D 28/6/99 12:21 pm Page 200




scientists who take polling seriously have, for some time, been aware of aslew of problems with the polling apparatus such as the disparity inresponse that can be generated by something as basic as question wording(see Schuman and Presser, 1981) or by apparently innocuous informationgiven by the interviewer. Sussman (1986), for example, found thatAfrican–American respondents were more likely to give negative evalu-ations of President Reagan if they were told they were taking part in asurvey of black American attitudes – thereby triggering a specific discur-sive context (Reagan’s treatment of the black community) in which Reaganwas less likely to appear favorable.

Some of the more notable recent work on public opinion has abandonedthe idea that polls are, in any simple sense, a reflection of independently con-structed, rationally formed views (see, for example, Page and Shapiro, 1992;Zaller, 1992). Delli Carpini and Keeter (1992), for example, argue that theresponses to polls depend upon an (unequal) distribution of knowledge, andtheir meaning – or lack thereof – is thus conditional. In short, we need tounderstand what people know to make sense of what they think. Thissignals an important move away from a long-standing empiricist distinctionbetween knowledge and opinion. Thus agenda-setting research, founded onsuch a distinction, has also begun to problematize the split between know-ledge and opinion as two independent categories of discourse (McCombs etal., 1995).

In another useful critical move, Salmon and Glasser (1995) haveextended Habermas’s approach by characterizing polls as affirmations (orrefusals) given in response to questions framed by elites, rather than therepresentation of more substantive or discursive processes. Pursuing aslightly different course, Wilson and Hodges (1992) have followed Bour-dieu to question the traditional notion that opinions are fixed, discretelyformed entities that we pluck from our mental files upon solicitation, whileJohn Zaller has suggested that we abandon ‘the notion that individualstypically possess preformed attitudes that they simply reveal when askedto do so’ (1992: 54).

Others have advanced a less negotiated position, questioning the verymotivation of such an endeavor – thus (following Foucault, 1977) themechanics of quantitative surveys are regarded as less an object of studythan an exercise of power and control. So, just as Ien Ang has argued thatthe TV ratings system is no longer understood as a way of finding out whatpeople think about television but as a mechanism for defining audiences inthe narrow terms of commercial television (Ang, 1991), Susan Herbst hasdiscussed polls as attempts at manipulation rather than enlightenment. Inthis vein, Susan Herbst’s work reveals the poignant irony that people them-selves feel alienated from – rather than represented by – the official discourseof ‘public opinion’ (Herbst, 1993).

Notwithstanding these critiques of a flawed, undemocratic, over-zealous

Lewis ● The opinion poll as a cultural form 201

03 Lewis (to) D 28/6/99 12:21 pm Page 201




202

technological determinism, the multiplication of poll cross-tabulationscontinues unabated. Indeed the widespread assumption that public opinionis no more or less than the aggregates recorded by opinion polls has takenroot in many areas of civil life. Polls of the crudest kind are now so ubiqui-tous that contemporary politics is almost unimaginable without them. Pollsdrive both political campaigns and the media coverage of those campaigns.In a spiral of imperfect circles, a distant public comments upon a discur-sive world shaped by the remorseless repackaging and representation ofitself.

The opinion poll as a form of representation

Most critical scholars, particularly within my own field of cultural studies,have tended to deride or ignore this multitude of narrowly defined responsesas a ceaseless barrage of apparently flimsy information. At best, publicopinion polls are cast aside as nothing more than persistently poor socialscience.

And yet the very persistence of polls indicates that this dismissal mightperhaps be a little too rarefied. We might, instead, interpret the abundanceof polls as part of popular culture, as carefully encoded media messages likeTV programs or advertising billboards. Cultural studies does not ignore tele-vision advertising simply because its representations are ideologicallyloaded. On the contrary, cultural studies begins with the assumption thatthese cultural forms reveal aspects of ideology and culture. They may notreflect the world they signify, but they do relate to it as well as to the shift-ing battleground of meaning in which the popular is configured. Indeed, itis partly because polls are taken seriously as a form of representation thatwe should not dismiss them.

I do not want to further rehash the critiques of Blumer, Bourdieu or othershere. Suffice to say that there is no homogeneous ‘public’ whose thoughtsare neatly captured in the columns and percentages derived from pre-codedquestionnaires. Unlike some forms of cultural production, any notion ofauthorship in mainstream polling is deeply suppressed. Polls are, in thissense, like the news in which they appear: legitimacy is often seen as depen-dent upon transparency. As Siune and Kline (1975) point out, the manu-facturers of polls are likely to be close, in an ideological sense, to media andpolitical elites. The discourse of public opinion is thus constructed from aparticular set of perspectives while the process of polling is hidden behindthe data it produces.

Lisbeth Lipari argues that polls often tell us more about the assumptionsof the questioner than the respondent (Lipari, 1996). If polls can be saidto be authored, then the author might best be identified as the pollsterrather than the public. In most opinion surveys, it is the questioner who


03 Lewis (to) D 28/6/99 12:21 pm Page 202




establishes the framework and sets the parameters for each response: mostof the ideological work has therefore been done before a single question isput. The respondent is merely asked to inhabit the questioner’s world fora few fleeting moments and push various buttons (Democrat/Republican,economy/crime/healthcare, yes/no, etc.).

And yet, in a Gramscian (1971) sense, polls are ideological devices thatrequire a degree of consent. If polling questions are often ‘closed’ rather than‘open’ texts, they not only depend upon a degree of acquiescence, but theless overtly manipulative surveys allow space for refusal. Moreover, likeother cultural forms they are imperfect expressions of dominant frameworksof thinking: they allow a great deal to seep through that does not easily ‘fit’those dominant frameworks.

As a cultural form, opinion polls provide partial, shorthand clues to theworld they signify, even as we lose a whole dimension and many of thedetails. Polls do not so much represent ‘the public’ as signify it within a care-fully structured framework. To understand those significations we need tocontextualize them: the answer to a pollster’s question has no absolute truth,it is dependent upon a set of ideas that give it a more specific, particularmeaning. An individual on the doorstep or on the telephone may give theresponse, but it signifies within a social realm full of ideological connections,a place where ideas are repeated and shared.

Although people are only allowed to respond within the tightly scriptednarratives of pre-coded questionnaires, and although these responses arelimited by the carefully sculpted moments in which they are uttered, theyare neither arbitrary nor meaningless. In a well-known discussion of ‘massculture’, Raymond Williams argued that the public ‘includes us, but yet isnot us. . . . There are in fact no masses, there are only ways of seeing peopleas masses’ (Williams, 1963: 289). As Williams demonstrated in his ownwork, this understanding does not prevent us from using fragments of dis-course to chart the ‘structures of feeling’ that allow us to describe a culture.What matters, in this regard, is that we acknowledge the conditions of pro-duction. Thus we read a novel, a speech, a newspaper, a film, an advertise-ment or an opinion poll as representations or accounts constrained byideology and style.

