Opinion Polls across Boundaries: The Early History of Northwestern European Opinion Polling beyond National Borders and Disciplinary Frameworks

Abstract

This article focuses on the early history of Northwestern European opinion polling (1940s–1950s), specifically the cases of the Netherlands and Sweden. The evolution of opinion polling and its influence on post-war politics and society should be understood in light of processes of international transfer and entanglement. The Dutch-Swedish comparison brings into focus the ways in which the national experiences of the Second World War influenced how opinion pollsters discursively linked the practice to ideas about democracy. Furthermore, the article highlights entanglements across the boundaries of science, as commercial survey methods were picked up by social scientists, and across national borders, as opinion pollsters across Western Europe were in frequent contact with each other.

Introduction

On 22 September 1950, Wim de Jonge from the Dutch Institute for Public Opinion (Nederlandsch Instituut voor de Publieke Opinie; NIPO) wrote the following to his colleague Sven Blomquist at the Swedish Gallup Institute (Svenska Gallupinstitutet):

Dear Sven, Thank you for your letter of September 20, 1950. Whole-hearted congratulations with the fine pre-election results you got. As I see you have the same difficulty as we have in the presentation of your pre-election figures. It is silly but necessary here too, to even out the results to halves of one percentage. … With kind regards, until September next year, Sincerely yours, Drs. W. J. de Jonge, Nederlandsch Instituut voor de Publieke Opinie¹

Over the course of the 1940s and 1950s, a group of enthusiastic men and women from all over Europe corresponded about an exciting new phenomenon: opinion polls. Inspired by all things American and by George Gallup in particular, the so-called ‘father’ of American opinion polling, pollsters such as Wim de Jonge from the Netherlands and Sven Blomquist from Sweden founded their own opinion polling organisations in the image of the American Institute of Public Opinion (AIPO). As the letter from De Jonge to Blomquist from September 1950 shows, the development of the opinion poll in Europe was not the sole product of this initial American transfer, but instead arose from a constant international network of Western European opinion pollsters. They exchanged best practices on statistical and sampling techniques, shared their results and client base, organised conferences and internationally comparative polls, and kept an eye out for new recruits – for example, actively meddling with the setup of opinion polling organisations in Germany after the war, where ‘democracy as we know it’ was absent.² In doing so, these fellow ‘fullblood statistician[s]’ also exchanged ideas on the possibilities of opinion polling for post-war democracy.³ From the United States, to Sweden, to the Netherlands: the first generation of European opinion pollsters pinned great hopes on this new statistical technique as a way to make sense of the uncertain times of the early postwar years.

Opinion polls have profoundly changed the concept of public opinion as they fostered the idea of public opinion as something quantifiable. Political scientist Susan Herbst notes that such ‘forms of quantification make the world seem less chaotic’ and that opinion polls held the promise of ‘rationalisation’. From the start of the eighteenth century, policy makers understood that numbers could help them map the ‘needs and opinions of the populace’ and consequently rule more efficiently.⁴ After 1945, the idea that opinion polls could serve as a way to gain knowledge about war-torn European societies, or even as a type of popular referendum, enthused some. Others feared the American practice fostered a superficial, simplified or even undemocratic perception of public opinion. As American and European pollsters would develop strong ties with the press, governments, companies and political parties, opinion polls would nevertheless gradually become part and parcel of post-war democracies in the Western world. Often, the rise of opinion polling is even seen as ‘a common characteristic’ of twentieth century democracies.⁵

Multiple historians have argued that in order to historicise the opinion poll, emphasis on its implementation and institutionalisation in different national contexts is essential, rather than treating polling as something that offers direct insight into the opinions of contemporaries.⁶ Taking heed from these authors, we treat opinion polls as being context-dependent while also having an effect on the world around them. The existing historiography on opinion polling in the United States and Europe shows that the development of the practice was heavily dependent on national contexts. So far, research has been conducted on, for example, the American, British, Dutch and French cases.⁷ Although Sweden was one of the first countries where Gallup-style opinion polling was introduced, historical research is lacking. Likewise, the limited historical research on the Dutch case is at odds with the important role that the country played in the European network of Gallup-affiliates throughout the post-war decades. Studying the connected early history of opinion polling in the Netherlands and Sweden thus brings into focus the profound impact of transnational and national political contexts, in particular national experiences of the Second World War, on how polling organisations introduced opinion polling and discursively linked the practice to ideas about democracy.

Despite the obviously international and transnational character of opinion polling, much of the previous research into its history has focused on national contexts. This article proposes to rethink the early history of Northwestern European opinion polling via its transfer from the United States by mapping the connected arrival, institutionalisation and public reception of Gallup-style opinion polling in Sweden (1941) and the Netherlands (1945), respectively. As the techniques of opinion polling transferred from the United States to Europe, so too did the discursive link between opinion polling and democracy. Our comparative approach shows the different ideas about democratic society prevalent at this time in Sweden and the Netherlands, which were a consequence of different experiences of the Second World War. Whereas in the Netherlands the Nazi occupation put an end to democratic government in the 1940–45 period, Sweden remained neutral and retained its democratic form of government. The different national war experiences had an impact upon how opinion polling and its possibilities for democratic society were understood and appropriated by various national opinion pollsters and their clients, such as the press and political parties.

