Does it Make a Difference?: Television’s Misrepresentation of the Working-Class as Cultivation Effects

ABSTRACT

Previous research has concluded that the working-class is largely invisible on television. When the working-class is displayed; however, common frames highlight moral shortcomings and lack of responsibility. This study asks what difference such representations make. The study relies on cultivation analysis and a survey of the adult Swedish population to understand the extent to which heavy television viewers provide “television answers” in their descriptions of and attitudes toward the working-class. While some results are inconsistent, heavy television viewers seem more prone to view social inequalities as the result of working-class people's failure to take responsibility for their own well-being.

Introduction

Research has shown that the working-class is largely absent from our television screens. In the instances when the working-class is represented, it is oftentimes depicted in negative ways. These tendencies within the symbolic milieu of television align with prevailing neo-liberal ideological frameworks that highlight individual responsibility over personal success and well-being (Ronsini, 2009; Stiernstedt & Jakobsson, 2019; Tyler, 2008). Studies on the televised output have documented working-class invisibility (Stiernstedt & Jakobsson, 2017) and lack of voice (Couldry, 2011; Jakobsson & Stiernstedt, 2018) as well as shaming (Hirdman, 2016; Kozma, 2018; Reifová & Hájek, 2021) and ridicule (Eriksson, 2015). A common trope, not least in the genre of reality television, is responsibilization – a type of framing according to which working-class individuals need to be disciplined into taking responsibility for their individual development and well-being (Couldry, 2008b; Stiernstedt & Jakobsson, 2019) and finding themselves in need of moral reform and improvement (Hirdman, 2016; Skeggs, 2009).

This study turns to a less studied question, namely what difference such representations make in terms of how television viewers perceive the social world. In building upon previous research, particularly two existing Swedish projects that have mapped the televisual representation of the working-class (Ross, 1988; Stiernstedt & Jakobsson, 2017), this study draws on cultivation analysis (e.g., Gerbner, 1998) to investigate the extent to which recurrent themes in televisions’ mode of representing the working-class are reflected in audiences’ attitudes and perceptions. To this end, we use a survey of the adult Swedish population (n = 2403) to study the extent to which heavy television viewers report “television answers” in their descriptions of a particular aspect of social reality – the working-class.

This study contributes with a relatively rare audience perspective (but see e.g., Yao et al., 2023) to the body of research on the televised depictions of the working-class. As such the present study provides initial answers to the ultimate concern regarding what difference ideological and pejorative representations and the invisibility of the working-class make. Hesmondhalgh (2017) suggests that the failure of the media to represent the complexity of the working-class likely has had damaging consequences, such as legitimizing political solutions that reduce support for people at lower social positions, as well as making it harder for the working-class to gain access to certain forms of education and employment. This study constitutes a step in turning these assumptions into something that can be studied empirically.

We take Sweden as our case – a society wherein television has kept its position as a dominant medium (Nordicom, 2022). The Swedish media system is marked by the strong presence of a well-funded, widely consumed and highly trusted public service television that operates on a remit explicitly emphasizing the obligation to represent the population as a whole (see e.g., Ohlsson, 2015). Swedish public service television exists alongside commercial channels and global streaming platforms with wide reach. Despite a strong public service alternative, however, the symbolic environment of Swedish television is marked by the symbolic annihilation (Jakobsson & Stiernstedt, 2021) and derogatory framing of the working-class (Eriksson, 2015). While the Swedish working-class makes up somewhere between 40 and 50% of the population it is, as we will discuss later on, more or less invisible on Swedish television.

It is against this backdrop we turn to the question about the extent to which the nature of how, and if at all, the working-class is represented on television mirrors in the attitudes of people who spend significant portions of their everyday lives watching television. To theorize this relationship, we turn to cultivation analysis.

Cultivation Analysis: Concepts and Assumptions

In many places around the globe, television remains a (if not the) dominant medium (Morgan et al., 2014). In Sweden, 89% of the population (9–85 years old) is reached by either streamed or linear television on a regular day and as such television is the medium with the highest reach (for comparison, social networking services have a reach of 83%). Linear television is furthermore the media format with the highest average use time (73 min/day), followed by recorded music (63 min/day) (Nordicom, 2022). To paraphrase George Gerbner, television, even today, still tells most of the stories to most of the people, most of the time (Gross, 2009).