We may read Charles Dickens as social commentary, but we do so in thecontext of the moral and aesthetic structures of Bleak House or Hard Times,a context enriched by all the other available accounts of Victorian England.An opinion poll is a rather less complex and less imaginatively crafted exer-cise in social description: it speaks from the narrow confines of elite con-ceptions of public affairs (thus ‘crime’ is usually on the list of issues forgovernment to address, while inequality of wealth is not), and in a staccatovoice that jumps from one gnomic expression of noisy terms (‘the economy’,‘freedom’, ‘government’) to the next. The opinion poll is partly an attemptto suppress the ambiguities and contradictions of public discourse, to


03 Lewis (to) D 28/6/99 12:21 pm Page 203




204

produce rational knowledge about a rational public, neatly displayed in theapparent purity of arithmetic terms (Edelman, 1995). As such, it is a kindof fiction. But, like most fiction, it bears a relation to the world it tries todescribe.

Reading polls critically has often been seen as an act of rejection, of focus-ing on what surveys do not tell us. What I shall attempt here is another kindof critical reading, to begin to think about what polls, in their circumscribed,confined way, do suggest about the state of public discourse. Thus, I wouldargue, while we can take the position that ‘the public’ is a socially con-structed category (Ang, 1991; Herbst, 1993) as all categories are, this doesnot mean it is not worth retaining. While the technology of polling is notneutral, neither is it ideologically fixed. There are, in short, many ways ofsignifying the ‘public’: the question is how one inscribes the public withinpolitical discourses. How, in short, is public opinion articulated with formsof political power?

Media representations

The political economy of public opinion polls falls broadly into three cat-egories: independent, quasi-public companies such as Gallup, Roper andHarris; private pollsters like MOR and the Wirthlin Group (who conductpolls for the US Republican Party); and in-house agencies for news corpo-rations – either full in-house operations like CBS and the New York Timesor services for the Washington Post and ABC (who do in-house design butsubcontract out for sampling and fieldwork). Of these three, it is the inde-pendent, quasi-public agencies whose influence has, in relative terms,declined, while private polling has become increasingly commonplace. Butin terms of the construction of a discourse about public opinion the mostsignificant producer/commissioner is undoubtedly the news media.

The news media’s use and commissioning of polls has grown precipitouslysince the 1970s. In 1976, major US media organizations conducted a (mean)average of four polls annually. By 1988 the number had risen to 32, whileall media outlets – large and small – are increasingly likely to commissiontheir own polls. In the four election years between 1980 and 1992 the NewYork Times published 697 stories based on its own polls, 30 percent runningon the front page (Ladd and Benson, 1992). This is by no means an Ameri-can preoccupation: the growth of polls for media use has been as global asother technologies. Even in France, where the publication of polls beforeelections is restricted, the annual number of opinion surveys conducted forthe national press doubled during the 1980s to an average of two a day(714) by the last year of the decade (Brule and Giacometti, 1990).

The technology of polling is thereby intertwined with the manufacture ofnews. On a purely technical level, academic public opinion researchers are,


03 Lewis (to) D 28/6/99 12:21 pm Page 204




on the whole, well aware of a litany of journalistic uses and abuses. Bradyand Orren, for example, describe the ‘inherent mismatch between the ethosof survey research and the ethos of journalism’. They proceed to count theways in which polling data is misinterpreted, whether in relation to the tech-nicalities of sampling error, the quasi-technicalities of measurement errors(such as the phrasing or context of questions), or in terms of the rather moreabstract category of ‘specification error’, in which the theoretical assump-tions behind a survey are flawed or insufficiently complex to describe thephenomenon being surveyed (Brady and Orren, 1992).

The political economy of polling is nevertheless complicated by the pro-duction of polls less clearly designed for media use. In the USA, for example,there are around 60 academic survey organizations, some of which – likethe National Opinion Research Center at the University of Chicago or theInstitute for Social Research at the University of Michigan – are fairly well-endowed, robust institutions. There are also Government agencies such asthe Department of Labor, the National Center for Health Statistics and theDepartment of Justice Statistics, which account for over a million surveyinterviews a year, as well as in-house industry surveys (AT&T’s TelephoneService Attitude Measurement program, for example, involved more thanfive million interviews annually). Agencies like the AC Nielson company(who measure program ratings) are billion-dollar enterprises whose func-tion is to transform public preferences into commodities to be bought andsold on the open market.

It is useful, in this context, to make a broad distinction between thosepolls that are given prominence in the reporting of public opinion – andwhich thereby contribute to the conventional wisdom espoused by politicaland media elites – and those that do not. In the first instance we have a massof complex data, much of which, like discarded film footage or video tape,is strictly the domain of industry analysts, archivists and academics. In thesecond instance we have an intermingling of discourses – the language ofpolls and the language about polls – that create dominant conceptions aboutthe nature of public opinion.

Differences between the two types of polling discourse can, at times, bestark. During the first two years of the Reagan Presidency, for example, thepress continually repeated the assumption that Ronald Reagan was anextremely popular President – an assertion culled from a few highly selec-tive nuggets of polling data, but one that was flatly contradicted by a seriesof comparative polls that suggested Reagan was one of the least popularpresidents in the post-war era. As King and Schudson argue, this was poss-ible because the dominant frame in which Reagan was interpreted – as anamiable, personable figure – made polls that contradicted this frame irrele-vant (King and Schudson, 1995). To this day, there remains a gulf betweentwo ‘public opinion’ discourses about the former President, one version sug-gesting a leader beloved, the other suggesting widespread dissatisfaction.


03 Lewis (to) D 28/6/99 12:21 pm Page 205




206

This distinction is an important one if we are to explore how the notionsof hegemony and resistance operate in relation to the ways in which ‘thepublic’ is represented. The technology of polling, by itself, guarantees adegree of control, but the popular significance of that control is contingentupon the ways in which polls are inscribed (or excluded) in media dis-courses.

Frameworks of interpretation

The reporting of polls is governed by a series of interweaving journalisticassumptions. These assumptions, I shall argue, push the representation ofpublic opinion toward a hegemonic frame in which public opinion is appro-priated within a center-right mainstream. I shall illustrate some of thefollowing points with analysis from a data set of 392 stories in the US mediabased on opinion polls in 1994–5. The sample was gathered from a Nexissearch of opinion polls, tracking coverage of all the Times Mirror pollsreleased in that year (the Times Mirror Center for the People and the Presshas since become the Pew Research Center). The sample was approachedqualitatively in order to identify the discourses and frameworks that typi-fied the coverage of polls.