Taking inspiration from previous research into the transfer of opinion polling from the United States to various countries, this article moves beyond comparative history to also highlight international connections.⁸ A focus on international transfer, be it on a cultural or political level, forces us to look beyond the national histories of phenomena such as opinion polling and recognise that they were, and continue to be, products of international networks. Where the comparative method allows historians to better understand the history of nations, studying cultural transfer makes it possible to ‘demonstrate that things did not only happen at the same time or resemble each other, but were also connected’.⁹ In other words, this study cannot be purely comparative in its approach; comparative history tends, as Kocka has noted, to presuppose that the units of comparison can be separated from one another.¹⁰

However, we also want to complicate the often quite linear narratives of transfer that surround the early history of polling, where polling seems to travel from the United States to various discrete national contexts. In the immediate post-war years, opinion pollsters and market researchers across national contexts worked hard to build wider international networks, such as the World Association of Public Opinion Research (WAPOR), founded in 1947, and the European Society of Market and Opinion Research (ESOMAR) that was set up in the following year.¹¹ In these processes of internationalisation, pollsters in Sweden and the Netherlands were in frequent contact with each other, and these international networks shaped the nature of polling in these two national contexts. We thus propose that the transnational history of polling should be written with inspiration from the entangled history approach. Here, the concept of entanglement – defined as interrelations in space, knowledge or time – points to how polling in these two, and other, national contexts were interrelated and affected each other.¹²

In sum, this article aims not only to compare how opinion polling was transferred from the United States to the Netherlands and Sweden, but also to show how this process was entangled across national boundaries.¹³ This article asks: how did Northwestern European opinion polling develop within a transnationally entangled framework (1940s–50s) and, specifically, how did ideas about the democratic potential of opinion polling transfer transnationally through the practice of opinion polling?

Beyond the transfer and entanglement of opinion polling across national borders, we further posit that this dynamic extended to the interplay across the boundaries of science. Previous research has often analysed opinion polling in relation to the notion of ‘the scientisation of the social’.¹⁴ The term was coined by Lutz Raphael in the mid-1990s and denotes the process by which experts from the social and human sciences, as well as their arguments, concepts, methods and research outputs, have become increasingly present in other societal spheres.¹⁵ According to Raphael, historical analysis of the role that social sciences played in society and politics must overcome the limits of discourse analysis by looking at the role of experts as protagonists, the clients and users of social sciences, the histories of techniques and methods of social sciences, and the history of relevant institutions.¹⁶ ‘Measuring’ people's views on the basis of ‘scientific’ methods, capturing them in numbers and representing them in percentages, graphs and charts, all contributed to the development of a scientific perspective on public opinion. Of additional importance is Igo's proposal to research the social and epistemological transformation triggered by opinion polling, and specifically its effect on ‘how citizens came to know the public mood but also how they placed themselves within the nation’.¹⁷

While the concept of scientisation thus highlights an important historical process of the twentieth century, its linearity can be problematised as it invites researchers to only focus on the one-directional transfer of experts, logics, methods and information from the sphere of science to other societal spheres. As Kaal and De Jong have argued, ‘the scientisation of the political’ was in fact a diffuse and multidirectional process that went hand in hand with a ‘politicisation of the social sciences’.¹⁸ We argue that the position of opinion polling in the borderlands between social science, politics and media can highlight the entanglement between these different spheres. Throughout the twentieth century, polls have been used and perceived in various capacities: as knowledge techniques or instruments of science, as propaganda mechanisms or marketing devices, as news events and as rhetorical figures.

Beyond travelling over the spatial boundaries of nations, opinion polling and the data created from it also travelled back and forth across the less tangible boundaries of different societal spheres. Polling thus has a hybrid quality: it is not purely scientific, journalistic, or commercial, but rather shifts between these positions.

This article is structured as follows. It starts with tracing the transfer process of opinion polling from the United States to the Netherlands and Sweden in the 1940s. Based on articles written by the first generation of pollsters in both countries, it explores how the idea of opinion polling as a democratic practice was transferred, emphasising the ways in which the national experience of war affected how opinion polling was understood. Second, we move on to discuss how ideas of opinion polling as a scientific way of understanding (and perhaps controlling) society were implemented in both countries. Based on articles written by both pollsters and journalists, the article points to the entangled aspects of knowledge dissemination at play in the early history of opinion polling. Third, this article maps the transnational network in which pollsters from both countries participated, based on sources such as international correspondence, conference proceedings and scientific publications.

International Transfer and Democratic Ideas

From a variety of historical studies, we can identify various common denominators in the history of opinion polling and how the practice influenced politics and society. One important theme is that opinion polling was often discursively linked to democracy and political representation.¹⁹ When it started in the United States in the 1930s, opinion polling was perceived as a ‘democratic science’, as it promised to make public opinion transparent for everyone and consequently protect democratic society from non-democratic leaders.²⁰ George Gallup had marketed it as a democratising tool and held that polling enabled politicians and policy makers to become more attuned to the views of the ‘common man’.²¹ With the transfer of opinion polling from the United States to the United Kingdom in the late 1930s, to Scandinavia in the early 1940s and to the European mainland after 1945, this link between opinion polling and democracy also transferred. Many studies prove the point. For example, Beers and Roodhouse show that British pollsters copied Gallup's rhetoric and promoted opinion polling as a democratising instrument.²² Ostrow argues that opinion polls were quickly embraced in Germany as an Anglo-American scientific antidote to Nazi-era thought patterns, and as a way to measure public opinion and show the Allies that German citizens were real democrats.²³ Opinion polls also influenced how political parties perceived themselves in relation to the electorate, as work by Kruke and Ziemann shows for the German case and work by Kaal and De Jong for the Dutch case.²⁴

