Abstract
The article investigates how the poor and welfare recipients are depicted in British, Danish and Swedish newspapers. The study was inspired by American media studies that have documented a negative stereotypic way of portraying the poor and welfare recipients, especially when they are African Americans. The article argues that there is an institutional welfare regime logic behind the way the poor and welfare recipients are depicted in the mass media. It is not only a matter of race. This argument is substantiated by showing that the poor and welfare recipients are (1) also depicted negatively in a liberal welfare regime, the UK, where most of the poor and welfare recipients are perceived to be white, and (2) depicted positively in two social-democratic welfare regimes, Sweden and Denmark, where the poor and welfare recipients have increasingly come to be perceived as non-white, especially in Denmark. The empirical analyses are based on a sample of 1750 British, 1750 Danish and 1750 Swedish newspapers covering the period from 2004 to 2009.
Introduction
Why public support for anti-poverty policies is so weak in liberal welfare regimes and so strong in social-democratic regimes is a puzzle that has occupied a number of scholars (e.g. Albrekt Larsen, 2006; Alesina and Glaeser, 2004; Svallfors, 2003). The comparative welfare-state literature suggests that there is an institutional logic behind the cross-national differences in public support for anti-poverty policies. These institutional effects have been illustrated by survey studies in different countries. For example, Albrekt Larsen (2006, 2008) shows that those living in liberal welfare regimes are much more likely to explain poverty by laziness than those living in social-democratic welfare regime (this is the link between 1 and 3 in Figure 1). He explains this result with the way the selectivism, the low generosity of benefits and the labour market of the liberal welfare regime fuel harsh deservingness discussions and how the opposite is the case in social-democratic regimes. The most central place for such deservingness discussions is the mass media. Several studies have documented that mass media content influence public opinion (i.e. the relationship between 2 and 3 in Figure 1; e.g. Zaller, 1992), but little has been done to document that the discussions in the mass media of liberal and social-democratic regimes do in fact vary from each other (the relationship between 1 and 2 in Figure 1). This is what this study intends to document: we analyse how British, Swedish and Danish newspapers describe and depict the poor and welfare recipients.
Suggested causal relationship between welfare regimes and welfare attitudes.
Theory
The stories and pictures in the mass media form an important basis for creating opinions toward recipients of welfare schemes. This is a fairly uncontroversial claim. The media content is likely to influence who we (politicians, journalists, social scientists and ordinary citizens) think the poor and welfare recipients are. This is the process of categorization and labelling described by Allport (1954) whereby we create images and stereotypes. These images and stereotypes are crucial for the moral judgements about whether the recipients are deserving or not. The American journalist Walter Lippmann (1889–1979) is said to be the first to use the term stereotype in its current meaning of a metaphor for ‘pictures in our heads’, arguing in the introduction to the 1961 edition that ‘whether right or wrong … imagination is shaped by the pictures seen … . Consequently, they lead to stereotypes that are hard to shake’ (1961 [1922]: 2). We will focus on stereotypes about ‘the poor’ and the receivers of means-tested social assistance, as it is within this field that previous studies find a clear regime pattern in public attitudes. Following the American research tradition (see below) we will have a focus on both text and visual elements. The central theoretical question is how welfare regimes, that is, the institutional configurations between state, market and family described by Esping-Andersen (1990, 2000), influence the media content. Our main argument is that the institutional structure of welfare regimes generates very different inputs to the mass media.
In liberal welfare regimes, the configuration of state, market and family structures is likely to generate a ‘negative’ media input. It is a classic thesis that systems dominated by universal benefits and services (the ideal-type policy of the social-democratic welfare regime) on the one hand, and systems dominated by selective benefits and services (the ideal-type policy of the liberal welfare regime) on the other, generate quite different public discussions and perceptions of recipients (e.g. Albrekt Larsen, 2006, 2008; Crepaz and Damron, 2009; Rothstein, 1998; Titmuss, 1974). Following Rothstein (1998: 158), ‘the public discussion of social policy in a selective system often becomes a question of what the well-adjusted majority should do about the less well-adjusted, in varying degrees, socially marginalized minority’. The general fairness of the policy is also open to challenge, as the majority might start asking ‘a) where the line between the needy and the non-needy should be drawn; and b), whether the needy themselves are not to blame for their predicament’ (1998: 159). It is also a classic thesis that selective policies increase the incentive to abuse benefits and services (by wrongly claiming to be among the entitled). These abuse stories are made even more newsworthy by the clear divide between ‘those’ who receive and ‘those’ who finance welfare benefits. It is not only a matter of the targeted nature of the schemes. It is also a matter of the unregulated labour markets generating poverty, which the residual welfare schemes cannot alleviate. The poverty and inequality found in liberal regimes increases the amount of social problems such as crime, teenage pregnancy and underclass culture (e.g. Wilkinson and Picket, 2010). Furthermore, the social problems in liberal regimes are especially concentrated in distinct neighbourhoods (in the US context often carrying the ‘ghetto label’) and among single mothers. This combination of selective benefits, neo-liberal labour markets and concentration of social problems produce a social reality that delivers a constant flow of negative newsworthy stories about ‘the deviant poor’. This is the perfect environment for building negative stereotypes about ‘the poor’. The only moderator on public opinion is the fact that ‘the poor’ are indeed very poor (by comparative Western standards).