1. The relevance of public opinion

Journalism is deeply tied to a ‘top-down’ political framework in which therange of legitimate political discourses is defined by political elites (see, forexample, Gitlin, 1980; Herman and Chomsky, 1988; Kellner, 1990;Murdock and Golding, 1973; Tuchman, 1978). For this framework to main-tain its coherence, public opinion must also be broadly defined in thoseterms: polls can then be seen as a vindication of dominant political dis-courses and of representative democracy in general. When majoritiesexpressed in opinion surveys place the public well to the left of political elites– as they do in many areas of government spending, for example – they tendto be dismissed as either inconsequential (since they do not appear to explainelection results) or irrelevant. So, for example, if polls show majorities infavor of a universal healthcare system while the main political candidatesare not, this majority is regarded as merely tangential to the political process(Canham-Clyne, 1994).

A Times Mirror poll reported on 14 November 1995, for example, sug-gested the presence of both anti-government and anti-corporate sentiments(only 4 percent of the survey said that corporations put their employees first,while an increasing percentage expressed an unfavorable view of business).While political elites repeatedly give voice to anti-government sentiments(the dominant discourse of Republican politicians since Reagan, one often


03 Lewis (to) D 28/6/99 12:21 pm Page 206




repeated in a kinder, gentler form by leading Democrats like Bill Clinton),the business funding base of both political parties makes American govern-ment far more pro-corporate than the American public (see, for example,Green et al., 1998). In my own survey of media outlets covering the poll(the Wisconsin State Journal, the Baltimore Sun, the Chicago Tribune, theDetroit News, the Atlanta Journal and Constitution, the Los Angeles Times,the Rocky Mountain News, the Tampa Tribune, the Phoenix Gazette, theArizona Republic and National Public Radio), most led with the anti-government views (as in ‘Jittery Americans Blame Congress’) while only one(the LA Times) even reported the anti-corporate sentiments, and then onlyin the last line.

A pro-corporate slant is, at the very least, a convenient position for cor-porate media to adopt. But the emphasis on antipathy to government in thereporting of these polls is also a response to the discursive world of Washing-ton politics. The notion of ‘political viability’ or ‘relevance’ in the rep-resentation of public opinion thus privileges those aspects of the publicopinion discourse that match the interests of political and corporate elites.This is solidified by the well-known journalistic tendency to focus on the‘horse race’ aspects of political life (who is ahead, who is behind – see, forexample, Diamond and Geller, 1995; Jackson, 1996), thereby vindicating ahierarchy of polling responses in which the choice of party or candidatebecomes the ruling grammar of the discourse of public opinion. The domi-nant narrative is based upon the shifting support for political parties orcandidates, obliging journalists to use other polling responses to explainthose shifts.

So, for example, a Times Mirror poll announced on 24 August 1995asked a number of specific policy questions and, in so doing, reported sub-stantial support for a range of government programs including Medicare,public housing programs and the Environmental Protection Agency. Whilehigh levels of support for such programs have been a fairly consistent featureof poll responses for some time (see Page and Shapiro, 1992), the popularityof many forms of government spending and intervention – so out of fashionin Washington – rarely receives much media attention. Of the 11 mediaoutlets in the sample to cover the poll (CNN, the Palm Beach Post, the LATimes, the International Herald Tribune, the Fresno Bee, the DallasMorning News, Newsday, National Public Radio, Reuters, the PhoenixGazette and the Washington Post), the responses indicating support forvarious government programs went unmentioned in all but two reports, andthen only as a footnote to the main story.

Instead, all eleven focused on the growing support indicated for an inde-pendent presidential candidate (e.g. ‘Unhappy Voters Back Independent’,‘Interest Rising for Independent Bid in ’96’, ‘Support for Third Party Risesin Poll’) – a ‘horse race’ story fueled by media speculation of a Colin Powellpresidential candidacy. Of the two brief references of support for government


03 Lewis (to) D 28/6/99 12:21 pm Page 207




208

programs, one was particularly interesting. At the end of an interview onNPR, Times Mirror pollster Andrew Kohut stated that:

we asked people to rate themselves on a scale of how much they supportcutting government versus maintaining government . . . they put themselvesmuch closer to where they think Bill Clinton and Gore is [sic] than to Doleand Gingrich, and that’s frustrating for many people, particularly on the leftor independent Democrats who don’t feel they have someone speaking totheir concerns about the paring down of government. (Morning Edition, 24August 1995; emphasis mine)

Kohut’s statement is suggestive. First, he implies that his survey puts amajority of the public to the left of the Washington consensus that (as BillClinton put it) ‘the era of big government is over’. This interpretation of thedata is not only absent from the press reports, disenchantment with thepolitical mainstream is reappopriated within that mainstream by the media’sfocus on a ‘horse race’ discussion of independents in which the likely candi-dates – such as Colin Powell – are situated somewhere between Bill Clintonand a conservative Republican Party. Second, the dominant discourse inwhich politics is reduced to the narrow range espoused by political elites isso powerful that it forces Kohut into a framework he must simultaneouslyundermine. People are thus closer to ‘where they think’ Clinton and Goreare: they are thereby slotted into the political mainstream (in which Clintonand Gore represent the left), even while they are unaware that Clinton andGore’s support for government is considerably more tepid than theirs.

This process is not, I would argue, limited to the USA. Thus when pollssuggested a high degree of opposition amongst the British electorate tofundamental aspects of the Conservative Party’s approach to taxation andspending during the 1980s and 1990s (Taylor-Gooby, 1995), this publicopinion discourse was ignored by most journalists partly because it did littleto explain a narrative in which the Conservative Party remained in powerfor 18 years. The dominance of ‘horse-race’ polling becomes part of thehegemonic process, legitimating a limited range of options and thereby pro-scribing which discourses are politically relevant and which are not.Hegemony thus works less by imposing specific political discourses than bydefining the discursive terrain.