According to Gerbner’s well-researched but much-debated cultivation theory, the long-term influence of the relatively coherent symbolic environment offered by television is that it shapes people’s perception of social reality (Gerbner, 1998). In the 1960s cultivation theory was fashioned as a three-step research process (Gerbner, 1966, 1998). While the first and oftentimes ignored task (Potter, 2014) is to study the institutional conditions of television production, the second and third steps are central to studying the extent to which the recurring stories and consistent patterns in television shape audience perceptions. This study puts analytical focus on the third and last step, although we base our study design on previous research that have mapped the “message system,” or television’s symbolic environment, with content analysis (the second step in cultivation analysis).

The hypothesis in cultivation analysis is that “the more time people spend watching television, the more their beliefs and assumptions about life and society will be congruent with the most stable and repetitive messages found in television’s entertainment programs” (Hermann et al., 2021, p. 515). In other words, heavy television viewers are expected to be more likely to provide “television answers” when asked to describe various aspects of social reality (Hermann et al., 2021).

Important to note is that cultivation analysis does not set out to explain swift attitudinal or behavioral changes resulting from exposure to isolated components of the televisual output. Rather, television is posited as a stabilizer – the recurring narratives (which are argued to be found across channels and genres) are understood to shape, or cultivate, audience’s world-views over time (Gross, 2009). Television, in this view, is “’reminding’ us, over and over, of what we are not supposed to forget” (Shanahan & Morgan, 1999, p. 191). A prime example is the so-called Mean World Syndrome which suggests that heavy viewers of television (a medium in which violence is prevalent across channels and genres) are more likely than others to believe that the world is a dangerous place and that other people are not to be trusted (Shanahan & Morgan, 1999). Cultivation is, furthermore, separated into two different types of estimates. First-order judgments refer to audiences’ factual assessments about the world (e.g., the estimated number of people in law-enforcement occupations), whereas second-order judgments are value-judgments or attitudes (e.g., the “mean world” notion that other people cannot be trusted) (Hermann et al., 2021; Shrum, 2004).

A key argument forwarded by contemporary proponents of cultivation analysis is that television, despite digitalization, globalization, and media fragmentation, still constitutes an organic whole that presents audiences to a relatively coherent symbolic environment that, in the long term, shapes audience perceptions about social reality (Gerbner et al., 2002; Hermann et al., 2021; Morgan et al., 2015). Indeed, meta-analyzes have indicated small but consistent cultivation effects across a range of cultural indicators (Hermann et al., 2021; Shanahan & Morgan, 1999). For others, the small effect sizes and the modest contributions to the variation in variables measuring people's world views that can be attributed exclusively to television viewership warrant skepticism (Hirsch, 1980; see also Potter, 1993, 2014). For many cultivation scholars however, including Gerbner himself, small effect sizes still matter since “A barely perceptible shift of a few degrees average temperature can lead to an ice age or make the desert bloom” (Gerbner et al., 1978, p. 193).

The hypothesis on the congruency between the perceptions among heavy television viewers and recurring television messages strikes many as outdated in today’s high-choice media landscape, in which television is losing its hegemonic position as a prime agent of socialization (Gross, 2009; Katz & Fialkoff, 2017). However, given the consistency of cultivation effects shown in previous meta-analyzes (Hermann et al., 2021), there is a certain attraction in testing the hypotheses with regard to television’s recurring and well-documented tendency to either simply hide the working-class or to depict it in ways that link working-class belonging to a failure to take individual responsibility (Jakobsson & Stiernstedt, 2021).

Previous Research and Hypotheses

Working-class persons appear relatively infrequently on television. This is well documented in classic (see, e.g., Smythe, 1954; von Feilitzen et al., 1989) and contemporary media research (Butsch, 2003; Ronsini, 2009; Stiernstedt & Jakobsson, 2017). The working-class is, however, not only made invisible and “voiceless” (Couldry, 2011; Jakobsson & Stiernstedt, 2018). When depicted, it is oftentimes outright ridiculed or shamed in the media, and not least on television (Hirdman, 2016; Kozma, 2018; Reifová & Hájek, 2021; Tyler, 2008). In Sweden, a previous large-scale content analysis of the televised representations of different social classes have concluded that members of the working-class make up only about 10% of all persons depicted, and that speaking working-class members account for only 9% of people talking on television. The study showed that the underrepresentation of working-class people is relatively consistent across different genres, from news programming to drama and entertainment (Stiernstedt & Jakobsson, 2017). Such findings align with a previous content analysis in the same country conducted in the 1980s (Ross, 1988). Furthermore, the working-class was found to be slightly more visible in reality programs (12% of the participants had a working-class occupation) (Stiernstedt & Jakobsson, 2016) – a genre that has been the object of study in many contemporary critical inquiries (e.g., Deery, 2015; Tyler, 2008). Important to note is that the same large-scale content analysis (Jakobsson & Stiernstedt, 2021) also found that reality television was the dominant genre on linear television in Sweden, accounting for more than 20% of the total content.