However, this American idea of opinion polling as a democratising practice did not transfer to the European mainland untouched. Factors such as national political culture and pre-war ideas about political representation played an important part in this. From previous research on a variety of national case studies, it is clear that practices and institutional arrangements of opinion polling differed significantly between countries and were dependent upon the expectations of democracy prevalent among both pollsters, as well as their clients such as the press and political parties. For example, Beers shows that while British pollsters embraced Gallup's rhetoric, political parties heavily opposed the idea that an entity other than a political party could represent the people democratically.²⁵ Likewise, Cowans demonstrates that in the French case, politicians and journalists both ‘were dragged kicking and screaming into the Polling Age’, partly because of their ‘revulsion and resentment toward America and its culture’.²⁶ In contrast, both Kruke and Kaal and De Jong show that electoral research caused Dutch and German political parties to rethink social-determinist conceptualisations of the electorate and political representation in the 1960s.²⁷ Although these studies do mention the relevance of the American origins of opinion polling, they do not help to explain the country-specific translation of ‘American’ ideas about opinion polling as a democratic practice to Northwestern Europe in depth.²⁸

Both Dutch and Swedish public discourse on opinion polling was initially marked by the idea of opinion polls as a tool to strengthen democracy. The first generation of Dutch pollsters was heavily inspired by both American and British polling initiatives.²⁹ Next to several one-time initiatives on a local and regional level, two national organisations emerged. First was the Dutch Foundation for Statistics (Nederlandsche Stichting voor Statistiek; NSS), which had been established in 1940 as part of the CBS (Central Bureau of Statistics) and promoted itself by means of the Dutch Association for Opinion Research (Nederlandsche Vereeniging voor Opinie-Onderzoek; NVO), mainly tailoring their efforts towards clients such as civil servants and administrators. Second was NIPO, established in 1945, which had a commercial purpose and initially focused on the press as their main client. While the NSS and the NVO were headed by university professors, NIPO was led by Wim de Jonge and Jan Stapel, both of whom had roots in advertising and corporate market research departments.³⁰ As Stapel and De Jonge had already visited the French Institute for Public Opinion (Institut français d'opinion publique; IFOP) and the British Institute for Public Opinion (BIPO) in 1945, they were strongly influenced by Gallup. Due to this early connection to the Gallup network and the fact that NIPO possessed a large arsenal of media contacts because of their commercial ambitions, it was NIPO that won Gallup's favour over the NSS and subsequently became an official Gallup institute in 1947.³¹

Like Gallup, NIPO promoted the opinion poll based on its democratic promise and considered it useful that pollsters could request citizens’ opinions more often and on a larger variety of issues than just on election day. NIPO claimed that polls could allow citizens to make their opinions known in the press and to the government and politicians. They emphasised that everyone could be asked for their opinion and that the opinion of each interviewee was legitimate enough to be included.³² This interpretation of opinion polling was more widespread: in 1945, the newly established Dutch State Information Service (Rijksvoorlichtingsdienst) even labelled opinion research a ‘demand of democracy and society’.³³ Moreover, in these early years polling was also adopted in several local and regional initiatives, for example in Kerkrade and Roosendaal. The initiators from Kerkrade were particularly inspired by the work of William Lydgate, who worked at the American Institute of Public Opinion founded by Gallup.³⁴ These grassroots pollsters departed from a certain distrust of government and politics: they believed that the people, not the government, should rebuild the country, as the leaders had not been able to prevent the war and German occupation. Moreover, they held that opinion polls allowed citizens who had previously been distant from social life and therefore susceptible to charismatic leaders, to express their opinions. Calling their initiative ‘a democratic deed’, they saw opinion polling not only as a democratising but also as an emancipatory practice.³⁵

The arrival of opinion polling in Sweden was far from a story of simple transfer from the United States. In 1939, Gallup was in Berlin to investigate the possibilities for a polling institute in Germany, as he was setting up an international network of polling institutes.³⁶ There, he was approached by a Danish market researcher, Haagen Wahl Asmussen, who received Gallup's permission to use Gallup's name and methods in Scandinavia.³⁷ The German occupation of Denmark soon complicated the work of the Danish Gallup Institute (Dansk Gallup Institut), but documents with methodological instructions were passed on to Swedish advertising executive Sven O. Blomquist during 1941. It was with these that he started the Swedish Gallup Institute, which published Sweden's first major opinion polls at the end of the same year.³⁸ Thus, the arrival of opinion polling in Sweden involved not only Sweden and the United States, but also Germany and Denmark, with the Second World War as an important context. The war furthermore limited the possibility of contact with American pollsters, so Swedish polling methods were initially characterised by a fair amount of trial and error in fitting the American instructions to the Swedish context. In 1945, the Swedish Gallup Institute could finally initiate contacts with the AIPO and obtain advice through letters, telegrams, phone calls and conferences.³⁹ Initially, the business model was similar to that of the AIPO – as well as that of NIPO – and relied heavily on newspapers signing up for weekly syndicated articles on the latest polls. The Institute also conducted various forms of specially commissioned polls and market research which would become increasingly important in the Institute's later years.⁴⁰

The highly influential couple Alva and Gunnar Myrdal – pioneering social scientists, public intellectuals and social democratic politicians – were the first to introduce the ideas of the democratic potential of opinion polling to the Swedish public.⁴¹ In March 1941, after three years in the United States, they published Contact with America (Kontakt med Amerika), a book in which they praised many aspects of American democracy, including opinion polls. For the Myrdals, opinion polling could provide direct access to public opinion. Moreover, it would foster better democratic citizens, as they allowed the ‘common man’ to understand himself as part of the democratic public that was represented in the polls and simultaneously make him more resistant to demagoguery. Thirdly, it could be used to map lacunas in political knowledge, making it clear where efforts of political education might be needed.⁴² These arguments were later repeated by others.⁴³ This position should be understood in relation to the context of the 1930s and early 1940s as a time when democracy was under threat. Partly, Social Democratic policy in 1930s Sweden was a response to the challenge from National Socialism, with the aim of demonstrating the redundancy of fascism in Sweden.⁴⁴ Furthermore, the 1930s saw discussions on democratic political culture and its forms of communication, with the Myrdals centre stage.⁴⁵ In 1941, with an ongoing world war of which the outcome was deeply uncertain, democracy was anything but secure. In Sweden, the idea of opinion polling as a tool for transforming and making democracy more effective thus seems to be related to the challenges to democracy posed by National Socialism and the ongoing war. This is supported by the fact that such arguments do not occur to nearly the same extent after 1945.