In the social democratic welfare regimes, the configuration of state, market and family structures is likely to produce a ‘positive’ media input. In a universal system no line needs to be drawn between the needy and the non-needy. As Rothstein (1998) argues, welfare policy does not, therefore, turn into a question of what should be done about ‘the poor’ and ‘the maladjusted’, but rather a question of what constitutes general fairness with regard to the relation between citizens and the state. Universalism also reduces the incentives to wrongly claim benefits and services. As everybody in social-democratic regimes is entitled to many of the welfare services and benefits, this institutional setup simply produces less abuse and fewer abuse stories. Again, however, it is not only a matter of universalism. It is also a matter of regulated labour markets, which by comparative standards produce high equality and little poverty. Furthermore, the social-democratic regimes have, through housing policy and generous benefits, avoided the concentration of social problems that can be found in the liberal regimes (Albrekt Larsen, 2013: Chapter 3). This combination of universal benefits, regulated labour market and absence of concentration of social problems produce a social reality that does not deliver a constant flow of negative stories about ‘the poor’. This is the perfect environment for building positive stereotypes about ‘the poor’. In this context, the most imminent threat to a ‘positive’ public opinion is the fact that ‘the poor’ are actually not very poor by comparative standards.
Comparing media content in different welfare regimes
Comparative studies of media content across countries are rare and to our knowledge no previous study has ever embarked on testing the regime argument discussed above. Even national media studies of how the poor and welfare recipients are depicted are rare. The exception is the US where one finds a research tradition in this area; most notable are the studies by Martin Gilens (1996, 2000; see also Iyengar, 1991). Gilens had a strong focus on the visual elements and documented that African Americans were overrepresented in pictures in articles about poverty. African Americans were often also shown in age categories and in connection with topics that made them seem less deserving than non-African Americans. American survey studies have also consistently shown a strong relationship between negative attitudes towards black people and the public’s lack of support for welfare policies (Alesina et al., 2001; Gilens, 1995, 1996, 2000; Lind, 2007). This focus on the racial logic behind the negative representation of the poor and welfare recipients is symptomatic for the American studies in this field. Given the American race history this is understandable, but it is also a matter of the empirical scope of the American studies. As only the US is included in them, it has been impossible to discover the institutional logic described above. To put it boldly, the regime argument suggests that the negative stereotyping found in the US might more be a matter of African Americans living a liberal welfare regime than a matter of African Americans being black.
This article investigates the possibility of regime effects by replicating Gilens’ analyses in another liberal welfare regime (UK) and two social-democratic regimes (Sweden and Denmark). If welfare regime type is the crucial determinant of media content on the poor and welfare recipients, we would expect to find harsh deservingness discussions (like those in the US) in the British media but not in Danish and Swedish media. The UK provides a good case for studying the ‘isolated’ effect of a liberal welfare regime as it is much ‘whiter’ (according to the 2001 census whites make up 92.8 percent of the population) and has not had the same level of racial tension as the US. As described below, most of the British ‘media poor’ are indeed of white skin colour. Despite this ‘whiteness’, a previous study leads us to believe that the British media poor are presented negatively. Golding and Middleton’s (1982) content analysis of all British newspapers in the second half of 1976 found that of all articles about ‘welfare’ (defined as stories dealing with aspects of welfare linked to the responsibilities of the Secretary of State for Social Services), about 31 percent dealt with abuse (Golding and Middleton, 1982: 68). As abuse was the most frequent topic, they argued that 1976 was a ‘scroungerphobia period’ in which the British welfare consensus was broken. Though time series are not available, it might be the case that ‘scroungerphobia’ increased in the mid-1970s in the UK. However, our findings below indicate that the abuse stories are still frequent in the UK, which supports the argument that their presence is systemic for liberal welfare regimes.