2. Democracy and coherence

The broad mass of opinion poll data is full of contradictions and ambigui-ties. It might thus be regarded as a relatively open text, a polysemy that isincompatible with the codes of news reporting that give polls their contem-porary significance. In the semiotics of the newsroom, ambiguity is regardedas unsettling to a discourse premised on notions of transparency and objec-tivity. The desire for closure is even more pronounced when those facts must


03 Lewis (to) D 28/6/99 12:21 pm Page 208




conform to notions of a rational public inside a functioning democracy –hence the invariable attempts to close off the meaning of opinion polls orelections (whereby pundits and reporters often try to reduce the poll to asingle meaning, as in ‘people wanted change’). To maintain faith in moderndemocracies the meaning of polls – especially electoral polls – must be seenas compatible with the political consequences that follow them. To thinkotherwise involves a radical rethinking of what democracy is and how itmight work. Salmon and Glasser put it well:

the quantification of public opinion appeals to, and simultaneously affirms,journalists’ faith in a free and enlightened electorate; it vivifies the politicalauthority of the citizenry and underscores the viability of each individual, sep-arate and sovereign, as the locus of democratic power. By recognizing the valueof individual opinion and by granting everyone, at least statistically, an oppor-tunity to be heard, public opinion polls foster what appears to be an entirelyopen and egalitarian form of democracy. (Salmon and Glasser, 1995: 444)

Nonetheless, as Salmon and Glasser suggest, polls are not simplydescribed, they come tightly wrapped in a discursive framework that pur-ports to make sense of them. As a cultural form, the opinion poll text issqueezed, pushed and molded into apparent coherence by an army ofreporters, experts and pundits – Walter Lippmann’s professional, educatedelite (Lippmann, 1922, 1925) – committed to the rationality of the process.All the more so as the political economy of polling becomes bound up withthe political economy of news production (Brady and Orren, 1992).

Voting is thereby expressed as a coherent expression of political ideology.And yet there is, as Page and Shapiro point out (Page and Shapiro, 1983),often only a weak connection between majority survey responses to a broadrange of policy questions and the candidates majorities vote for. So, forexample, Ronald Reagan was elected to a second term in 1984 at a timewhen more specific, policy-oriented polls were generally registering a shiftto the left (Ferguson and Rogers, 1986; Mayer, 1992).

The structure of a number of polling questionnaires contains implicitassumptions about why people vote: we are asked to say which are the mostimportant issues, and then to say which candidate or party is ‘better’ onthose issues. Voting in this ideal, rationalist model is the apotheosis of aseries of structured political calculations. But for a person to vote in such away is to ignore the ways in which people are generally addressed. Themainstream news media’s reluctance to devote time and energy to policypositions means that most people simply do not have the information avail-able to make policy-driven judgments (see, for example, Delli Carpini andKeeter, 1992; Kuklinski and Quirk, 1997; Lewis and Morgan, 1996; Lewiset al., 1998; Morgan et al., 1992). We are encouraged instead to make judg-ments based on the less tangible clusters of images and symbols that fill cam-paign pitches and television news reports.


03 Lewis (to) D 28/6/99 12:21 pm Page 209




210

If many journalists are aware that the act of voting is not always com-mensurate with a person’s political or economic interests, such thoughts aregenerally abandoned in subsequent discussions. So, for example, whether itwas proclaimed with satisfaction or reluctance, the Republican victory incongressional elections in November 1994 was widely interpreted as a clearmandate for a conservative program. The LA Times, for example, reporteda Times Mirror poll thus: ‘With an activist, Republican Congress poised atthe reins, most Americans strongly support many of the GOP’s top pri-orities, including crime control, welfare reform and further deficit reduction,according to a Times Mirror poll released Wednesday’ (LA Times, 12August 1994). While this is, on one level, a perfectly accurate representation,the same poll also found majority support for increases in spending ondecidedly ‘liberal’ projects such as public education (64 percent) and AIDSresearch (55 percent), as well as low levels of support for increasing defensespending (31 percent) or cuts in capital gains taxes (27 percent). A majorityalso said that they had never heard of the ‘Contract with America’ or didnot know enough about its contents to be able to comment upon it.

This is not to imply that reporters deliberately suppress polling data thatdoes not fit the rational ideal of representative democracy. Nevertheless, thesurvey’s political ambiguities are lost in the gap between the polling dataand the LA Times’s representation of that data. The LA Times choice ofemphasis reflects the dominant culture’s commitment to what Edelman(1995) has described as a discourse of rationality, whereby rational votersmake rational ideological choices in a way that makes elections coherentdemocratic enterprises. As Edelman writes:

The focus on rational choice serves quite often to justify actions that power-ful groups or officials favor but which much of the public disapproves. . . .Majorities have long favored gun control, abortion rights, a higher minimumwage, and many other specific objectives that legislatures and administrativebodies have often rejected. . . . The focus on rationality muddles discussionof goal priorities while helping to rationalize the specific actions of bureau-cracies. (Edelman, 1995: 408–9)

Even attempts to invalidate the 1994 Republican victory invariably clungto these notions of systematic rationality. So, for example, an article in theNew Yorker by Alan Brinkley argued that the Republican mandate wasflawed because: ‘Republican candidates won 51% of the popular vote in ayear in which 39% of the eligible electorate went to the polls; their “mandate”thus comes to slightly less than 22% of eligible voters’ (New Yorker, 29January 1996). While the voters in that election were certainly somewhatricher and less ethnically diverse than the population as a whole, it is also truethat rather more representative (if much smaller) polls also found majoritysupport for the Republican Party (a Times Mirror poll, for example, reported57 percent of the general population content with the sweeping Republican


03 Lewis (to) D 28/6/99 12:21 pm Page 210




victory). What is interesting about Brinkley’s claim is the refusal to questionthe notion that people might choose to vote for parties that reflect neithertheir economic interests nor many of their professed wishes.

If the fragments of discourse suggested by opinion polls do add up to some-thing tangible and binding, they do so in ways that consistently elude themoments the political culture has chosen to construct such unities. To returnto the British example, the election of four successive Conservative govern-ments from 1979 to 1997 was generally assumed to be an endorsement ofan economic policy of lower taxation and cuts in social spending, and yet infive surveys taken between 1983 and 1994, the number of people who feltthe government should ‘reduce taxes and spend less on health, education andsocial benefits’ fluctuated between 3 percent and 9 percent, while in three ofthe surveys majorities (between 54 percent and 63 percent) actually favoredincreasing taxes to pay for more social spending (Taylor-Gooby, 1995).

As Peter Miller (1995) points out, the logic of the opinion poll is notentirely the creation of pollsters, but of modern democracies. If polls areseen as problematic expressions of public opinion, elections are, in manyways, even more so. In this context, I would argue, modern elections are notideologically metonymic or symbolic events – they may signify very littlebeyond themselves. This is such an uncomfortable notion to address that itis hardly surprising that the main criticism made of opinion surveys is onethat legitimates rather than undermines electoral polls: namely, that pollsare insufficiently accurate or representative of the voting population. To beaccurate, in this framework, is not to capture public discourse about poli-tics, but to reflect as precisely as possible the electorate’s highly ambiguousanswer to an extremely limited question.