Although both NIPO and the Swedish Gallup Institute were commercial companies that had economic gain as their primary purpose, national wartime experience caused different interpretations of the American idea of opinion polling as a democratising practice. The Swedish Gallup Institute did not subscribe publicly to the idea that opinion polling was a democratic tool. In 1942, in two long articles meticulously outlining the methods and purposes of the new ‘Gallup polls’ to the readers of the newspapers with a subscription to the polls, Tord Palander, professor of macroeconomics with statistics (Gothenburg University) and ‘scientific controller’ of the Swedish Gallup Institute, argued that it was still too early to determine the political importance of opinion polling. Furthermore, he underlined that these polls were not to be seen as purely scientific. Even though they should of course use the best methods available, opinion polls should primarily be characterised and judged as ‘popular sociology’ or ‘journalistic sociology’, with the creation of interesting journalistic material as their primary purpose.⁴⁶ The Swedish Gallup Institute seems to have endorsed Palander's views regarding this, as the two articles were reprinted as a brochure used by the Institute to market the polls.⁴⁷ By stating that the polls were in fact to be seen as neither direct representations of the voice of the people nor purely scientific studies of it, they could argue that criticism facing the polls was in fact not directed at the polls themselves, but at the ‘possibly exaggerated’ idea that others might have regarding their importance for democracy.⁴⁸

International acceptance of his methods was a core part of Gallup's marketing strategy because it generated the trust in Gallup ‘as a brand’, as noted by Fulda.⁴⁹ The authority derived from the Gallup brand worked both ways, as European pollsters used their connection to Gallup to gain authority in their respective national contexts. NIPO's magazine The Public Opinion (De publieke opinie) called Gallup-style opinion polling the embodiment of the American promise and continuously emphasised the connection to the Gallup-network, as it gave opinion polling an aura of progress and modernity.⁵⁰ Similarly, in Sweden, just the name of the Swedish Gallup Institute (Svenska Gallupinstitutet) – which would effectively have a monopoly on polls in Sweden until the mid-1950s – suggested a strong connection to the American pollster. In reality, as was also the case in the Netherlands, the Institute was an independent enterprise with institutional links mostly to the other Nordic polling companies.⁵¹ Furthermore, when the Institute was founded in 1941, the first two subscribing publications presented it to their readers by repeating the origin myth of modern opinion polling: the story of how George Gallup's clever sampling beat the respected Literary Digest straw poll in predicting the outcome in the American presidential election of 1936.⁵² Seved Apelqvist, editor in chief of Vi, Sweden's most popular magazine at the time, wrote that politicians were in need of an instrument by which ‘the will of the people could be exactly and continuously monitored’, but then stated that such an instrument already existed and it bore the inscription ‘Made in U.S.A’.⁵³ Also in the Netherlands, the magazines of both NIPO and NVO/NSS referred to Gallup, and when NIPO became an official Gallup-Institute in 1947, The Public Opinion printed a similar story to those published in Swedish newspapers in 1941.⁵⁴ This emphasis on both the connections to Gallup and the American origins of polling methods was again intended to associate the polls with progress and democracy, and to make the work of polling organisations attractive to possible customers within the spheres of media and politics.

The degree to which the introduction of polling was met with hopes of democratic reinvigoration seems to have been conditioned by the national contexts. In Sweden, arguments on the democratic potential of polls were posed by some important political actors – most notably the Myrdal couple – in the very early 1940s, in the context of the war and threats against Swedish democracy. However, the pollsters themselves tried to legitimise and, indeed, market their polls as a form of interesting content for their newspaper customers. By the end of the war, the explicitly democratic arguments had become uncommon. In the Netherlands, on the other hand, polling was introduced in the wake of the German occupation, establishing its status as an enterprise with a democratic purpose. Emphasising the democratic potential and the scientific nature of opinion polling was an important part of the legitimisation strategies of Dutch pollsters, while their Swedish counterparts seemed more oriented towards presenting an image of opinion polls as entertainment.

International Transfer and Knowledge Dissemination

Next to hopes of democratic reinvigoration, ideas of the scientific potential of opinion polling travelled across the Atlantic as well. After the war, opinion pollsters in multiple countries believed that social scientific methods would generate seemingly objective and neutral information on society and hence help deal with the chaos that was post-war Europe. Hence, the history of opinion polling mirrors not only a conceptualisation of post-war democracy in favour of bottom-up, popular democracy, but also shows one in favour of a controlled, ‘formal’ democracy based on representative institutions, which make use of opinion polls to predict and consequently shape popular preferences.⁵⁵ Opinion pollsters, active on a global scale, shared a strong belief in the ability of social scientific techniques to create knowledge about society. International organisations such as UNESCO (United Nations Educational, Scientific and Cultural Organisation), established in 1945 as part of the United Nations, also promoted international cooperation regarding social scientific research. This was based on a belief that such knowledge would benefit humankind, solve social questions from the past and hence guard the hard-fought-for peace.⁵⁶ UNESCO's Department of Social Sciences (SSD) functioned as what Wisselgren calls an ‘international boundary organisation’, that mediated and stabilised the relationship between social science and politics in various ways, while also making collaboration between these worlds possible on an international scale.⁵⁷ Next to organising international conferences on the topic, SSD – under the leadership of Alva Myrdal – published a magazine (International Social Science Bulletin), which even had a special issue on ‘Public opinion research’ in 1953.⁵⁸