The depiction of the poor and welfare recipients in social-democratic welfare regimes has not been investigated systematically previously. Our theoretical argument and previous survey results (see above) lead us to expect positive stories and pictures of the poor and welfare recipients in Sweden and Denmark (see introduction). Another possibility is that the mass media content has changed due to increased ethnic heterogeneity (e.g. Alesina and Glaeser, 2004). The number of non-Western immigrants has indeed increased dramatically in Sweden and Denmark in the last three decades, which has made the poor and welfare recipients much more ‘coloured’ (see below). Recent survey studies have also documented that Swedes and Danes perceive the out-group of non-Western immigrants as negatively as non-black Americans perceive African Americans (Albrekt Larsen, 2011). Thus, it would support the regime argument if – despite the salience of the ethnic logic – one could still find ‘soft’ media content in Denmark and Sweden.
Data material
The collection of British, Swedish and Danish media content was designed to match data collections conducted in the US. Gilens studied the pictures and stories in three leading weekly news magazines in the 5-year period from 1988 to 1992 (Gilens, 1996) – Time, Newsweek and U.S. News and World Report. Clawson and Trice (2000) conducted a follow-up study covering the 5-year period from 1993 to 1998. Besides Gilens’ three magazines, Clawson and Trice included the stories and pictures in Business Week and the New York Times Magazine. Finally, Gilens in his 2000 book expanded the 1996 analysis with a rough coding of pictures from 1950 to 1992. Because news magazines are not as common in Europe as in the US, we sampled newspaper stories on poverty and welfare recipients that included pictures of the ‘target group’ (see below). In each of the three countries, we selected five of the largest nationwide newspapers (local and regional papers were omitted). Papers available for free were excluded because many are local and their archives are scattered.
The study balances the newspaper selection in each country. This was done by including three of the largest newspapers from what can be labelled ‘semi-serious’ or ‘serious popular’ press and two of the largest from what can be labelled the ‘newsstand tabloid’ press (Sparks, 2000). We use the terms ‘broadsheet’ and ‘tabloid’ for these two groups. The classification was made by reading national media researchers’ descriptions of the content of the different newspapers. The media study did not include what Sparks labels the ‘serious’ press because its readership is more modest than that of the other papers and because these papers have a strong focus on business (except for the Danish ‘Information’), by which we assumed that they contain few portraits of poverty. Nor did the study include the ‘supermarket tabloid’ press papers because they feature so little concentration on politics, economics and society that we also assumed that they contained very few portraits of poverty.
The selected Danish broadsheet newspapers were Jyllandsposten, Politiken and Berlingske tidende. The Danish tabloid newspapers selected were BT and Ekstrabladet. These are also the five biggest newspapers in Denmark (Dansk Oplagskontrol, 2009). The selected Swedish broadsheet newspapers were Dagens nyheter, Svenska Dagbladet and Göteborgs-Posten, and the tabloid newspapers, Aftonbladet and Expressen. These are also the five biggest newspapers in Sweden (Tidningsutgivarna, 2009). In the UK the broadsheet newspapers selected were The Times, Daily Telegraph and the Daily Express, and the selected tabloid newspapers, Daily Mirror and Daily Star (Sunday edition included). The Daily Mail and the Sun are larger than the Daily Express and Daily Star but the latter two were chosen because they have better archives. The Daily Express and Daily Star are still among the top six largest UK newspapers (Audit Bureau of Circulations, 2009).
The media study operated over the 5-year period between 2004 and 2009. The fairly long time period was chosen because it enabled us to make comparison with the studies by Gilens (1996) and Clawson and Trice (2000), and enabled us to catch general country effects. As European newspapers are published daily (in contrast to American magazines that are published weekly), a complete search in all five newspapers (see procedure below) over a 5-year period would result in a very large amount of relevant articles, so that instead of a complete search we sampled random days (stratified on year and weekday). In practice we sampled one random week at a time, meaning that we drew a random Monday, a random Tuesday, and so on, until we had a constructed a week. Ten such constructed weeks were sampled within each of the 5 years: in each country we sampled 50 constructed weeks, which added up to 350 days and 1750 newspapers for each country (see technical report, Dejgaard and Albrekt Larsen, 2011, for further information about the selected newspapers, the actual dates sampled, picture archives used, and so on).