The recurrent findings from previous studies on the representation of the working-class point to what cultivation researchers refer to as the “broad structures and consistent patterns” in television’s message system (Morgan & Shanahan, 2010, p. 338). From a cultivation analysis perspective, one would thus expect certain consequences for people who form their world-views by watching television. However, cultivation theory does not tell us exactly what we should expect those effects to be. One hypothesis might be that if the working-class is virtually invisible on television, this should have consequences for heavy television viewers’ perception of the size of the working-class, in comparison with light television viewers who form their world-views elsewhere. While such an expectation is broadly aligned with cultivation analysis and effects on so-called first-order judgments, there is something counter-intuitive about it. It is safe to assume that television viewers to some extents are aware of the constructed nature of reality as depicted by television and the genre conventions that guide the production process of televisual content. People understand that there are proportionally more news and weather presenters on television than in society at large, and heavy-television viewing does not lead people to radically overestimate the number of news and weather presenters in the general population. Thus, even if cultivation effects have been documented in relation to some recurrent trends in world on television (see e.g., the Mean World Syndrome) we should be cautious with regard to expectations that the invisibility of the working-class on television would lead television viewers to drastically fail in estimating the actual size of the working-class.

We find it more reasonable to suggest that long-term heavy television viewing can have consequences when it comes to how audiences explain and understand social inequalities. The reasons behind social inequalities are by nature opaque and even social science researchers disagree on the underlying causes behind social inequality. Thus, the ways in which social classes are depicted in a dominant medium such as television can constitute important references for how heavy-television viewers understand the societies they inhabit. According to cultivation analysis, as argued above, television functions as a social stabilizer and of social beliefs and values (Gross, 2009; Shanahan & Morgan, 1999). This aligns with many different strands of arguments in the literature on media power, which converge on the point that the power of the media lies not in teaching specific facts, but in highlighting what to think about and how to think about it. Although this is not the place to review this literature, a quote from Couldry (2008a) in his discussion on the social consequences of media texts, in conversation with cultivation theory, audience reception theory, critical discourse analysis and more, catches this point nicely. Couldry maintains that what is crucial with regard to media power and media effects is as follows:

not the factual content of media (as remembered by audiences or registered in their persistent patterns of story-telling, important though that is), but rather a longer process of naturalisation whereby elements and categories of media discourse [become] taken-for-granted as “natural.”

(Couldry, 2008a, p. 78)

Conclusions from previous studies on cultivation effects with regard to social class prejudice and other socio-economic issues nonetheless remain inconsistent. While Yao et al. (2023) found no significant relationship between television viewing and social class prejudice other cultivation analyses have identified weak albeit consistent differences in how heavy television viewers perceive social reality with regards to socio-economic issues such as affluenza and materialism (Harmon et al., 2019; Hennigan et al., 1982; Shrum et al., 2005, 2011).

Television, as demonstrated in previous research, and reality television in particular, forwards a narrative of individual responsibility when it comes to explaining a person’s social position. The poor and the working-class are thus often depicted as failing or lacking in individual qualities and inner motivation to succeed and rise above their current social position. This is part of the “public pedagogy” of reality television (Skeggs & Wood, 2012). As discussed by Ouellette and Hay (2008), among others, this theme of responsibilization has been a central feature in (neoliberal) social and economic policy for decades. In accordance with this ideological theme, poverty is categorized as an individual failure, rather than as a structural economic and social issue. What television does then, if anything, is that it reproduces central features of the dominant ideology.

A difficulty in hypothesizing cultivation effects is that the patterns described above regarding how the working-class is portrayed on television arguably are the same, or similar, in other media formats. At the very least we cannot expect representations of the working-class on YouTube, on the radio, in newspapers, and so on, to be wildly different from the representations on television. As such, whatever cultivation effects that might be found with regard to heavy television viewing are likely to be small. Nevertheless, and in line with (the somewhat problematic) discussions on first-order estimates, we hypothesize that heavy viewers are more likely to hold misconstrued views about the size of the contemporary working-class. This regards a cultivation effect stemming from the invisibility of the working-class as a recurrent pattern in Swedish television (Ross, 1988; Stiernstedt & Jakobsson, 2017; von Feilitzen et al., 1989):

H_1a:

Heavy television viewers will be more likely than others to incorrectly estimate the size of the contemporary Swedish working class.