Influenced by the global enthusiasm for social scientific methods, both Dutch and Swedish pollsters adopted the idea that opinion polls were ‘a useful tool’ that could help elites find out what citizens thought.⁵⁹ In contrast to NIPO, for NSS and NVO a certain distrust of the masses was often at the base of this line of reasoning. Philip Idenburg, NVO board member and director-general of the Central Bureau of Statistics, argued that using opinion polls as a democratising instrument would mean equating democracy with self-government of the masses, who did not have the knowledge to govern themselves. In a parliamentary democracy, Idenburg argued, citizens should elect a representative who can then conduct a political discussion on their behalf.⁶⁰ His fellow board member and professor of law Hendrik Hoetink (University of Amsterdam) agreed with this interpretation of democracy: in his view, the people were too irrational to make decisions on a national scale. Democracy, he wrote, was a form of state for ‘mature and more or less sensible people’ and would not be safe with people ‘who are the playthings of their unconscious feelings’.⁶¹ In a 1946 report, NSS/NVO stated that the opinion of ‘the masses’ was often badly informed: before opinion research could be of use to the government, information or ‘educational propaganda’ was needed.⁶²

More specifically, opinion polling could serve as a thermometer of democracy. The aftermath of the war caused a craving for knowledge about European countries and their attitude towards friend and foe.⁶³ UNESCO commissioned opinion polls on interrelationships and prejudices within and between countries, in order to gain useful knowledge upon which the UN and its members could base their policies.⁶⁴ Following these international initiatives, while also promoting itself to potential clients within the Dutch government, NIPO mapped how post-war democracy was valued among the Dutch public. NIPO believed opinion polls could measure ‘citizen morale’ and regularly polled citizens’ attitudes towards foreign powers and their appreciation of democracy. The idea was that polls could serve as a gauge of unwelcome opinions among the population, to which the government could respond.⁶⁵ For example, NIPO regularly asked whether respondents supported a particular government policy and how respondents felt about certain ministers. The organisation also often published the results of opinion polls held in other countries, on topics such as national public opinion on other nations or politicians.⁶⁶ However, NIPO admitted that it was ‘somewhat ironic’ that leaders of a democratic country needed polls to find out what the people wanted.⁶⁷

Also in the Swedish case, opinion polls served as a useful tool that could assist government elites with knowledge of the people, albeit in a slightly different way. During the 1940s and early 1950s, the Swedish Gallup Institute conducted polls and other surveys for various governmental agencies and commissions. These polls did not primarily concern political opinions but instead focused on phenomena such as the Swedish population's reading habits, alcohol consumption and knowledge of recent reforms of the postal system.⁶⁸ Additionally, some reports of government commissions quoted quantitative data from polls published in the press or elsewhere, on things such as young people's occupational satisfaction, women's participation in the workforce or the population's views on kindergartens.⁶⁹ Wisselgren has studied the government commissions in Sweden as a ‘boundary organisation’ where the relationships and boundaries between the knowledge cultures of science and politics were negotiated.⁷⁰ With increasing use of commercially ordered polls and surveys in governmental commissions, boundaries also had to be established in relation to another knowledge culture: the opinion polling and market research industry. In the early 1950s this boundary was re-negotiated when the state established their own survey organisation – dubbed the ‘state's Gallup’ in several newspapers – as part of the government agency Statistics Sweden and required state agencies and committees to use this instead of commercial survey institutes like the Swedish Gallup Institute.⁷¹

Although the perception that opinion polling used scientific methods played a role in opinion polling gaining credibility with the state, the connections between opinion polling organisations and the field of social science were often fuzzy. From the late 1950s onwards, opinion polling would gradually be incorporated into Dutch social scientific research, especially when a new generation of social scientists set their sights on election research. Inspired by NIPO's work and a visit to the World Conference of the International Political Science Association, in 1956 Jelle Jan de Jong, legal scholar and professor of political science (Free University of Amsterdam), conducted electoral research in the municipality of Nieuwer-Amstel. His enthusiasm was based on remarkably similar arguments as those used by pollsters in the 1940s: he believed that electoral opinion research had the potential to bridge the distance between citizens and the government and correct existing instances of public opinion communicated through protests or the press.⁷² Based on Anglo-American predecessors, the first instances of Dutch election research used the same methods (samples, surveys and interviews) as pollsters, while developing the theoretical frameworks necessary for anchoring it within the discipline of social science. Influential social science theorising on the ‘floating’ voter, developed by academics such as Hans Daudt (University of Amsterdam) and based on collaborations with NIPO, provided the framework for the scientific application of opinion research.⁷³ In the course of the 1960s, election research would mark the breakthrough of opinion research and perpetuate the reputation of such research as scientific. Opinion pollsters and university researchers would work together on explaining voter behaviour based on survey research from 1966, resulting in the yearly National Electoral Research project (Nationaal Kiezersonderzoek, NKO) from 1971 onwards.