Selection of articles and pictures and the coding
Gilens (1996) and Clawson and Trice (2000) used an index (the Reader’s Guide to Periodical Literature) to locate the relevant articles. They selected the core categories of ‘poor’, ‘poverty’ and ‘public welfare’. In the US, ‘public welfare’ primarily referred to the former Aid to Families with Dependent Children (AFDC) program (the current Temporary Assistance for Needy Families (TANF) program), food stamps, Medicaid and General Assistance: all targeted, means-tested benefits. For Britain, Sweden and Denmark we could not find a reliable index to locate the relevant articles; we therefore conducted a full text search for Britain, Sweden and Denmark. We searched for the words ‘poor*’ (i.e. poorest, poorer also included), ‘poverty*’ and the dominant social assistance scheme(s) in the country (the term ‘welfare’ cannot be used as it has a much broader meaning in Europe than in the US (see technical report for the full search string).
The many articles from this full text search (from the sampled dates) were sorted manually. Articles that dealt with poor, poverty and social assistance clients in the relevant country were selected. Despite differences in the methods of data collection, we ended up with articles that largely covered the same topics as in Gilens’ articles (see below and Gilens, 1996: 525). Among the relevant articles, we selected those that were accompanied by a picture or drawings of relevant persons. A relevant person was defined as a person that potentially is, has been or might become poor. Pictures of authors, commentators, researchers, caseworkers, and so on, were not perceived as relevant persons. The term ‘potentially’ was used because often it could not be assumed that the person depicted was poor. If an article about child poverty was accompanied by a picture of a child outside an apartment complex, the reader of the newspaper might perceive the child as poor, but naturally one cannot be sure (except in cases where information was provided in the text). The term ‘has been poor’ was used because pictures of previously poor people and citizens previously receiving social assistance were also included. In most cases, it was easy to see whether the picture contained relevant persons or not. Relevant foreground persons as well as relevant background persons were coded (see the technical report for details on the picture archives used).
Finally, the pictures and stories were coded. The people in the pictures were coded for gender, age, skin colour and work status. Below we only make use of the skin colour coding, for which Gilens reported an intercoder reliability of 0.97. With an expansion of the categories (see below) and an inexperienced coder, we reached an intercoder reliability of 0.89, which was still acceptable. Furthermore, the article uses a topic coding, which is discussed below.
The presence of the poor and welfare recipients in the mass media
The material was primarily used to explore qualitative differences in how the poor and welfare recipients are depicted, but it also allowed us to establish a rough measure of the ‘media-salience’ of the poverty and welfare issues. In the 780 magazines studied by Gilens (three magazines over 52 weeks for 5 years, 1988–1992), he found 182 stories accompanied by 206 pictures showing 560 poor people. These persons will be labelled the ‘media-poor’. Thus for every 100 issues, on average, there were 23 stories related to poverty and 26 pictures (see Table 1). Using the same procedure but two more journals, Glawson and Trice found a less intense coverage in the period from 1993 to 1998, and the difference was not caused by Business Week and the New York Times Magazine showing fewer pictures than Newsweek, Time and U.S. News and World Report. Of the 1300 issues they studied (five magazines over 5 years, multiplied by 52 weeks), they found 74 stories accompanied by 149 pictures showing 357 media-poor subjects. Thus in this period there were on average six stories and 11 pictures per 100 magazines. Apparently the media coverage prior (1988–1992) to the welfare reform of 1994 was greater than the media coverage during and after the reform period (1993–1998).
The presence of the poor and welfare recipients in mass media in the US, UK, Sweden and Denmark.
Because the American studies used magazines and articles selected by means of an index, it was difficult to compare directly these intensity measures with our results. However, in the 1750 selected British newspapers we found 188 stories related to poverty, each accompanied by one or more relevant pictures. The 257 pictures depicted 545 media-poor people. Thus, in the UK there were on average 11 poverty-related stories and 15 pictures of media-poor people per 100 newspapers. Whether this was more or less than occurs in the US is difficult to judge, but British figures can be compared with the Danish and Swedish ones as we used the same selection method in these countries.
As one might expect from the low poverty levels found in Scandinavia, the Swedish and Danish newspapers contained fewer pictures of poor persons than did the British. In the 1750 Swedish newspapers we found only 73 stories with 102 pictures that depicted 180 persons. Thus, the British papers contained 2.5 times more pictures of poor persons than did the Swedish newspapers. So the British newspaper readership is much more exposed to pictures of the poor than is the Swedish. This is what one might expect. The Danish case is more challenging. In the selected 1750 Danish newspapers we found 152 stories accompanied by 190 pictures that depicted 376 poor persons. Thus, the Danish newspapers contained 1.9 times more pictures of the poor than did the Swedish newspapers. The figures reflect that, despite low poverty levels, Denmark had a rather intense discussion about poverty in the period between 2004 and 2009. The 190 Danish pictures could not match the 257 pictures found in UK, but the Danish figures indicate that the lack of resistance to spending on anti-poverty measures in Sweden and Denmark cannot simply be explained by Swedes and Danes not seeing pictures of the poor and social assistance recipients. It is also a matter of how these groups are described and portrayed.