In this interpretative framework, scrutiny is reserved only for questionsof mathematical or technical detail. The question ‘is the poll accurate?’ isnot directed at the philosophical or ideological assumptions that reduce thecitizenry to a few percentages, but to the poll’s statistical reliability. Pollstersare taken to task only when a large sample (the voting public) does not quitereplicate the preferences of a smaller one. So, for example, the failure ofpollsters to accurately predict the results of the 1992 elections in Australiaor the United Kingdom induced a bout of journalistic skepticism in whichthe underlying assumptions about public opinion remained intact, leavingthe pollsters with an admonition to get it right next time. The assumptionbehind this admonition can be seen as an inversion of the logic of repre-sentative democracy: the role of polls is to reflect – as perfectly as possible– the make up of the governing body.

3. The suppression of difference

In contrast with other news values, the premise underlying the represen-tation of public opinion tends to be an absence of conflict. While pollsters


03 Lewis (to) D 28/6/99 12:21 pm Page 211




212

will often break down responses by region, gender, race, party, education orincome, these differences are routinely ignored in media representations.Any divergence of interests tends to get lost in the undifferentiated mush ofstatistical majorities. Conflicts that underpin and even define social lifebecome mere hiccups. A careful reading of polling data in recent yearswould reveal that people separated by class, race or gender tend to havedifferent interests and often view the world differently, and yet these div-isions sit uneasily within a framework that suppresses such an overtly politi-cal fracturing of ‘the public’. When polls are reported as expressions ofconflict, it is often in response to a story in which divisions are alreadyinscribed. Responses to the O.J. Simpson trial are a case in point: the racialconflicts manifested by the trial made differences between black and whiteresponses relevant: thus a report of a Times Mirror poll in the Rocky Moun-tain News (14 October 1994) covered a number of issues but only chose todisaggregate responses when reporting on interest in the Simpson case.Although it is worth noting that even while reporting that ‘Blacks were morethan twice as likely to retain an intense interest in the case’, the story beganwith a headline that masked this distinction: ‘Americans’ Interest in O.J.Trial Fades, Poll Says.’

Similarly, in April 1995, National Public Radio, discussing the findings ofa recent poll, reported that: ‘Groups that approve of what the new Republi-cans in Congress are doing include whites, men, people with money. Groupsthat disapprove include women, blacks, and people who earn less than$30,000’ (All Things Considered, NPR, 12 April 1995). They were, in sodoing, tracing the outlines of a discourse in which clear social and politicaldivisions are drawn. And yet, in the ensuing discussion between reporter andpollster, these divisions were repeatedly over-ridden by the discourse ofhomogeneity: ‘We’re [the public] glad that the Republican [sic] are in charge. . .’, ‘many people believe . . .’, ‘the American people don’t seem to be feeling. . .’, ‘many people feel that . . .’ (emphasis mine).

In this discursive framework the opinion poll – with its ability to reducedivisions to broad aggregates – is often the ideological glue that masks socialdivision and creates the mythical notion that we are ‘all in it together’. Thisis tied to a broader political discourse in which politics is reduced to a mer-itocracy in which the public support the ‘better’ candidate (often the onewho, from the same meritocratic lexicon, ‘wins’ debates), rather than thecandidate who best expresses particular interests (such as men/women,poor/rich people, etc.). This is more than a process of reducing a complexpolitical engagement to a single digit. The idea that politicians representcoalitions of different and conflicting interest groups is entirely absent fromthis formulation: candidates are, on some absolute and quantifiable scale,simply better or worse. In this instance, the opinion poll not only fails toreveal a political response, it masks it beneath a sporting metaphor in which‘the best man/woman wins’.


03 Lewis (to) D 28/6/99 12:21 pm Page 212




In this framework, the ‘public’ itself becomes a device that binds society’ssub-groups into one undifferentiated lump. The public becomes a coherentaggregate subject, who thinks and wants without contradiction and whosepluralities are, in the end, muffled by the search for monochromatic majori-ties. Typically we read or hear statements such as: ‘Americans consider con-gressional leaders Newt Gingrich and Bob Dole more powerful thanPresident Clinton, according to a poll released Wednesday’ (Reuters, 12April 1995; emphasis mine). This statement is based on a poll in which Gin-grich and Dole were seen as more powerful than Clinton by 57 percent to35 percent and 55 percent to 36 percent respectively. The slippage from‘55% to 57% of Americans consider . . .’ to ‘Americans consider’ is morethan mere shorthand in which minorities – in this case, substantial minori-ties – are apt to simply disappear, it is a philosophical presumption that onecan speak of public opinion as an undivided, coherent being.

While there are occasional exceptions (such as discussion of ‘gender gaps’in party support or moments of racial conflict), this consensual frameworklends itself to an analysis in which public opinion can be unified around aseries of generalities. It is an apparently egalitarian image that makes anyintroduction of questions of power and structural inequality difficult toformulate. This, in turn, facilitates attempts to articulate the public as awhole within a hegemonic framework of concerns.

The unity of public opinion is often linked to the ideology of politicalelites by the use of metonyms in descriptions of political allegiance. Themost common manifestation of this is the tendency to use the ‘swing’ votersin elections as symbols of the electorate en masse. If one in five voters switchfrom one party to another to precipitate a decisive shift in political rep-resentation, the motives and predilections of this group are assumed to rep-resent the ‘public mind’. The 80 percent who did not change their mindshave little symbolic value and are consequently rarely discussed, giving us asyllogism that frames much of the media coverage of politics: swing votersdecide elections, therefore swing voters express the will of the electorate.This, in turn, creates a tendency to over-estimate the popularity of those cen-trist political discourses likely to appeal to this (minority) group.

It is a metonym in which the center becomes the whole. In 1996, electionsin the USA installed a Democrat in the White House and Republican majori-ties in Congress. If this result depended on a degree of vote-splitting, thepercentage who actually voted both for Clinton and Congressional Repub-licans was fairly small – over 90 percent voted straight Republican or Demo-crat tickets. The question repeatedly asked by journalists ignored thispartisan consistency and instead pondered on ‘the message’ that the elec-torate were sending by voting for a divided government – something that,in fact, the great majority did not vote for.

There are, accordingly, moments when symbolic power is granted tothose who change their opinions even though they remain in a minority.