However, if the working-class appears relatively infrequently on television (making up only about 10% of the people on screen) a more precise hypothesis would posit that heavy viewers are more likely to underestimate the size of the working-class. This would, in our view, constitute the more clear-cut “television answer” connected to tendencies and recurring patterns in the televised output. As such, we pose a sharper version of our first hypothesis:

H_1b:

Heavy-viewers of television will be more likely than others to underestimate the size of the contemporary Swedish working-class.

With regard to the individual responsibility framing of the working-class (Eriksson, 2015; Stiernstedt & Jakobsson, 2019) cultivation analysis leads us to expect that heavy television viewers are more prone to hold views that people in vulnerable or precarious social circumstances only have themselves to blame for their well-being and socio-economic position, as a second-order, and in our view more reasonable, estimate.

H₂:

Heavy-viewers of television will be more prone to agree with statements that posit that people that are poor, or otherwise find themselves in disadvantaged social positions, have themselves to blame.

In what follows, we use an online survey (n = 2403) to test the above hypotheses on both first-order and second-order cultivation effects, other things held constant.

Data and Method

A key step in cultivation analysis involves asking a sample of the population about their total television consumption, and how they perceive social reality (Gerbner, 1966, 1998). This allows studying congruencies between television’s recurrent themes and the perceptions and world-views of individuals that are most exposed to this symbolic environment. Cultivation effects are claimed to be identified in the cases where “heavy viewers” are shown to be more prone to provide “television answers” when asked about the character of social reality, other things held constant (Gerbner, 1998; Morgan & Shanahan, 2010).

This final step in cultivation analysis presupposes knowledge about the symbolic environment of television. Before surveying respondents, we need, in other words, to know what would constitute a “television answer.” Here, we draw upon two previous projects (detailed in the previous section) on how social classes are depicted on Swedish television (Ross, 1988; Stiernstedt & Jakobsson, 2017), as well as adjacent research on how the working-class is represented in the media. A key insight here is the striking similarity between the 1980s (Ross, 1988) and the 2015 (Stiernstedt & Jakobsson, 2017) content analyses. This would suggest that the ways in which the working-class is represented in Swedish television remain relatively stable over time, and that we have little reason to expect that things have dramatically changed in the years leading up to the present cultivation analysis. For this study, we designed an online survey for the adult Swedish population (18 and above) administered by the research institute Kantar-Sifo. The survey was put in the field during the first 2 weeks of March 2023 and retained a response rate of 18.15%. The dataset includes 2403 individuals. For the present analysis, data were weighted to deal with the low response rate and skewness toward older and relatively well-educated individuals.

Dependent Variables

In order to test H_1a regarding the first-order cultivation of real-world perceptions, we asked respondents to estimate the size of the contemporary Swedish working-class in percentage deciles. Analyses of the actual size of the working-class provide different answers and such distinctions and measurements are difficult to make. Depending on exactly where lines are drawn contemporary estimations range from 39 to 49% (Ahrne et al., 2018). Since opinions and statistics diverge, we categorized the “correct” estimate generously as located between 30 and 60%, while the “incorrect” answer was classified as less than 30% or more than 60% of the Swedish population, as well as “I don’t know.” This variable is binary where the “correct” estimate on the size of the Swedish working class was coded “0” (49.1%) and the “incorrect” estimate as “1” (50.1%) for use in a binary logistic regression analysis.

As discussed above, a first-order perception that would more clearly correspond to a “television answer,” and better align with the presuppositions of cultivation analysis, would be that heavy viewers underestimate of the size of the working-class. Thus, to test H_1b we re-coded the dependent variable whereby underestimating the size of the working-class (0–30%, including “don’t know”) was coded “1” (25.5%) and not underestimating the size of this class as “0” (74.5%).

In order to test H₂ on second-order value-judgments, we presented respondents with four different statements on 5-point Likert-scales on which respondents could agree or disagree. These items captured attitudes aligned with the trope of responsibilization whereby poor and disadvantaged people are viewed as having failed in terms of taking responsibility for themselves and their own well-being, while the rich and successful are framed as deserving (e.g., Eriksson, 2015; Ronsini, 2009; Stiernstedt & Jakobsson, 2019; Tyler, 2008). The statements read (1) “Poor people can change their lives if they just work hard enough”, (2) “If you fail in life, you only have yourself to blame”, (3) “That people with lower incomes have bad health is only because they cannot take care of themselves” and (4) “Rich and successful people have, in most cases, made themselves deserving of their success.” These items were collapsed into an “responsibilization index” (Cronbach’s alpha = .765) and used as the dependent variable in an ordinary least squared regression analysis (scale 1–5, M = 2.80, SD = 0.79).