The colour of the media-poor
The most prominent finding in Gilens’ study was that African Americans were strongly overrepresented in the media pictures. Black people made up 62 percent of the media-poor, while the official share was 29 percent. Clawson and Trice also found African Americans to be overrepresented in magazine pictures from 1993 to 1998. In this period, the share was 49 percent compared with the census estimate of 27 percent. As discussed in the Introduction, the perception of the poor and welfare recipients as belonging to an ethnic minority might matter quite a lot in deservingness judgements.
To judge by the media pictures, the American scholars are right in arguing that the racialization of the poverty issue is extreme in the US. None of the other three countries come close to portraying as many black persons as was the case in the US. Black people made up 9 percent of the British media-poor, 5 percent of the Swedish and 3 percent of the Danish (see Table 2). Thus, if one believes that the negative stereotypes are directly associated with those of African origin, then the poor are presented as somewhat less deserving in the UK than in Sweden and Denmark. However, in a European context, it is more a matter of white versus non-white. It is a matter of the white natives and the non-white immigrants. Therefore we not only coded ‘black’ and ‘non-black’, but also distinguished between ‘white’, ‘black’ and ‘other non-white’ subjects. Clawson and Trice (2000) also expanded Gilens’ colour coding. They found that the white people made up only 34 percent of the American media-poor. In the British media pictures, the ‘other non-white’ made up 5 percent, that is, in total 14 percent of the British media-poor were non-white. Thus, the vast majority, 86 percent, of the British media-poor were white, so that, as argued in the Introduction, if public attitudes are dominated by the ethnic logic, the British poor and welfare recipients should be depicted and perceived as rather deserving.
The skin colour of the media poor in UK, Sweden and Denmark (2004–2009) and the US (1988–1992).
The Swedish and Danish media-poor are also less ‘non-white’ than those in the US. They are, however, more ‘non-white’ than in the UK. The ‘other non-white’ people made up 20 percent of the Swedish media-poor and 30 percent of the Danish, that is, the non-white people made up 25 percent of the media-poor in Sweden and 33 percent in Denmark. That the Danish media-poor are ‘more non-white’ than the Swedish (the difference is significant) was expected, due to the greater salience of this issue in Denmark. In the European context it is difficult to find poverty statistics broken down by skin-colour, but in Sweden and Denmark non-white persons are probably somewhat overrepresented. In Denmark non-Western immigrants made up about 12 percent of the relatively poor in 2007 (calculation based on AE, 2010). However, if one uses instead the yardstick of how large a share non-white people make up of social assistance clients, then the distribution among the media-poor is almost in accordance with the Danish estimates from 2008 (DA, 2008). In any case, the findings indicate that if the ethnic logic is the dominant driver behind the representation of the poor and welfare recipients, then we should expect the media content to be harshest in the US and softest in the UK. When we look at the character of the stories, however, the UK media content is by no means ‘soft’.
Categorizing the stories about the media-poor
Table 3 shows the main topics grouped into two main categories, ‘positive’ and ‘negative’, and two subcategories, ‘out-of-poverty stories’ and ‘miscellaneous others’. We started with describing the positive stories. Child poverty is one of the topics that typically puts the depicted media-poor in a positive light. For Britain, we found 12 such stories (6 percent). For Sweden and Denmark, we found six stories in each country (8 percent and 6 percent, respectively). Another positive topic is ‘Christmas help’, an especially Danish phenomenon. Ten Danish stories (7 percent) described how needy families were helped during the Christmas period. For Sweden we found one such story, and for Britain, we did not find any. Pension poverty is also a positive topic, more frequent in the UK than in Sweden and Denmark. We found 19 stories about old-age poverty in the UK (10 percent), but only three stories in Sweden and one story in Denmark (4 and 1 percent, respectively). Old-age poverty was the most frequent ‘positive’ story in the UK. Stories about ‘bad labour-market conditions’, stories about ‘economic need and social stratification in general’ and stories about ‘housing prices and homelessness’ also tend to leave a positive impression of the media-poor. On these topics there were no clear country variations; together they made up 14 percent of the British stories, 29 percent of the Swedish and 14 percent of the Danish.