03 Lewis (to) D 28/6/99 12:21 pm Page 213




214

On 25 September 1994, the press reported on a poll that found, as theHouston Chronicle put it, ‘the nation has grown more cynical, sour andmean-spirited’, or, as Newsday declared, ‘an angry disenchantment isunsettling the public, turning American against government and . . . eachother’. The evidence in the survey that justified such a claim was based onfindings that:

. . . attitudes hardened on issues potentially affecting taxes and job security.In 1987, for example, 71 percent said that the government should take careof people who can’t take care of themselves. That fell to 57 percent this year.(The Columbian, 25 September 1994)

The minority – less than one respondent in six – who apparently changedtheir minds thereby come to represent a general shift in the state of thepublic mind towards an ideology that, as the poll itself indicates, mostpeople (57 percent) would appear to reject.

The closing of the public mind

These frameworks of interpretation are not unbending or monolithic insuppressing ambiguities. On 10 March 1998, the Washington Post reportedthat: ‘Trust in Government Edges Up . . . Survey Finds’, while the New YorkTimes summarized the same data with the headline: ‘Americans Take a DimView of the Government, Survey Finds.’ Nonetheless, divergence in theinterpretation of public opinion poll data generally takes place within thenarrow range of elite political discourse.

If opinion polls themselves are often limited, they nonetheless reveal ideo-logical possibilities that the media’s interpretative frameworks tend to sup-press. Amidst a swathe of contradictions, we can identify strong or majoritysupport for ideas spanning a wide range of ideological positions from leftto right. Particularly notable is the degree of support for a variety of politi-cal positions on the left – from gun control to social justice issues. Majori-ties consistently support increased government spending in traditionally‘liberal’ areas such as healthcare, education and environmental protection.(This has been well documented in a number of comprehensive studies suchas Ferguson and Rogers, 1986; Mayer, 1992; Page and Shapiro, 1992.)

Scattered among the 392 articles I examined were polls showing majoritysupport for a range of progressive positions than ran directly counter to therecently elected Republican Congress and sometimes to the left of a broaderWashington consensus (e.g. support for raising the minimum wage, opposi-tion to increasing spending on defense and particularly to a ‘Star Wars’missile defense system, support for more federal spending on public edu-cation, support for stricter government regulation on environmental pro-tection, support for more spending on government run healthcare, support


03 Lewis (to) D 28/6/99 12:21 pm Page 214




for increased gun control legislation, support for programs like the NationalEndowment for the Arts and the Corporation for Public Broadcasting,support for the United Nations, and a suspicion of corporate America). Butin none of the 392 articles were such attitudes straightforwardly articulatedas support for anything further to the left than Bill Clinton (the possibleexception being Andrew Kohut’s cryptic statement on National PublicRadio, quoted previously). The frameworks I have described tie publicopinion closely to political elites, making such an articulation extremelydifficult. As I have tried to show, even when disillusionment with the twomain parties becomes newsworthy, it is rearticulated to the political main-stream by tracking support for a ‘middle of the road’ candidate like ColinPowell.

It is possible to identify a number of indistinct, almost romantic notionsthat are rarely interrogated: an attachment to individual freedom; modera-tion and an antipathy to self-defined socialist notions of government. Theseabstractions tend to lend themselves to politicians on the right rather thanthose on the left, and polls will dutifully show that majorities within theAmerican public are prepared, when asked, to embrace all of them (Pageand Shapiro, 1992). And yet it is a weak, gestural embrace that tells us verylittle about what most Americans think about political practicalities.

The ability to tap into suspicions of government spending and regulationthus depends upon its discursive form. When government is placed in oppo-sition to the private individual – whether in relation to abortion or propertyrights – it is easy to create majorities for unfettered free agency or enter-prise. A Harris poll in April 1995 asked the question: ‘Do you think thatthe federal government should have the right to set certain regulationsaffecting the use of private property, or do you think the use of private landsshould be left solely up to the property owner?’ As we might expect, theresponse showed only a minority – 38 percent – supporting a government’spower to regulate, with 59 percent opposed to it. This would appear toaffirm an ideology that promotes the sanctity of private property. Harristhen asked the same question but with a caveat: ‘Do you think that thefederal government should or should not have the right to prevent ownersof private land from developing the land if that development would involveharming or polluting the environment?’ The response to this question wasquite different: 79 percent now supported the government’s power to regu-late, and only 20 percent continued to affirm the sanctity of private prop-erty (The National Journal, 20 May 1995). While this does, in some ways,invalidate the first response, it is perhaps more useful to see it as indicativeof the presence of another powerful – in this instance more powerful – com-peting ideology about environmental protection. In this case the commit-ment to the sanctity of private property was compromised and overriddenby the appeal of an environmentalist discourse.

In an instructive study over three decades ago, Free and Cantril (1967)


03 Lewis (to) D 28/6/99 12:21 pm Page 215




216

found that while many Americans were inclined to support conservativeprinciples in abstraction, they tended to lean leftward when those principleswere given an immediate and specific context. So, for example, nearly halfof those who had endorsed abstract conservative philosophies were com-pletely or predominantly liberal in their support of specific New Deal stylegovernment programs.

The failure of polls to elicit more ideologically coherent responses tells ussomething about the mismatch between frameworks of encoding and decod-ing – what makes sense in one place seems contradictory in another.Nonetheless, as Page and Shapiro (1992) and Zaller (1992) argue, thesepolls do not collapse under the weight of apparent contradictions: theyclearly identify different ideological strands in the political culture, and theysuggest that the pertinence of a particular political idea is dependent upona specific discursive context.

The notion of articulation is especially useful here: fragments – or evenwhole chunks – of one political discourse may be articulated or separatedfrom another. Politics is therefore a matter of making some articulations stickwhile those discourses that might contradict it are scattered or ignored. Inthis landscape, the construction of a view of the American public in Washing-ton’s image – as generally moderate or conservative (and certainly distrust-ful of positions on the left) – is a hegemonic endeavor, one that works toexclude the oppositional discourses present in the broad mass of polling data.

We should not underestimate the power of this redefinition. It has ledmany, including those on the left, to characterize recent history as a periodin which a resurgent free market ideology has, on a global scale, built a con-sensus, albeit opportunistic, for the gradual dismantling of the welfare state.

And yet the evidence for such a consensus is based on a highly selectivereading of the opinion poll discourse. There is instead the appearance ofconsensus, one that a preferred reading of opinion polls has helped manu-facture. Right-wing parties (like the Republican Party or the British Con-servative Party) or influential right-wing factions within parties (like theDemocratic Leadership Council and Labour Party ‘modernizers’) have suc-ceeded in capturing public opinion, but only in the narrow terms in whichpublic opinion is usually defined and described.

