In the current day Netherlands, opinion polls or surveys are part and parcel of the media coverage of elections, as is the case in most Western democracies. Nonetheless, the omnipresence of polls in newspapers and on radio and television is a relatively recent phenomenon. Whereas Dutch mass media outlets had been hesitant to use polling data in their coverage of election news in the early 1960s, they had turned into one of the most prominent users of polling institute surveys by the 1980s. The effects of this rapid change on political journalism can hardly be overstated. This article contends that the Dutch mass media’s use of opinion polls has contributed to the emergence of the image of Dutch voters as emancipated and opinionated citizens and to the popularisation of a ‘marketised’ imaginary of the dynamics of democratic politics. Polls promoted the notion of Dutch voters as consumers in an open marketplace, who were no longer bound by traditional loyalties such as religion or class. As such, this article contributes not only to our knowledge of Dutch media and political history, but also to the ever-growing scholarship on the wider history of post-war mediatised politics.

In this article, I investigate how Dutch broadcasters appropriated political opinion polls and deployed these in radio and television formats since the 1960s. My analysis continues up to the end of the 1980s when the introduction of commercial television marked the beginning of a whole new chapter in Dutch media history.² While not completely ignoring the role newspapers and magazines played in this development, my main focus lays on the audio-visual media, as they were the trendsetters in practicing political journalism and influencing political culture in innovative ways.³ In this respect, the Dutch history of the media’s use of opinion polls very much mirrors developments in West Germany. ard and zdf, Germany’s first and second broadcast company respectively, played a key role in popularising opinion polling among the German public in the 1960s and 1970s, more or less around the same time as their Dutch counterparts did.⁴ The Dutch and German cases, in turn, somewhat contrast with other Western European countries, most notably France and Great Britain, where political surveys have been strongly associated with print rather than with broadcast media.⁵

Communication and media scholars tend to conceptualise the media’s appropriation of opinion polls as a process they refer to as the ‘mediatisation’ of politics. This concept describes a situation in which the mass media control political communication.⁶ In such a state of affairs, they argue, neither ‘partisan’ nor ‘public’ logic, but ‘media logic’ shapes the relation between mass media and politics or the so-called political-media complex. Informed by the ever more commercialised and competitive media landscape since the late twentieth century, political journalists adjust the form and content of information to what they believe to be the wishes of the public. In their quest for ever larger numbers of viewers, media makers frame political news as something entertaining.

According to mediatisation scholarship, opinion polls are the linchpin of mediatised political communication. It is argued that they inform ‘horse-race journalism’, that is the framing of politics and elections as a thrilling competition between opposing candidates. As such, they shift the focus from the actual content of political parties’ policy proposals to the performances of political leaders and their relative chances of winning elections.⁷ Political scholars engaging with this narrative argue that mediatisation affects the core parameters of modern political representation and has resulted in the establishment of an ‘audience democracy’, as philosopher Bernard Manin has called it. In this new political landscape, politicians are left with no alternative than to adhere to this media logic and become media personalities, hoping to claim a substantial part of the volatile electoral market.⁸

However, the question is whether the mediatisation narrative sufficiently clarifies the dynamics behind the Dutch audio-visual media’s embrace of opinion polls. Even though the concept of mediatisation suffers from a certain conceptual and operational fuzziness – as critical media scholars have already argued – I do not necessarily want to dispute that media logic is a useful concept for characterising specific changes in the political-media complex.⁹ Nevertheless, the history of Dutch broadcasters’ appropriation of opinion polls does not correspond with the chronology that is usually attributed to the mediatisation of Dutch political communication. Commonly, media historians contend that media logic came to shape Dutch political journalism only from the 1990s onwards.¹⁰ Yet, polls have been dominant media phenomena since the late 1960s. By focusing on the swift rise of political opinion polls on radio and television before the 1990s, this article aims to unravel the arguments that actually fuelled this development, and studies which motivations – besides entertaining the audience – were responsible for broadcast editors’ eager adoption of opinion polls.