Topic of articles in Denmark, Sweden and UK.
What we have labelled ‘stuck-in-the-system’ stories also presented the media-poor in a positive light. These stories seem to be a particularly Scandinavian phenomenon. The typical framing is that a person cannot be helped because of red tape or problems with the welfare system. Thus, the blame is put on the system and the motivation of the citizen is not questioned. Together with general stories about economic need, this is the most frequent positive story found in Sweden and Denmark. We found nine of these stories in the Swedish newspapers (12 percent) and 16 of these stories in the Danish newspapers (11 percent). In the UK newspapers, we only found two such stories (1 percent). There is also a Scandinavian tendency to write ‘positive’ stories about poor immigrants. The deservingness literature suggests that just depicting immigrants hardens attitudes towards the poor and welfare recipients; however, this effect might be of less importance where the attached story deals with structural problems (typically discrimination) that cause economic hardship for immigrants. In the Danish newspapers we found 12 such stories (8 percent), in the Swedish papers three such stories (4 percent) and in UK newspapers none. Single-mother poverty is another topic that can have ‘positive’ and ‘negative’ framing: a number of the stories present single mothers as abusing the system (see below). However, particularly in the Danish and Swedish newspapers, we did not find stories on the topic of abuse. Instead, the main topic is the economic hardship experienced by this group, which by comparative standards is actually very well protected in Sweden and Denmark. We found nine such stories in the Danish newspapers (6 percent), four in the Swedish (4 percent) and four in the UK (2 percent). Finally, stories about public policy initiatives that ease the living conditions of ‘the poor’ also tend to leave the reader with a positive image of the target group. Descriptions of such policies were most frequent in the UK. We found 11 such stories for the UK (6 percent), three for Denmark (2 percent) and zero for Sweden. It is open to discussion whether students are a deserving or undeserving group. However, the few stories we found (three for the UK, two for Sweden and four for Denmark) were generally positively framed. Therefore they were added to the ‘positive’ bracket.
Categorized this way, the overall finding is that 41 percent of the British stories depict the media-poor in a ‘positive’ light. As expected, the share is larger for Sweden and Denmark. For Sweden, positive stories make up 62 percent, for Denmark, 55 percent. The difference between the UK and the two social-democratic regimes is clearly statistically significant (0.01 level). 1 Thus as expected, the Swedish and Danish media-poor are framed in a more positive way. Whether the size of this difference is larger or smaller than expected is difficult to judge. On the one hand the studies that claim that the discourse of liberal regimes is particularly ‘harsh’ (see the Introduction) would lead one to expect larger differences. What keeps the British stories somewhat positive is the overrepresentation (compared with Scandinavia) of the issue of old-age poverty and anti-poverty policies that ease the living conditions of those at the bottom of society. On the other hand, one could argue that, as the living conditions of the poor and welfare recipients in the UK worsen, it is quite telling that the British media still publish fewer ‘positive’ stories than do the Scandinavian. However, in any case it is not the ‘positive’ topics that cause a public uprising. It is the negative stories that have the potential to cause what Stanley Cohen (1972) labelled ‘moral panic’.
The negative stories about the media-poor
The most negative topic on poverty is properly the abuse of benefits and services. This abuse can take various forms. A frequent story about the British media-poor deals directly with fraud. These stories are typically based on a preliminary charge or court conviction of persons that have cheated on welfare benefits. For the UK we found 19 such stories (10 percent). For Sweden, we found one. For Denmark, we did not find a single story. Another frequent British story is single mothers abusing the welfare system. This abuse is typically not illegal, but the stories clearly indicate that these single mothers have made a living out of producing children to be supported by the welfare state. Thus, the black American ‘welfare queen’ is found in a ‘white’ version in the UK. We found 12 such stories in UK, one in Denmark and none in Sweden. Finally, we found five other stories about legal abuse of welfare benefits in the UK (3 percent). In Denmark we found two, and in Sweden, none. Together, stories about abuse of benefits made up 19 percent of the British stories about the media-poor (36 stories). In contrast, these abuse stories made up only 2 percent of the Danish stories and 1 percent of the Swedish stories. Thus, here we found the expected difference between the liberal and social-democratic welfare regimes. This supports the classic argument that it simply lies in the nature of selective benefits that they fuel a discussion about abuse, while more universal benefits hamper such discussions.