Thus, in the USA, we see simultaneously a decline in the willingness ofpeople to embrace the term ‘liberal’ and consistent support for a range ofliberal programs and positions. Both can be tied to poll responses, but it isthe first that is generally signified as an expression of public opinion. Theleft-wing politician advocating a program of increased social spending andgovernment intervention that matches poll responses is thus cast adrift fromthose responses so that they appear to be ‘out of touch’ with public opinion.Liberal Democrats in Congress can thus be defined – in the words of a politi-cal correspondent on National Public Radio – as ‘extremists’ (WeekendEdition, 2 August 1997).


03 Lewis (to) D 28/6/99 12:21 pm Page 216




In the postmodern age we cannot, of course, underestimate the import-ance of appearance – election campaigns, after all, are dominated by suchappearances. But the closure of public opinion around a center-right axis isa political rather than an ideological victory. The center-right have, throughthe appropriation of media frameworks, succeeded in defining publicopinion and manipulating its terms, but they have failed to move publicopinion in a broader sense.

So, for example, as Page and Shapiro point out: ‘questions about pro-grams for “the poor” . . . often get a remarkable thirty or forty percentagepoints more support than questions about “welfare” that are asked at thesame time’ (Page and Shapiro, 1992: 126). This is, for all practical purposes,an expression of contradictory feelings about welfare, although as Page andShapiro argue, it implies that the word ‘welfare’, for many, establishes anoperational distinction between the poor and the undeserving poor. In Aus-tralian slang, it is the difference between a battler (a poor person strugglingagainst the odds) and a bludger (a shiftless recipient of government hand-outs).

The desire to, in contemporary political parlance, ‘end welfare as weknow it’ is thus contingent upon the way we know it. While there areundoubtedly complexities behind these figures (not least the racial inflec-tions of stereotypically idle welfare recipients, see Gray, 1996) they mightjust as easily signify a desire for fairness as support for a radical curtailmentof government assistance to the poor.

This discursive construction of welfare has been illuminated by Kuklin-ski and Quirk, whose research suggests that attitudes about welfare relateto a set of media stereotypes (who the welfare recipient is, how much theyreceive, the burden on the taxpayer, etc.). When survey respondents weregiven certain facts about welfare programs – when, in one sense, welfarewas rearticulated as a program for the poor – they were more likely toexpress support for welfare spending (Kuklinski and Quirk, 1997).

The stigmatization of ‘welfare’ can thus be seen as a piecemeal rhetoricaldevice. It allows the right to disarticulate the connection between welfarerecipients and the deserving poor at the ideological level while more quietlyarticulating them at the political level. Programs for poor people – whichare generally popular – are thereby reduced in the name of reforming anunpopular welfare system.

In much the same way, attacks on government largesse (linked to notionsof waste, bureaucracy, meddling and indulgence) reflect an ideologicalproject that neatly circumnavigates the popularity of most governmentspending (on education, healthcare, the environment, policing and so forth).The attack on ‘big government’ can be linked to a well-established ideo-logical frame in which individual freedom is pitched against a sea of face-less, interfering bureaucrats. Popular government programs can therefore becut in the name of this abstract, populist discourse.


03 Lewis (to) D 28/6/99 12:21 pm Page 217




218

The center-right’s success in these endeavors is not to capture publicopinion but the agencies that define it. Thus the unpopularity of welfare isreported, discussed and allowed to adorn what Robert Parry has called the‘conventional wisdom’ of media and political elites (Parry, 1992). The popu-larity of programs for poor people is, in turn, a scrap of polling data that isdiscarded, unused and unintelligible in the dominant framework.

Discourses about public opinion polls are therefore just as integral to thecultural form as the polls themselves. And if the right have been ratherpatchy at influencing the latter, they have been singularly successful atshaping the former. Thus, for example, when Sidney Verba argues thatsurveys are, in fact, a more democratic means of expression than citizen par-ticipation because pressure groups require significant resources to be effec-tive (Verba, 1996), I would argue that he takes insufficient account of hiscaveat about the publication, presentation and interpretation of those polls.

If public opinion, in its broader sense, is to be won or lost, it is necessaryto understand not only how it is signified but the political cracks in that sig-nification. What remains, as Stuart Hall has argued in another context (Hall,1988), is not simply a question of highlighting the columns and percentagesthat are routinely ignored – of reading against the grain – but of using thesediscursive fragments to articulate a progressive politics. Or, to borrow JaniceRadway’s metaphor, of sewing ideological seams between the cluster ofsocial democratic sentiments that polls tap into and a more general pro-gressive vision (Radway, 1986). This involves using polls as well as decon-structing them.

Polls are, in this sense, an indication of the complexity of hegemony: theysignify both dominance and resistance. While polls suggest strong areas ofresistance to a center-right corporate ideology, hegemony is achieved at thepolitical level by strategic forms of intervention. These interventions arebased upon a particular political economy in the representation of publicopinion and, in a more general sense, the recognition that polls are a cul-tural form.

References

Ang, I. (1991) Desperately Seeking the Audience, London: Routledge.Blumer, H. (1948) ‘Public Opinion and Public Opinion Polling’, American

Sociological Review 13: 242–9.Bourdieu, P. (1979) ‘Public Opinion Does Not Exist’, in A. Mattelart and S.

Siegelaub (eds) Communication and Class Struggle, pp. 124–30. New York:International General.

Brady, H.E. and G.R. Orren (1992) ‘Polling Pitfalls: Sources of Error in PublicOpinion Surveys’, in T.E. Mann and G.R. Orren (eds) Media Polls in Ameri-can Politics. Washington DC: Brookings Institution.


03 Lewis (to) D 28/6/99 12:21 pm 8




Brule, M. and P. Giacometti (1990) ‘Opinion Polling in France at the End of the80s’, Public Perspective (Mar.–Apr.).

Canham-Clyne, J. (1994) ‘When Both Sides Aren’t Enough: The RestrictedDebate over Healthcare Reform’, Extra (May–June).

Delli Carpini, M.X. and S. Keeter (1992) ‘The Public’s Knowledge of Politics’,in J. Kennamer (ed.) Public Opinion, the Press and Public Policy, pp. 19–40.Westport, CT: Praeger.

Diamond, E. and G. Geller (1995) ‘Campaign ’96: Never Have So Many WrittenSo Much About So Little’, The National Journal (29 Apr.).

Edelman, M. (1995) ‘The Influence of Rationality Claims on Public Opinionand Policy’, in C. Salmon and T. Glasser (eds) Public Opinion and the Com-munication of Consent, pp. 403–16. New York: Guilford Press.

Ferguson, T. and J. Rogers (1986) Right Turn: The Decline of the Democratsand the Future of American Politics. New York: Farrar, Straus & Giroux.

Foucault, M. (1977) Discipline and Punish: The Birth of the Prison. New York:Pantheon.