Although H_1a and H_1b deal explicitly with the working-class, H₂ is formulated in a way as to make it about “poor people” and in one respect “rich and successful people.” This choice was made in order to make the survey questions less open to interpretation, since there is arguably less divergence concerning opinions on what it means to be poor (or rich and successful), in comparison to what it means to be working-class. However, income, wealth, and class are correlated, implying that lower income/less wealth corresponds to a lower class position and vice versa. It is also known from previous qualitative audience research that television audiences tend to conflate the concepts of occupation, status, income, and wealth with the notion of social class (Ross, 2008).

Independent Variable

In following the “traditional” approach in cultivation analysis, we relied on an overall, self-reported, estimate on the number of hours spent watching television on a regular day. As discussed by Hermann et al. (2021), the digitalization and the transition from a “low-choice” to a “high-choice” media landscape has led a group of reformists to argue for the need to renew approaches to cultivation by studying viewership of different genres or platforms, not least because the prevalence of social media in people's media repertoires (Breuer et al., 2015; Gross, 2009; Tsay-Vogel et al., 2018). Others, including those who have compared the different forms of measurement (Hermann et al., 2021), have maintained that the total amount of television viewing remains a robust measurement that more clearly links to cultivation researchers’ concern with the overall message system provided by television (Morgan & Shanahan, 2010). As discussed above, television remains a dominant medium in Sweden (Nordicom, 2022) that, at least with regard to the representation of the working-class, provides a relatively coherent symbolic environment. As such, this study followed the meta-analysis by Hermann et al. (2021) and used a variable measuring total amount of television viewing across devices. We asked respondents to estimate the number of hours spent on viewing television (on a tv-set, a mobile phone, a tablet or a computer) on a regular day. This variable used an ordinal scale ranging from 1 (“I never watch television”) to 7 (“More than 5 hours per day”). We classified “heavy viewers” (coded as “1”) as those spending more time on television than the sample mean category of 4 (2–3 h/day). This implies that the category of “heavy viewers” comprises people that watch television for more than 3 h per day (36.6% of the sample). This group is contrasted by those spending average or less time on watching television (<3 h/day) (63.4%, coded as “0”).

Control Variables

The control variables include age, a continuous variable (18–94, M = 49, SD = 17), a binary gender variable separating men (50.1%) from women (49.9%), a binary education variable separating respondents holding a university degree (45.2%) from those without a degree (54.8%) and finally subjective class affiliation separating blue-collar workers as well as unemployed people (in total 35.8%) from other occupational categories including white-collar workers, the self-employed and farmers (64.2%). This implies that we compare social groups with relatively lower social status (blue-collar workers and the unemployed) to a group dominated by white-collar workers, but also includes entrepreneurs and the self-employed. Table 1 details the key demographic traits of the sample. In terms of collinearity diagnostics of the regression models used to test the hypotheses, no VIF-value was higher than 1.3. Table 2 provides a correlation matrix for all included variables.

Table 1. Demographic characteristics of the sample.

Variable	% (n)
Gender
Male	50.1 (1203)
Female	49.9 (1200)
Education level
Not finished mandatory education/unfinished upper secondary school	12.4 (298)
Upper secondary school (gymnasium)	17.5 (419)
Post-upper secondary education	10.6 (256)
Studies at tertiary education	14.2 (342)
Tertiary degree	45.2 (1087)
Age
18–29	18.4 (428)
30–49	38.7 (902)
50–64	25.5 (594)
65+	19.8 (478)
Subjective social class
Blue-collar/unemployed/sick-leave	35.8 (842)
Other (incl. white-collar, self-employed)	64.2 (1508)

Table 2. Correlation matrix.