On a more fundamental plane, I take issue with the normative conception that is at the heart of the mediatisation narrative. The argument implicitly goes that ‘a well-functioning representative democracy’ is in need of ‘well-informed citizens’ who make their decisions on whom to vote for in a rational manner.¹¹ Mediatisation scholarship also often echoes popular grievances about the mass media’s ‘colonisation’ of politics, the ‘degradation’ of political news coverage, and the decline of democracy.¹²

Yet, as several prominent political historians argue, politics and democracy are never fixed or everlasting categories. On the contrary, what is grasped as ‘democratic politics’ is very much dependent on socio-cultural contexts.¹³ I agree with them and believe that historians should therefore historicise past notions of politics, unravel contemporary conceptions of politics and analyse how these beliefs have changed over time. This article aims to approach the phenomenon of opinion polls on Dutch radio and television in precisely this manner. By studying both the media makers’ considerations behind the deployment of polls in radio and television formats, as well as the reception of those formats by politicians and other viewers and listeners, we gain a deeper understanding of how these groups conceived of – and thus symbolically shaped – the political reality they were part of.¹⁴

This article maintains that the concept of ‘scientisation of the social’ is more useful than ‘mediatisation’ to grasp the earliest polling programmes on Dutch television and radio. Historian Lutz Raphael uses this concept to refer to the direct and indirect consequences of the ‘continuing presence of experts from the human sciences, their arguments, and the results of their research (…) in administrative bodies and in industrial firms, in parties and parliaments’.¹⁵ More specifically, I build upon to the growing scholarship on the ‘scientisation of the political’. So far, historians have mainly analysed the gradual introduction and often hesitant embrace of scientific expertise into the domains of governance and policy making, and of political parties and electoral strategies.¹⁶ For the Dutch case, historians Wim de Jong and Harm Kaal have demonstrated the profound influence of party think tanks and electoral experts on the strategies of political parties in the post-war decades. They have shown how these experts interpreted opinion polling as a ‘democratic science’ that allowed parties to get to the hearts and minds of the people they sought to mobilise.¹⁷ In its aim to investigate the role of Dutch public broadcasters with regards to the scientisation of the political, this article especially draws its inspiration from the research of historian Anja Kruke. She has argued that the use of polling data and the staging of polling experts became an important means for the West-German mass media to win over the prerogative of interpreting politics from political parties. As such, this development helped mass media to redefine their position in the West German public sphere after the 1960s.¹⁸ This article analyses if – and if so, in which ways – Kruke’s conclusions apply to the Dutch case.

A selection of sources produced by Dutch media makers are the basis of my analysis. Firstly, I examined the minutes of meetings of various committees of the nts (Dutch Television Foundation) and its successor, the nos (Dutch Broadcast Foundation). These public broadcasters were responsible for regulating and overseeing all content that was made for Dutch radio and television, as well as producing their own news formats. Secondly, I studied the written press, which both made news out of the broadcast opinion polls and offered a platform for critical observers to discuss the practice of opinion polling.¹⁹ Thirdly, I analysed episodes of various radio and television formats in which opinion polls were deployed.²⁰

Taking a chronological approach, this article opens with a section that unravels how opinion polling, emerging as an American science, entered Dutch society after the Second World War, but was only very reluctantly embraced by the mass media as an instrument for making news in the immediate post-war years. In the following section, I examine how broadcasters started instrumentalising opinion polls in the changing mass media regime of the 1960s. Next, I demonstrate how broadcasters overcame sceptical responses to their use of polling and how regular surveys swiftly became central elements of electoral discourse, popularising new ways of looking at political representation. Lastly, I look at the ‘polling debacle’ of 1986, when all polling institutes had predicted the outcome of the parliamentary elections incorrectly. I investigate how this controversy marked a definitive shift in the reception of the media’s use of opinion polls, yet did anything but put an end to it.

The early twentieth century was the period in which the practice of polling manifested itself in industry and politics.²¹ The technique of statistical sampling in particular distinguished the modern opinion poll from older forms of social research. This method was and still is based on the notion that opinions are distributed evenly throughout society. Consequently, a properly constructed random sample of citizens can yield accurate information about the distribution of opinion throughout the population as a whole.²² It was the American statistician George Gallup who paved the way for the application of this new technique to map voting preferences. When his aipo (American Institute for Public Opinion) correctly predicted Franklin D. Roosevelt’s victory in the 1936 elections, it marked a milestone for the modern opinion poll and definitively put Gallup-style polling on the map as the method par excellence to chart public perceptions on all topics imaginable.²³ Around this time, Gallup started promoting public opinion polling in Western Europe, resulting in Gallup-licenced polling institutes in Great Britain and France that consistently monopolised the national market for opinion research.²⁴ In marketing opinion polling, Gallup adopted a discourse that stressed its alleged democratic advantages, making it an appealing asset for those working in the sphere of politics. He considered the opinion poll an ‘instrument which may help to bridge the gap between the people and those who are responsible for making decisions in their name’.²⁵