Likewise, in Sweden, opinion polling was not connected to the academic social sciences from the start. Although the Swedish Gallup Institute employed a professor of macroeconomics and statistics – Tord Palander – as their ‘scientific advisor’, the connections between the first Swedish pollsters and the early Swedish sociologists who had recently introduced survey research into the academic sphere were scarce.⁷⁴ Much like in the Netherlands, the major Swedish points of contact between polling and science came some years after its introduction, when political scientists started exploring how the material created by commercial pollsters could be utilised to study political processes in new ways. In 1946, a group of political scientists at Stockholm University led by Elis Håstad – professor and member of parliament for the Conservative Party – initiated a seminar series on polls conducted by the Swedish Gallup Institute. In 1950, this series resulted in an edited volume titled ‘Gallup’ and the Swedish Electorate (‘Gallup’ och den svenska väljarkåren), discussing what new ways of studying political life could be offered by the Gallup poll.⁷⁵ This interaction between the commercial pollsters at the Swedish Gallup Institute and the political scientists at Stockholm University – which at the time was a hotbed for new, American quantitative sociology – was pivotal in the institutionalisation of opinion polling as a scientific practice in Sweden.⁷⁶ The work of Jörgen Westerståhl – one of the participants in the project – would be particularly important. In 1951, he went to the United States on a Rockefeller grant to visit several important opinion researchers, and in the following year he became a professor at the University of Gothenburg, where he would lay the foundation for electoral research as an academic field in Sweden.⁷⁷

The press, however, constituted the main arena for Gallup polls. Throughout the 1940s, several of Sweden's biggest papers published weekly syndicated polls on a very wide range of themes and issues, ranging from political opinions to personal habits. Palander described the wide-ranging object of study of these polls as ‘the living conditions, knowledge, views, or actions of a population’.⁷⁸ Published with diagrams illustrated with photographs and cartoonish figures, these polls functioned as a statistical attraction in the press, by which the reading population could see themselves and their countrymen represented quantitatively.⁷⁹ By comparison, in the Netherlands polls were only presented in tables with percentages and more extensive visualisations such as graphs and pie charts did not emerge until the 1950s.⁸⁰ The presentation of polls in the Swedish press could be seen as the popularisation of social scientific methods, and as a scientisation of societal phenomena such as public opinion and political preferences, which were now increasingly understood through the framework of quantitative survey research. Such narratives can also be problematised, however. The Swedish polls were, after all, not defined as science by their creators at the Swedish Gallup Institute, which was a commercial company rather than a research institute. Even if the techniques of sampling used by these pollsters originated in American academic survey research, the roots of this research tradition can be found not only in psychological and sociological attitude research, but just as much in commercial market research and policy oriented social surveys.⁸¹

The history of polling clearly defies common narratives of knowledge dissemination as social scientists in both Sweden and the Netherlands would eventually take inspiration from the commercial practice of opinion polling – rather than the other way around. This complicates the narrative implicated in the scientisation of the social in two ways. Firstly, it seems that opinion polling does not quite conform to the idea of scientisation as the transfer of practices, logics and experts from social sciences to other spheres of society. Rather, opinion polling seems to move back and forth across the boundaries of science, changing shape as it does so. Not only were commercial opinion polling and market research influenced by social scientific methods, but social scientists were also influenced by commercial survey methods, as Jean Converse has highlighted for the American context.⁸² Secondly, the concept of science, especially social science, can be problematised from a historical perspective. The concept of scientisation is underpinned by a specific definition that presupposes that social science is a sphere with a somewhat fixed contour, a clear-cut border over which knowledge can move in a unidirectional manner. However, one might very well question whether the delimitations of science are as fixed as one might think.⁸³ Earlier research on scientisation seems to almost take for granted that some forms of knowledge and knowledge production – such as opinion polling – are essentially scientific. Studying these two different national contexts, however, makes it clear that opinion polling should rather be thought of as a shapeshifter or a hybrid that, on the one hand, mirrors and, on the other hand, constructs the world around it.

Looking at the relation of early Dutch and Swedish opinion polling to social science, it appears that the migration of practices, logics and knowledge across the boundaries of social science is not unidirectional. Rather than a linear transfer of practices and logics from a clearly demarcated (social) scientific sphere into the surrounding society, this points to a more complex entanglement where changing research methods move back and forth over the porous boundaries of an emerging academic field. Moreover, rather than social science being merely popularised through opinion polls in the press, commercial opinion polls originally made for the press made up the empirical material for a new direction within social science, albeit in a specific national context.⁸⁴

Transnational Entanglements: Benefits and Challenges

After the initial transfer from the United States, an extensive international network connected the opinion polling organisations in the Netherlands and Sweden to each other and to pollsters around the world. Sharing a strong enthusiasm for the promise of opinion polling to make sense of post-war societies in a changed world, pollsters shared data and intelligence on possible customers and best practices. Furthermore, they cooperated on conducting international opinion polls. This section will pay attention to transfer by both interpersonal contacts, such as the correspondence between and conferences attended by Gallup licensed pollsters, as well as diffusion through newspapers and magazines.

As discussed above, Gallup and the AIPO worked in the late 1930s to establish an international network of polling institutes. After being hampered by the Second World War, this project resumed as the war ended. From 1944 and onwards, Gallup's ‘manual’, A Guide to Public Opinion Polls (1944), was distributed across Europe.⁸⁵ Already in May 1945, Gallup started coordinating questions to be asked by institutes across many countries, the first one asking people for their opinions on the ongoing San Francisco conference, on which of the Allied powers would have the most influence in the post-war world and on whether Hitler actually was dead.⁸⁶ In 1946, the AIPO entered into an agreement with nine other public opinion institutes (in Australia, Brazil, Canada, Denmark, the United Kingdom, Finland, the Netherlands, Norway and Sweden) to cooperate and conduct international comparative opinion surveys. Through correspondence, international publications and conferences, the organisations kept each other informed about the latest research and techniques.⁸⁷ In 1948, the European Society for Opinion and Market Research (ESOMAR) was founded in Amsterdam, under the auspices of NIPO.⁸⁸ In 1953, the World Association of Public Opinion Research (WAPOR), founded in 1947, was even attached to UNESCO as a nongovernment consultant organisation. This infrastructure provided opportunities for opinion pollsters from all around the world to discuss new methods, share data sets and compare national results.⁸⁹ In 1953, Gallup himself emphasised that the development of opinion polling happened in constant interaction between the United States and Europe.⁹⁰ In the Netherlands, for example, the magazines distributed by NIPO and NVO regularly reported on the state of the art in opinion research in other countries. The magazines debunked criticism of opinion research and reprinted international articles on the topic, such as pieces from the American magazine The Public Opinion Quarterly.⁹¹ Dutch newspapers such as Het Parool, an early customer of NIPO, also reported extensively on both the outcomes as well as the methods of international opinion research.⁹²