	1	2	3	4	5	6	7	8
(1) Age	−
(2) Heavy television viewing (ref: <3 h/day)	.102**	−
(3) Woman (ref: man)	−.001	.058**	−
(4) University degree (ref: no degree)	−.039	−.142**	.137**	−
(5) Lower social position (working-class and unemployed, ref: other classes)	−.286**	.059**	.010	−.303**	−
(6) Responsibilization-index	.082**	.043*	−.186**	−.059**	−.087**	−
(7) Underestimates size of working-class (ref: does not underestimate)	.016	−.049*	−.006	.060**	−.080**	−.004	−
(8) Wrong estimate of size of working-class (ref: correct)	−.058**	.053*	.004	−.073**	.110**	−.009	.575**	−

**= correlation is significant at the 0.01 level, *= correlation is significant at the 0.05 level.

Results

Table 3 turns to H_1a – Heavy television viewers will be more likely than others to incorrectly estimate the size of the contemporary Swedish working-class. This expectation stems from large-scale content analysis in the 1980s and in 2015 that have concluded that the working-class is heavily underrepresented on Swedish television (Ross, 1988; Stiernstedt & Jakobsson, 2017). According to cultivation analysis such recurrent patterns should reflect in heavy television viewers’ world-views. This was tested using binary logistic regression on the variable predicting the odds for providing the wrong estimation on the size of the contemporary working-class. Important to note in terms of the binary logistic regressions (Tables 3 and 4) is that an odds ratio above 1 indicates increased likelihood of an event happening (e.g., incorrectly estimating the size of the working-class), whereas an odds ratio below 1 indicates decreased likelihood for the event taking place. Model 1 (Table 3) includes only the control variables (showing that university educated individuals are significantly less likely to provide the incorrect estimate, and that people at lower social positions [working-class people and the unemployed] are more likely to be incorrect). In Model 2, we add heavy television viewing as a predictor, which is significantly correlated with providing an incorrect estimation of the size of the working-class (ExpB = 1.198, p = .040). This implies that heavy television viewers are estimated to be about 1.2 times more likely to provide incorrect descriptions of the size of the working-class. In this second model, level of education is no longer statistically significant. Thus, our first hypothesis is supported. It should be noted, however, that the model explains only about 2% of the variation in the dependent variable, which taps into long-standing discussions on what a significant or meaningful effect-size in terms of cultivation analysis really is (e.g., Hermann et al., 2021; Hirsch, 1980).

Table 3. Odds for providing incorrect estimates on the size of the contemporary Swedish working-class. Binary logistic regression.

	Model 1		Model 2
	Exp(B)	Sig.	Exp(B)	Sig.
Heavy television viewing (ref: <3 h/day)			1.198	.040
Age (18–94)	.199	.997	.996	.131
Woman (ref: man)	1.053	.543	1.039	.650
University degree (ref: no degree)	.822	.029	.841	.055
Lower social position (working-class and unemployed, ref: other classes)	1.446	<.001	1.433	<.001
Constant	1.120	.479	.638	1.078
Nagelkerke R^2	.019		.022
N	2348		2348

Bolded values are significant at p < .05.

Table 4. Odds for underestimating the size of the contemporary Swedish working-class. Binary logistic regression.

	Model 1		Model 2
	Exp(B)	Sig.	Exp(B)	Sig.
Heavy television viewing (ref: <3 h/day)			.812	.043
Age (18–94)	1.001	.673	1.002	.519
Woman (ref: man)	.948	.582	.962	.688
University degree (ref: no degree)	1.236	.038	1.204	.072
Lower social position (working-class and unemployed, ref: other classes)	.729	.005	.736	.007
Constant	.329	<.001	.344	.043
Nagelkerke R^2	.012		.015
N	2348		2348

Bolded values are significant at p < .05.

In Table 4 we test the “sharper” hypothesis on the first-order estimate, that is, the extent to which the invisibility of the working-class on television translates into heavy viewers being more likely to underestimate the size of the working class (H_1b). In Model 1 (Table 4) only university degree and belonging to the working-class display significant effects on the odds for underestimating the size of the working-class. People holding a university degree are significantly more likely to underestimate the size of this social group, whereas people at lower social positions are less likely to do so. The second model of Table 4 includes our independent variable heavy television viewing. Results run contrary to the cultivation hypothesis, as heavy viewers are significantly less likely than others to underestimate the size of the working-class (ExpB = .812, p = .043). The odds for respondents’ underestimation are, in other words, almost 20% lower among heavy television viewers. Like the previous model, this model’s predictive power is minimal (Nagelkerke R² = .015). It is nonetheless worth noting that this “sharper” model does indeed runs contrary to what should be expected according to cultivation analysis. We thus arrive at incoherent results with regards to the first-order cultivation effects stemming from heavy exposure to television.

The results from the testing of H_1a and H_1b go partially against what we would expect from the perspective of cultivation analysis, when it comes to first-order effects. The fact that there are few working-class people on television does not seem to lead heavy television viewers to underestimate the actual size of the working-class. This ties into our theoretical discussion, where we argued that first-order effects regarding the cultivation of factual perceptions seem much more far-fetched than more reasonable arguments on the reinforcement of already existing attitudes.