Crime, gang issues, anti-social behaviour and terrorism are other clearly ‘negative’ topics. Though neither of these words was included in the search string, we found 21 such stories in the UK (11 percent) in which this was the main topic of the article. In contrast, we found only five such stories in Denmark (3 percent) and one in Sweden (1 percent). As the living conditions of the poor and welfare recipients are worse in the UK than in Sweden and Denmark, this pattern is also what could be expected. In the British context, many of these stories had a reference to ‘deprived areas’ but that was not the main topic. We found only one British story (1 percent) where the ‘ghetto topic’ was the main topic. In contrast, the ‘ghetto topic’ was the most frequent negative story found in Sweden and Denmark, with nine ‘ghetto-stories’ in Sweden (12 percent) and 12 in Denmark (8 percent). This issue is definitely also a ‘negative’ frame, but there was a qualitative difference between the British ‘crime stories’ and the Scandinavian ‘deprived-area stories’.
When they are not describing structural problems that hinder the integration of immigrants, stories about immigration are typically also a ‘negative’ topic. We found 10 British stories (5 percent) in which the main topic was immigration (but without reference to structural problems such as discrimination). We found five such stories for Sweden (7 percent) and three such stories for Denmark (2 percent). Stories in which the main topic is new public policies that ‘press’ the target group typically also leave a ‘negative’ impression of the media poor. These stories typically focus on new requirements for job-search activity, and so on. We found three such stories for the UK (2 percent), five such stories for Sweden (7 percent) and eight such stories for Denmark (5 percent).
Finally, there were a number of negative stories that, though being ‘soft’ in tone, still leave a negative impression of the media-poor. For Britain we found five stories (3 percent) about the erosion of the middle-class family. Despite the fact that Sweden and Denmark have experienced the same kind of break-up in the nuclear family structure, we did not find a single story on this topic for Sweden and Denmark. Another British topic was about incentive problems caused by welfare benefits. This is a classic academic debate, and in the four British articles on the ‘incentive-topic’ the tone taken towards welfare recipients was decidedly neutral. However, the very topic questions the work ethic of the poor, and therefore we grouped these stories as having a negative topic. Finally, we found eight Danish stories about very good labour market conditions: the Danish economy picked up in 2007 with an unemployment level as low as 3.8 percent (Organisation for Economic Co-operation and Development (OECD)-standardized). Though being ‘soft in tone’, these stories also leave the impression that the poor do not take the available jobs (a lot of the stories were about employers unable to recruit enough labour). Therefore these were also grouped as examples of a ‘negative story’.
In this way, 43 percent of the British stories were ‘negative’, which should be compared with 27 percent in Sweden and 26 percent in Denmark (the difference is statistically significant at the 0.01 level). 2 So not only do the British media show more pictures of the poor and welfare recipients than the media of the other countries, when they do so, they also tell more negative stories about the media-poor. Furthermore, if one looks at the distribution of negative stories, the most negative ones, that is, about abuse and crime, are much more frequent in the UK than in Sweden and Denmark. In Sweden and Denmark the topics of abuse and crime were almost completely absent. Finally, there were a number of ‘out-of-poverty’ stories about previously poor, unemployed or homeless persons (not described here).
The harsh British versus the soft Swedish and Danish tabloid press
It is often claimed in media research that a negative story is more sensational than a positive story. Thus, as the tabloid ‘genre’ is defined by its tendency to be sensational, one could expect that the tabloid press features more of the stories that we categorize above as negative. This turns out to be the case in the UK: when the British tabloid press brought out a poverty story with one or more pictures included, the main topic was ‘negative’ in 56 percent of the cases (see Table 4). Forty percent had a ‘positive’ main topic, and 4 percent were in the ‘others’ category. In contrast, 39 percent of the stories in the British broadsheet press had a ‘negative’ main topic, 41 percent had a ‘positive’ main topic and 20 percent belonged among ‘others’. Thus, there is a tendency to bring out more ‘negative’ stories in the British tabloid press. This turns out to be a very clear difference if one looks specifically at abuse and crime stories. These make up 49 percent (22 stories) of the tabloid stories, but only 25 percent (35 stories) in the broadsheet press. The difference is significant at 0.01 level.
The relation between newspaper genre and type of poverty story.
UK country-difference significant at 0.03 level (‘negative’ versus ‘positive’ plus ‘others’), one-sided.