Opinion research mirrors the quest of the Western European nations for a post-war narrative of unity and democracy as, from the late 1940s onwards, pollsters in different countries joined forces. Pollsters started to coordinate international surveys, for example in 1947 on the opinion of American, Canadian, English, French and Dutch citizens on ‘the German people’, in 1948 on the partition of Palestine and in 1949 on the desirability of permanent employment.⁹³ The Gallup associates also organised international conferences and at the first conference in 1947 at Loxwood Hall, Sussex, it was decided that all affiliates would ask the same questions to their respondents at least once a month.⁹⁴ European pollsters additionally met regularly at conferences which focused on a particular strand of opinion research: for example, radio and newspaper research was the topic of the conference arranged by the Scandinavian Institutes of Public Opinion in Stockholm in 1947, to which British and Dutch colleagues were also invited.⁹⁵ The Gallup affiliates also conducted research requested by UNESCO in 1948, in which the opinions on other countries were compared: allegedly, the Dutch were seen by other nations as ‘peaceful, hard-working and sensible’ and the Americans as ‘practical and progressive’, whereas Russians were seen as ‘cruel and despotic’.⁹⁶ NIPO followed up on this topic in 1950 and concluded that as sympathy for Germans was rising, sympathy for Russians was waning. According to this research, the United States remained the ‘favourite’ of Dutch respondents.⁹⁷ Another recurring theme in Dutch opinion polls was that of the American and the Western European relationship, with polls on topics such as the people's opinions on the Marshall Plan and their definitions of democracy.⁹⁸

The national opinion polling organisations working together within the framework of the Gallup International Research Institutes had shared interests in maintaining the good reputation of opinion polling, which served as an impetus for sharing best (and worst) practices.⁹⁹ Therefore, national opinion polling organisations were in continuous correspondence with each other. Besides the formal international networks centred on the AIPO, it is clear that the individual polling institutes sought contact with each other in a more decentralised way. For the Swedish pollsters, the most immediate partners for international cooperation were found in the Nordic Region. The four polling institutes in this region performed transnational polls and there were institutional connections as well, as they had a ‘Central Co-operative Office’ in Copenhagen, led by Asmussen.¹⁰⁰ Beyond the Nordic region, there was also substantial correspondence with American, Australian and Dutch polling colleagues.

The personal relations between the directors of the various polling institutes constituted an important part of the international polling networks. For example, Sven O. Blomquist was on a first name – or nickname – basis with most of the international polling executives. In correspondence with George Gallup, for example, Blomquist generally addresses him either with his nickname Ted or as ‘Papa’, the latter presumably being a joke based upon Gallup being seen as the father of opinion polling.¹⁰¹ There is more evidence of personal friendship between some of the Nordic Gallup executives, such as Blomquist inviting Artturi Raula, director of the Finnish Gallup Institute, to stay in his home during a conference in Stockholm or Blomquist sending Asmussen an amicable Christmas poem by telegram.¹⁰² The Swedes and Dutch were also on first-name basis: for instance, NIPO's De Jonge started his letters to Blomquist with ‘Dear friend’ or ‘Dear Sven’, and Stapel invited his Swedish colleagues to visit his new house in Aerdenhout.¹⁰³ Blomquist also addressed his Dutch colleagues with ‘Dear friend’ or ‘Dear Sirs, Jan and Wim’.¹⁰⁴ United in their enthusiasm and aware of their collaboration being necessary for perpetuating the reputation of opinion polling, these pollsters nourished the international network set up by Gallup. This personal correspondence also served as a vehicle for sharing data, as well as intelligence on possible customers and best practices. In 1949, Stapel was only too happy to share his experiences after a trip to Allied polling institutes in Germany with his international Gallup colleagues, under the telling title ‘Confidential gossip about polling in Germany’.¹⁰⁵ Stapel also referred Dutch clients, such as Philips, to his foreign colleagues, and Blomquist in turn warned him about potentially problematic clients, such as Standard Oil.¹⁰⁶ The correspondence was furthermore used for sharing and developing polling methods. After the Dutch elections in 1948, for example, de Jonge made sure to send to Blomquist the data on which they had built their successful election predictions.¹⁰⁷ In the following year, de Jonge asked Blomquist for some of the unweighted survey results from the Swedish Gallup Institute's prediction before the Swedish 1944 general election, as he wanted to use these data when developing his Amsterdam sample before the Dutch municipal elections in July 1949.¹⁰⁸