Inspired by the example of aipo, opinion polling also gained ground in Dutch society in the immediate post-war years, most notably as a commercial tool for industries to acquire a better understanding of consumer behaviour.²⁶ Yet, not one but multiple polling institutes formed the foundation for the Dutch polling industry, much as was the case in West Germany.²⁷ Besides the Gallup-licenced nipo (Dutch Institute of Public Opinion, founded in 1945), there was the nss (Dutch Foundation for Statistics, founded in 1940), while institutes such as Lagendijk, Veldkamp and Intomart emerged in the 1950s and 1960s.²⁸

As soon as polling agencies started conducting surveys on Dutch public opinion after the Second World War, newspapers and magazines from time to time made news out of the results. Yet, whereas British newspapers started publishing the results of electoral surveys in the immediate post-war period, the scope and scale of this type of opinion polling in the Dutch press was very limited. Most published polls dealt with non-political, social and cultural topics, while newspapers only occasionally published the results of surveys concerning electoral behaviour. Unlike the United States and Great Britain – with their straightforward two-party systems, enabling the possibility of creating an exciting ‘horse-race’ frame – the Dutch multi-party elections were too complex to be captured in such an easily-understood image. Even when institutes did try to survey voting behaviour, Dutch opinion polling was too underdeveloped to come up with a correct prediction. This became clear when opinion magazine Elseviers Weekblad and newspaper Het Parool made use of nss and nipo data respectively and were not able to correctly foresee the outcome of the first post-war general elections in 1946.²⁹

One must also realise that, until the 1960s, the Dutch political-media complex was characterised by what is commonly known as ‘pillarisation’.³⁰ To a large extent, Dutch civil society was organised along the lines of religion and ideology, resulting in closely knit networks of social, political, religious and economic organisations. Within each of these ‘pillars’, strong ties existed between media outlets and political parties, resulting in a culture of journalism that was very much characterised by appeasement and depoliticisation. The real political events took place behind closed doors, where the political elites of competing political persuasions fought out their battles, while journalists deliberately stayed out of these backrooms, preferring not to shake things up by taking a critical stance. Journalists considered it as one of their most important tasks to contribute to the consensus within their own socio-political segment of society.³¹ Therefore, the written press was very careful when it came to publishing surveys on electoral behaviour: they could raise political discussion or delegitimise a political party’s authority.³²

Initially, Dutch broadcasting companies were even more restricted by paternalist ideas about political journalism. Until 1963, they were prevented by the Dutch government from broadcasting propaganda or debates in the three weeks preceding an election. This arrangement limited the ability of editors to experiment with innovative ways of reporting on electoral campaigns on radio and television, such as developing polling-based formats. In this way, the government hoped to create a level playing field for all political parties. They attempted to prevent parties that were closely aligned to a public broadcaster from having an advantage over parties that did not have a broadcaster at their disposal.³³

Such concerns were not unwarranted, as Dutch broadcasting companies – just like the press – were in some cases closely connected to specific political parties.³⁴ This explains why broadcast editors probably did not even consider experimenting with polls in the first place. They considered it among their most important tasks to offer a platform to politicians of the political party affiliated to their broadcasting network and to educate the members of the socio-political segment in society they represented.³⁵

The ties between parties and broadcast companies began to loosen over the course of the 1960s. Mirroring developments in the press, a new generation of broadcast media makers brought along new notions of political journalism. Instead of manifesting themselves as the ‘subservient supporters’ of political elites, they aimed at critically confronting these elites and acting as the autonomous and critical interpreters of political life.³⁶ They intended to establish themselves as the crucial mediators between politics on the one hand and the Dutch citizenry they sought to emancipate on the other.³⁷ Broadcasters did so, for example, by developing critical media formats in which political leaders were confronted with the views of the audience or in which they had to debate each other, so that the viewers at home could get a clear idea of the different party positions.³⁸ Journalists’ need for new ways of analysing politics was confirmed by the growing electoral volatility of the 1960s, indicating that traditional people’s parties had lost their grip over their core electorates.³⁹