Free, L. and H. Cantril (1967) The Political Beliefs of Americans: A Study ofPublic Opinion. New Brunswick, NJ: Rutgers University Press.

Gitlin, T. (1980) The Whole World Is Watching. Berkeley: University of Cali-fornia Press.

Gramsci, A. (1971) Selections from the Prison Notebooks. London: Lawrence& Wishart.

Gray, H. (1996) Watching Race: Television and the Struggle for Blackness. Min-neapolis: University of Minnesota Press.

Green, J., P. Herrnson, L. Powell and C. Wilcox (1998) Why Donors Give.Washington, DC: The Center for Responsive Politics.

Habermas, J. (1989) The Structural Transformation of the Public Sphere: AnInquiry into a Category of Bourgeois Society. Cambridge, MA: MIT Press.

Hall, S. (1988) The Hard Road to Renewal. London: Verso.Herbst, S. (1993) Numbered Voices. Chicago, IL: University of Chicago Press.Herman, E. and N. Chomsky (1988) The Manufacture of Consent. New York:

Pantheon.Jackson, J. (1996) ‘Sex, Polls and Campaign Strategy: How the Press Missed the

Issues of the 1992 Election’, in J. Jackson and J. Naureckas (eds) The FairReader: A Review of Press and Politics, pp. 57–65. Boulder, CO: WestviewPress.

Kellner, D. (1990) Television and the Crisis of Democracy. Boulder, CO: West-view Press.

King, E. and M. Schudson (1995) ‘The Press and the Illusion of Public Opinion’,in C. Salmon and T. Glasser (eds) Public Opinion and the Communicationof Consent, pp. 132–55. New York: Guilford Press.

Kuklinski, J. and P. Quirk (1997) ‘Political Facts and Public Opinion’, paperpresented at the ‘Future of Fact’ conference, University of Philadelphia, Feb-ruary 1997.


03 Lewis (to) D 28/6/99 12:21 pm Page 219




220

Ladd, E.C. and J. Benson (1992) ‘The Growth of News Polls in American Poli-tics’, in T.E. Mann and G.R. Orren (eds) Media Polls in American Politics.Washington, DC: Brookings Institution.

Lewis, J. (1997) ‘What Counts in Cultural Studies’, Media, Culture & Society(19): 83–97.

Lewis, J. and M. Morgan (1996) ‘Images/Issues/Impact: The 1992 PresidentialElection’, in J. Jackson and J. Naureckas (eds) The Fair Reader: A Review ofPress and Politics, pp. 87–93. Boulder, CO: Westview Press.

Lewis, J., M. Morgan and S. Jhally (1998) ‘Polling Clinton’s Appeal’, TheNation (9 Mar.).

Lipari, L. (1996) ‘Rituals and Identification: A Critical Discourse Analysis ofPublic Opinion Polls’, PhD dissertation, Stanford University.

Lippmann, W. (1922) Public Opinion. New York: Macmillan.Lippmann, W. (1925) The Phantom Public. New York: Harcourt Brace.McCombs, M., L. Danielian and W. Wanta (1995) ‘Issues in the News and the

Public Agenda’ in C. Salmon and T. Glasser (eds) Public Opinion and theCommunication of Consent, pp. 281–300. New York: Guilford Press.

Mayer, W.G. (1992) The Changing American Mind. Ann Arbor: University ofMichigan Press.

Miller, P. (1995) ‘The Industry of Public Opinion’, in C. Salmon and T. Glasser(eds) Public Opinion and the Communication of Consent, pp. 105–31. NewYork: Guilford Press.

Morgan, M., J. Lewis and S. Jhally (1992) ‘More Viewing, Less Knowledge’, inH. Mowlana, G. Gerbner and H. Schiller (eds) Triumph of the Image: TheMedia’s War in the Persian Gulf – A Global Perspective, pp. 216–37. Boulder,CO: Westview Press.

Murdock, G. and P. Golding (1973) ‘For a Political Economy of Mass Com-munications’, in R. Milliband and J. Saville (eds) Socialist Register. London:Merlin.

Page, B. and R. Shapiro (1983) ‘Effects of Public Opinion on Policy’, AmericanPolitical Science Review 77: 175–90.

Page, B. and R. Shapiro (1992) The Rational Public. Chicago, IL: University ofChicago Press.

Parry, R. (1992) Fooling America. New York: Morrow.Radway, J. (1986) ‘Identifying Ideological Seams: Mass Culture, Analytical

Method, and Political Practice’, Communication (9): 93–121.Ruddock, A. (1998) ‘Doing it by Numbers’, Critical Arts 12(1–2): 115–37.Salmon, C. and T. Glasser (1995) ‘The Politics of Polling and the Limits of

Consent’, in C. Salmon and T. Glasser (eds) Public Opinion and the Com-munication of Consent, pp. 437–58. New York: Guilford Press.

Schuman, H. and S. Presser (1981) Questions and Answers in Attitude Surveys.New York: Wiley.

Siune, K. and F.G. Kline (1975) ‘Communication, Mass Political Behavior and


03 Lewis (to) D 28/6/99 12:21 pm Page 220




Mass Society’, in S. Chaffee (ed.) Political Communication: Issues and Strat-egies for Research. Beverly Hills, CA: Sage.

Sussman, B. (1986) ‘Do Blacks Approve of Reagan? It Depends on Who’sAsking’, Washington Post (weekly edition, 10 Feb.).

Taylor-Gooby, P. (1995) ‘Comfortable, Marginal and Excluded: Who ShouldPay Higher Taxes for a Better Welfare State?’, in British Social Attitudes, pp.1–18. Aldershot: Dartmouth.

Tuchman, G. (1978) Making News. New York: Free Press.Verba, S. (1996) ‘The Citizen as Respondent: Sample Surveys and American

Democracy’, American Political Science Review (90): 1–7.Williams. R. (1963) Culture and Society. New York: Columbia University Press.Wilson, T. and S. Hodges (1992) ‘Attitudes as Temporary Constructions’, in A.

Tesser and L. Martin (eds) The Construction of Social Judgement, pp. 37–65.Hillsdale, NJ: Erlbaum.

Zaller, J.R. (1992) The Nature and Origins of Mass Opinion. Cambridge: Cam-bridge University Press.

● JUSTIN LEWIS is Professor of Communication at the Universityof Massachusetts, Amherst. He has written a number of books on mediaand culture, and is currently writing a book on public opinion forColumbia University Press. Address: Department of Communication,University of Massachusetts, Machmer Hall, Box 34815, Amherst, MA01003-4815, USA. [email: [email protected]] ●


03 Lewis (to) D 28/6/99 12:21 pm Page 221




international journal of cultural studies 1999 lewis 199 221

Documents