Thus, hypothesis H₂ is, in contrast to H_1a and H_1b, about second-order cultivation effects, similar to previous studies on materialism and affluenza (Harmon et al., 2019; Hennigan et al., 1982; Shrum et al., 2005, 2011). These are not about factual statements but rather about value-judgments and attitudes. In Table 5 we turn to television’s ability to shape attitudes and judgments among people that spend significant portions of their lives viewing television, thus forming significant portions of their world-views from this medium. Following previous research emphasizing the individual responsibility trope connected to the well-being of poor people and the working-class (e.g., Couldry, 2008b; Stiernstedt & Jakobsson, 2019) we hypothesized that heavy television viewers would be more likely than others to agree with statements capturing notions linked to the recurring televisual thematic of responsibilization.

Table 5. Predicting responsibilization-attitudes. Ordinary least squared regression.

	Model 1		Model 2
	B	Sig.	B	Sig.
Heavy television viewing (ref: <3 h/day)			.083	.019
Age (18–94)	.002	.056	.002	.096
Woman (ref: man)	−.283	<.001	−.288	<.001
University degree (ref: no degree)	−.085	.017	−.075	.038
Lower social position (working-class and unemployed, ref: other classes)	−.152	<.001	−.155	<.001
Constant	2.943	.000	2.924	.000
Adjusted R^2	.046		.048
N	2106		2106

Bolded values are significant at p < .05.

To assess H₂, we relied on Ordinary Least Squared regression where the “responsibilization”-index (scale 1–5) was used as the dependent variable. Model 1 (Table 5) includes only the control variables and we observe that women, university-educated individuals and people at lower social positions are significantly less prone to agree with this attitudinal constellation. When we include the independent variable on television viewing the explanatory power of the model increases slightly, albeit from very low levels (Model 1, Adjusted R² = .046, Model 2, Adjusted R² = .048). Heavy television viewing is positively and significantly correlated with the responsibilization-index (B= .083, p = .019). As such, H₂ is supported, but minimally so. This runs contrary to a recent study failing to find statistically significant associations between television exposure and prejudice toward low-income individuals (Yao et al., 2023) but aligns with previous studies on television’s capacity to cultivate various attitudes with regard to affluence and social class (Harmon et al., 2019; Hennigan et al., 1982; Shrum et al., 2005, 2011). Still, we are confronted with a limited capacity to explain variation in the dependent variable, and small effects-sizes.

For the critics of cultivation analysis, the small effect-sizes and modest contributions to the explained variation across various cultural indicators imply that we should move on and identify other variables, or revise cultivation analysis as such (Hirsch, 1980; Potter, 1993, 2014). It is reasonable to expect that attitudes on the lifestyles and socio-economic situations of the working-class is better explained by political and ideological position-takings than by heavy exposure to television. For the defenders of cultivation analysis on the other hand, small effect sizes still matter (Hermann et al., 2021) as “[a] barely perceptible shift of a few degrees average temperature can lead to an ice age or make the desert bloom. A slight but pervasive tilt in the cultural climate can have major social and public policy implications” (Gerbner et al., 1978, p. 193). In this context, it is worth highlighting the discussion on the tendency to over rely on measurements on variation (Abelson, 1985). It should also be noted that social class is a different kind of phenomenon compared to, for example, violence or other phenomena previously studied from the perspective of cultivation analysis. People are confronted with the realities of social class, status, hierarchies, and stratification on a first-hand basis in their everyday lives. From the outset, and in line with decades of empirical research into cultivation effects (Hermann et al., 2021), it is to be expected that television in itself only plays a small role in the formation of beliefs about social class.

Discussion

Numerous studies have critically assessed the representation of the working-class on television (Couldry, 2011; Smythe, 1954; Tyler, 2008). In analyzing a sample of the adult Swedish population (n = 2403) this study has translated the tendency of television to render the working-class invisible and the propensity to rely on individual responsibility frames when depicting the life-situation of poor or otherwise socially disadvantaged people, as hypotheses for cultivation analyses on audiences’ world-views. We have, simply put, asked to what extent the recurrent patterns in televised representations of the working-class matter for viewers’ perceptions of and attitudes toward the working-class and socio-economically disadvantaged people.