This relationship, however, cannot be found in Sweden and Denmark, which is somewhat surprising. In Sweden, the ‘negative’ stories make up 23 percent (six stories) of the tabloid stories and 30 percent (14 stories) of broadsheet stories. In Denmark, the equivalent figures are 18 percent (six stories) and 29 percent (34 stories). Thus, the Swedish and Danish broadsheet press actually seem to bring more ‘negative’ stories than the tabloid press (though not statistically significant due to the limited sample size). The expected pattern did not emerge even though we specifically looked for abuse and crime stories. The two Swedish tabloid newspapers do not bring out a single story on abuse or crime in the sampled period. The two Danish tabloid newspapers feature four such stories. Thus, in contrast to the British tabloid press, the typical Swedish and Danish tabloid poverty story is actually ‘positive’. Seventeen of the 26 Swedish tabloid poverty stories have a positive main topic. For Denmark, 25 out of 33 stories are positive. This is a remarkable difference between the British, and the Swedish and Danish tabloid press. These newspapers largely work under the same conditions. They all have a focus on sensational stories and they all have their readership primarily among lower social classes. Still, they feature stories and pictures that are very different.
Conclusion
The article is a first attempt to provide a systematic cross-national analysis of the media stories and pictures of the poor and welfare recipients in liberal and social-democratic welfare regimes. The results broadly support the classic argument that the public discourse about the poor and welfare clients is harsh in liberal regimes and soft in social-democratic regimes. In the sampled newspapers, we estimated that the ‘positive’ stories made up 62 percent of the Swedish stories, 55 percent of the Danish stories and 41 percent of the British stories. What contributed to the relatively large share of ‘positive’ stories in the UK were stories about poverty among pensioners and stories about public policies that ease the living conditions of the poor. Still, the difference between the UK and the two social-democratic welfare regimes is clearly significant. The difference is remarkable when it comes to able-bodied persons of working age.
The regime differences are most strongly reflected in the ‘negative’ stories. As expected, there were many more stories on abuse of benefits in the UK than in Sweden and Denmark. In the UK, these stories made up 19 percent, whereas such stories were almost absent in Sweden and Denmark (one and three stories in the sampled newspapers, respectively). This supports the classic argument that welfare systems dominated by selective benefits and services tend to produce public discussion about abuse. Another classic argument is that the economic inequality and high poverty rates in liberal welfare regimes tend to generate negative stories. Our analyses also support this argument. The stories about crime, gangs, terrorism and anti-social behaviour made up 11 percent of the British sample compared with, respectively, 1 and 3 percent of the Swedish and Danish samples. The overall finding was that, if a range of other ‘negative’ topics are also included, the ‘negative’ stories make up 43 percent of the British stories compared with 27 percent and 26 percent of the Swedish and Danish stories, respectively – again, a statistically significant difference.
This institutional logic seems to be present despite variations in ethnic divides. We found harsh British media content despite the fact that the poor and welfare recipients are primarily white. We also found soft Swedish and Danish media content despite the fact that the poor and welfare recipients increasingly are non-white. This finding supports the argument that there is an institutional logic behind the negative American attitudes towards the poor and welfare clients, which is more or less ignored in the American literature focusing on race. This is not to say that ethnic divides are of no importance. The data material also document that the American tendency to depict non-whites in negative stories and whites in positive stories are present in UK as well as in Sweden and Denmark (see Albrekt Larsen, 2013). This effect, however, does not alter the overall regime effect.
Our interpretation of the results is naturally open to discussion. One question is whether these remarkable differences in media content really are caused by the institutional structure of welfare regimes. The study was designed to out-rule the argument that it all comes down to ethnic diversity. However, the study does not out-rule the interpretation that the ‘regime differences’ could be caused by deeper cultural values related to liberalism and social-democratism (e.g. Oorschot et al., 2008) or differences in the structure of the media system (e.g. Hallin and Mancini, 2004). Finally, following the media scandals, in the British context it is sometimes suggested that the negative stories found in the tabloid press comes down to the political orientation of Rupert Murdoch. However, the two tabloid newspapers from UK included in our sample were not part of the Murdoch empire. Furthermore, the British ‘scroungerphobia’ can also be found in the media of the 1970s. Therefore we adhere to the more structural regime interpretation. A more qualitative reading of the material (Albrekt Larsen, 2013) indicates that, in liberal regimes, the ‘moral panic’ of the tabloid press is often constructed around the stereotypes about the deviance of the poor. In contrast, in social democratic regimes, the ‘moral panic’ is constructed around the stereotypes about the ‘ordinary citizen’ not helped by the state. Therefore we recommend that future research on welfare attitude research focuses on stereotypes as these ‘pictures in our heads’ of fellow citizens are indeed both crucial and difficult to change.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.