In this changing media regime, broadcast companies started to commission and present survey results on electoral behaviour. Polls were not objective lenses, as media makers very much used them to create a narrative of political crisis and to echo and popularise the very narrative of ‘depillarisation’.⁴⁰ In this sense, they were making the political weather as much as reporting on it.⁴¹ The first polling programmes on Dutch television served as a means to democratise political communication and redefine the relation between politics, the mass media and the public – just like journalists in West Germany had intended when they started commissioning opinion polls around the same time.⁴²

The first Dutch format to completely revolve around opinion polls was nipo-these, a monthly television programme produced by the ncrv (Dutch Christian Radio Association), which aired six times in the run-up to the parliamentary elections of 1967. Every episode staged a prominent politician of one of the major political parties responding to the results of surveys nipo had conducted for this Protestant broadcaster. ncrv’s press release stated that the aim of nipo-these was to ‘look through’ politics and ‘confront’ the attending politicians with the perspectives of the electorate.⁴³ This type of language is exemplary of the democratic discourse that underpinned opinion polling from the outset. The broadcaster subscribed to the notion that polls were tools to narrow the gap between politicians and the Dutch voters. This was illustrated by the type of surveys they commissioned: these did not only gauge voting behaviour, but also mapped people’s trust in political leaders and public perceptions of political parties, allowing them to indicate which politicians failed to empathise with the electorate.⁴⁴

Yet, in contrast with the emancipatory discourse invoked by the broadcaster, the set-up was rather modest. nipo’s survey results incited friendly discussions between the studio attendees rather than a confrontation. This can be explained partly by the fact that nipo’s polls did not yield any spectacular, ground-breaking results.⁴⁵ Moreover, some political parties shielded their political leaders from going to the ncrv studio. Instead of its prominent and much-talked-about political leader Norbert Schmelzer, the kvp (Catholic People’s Party) only sent its chairman.⁴⁶ In response to Schmelzer’s absence, a critical observer noted that it would take the broadcaster more than ‘some survey statistics and some little interviews’ to fulfil its promise of ‘looking through politics’.⁴⁷

In the ranks of the nts, somewhat paternalist perceptions with regard to political communication seemed to continue to exist. One of its broadcast managers proved to be reluctant when it came to presenting prognoses too close to election day and proposed forcing ncrv to air nipo-these’s final episode at least one week before the elections.⁴⁸ He was probably afraid that the survey results would incite citizens to change their voting behaviour. As it turns out, other editors did not share his standpoint, for the final episode was actually aired not seven, but only five days before the elections.⁴⁹

That same year, the nts itself also started to experiment with opinion polls in its election programming. After the closing of the polls on election day in 1967, the broadcaster presented a prognosis drawn up by a team of sociologists of Utrecht University. This so-called exit poll was devised, suggested and hosted by the young sociologist Marcel van Dam. Polls were conducted at four of Utrecht’s polling stations and the collected data were transported and analysed swiftly by the use of punch cards. Van Dam succeeded in giving a rather accurate indication of the national election results, hours before these were made public.⁵⁰ Unlike nipo, Van Dam and his team did actually foresee the unprecedented electoral losses the traditional people’s parties – kvp and PvdA (Labour Party) – would suffer during the 1967 elections, consequently raising their polls’ newsworthiness. Besides this, they were able to show how voting behaviour varied along the lines of age, confessional choice and previous voting behaviour.⁵¹

Needless to say, it is hardly surprising that Van Dam was successful in predicting the outcome of the election when nipo was not. nipo conducted its final poll days before election day and could not account for last-minute changes in voting intention. Van Dam and his team, in turn, intentionally surveyed Dutch voters after they had already cast their ballots. Today, it is generally known that exit polls, due to their methodology, guarantee a more precise prognosis of election results than ‘regular’ opinion polls. Yet in 1967 the Dutch public was very much impressed by the accuracy of Van Dam’s exit poll. More than nipo-these, nts’s election night exit poll, therefore, raised public awareness about opinion polling as a tool to map and interpret voting preferences. The day after election day, various newspapers devoted attention to Van Dam’s survey and praised its novelty: they wrote of ‘an interesting and unique phenomenon’ and ‘an informative analysis’.⁵² nts’s successor nos decided the exit poll was here to stay. In the years to come, the public broadcaster would innovate the concept: telex connections and high-speed printers were used to collect polling data from no less than twenty-one municipalities during the parliamentary elections of 1972.⁵³ Exit polls are used to this very day in the Netherlands, but are now hosted by nos journalists themselves.