In its early years, the reputation of opinion polling was also vulnerable, not only because of its attachment to the individual of George Gallup, but also because of its international network.¹⁰⁹ Failed attempts by pollsters worldwide, be they connected to Gallup or not, could affect the reputation of opinion research as a whole. A telling example of the pitfalls of the international character of polling was Gallup's inability to correctly predict the result of the 1948 American elections. The national election was ‘a referendum on polling’ and criticism arose from within the social scientific field. Because the pollsters used social scientific methods such as statistics, it was believed that ‘if the pollsters took a fall, social science would fall with them’. Nevertheless, the next US elections would see an even greater amount of opinion polls.¹¹⁰ The 1948 incident had a negative impact on the reputation of polling in multiple European countries. The Dutch press reported extensively when Gallup wrongly designated candidate Thomas Dewey as the new American president.¹¹¹ Moreover, Gallup's blunder plus NIPO's own inaccurate predictions in 1946 and the summer of 1948 caused a surge of criticism in the printed press.¹¹² Prior to the elections of 1952, the Dutch newspapers often accompanied the results of NIPO's electoral predictions by mentioning George Gallup's ‘blunder’.¹¹³ When NIPO then also failed to predict the outcome of the 1952 elections, the De Telegraaf newspaper argued that such ‘suggestive predictions were better left out’ from now on.¹¹⁴ As a result, no more electoral predictions were organised prior to the 1956, 1959 and 1963 elections and, despite newspapers printing results of other opinion polls in 1956, the discourse on opinion polling was mostly negative.¹¹⁵ It was not until 1967 that NIPO again conducted electoral predictions on a large scale.

Likewise, the 1948 polling fiasco hurt the legitimacy of the Swedish Gallup Institute.¹¹⁶ To make matters worse, their predictions ahead of the Swedish general elections two months earlier had also been off, underestimating the result of the Social Democrats while overestimating the Conservative Party. This was useful material for some Social Democrats who were already suspicious of the connections between the Swedish Gallup Institute and anti-socialist business interests.¹¹⁷ This has to be seen in the context of ‘the planned economy debate’ (planhushållningsdebatten), a fierce political conflict in post-war Sweden over the degree of state control in the economy, which transformed the election of 1948 into one of the most contentious in Swedish political history.¹¹⁸ Ahead of the local elections in 1950, the Swedish Gallup Institute once again underestimated the Social Democrats, while overestimating the liberal People's Party.¹¹⁹ In the following years, the Institute met this series of setbacks by slowly reorienting their activities away from interview surveys and towards market research by the means of stock taking.¹²⁰ By 1956, the Swedish Gallup Institute had completely ceased to perform opinion polls and was acquired by and merged with A.C. Nielsen, an American company focused precisely on stock taking. Their role as the main opinion polling institute had then been taken over by the Swedish Institute for Opinion Polls (Svenska institutet för opinionsundersökningar; SIFO), founded in 1954 by Sten Hultgren, a former employee of the Swedish Gallup Institute.¹²¹ In striking similarity to the Dutch case, the failures of 1948 and 1950 resulted in a long period in which no election predictions were published. It was only in 1964 that election predictions were once again made public, this time by SIFO. Despite the prevalence of Swedish pre-election polls in the 1940s, the 1964 polls were presented as something completely new, ignoring rather than discussing the Gallup polls of the 1940s and the failures of 1948. Perhaps this rhetoric of novelty was bolstered by the fact that the results were publicised through the relatively new medium of television, rather than just through newspapers.¹²²

Beyond the different conditions of countries, the entanglement of these national contexts through international networks must be taken into account. Earlier histories of opinion polling generally focus on the transfer of polling technologies from the United States to other countries. However, sources clearly show that wider international networks were soon established that worked more or less independently of American influence. Through these networks pollsters exchanged information, methods and contacts with potential customers. The connections to the American origins of polling were at times emphasised to benefit from the equation of Americanness with democracy and modernity. However, such connections could also become a liability, as is evident after the failures of 1948 and the ensuing blow to the legitimacy of polling in both Sweden and the Netherlands. Despite these challenges and the ensuing slump of the 1950s, opinion polling was here to stay. As the media, politics and the social sciences embraced the practice of opinion polling, over the course of the 1960s people all over the Western world would be presented with opinion polls on a plethora of topics on a daily basis.

Conclusion

The comparison of the introduction of opinion polling in Sweden and the Netherlands shows that the different national political contexts and, in particular, experiences of the Second World War had a profound impact on how opinion polling was introduced and how it was discursively tied to ideas about democracy. The Dutch experience of German occupation and the subsequent need to reestablish and rejuvenate democracy created an idea of opinion polling as tied to contemporary demands for more direct democratic participation. Similar arguments had been seen in Sweden in 1941 and 1942 – when democracy was severely threatened on a global scale – but never from the pollsters themselves, whose focus seemed more commercial than political. After the war, when Swedish democracy seemed secured, such arguments were uncommon. Dutch pollsters stressed the scientific and democratic nature of their work, while their Swedish counterparts more often portrayed their polls as merely journalistic in their scope, offering readers popular sociological content.

This article has highlighted the entanglement of this process across two different types of boundaries: national borders and the boundaries of social science. Regarding the national borders, opinion polls mirror not only post-war transnational interconnectedness in the Western world through brand new international networks such as UNESCO and WAPOR, but also a shared desire among pollsters to make sense of post-war society. Although the American origins of opinion polling were often rhetorically emphasised, it is evident that this practice developed through more decentralised transnational networks than one might assume. Pollsters in the Netherlands and Sweden, among other countries, frequently shared methods, customer contacts and other information with one another, often independently from the perceived centre of opinion polling in the United States.

Beyond adding a more complex transnational perspective to the historiography of opinion polling – often signified by a focus on national contexts and unidirectional transfers between them – the framework of entanglement can also provide new insights on the dynamics of scientisation across the boundaries of science. Studying these two different national contexts makes it clear that opinion polling should rather be thought of as a shapeshifter or a hybrid practice: it is not only used by a variety of actors, from social scientists to journalists and politicians, but is in itself a complex product of both commercial, political and scientific endeavour. The transnational entanglement between commercial polling and social science not only scientised conceptions of public opinion and popularised survey research methods. It is also true that the practices of social science in post-war Europe were reconfigured through the introduction of methods of opinion polling, developed within a framework of constant interaction between the spheres of science and society.