Our hypotheses receive mixed and contradictory support. Heavy television viewers are more prone to hold misconstrued notions about the size of the contemporary Swedish working-class but they are, at the same time, less likely to underestimate its size, which would be the clear-cut “television answer” given the lack of working-class visibility on television. We thus arrive at inconclusive findings regarding so-called first-order estimates of cultivation effects (Shrum, 2004). Our analyses provide, however, some modest support for cultivation effects when it comes to people’s attitudes toward the socio-economic circumstances of poor people. Heavy television viewers seem more likely to adopt the ideological viewpoint that being poor is mainly a matter of individual failings, rather than a symptom of wider structural issues. This viewpoint, although perhaps not ubiquitous throughout the totality of the symbolic landscape of television, is nevertheless a dominant frame in much television content (Reifová & Hájek, 2021), and especially in reality television which is the largest genre in Swedish television (Stiernstedt & Jakobsson, 2019). This, it should be noted, is also an ideological belief that has been supported by both right- and left-wing parties in Sweden for a long time. The role of television in this regard is thus mainly as a “reminder” and stabilizer of the dominant ideology. Thus, in relation to research on representations of social class on television this study provides some modest support to the often-implied understanding within these studies – that representations do matter in some sense. One key debate within research on social class and media revolves around the question of whether the invisibility and misrepresentation of the working-class is obstructing the formation of a working-class identity, or a “class consciousness.” The findings of this study concerning the issue of responsibilization emphasize precisely this point and are especially interesting since we know that working-class people tend to spend more time watching television compared to people at other social positions (Auhuber et al., 2019; Lindell & Hovden, 2018). Given the small effect-sizes, the limitations in the present study, however, we call for additional research that can supplement, confirm, or debunk our findings.

Limitations

While this study followed the research design by previous cultivation analysis it is not, like cultivation analysis itself, without limitations. First, the study was based on self-reported data with low response rates. Despite the data being weighted, this limitation should warrant caution. Second, the idea of cultivation effects is inherently about long-term effects of televised socialization yet this and most other cultivation studies rely on cross-sectional data that remain unable to capture long-term exposure to television (see also Potter, 2014). In other words, the present study design presupposed that self-reported heavy exposure to television on “a regular” day is more or less stable over time. Third, by focusing exclusively on overall volume of television viewing we have not studied potential differences in the world-views of people who are exposed to more or less programming infused with the “responsibilization” framing. Experimental designs (see e.g., Herrett-Skjellum & Allen, 1996) measuring differences in exposure to various television representations of the working-class would provide valuable extensions of the present study. Fourth, a key intervening variable with regard to attitudes toward social success in particular and the working-class in general is political affiliation and ideological preferences which our dataset does not include. Future studies would benefit from including such variables as controls. Fifth, one can on good grounds critique cultivation scholars that insist that television remains the dominant storyteller of our time (e.g., Gross, 2009; Morgan et al., 2015). We stress, however, that while many people arguably form their world-views in other fora, such as social media, television remains a dominant medium in many parts of the world, including Sweden. We have here been concerned with people who spend a significant portion of their everyday life watching television – a symbolic milieu that tends to exclude the working-class, or to represent it in certain recurring frames.

Lastly, it must be stressed that our findings align with previous observers that have emphasized that in cultivation analysis television accounts for very small shares of the variation in cultivation indicators/audience perceptions (Potter, 2014, p. 1027; but see Abelson, 1985 for a discussion on the dangers of over relying on share of explained variance). In terms of our overarching question about the extent to which the tendencies and recurring patterns of (mis)representation of the working-class actually matter for how audiences perceive the social world we must conclude on ambiguous terms: while it is difficult to say anything about heavy viewers’ real-world perceptions about the size of the working-class we have identified modest support for the notion that heavy viewers are slightly more aligned with the neoliberal morality inscribed in how television, and particularly reality-TV, depicts the working-class and people at precarious social positions. We thus call upon future research to establish whether or not we are dealing with statistical constructs or actual associations.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Johan Lindell

Johan Lindell (Ph.D., Karlstad University) is an Associate Professor of Media and Communication Studies at Uppsala University, Sweden. His research interests include fields of cultural production, media use and social inequality and the Nordic media system.

Peter Jakobsson

Peter Jakobsson (Ph.D., Södertörn/Örebro University) is an Associate Professor of Media and Communication Studies at Uppsala University, Sweden. His research interests include media policy, media and social class and media trust.

Fredrik Stiernstedt

Fredrik Stiernstedt (Ph.D., Södertörn University) is a Professor of Media and Communication Studies at Södertörn University, Sweden. His research interests include media policy, media history and media trust.