Van Dam’s endeavours in the field of what he called ‘electoral sociology’ were motivated by an engagement with progressive politics and political renewal. By means of the polls, he aimed at delegitimising traditional and social determinist notions of political representation. The political identities of voters were no longer formed along socio-religious lines, Van Dam demonstrated, and an increasing number of voters had become active and opinionated citizens who based their electoral choices on the political parties’ agendas. ‘25 to 30 per cent of voters belong to the “floating vote”’, he told a newspaper journalist after his first exit poll in 1967.⁵⁴ The concept of ‘floating voters’ already existed among political scientists – both in the Netherlands and beyond – and he now used it to popularise a narrative of political crisis and promote the idea of political renewal.⁵⁵ ‘Centrist’ political parties could only get by if they learned to take the perceptions of the electorate more seriously and became more transparent in their political positioning, or so Van Dam believed.⁵⁶ ‘PvdA should become more radical if they wish to improve their relationship with the people,’ he would tellingly explain to the same newspaper journalist.

In addition to the fact that the exit polls helped the broadcaster to fulfil its statutory task of producing informative television, the nos had more instrumental motivations for continuing with the election night exit poll. Editors also perceived the broadcasting of polling results as a tool to combat the waning public interest in political television that set in since the 1960s, a development that concerned them greatly. One broadcast manager argued that the segment added ‘a valuable element of competition’ that gave the election night broadcast more allure. Nonetheless, others claimed that the exit poll distracted from the interviews with politicians attending the election night broadcast and, as such, slightly ‘missed the mark’. As it turns out, the early manifestations of the media logic mentality of some of the nos managers were not uncontested within the ranks of the broadcaster.⁵⁷

Over the course of the late 1960s and the 1970s, more broadcasters started to use opinion polls to present prognoses on the public’s electoral preferences. Current affairs television programmes of traditionally ‘pillarised’ broadcasters – such as Hier en Nu (Here and Now) by ncrv, Achter het Nieuws (Behind the News) by vara (Association of Worker Radio Amateurs) and Brandpunt (Focal Point) by kro (Catholic Radio Broadcaster) – occasionally commissioned polls at institutes like Bureau Veldkamp, nipo and Intomart. The radio programme Delta by ‘neutral’ broadcaster avro (General Association of Radio Broadcasting) followed their example. Broadcasters sometimes continued to use polls as tools for chronicling a narrative of political crisis and as a means for manifesting themselves as the most important representatives of the Dutch electorate. In 1972, avro’s Delta, for example, reported that two out of three Dutch voters considered themselves insufficiently informed about the intricacies of Dutch politics and contended that they could not arrive at an informed political opinion.⁵⁸ Most dominant, however, were polls about voting preferences and prognoses of elections to come. Newspapers, in turn, were often eager to report on the results of these surveys.⁵⁹

Yet, as Dutch electoral opinion polling was still in its early days, the projections of the various institutes often significantly varied and frequently failed to correctly predict election results.⁶⁰ As a result, the reception of these polls was somewhat lukewarm, and critical journalists and social scientists were rather sceptical about polling as a way of measuring public opinion.⁶¹ While a psychologist argued that pollsters just had to improve their methodology and interpretative frameworks, a journalist from newspaper nrc Handelsblad remarked that the problem ultimately lay with the electorate.⁶² The latter believed that most prognoses had no predictive value at all, as most voters were not nearly as opinionated as pollsters believed them to be.⁶³ The concept of the opiniated ‘floating voter’ Van Dam sought to popularise with his exit polls was clearly not shared by all Dutch political commentators.

In addition, broadcasters were frequently confronted with accusations that they handled polls as a means of propaganda. Indeed, clear political alliances became visible between broadcasters, political parties and polling institutes, especially between vara, the PvdA and nipo. Both the left-wing broadcaster and the Social Democrats made use of nipo data and sometimes even seemed to share survey results.⁶⁴ More than that, nipo’s director even had a seat on one of the party’s electoral committees in the 1960s.⁶⁵ Because of these affiliations, vara was repeatedly confronted with allegations about the manipulation of polling data to show that important political opponents were losing electoral support.⁶⁶

Worries about the reliability of opinion polls and their added value for democracy were voiced in other countries as well. In West Germany conservative critics had deemed the ‘galluping’ consumption of polls a possible threat for political leadership, while in France the biggest criticism was formulated by progressive political observers.⁶⁷ Well-known sociologist Pierre Bourdieu argued that public opinion did not even exist and that polls benefitted select elites such as the politicians and journalists that commissioned them rather than the citizenry at large.⁶⁸ Likewise, communists in the Netherlands interpreted opinion polls as manipulative tools for the bourgeois powers that be.⁶⁹

It was vara’s radio programme In de Rooie Haan (In the Red Rooster) that marked a shift in the public perception of opinion polls on Dutch television and radio. As of 1976, the left-wing political radio programme broadcast the results of opinion research on a monthly basis. Young, media-savvy social geographer Maurice de Hond was asked to analyse the data and disclose the results in the Saturday afternoon broadcasts. Making use of the data of a continuous nipo-poll but applying his own method of analysis, De Hond claimed he could follow the development of electoral behaviour over time more accurately than other pollsters. Not only did he conduct his poll on a more frequent basis, namely once a month (and later even once a week), and did he make use of a larger sample than other pollsters, he also aimed at correcting certain over- or under-representations in his data.⁷⁰ His first prognosis in which he foresaw a significant loss of electoral support for the progressive parties, which formed the backbone of the left-wing government at the time, harvested great attention and criticism from politicians and electoral researchers alike.⁷¹ Yet, accusations of manipulation and imprecise methodology swiftly vanished when the outcome of De Hond’s prognoses turned out to be accurate a few times in a row.⁷² The polls quickly turned into one of the radio show’s most popular segments, with De Hond becoming somewhat of a polling celebrity.⁷³

Once more, the opinion polls were, to an extent, presented as a project to democratise political communication. In a newspaper interview after the first In de Rooie Haan opinion poll, the programme’s editor Jan Nagel implied that the Dutch audience had the right to know these numbers and argued that it was ‘a good thing’ that the left-wing government leaders were publicly confronted with their waning electoral support.⁷⁴ Yet, he later admitted that publicity purposes had played at least as big a role in launching the In de Rooie Haan polls, a motive newspaper journalists had already recognised from the outset.⁷⁵

In several regards, the developments surrounding vara’s electoral prognoses represent the further institutionalisation of broadcasters’ use of opinion polls in the 1980s. Firstly, prognoses proved to be increasingly accurate as the practice of opinion polling professionalised. A significant development, amongst others, was the implementation of telephone surveys in the second half of the 1970s instead of interviewers that went from door to door, which enabled more frequent and thus more precise gauging of political preferences.⁷⁶ As a result, criticism of opinion polling grew somewhat silent. Political observers started to accept that opinion research was here to stay and that it was an important aspect of representative democracy.

Secondly, through their even more frequent use in radio and television programmes, opinion polls turned from individual news events into recurring political ‘forecasts’ of the political landscape and became fixed parts of the country’s public electoral discourse. Broadcasters such as avro and tros (Television and Radio Broadcast Foundation) followed vara’s example and bought fixed subscriptions to surveys of polling institutes, presenting the results in their political radio programmes.⁷⁷ The establishment of recurring polling segments was not unique to the Netherlands. Around the same time, zdf in West Germany started airing Politbarometer, a monthly programme that aimed – and still aims – at providing information about electoral preferences, presenting shifts and alterations in voter opinion on a regular basis, like a type of political ‘weather report’.⁷⁸

Apart from the fact that recurring polls further accustomed the Dutch public to imagining politics through the scientised language of percentages and bar graphs, they also popularised a marketised perception of political representation. Broadcasters presented and interpreted the shifts in political preference these polls uncovered as if the electorate were a stock market: opinion polls publicly mapped which political party had the greatest ‘market demand’ at any given time. Newspaper articles about these polls are indicative of the type of commercialised language that was used at the time. Headlines like ‘Survey Shows Christian Democrats and Liberals Drop Below 50 percent’ and ‘Popularity of Social Democrats Somewhat Wanes’ were not uncommon in the 1980s.⁷⁹ This discourse contributed to the image of voters as ‘citizen consumers’, who are individuals ‘that will question every aspect of elite provision and will no longer accept being told by the elite what is good for them’.⁸⁰ Opinion polls helped spread the notion that citizens no longer perceived their party choice as a fixed given, but critically and constantly compared politicians’ claims to their own needs and demands and, if necessary, switched party – as if it were a consumer good. This conceptualisation of politics as an open marketplace popularised the idea that political parties could only win elections if they were able to get themselves noticed among floating voters and capitalised on the political preferences of this group.⁸¹

The tacit approval of opinion polls made way for the definitive return of a more critical attitude after the general elections of 1986. In the run-up to these elections, opinion polls were more present than ever. All polling institutes, including De Hond’s recently founded market research institute Inter/view, conducted surveys on the popular support of political parties. They saw their polls being broadcast on radio and television, while newspapers just as eagerly reported on all the various survey results.⁸² Polling experts such as De Hond and Ger Schild (director of nipo) were frequently invited to radio and television studios, where they were presenting the latest survey results of their institutes and clarifying what this meant for parties’ chances at the ballot box.⁸³ As the voting results came in on election night, however, it turned out that none of the polling institutes had successfully predicted the outcome of the elections. Whereas they had foreseen a victory for the Social Democrats, the Christian Democrats eventually ended up with two more parliamentary seats than their left-wing opponent.⁸⁴

This polling ‘debacle’ very much exposed the limits of opinion polling in Dutch society. Paradoxically, the mass media themselves played a central role in fuelling the debate. Programmes that, only days before, had staged De Hond and Schild as their ‘own’ polling experts now critically interrogated them about their shortcomings.⁸⁵ Likewise, newspapers offered a platform to critical observers discussing the ‘possibilities and pitfalls’ of opinion polling.⁸⁶ The controversy heralded the definitive consolidation of a culture of what scholars call ‘secondary scientisation’, which is characterised by a more critical and reflexive attitude towards scientific expertise in society.⁸⁷ From then on, opinion polls were no longer straightforwardly accepted as information on the electorate, but constantly scrutinised for their inaccurate and disruptive nature.

Some older complaints with regard to the use of opinion polls also resurfaced, such as the idea that polls were mainly propaganda devices.⁸⁸ Yet, it was another discourse that set the tone in 1986. Critics, most notably social scientists, now predominantly had qualms about the corrupting effect of broadcasting and publishing opinion polls on political communication in general. The fact that the polls had virtually declared the Social Democrats the winners of the election had encouraged citizens to cast their votes for other parties, it was argued.⁸⁹ Critics believed that the ways in which opinion polls had become mass media phenomena had enhanced their performative value in such a way that they had actually influenced – instead of just measured – voting behaviour.⁹⁰ In addition, they claimed that the constant gauging of voting preferences had rendered political journalism superficial.

For those reasons, political scholar Andries Hoogerwerf and communication scholar Anne van der Meiden argued in 1986 for the prohibition of the publication of polls in the two weeks preceding an election. They asserted that polls distracted from the actual content of politicians’ arguments and focused, instead, on the talking heads of the political parties’ campaigns and their relative chances of winning the elections.⁹¹ In doing so, they were urging for legislation that was similar to that in France. There, the publication and broadcasting of opinion polls in the week before elections had been banned since 1977, for polls allegedly produced ‘sheep-like movement’ and ‘collective madness’, or so critics argued.⁹² Hoogerwerf’s and Van der Meiden’s arguments formed the building blocks for the mediatisation narrative that – as I have shown in the introduction – would eventually dominate scholarship on the political-media complex in general and discussions about opinion polls in particular.

Even though the nos recognised that it had to make sure to more cautiously present polling results in the future, things did not really change. The polling debacle gave rise to a discourse that stated that the share of floating voters was bigger than ever and that elections could have unforeseen outcomes. One of the people to engage with this discourse was De Hond, who maintained that his wrong prediction was caused by the fine performance of Christian Democratic leader Ruud Lubbers in the televised debate that was aired on the Sunday before election day.⁹³ Had De Hond been able to conduct a poll after this debate, he stated, then he probably could have indicated that Lubbers had swayed a great deal of the floating voters into voting for him.⁹⁴ This line of reasoning propagated the idea that elections were ultimately won on television and only pressed opinion polls more firmly into the hands of media makers, as surveys were the instruments par excellence to keep the finger on the pulse of the ever more volatile electorate.

Historians have shown that the ‘scientisation of the political’ entailed the ways in which political elites changed their electoral tactics and governance practices. This article demonstrates that the scientisation of Dutch society through the media’s use of polls aimed at transforming the very self-image of the Dutch voter who had to come to see itself as a democratic powerholder. I have traced the appropriation of opinion polls by Dutch broadcasters in order to provide insight into the shifting ideas and imaginaries of the dynamics of democratic politics. Growing electoral volatility in the second half of the 1960s had disproved social determinist notions of political representation and delegitimised the image that the core electorates of the traditional people’s parties consisted of stable communities revolving around fixed identities. Polling segments on radio and television began to use surveys as a means to map and interpret the ever more unpredictable parameters of voting behaviour and the electorate’s changing demands.

Yet, polls also formed the building blocks for ‘depillarised’ ways of conceiving of democracy. Within less than twenty years, opinion polls became the focal point of a dominant marketised electoral discourse, revolving around recurring updates about the electoral ‘stock market’ and popularising the notion that elections were ultimately won by those parties that got through to the ever-growing share of floating voters. Simultaneously, the mass media offered a platform for meta-perspectives on the position and function of polling in Dutch democracy and often fuelled debates on the shortcomings and manipulative nature of opinion surveys, for example in the early 1970s and more notably after the elections of 1986.

Furthermore, the earliest polling programmes on Dutch radio and television have played a key role in the changing nature of the political-media complex, yet not in the way scholars generally argue. Just as Anja Kruke has done for West German broadcasters, this article argues that, by constantly mapping and disclosing the peculiarities of the voting preferences of the Dutch electorate, Dutch broadcasters were able to claim a crucial role for themselves as the intermediaries between political elites on the one hand and the electorate on the other. This argument does not comply with the notions of ‘media logic’ and ‘mediatisation’ that media and communication scholars usually employ when qualifying the role of polls as part of the relation between media and politics. If anything, not media logic but the notion of ‘public logic’ is more suitable to grasp the dynamics of the earliest polling programmes on Dutch radio and television. Characteristic of this logic is a robust sense of autonomy among political journalists, who strongly identify with the public good and the democratic process.⁹⁵ Indeed, polls would eventually play an important role in making political journalism more exciting, especially by frequently mapping the popularity of political leaders. Yet, this article demonstrates that the need for entertainment was only one – and not even the most prominent one – of the various arguments that media makers had for developing polling programmes in the 1960s, 1970s and 1980s.

Far more decisive were the progressive notions of political journalism that a new generation of media makers brought along. Their emancipatory ideals were shared by the young and media-savvy experts that these media makers staged in their polling programmes. Using statistical analysis and scientific concepts such as ‘the floating voter’, they aimed to demonstrate that Dutch voters were anything but the passive playthings of political elites, but rather critically followed what happened in The Hague politics and adjusted their voting choice accordingly. Besides Van Dam, De Hond is an example of such an expert. In the decades following his first performances on vara radio, the latter became the textbook example of an expert who used his expert persona to combat the supposed hegemony of political, media and academic elites.⁹⁶

How does all this compare to developments in other Western European countries? The appropriation of opinion research by the Dutch mass media happened more gradually in comparison to the United States and Great Britain. This ‘delay’ can be partly explained through the ‘pillarised’ nature of Dutch democracy in the immediate post-war period. Unlike journalists in the United States, Great Britain and West Germany – all countries with a more-or-less two-party system – Dutch political journalists only to some extent embraced the ‘horse-race’ frame in their coverage of elections in the early decades of polling. In other regards, however, it turns out that the Dutch case hardly stands out. The ways in which the West German media devised polling programmes and quickly became key players in the representation of polling data actually mirror the trends I analyse in this article.

Furthermore, just like in many other countries – most prominently France – opinion polls were never completely accepted as unproblematic and fully objective representations of public opinion. In the Netherlands, the ‘debacle’ of the 1986 elections played a pivotal role in popularising the notion that opinion polls did not just measure, but also affected voting behaviour. Yet, the aftermath has proven that the opinion poll was here to stay. Unlike in France, no bans were installed and while the scepticism endured, the pace and scope of opinion polling only intensified at the turn of the century.