Maintainers of Ethnic Hierarchies? Investigating the Relationship Between Media Use and Attitudes Toward Perceived Remote Versus Perceived Close Immigration

Abstract

The immigration issue seems to have consolidated a top position on the European political agenda, and the body of literature devoted to explaining attitudes toward immigration continues to expand. Alongside established models focusing on threat perceptions as the central mechanism behind anti-immigration attitudes, increased attention has been paid to the role of media. The aim of the present study is to test the effects of media use on attitudes toward immigration from regions that are perceived to be remote versus those close to the host society. Moreover, the study goes beyond previous research by investigating reciprocal relationships between the use of specific media types and attitudes toward immigration. While the empirical findings suggest that the effects of using traditional news media is weak, right-wing alternative media appears more powerful in influencing immigration attitudes. Specifically, the results indicate that the use of right-wing alternative media reinforces negative attitudes toward perceived remote immigration. At the same time, the use of right-wing alternative media appears to stimulate more positive attitudes toward perceived close immigration. Further, reciprocal relationships were found between right-wing alternative media use and attitudes toward immigration.

INTRODUCTION

In light of increased refugee immigration and the continued success of radical right parties in many countries, the immigration issue seems to have consolidated a top position on the European political agenda (Jacobs, Damstra, Boukes, & De Swert, 2018). The issue is also characterized by a high degree of polarization (Demker, 2017; van Klingeren, Boomgaarden, & de Vreese, 2017). Consequently, the body of literature that focuses on explaining why some individuals have negative attitudes toward immigration while others have positive attitudes continues to expand. Alongside established models focusing on threat perceptions as the central mechanism behind opposition toward immigration, increased attention has been paid to the role of media. For instance, evidence indicates that news media, by linking immigration to problems, might reinforce negative attitudes toward immigrants and immigration while content focusing on the positive aspects of immigration is likely to have the opposite effect (Eberl et al., 2018; Igartua & Cheng, 2009; Jacobs & van der Linden, 2018). Moreover, research suggests that the use of specific media types can stimulate different effects depending on content characteristics (e.g., Beyer & Matthes, 2015; Watson & Riffe, 2013). It is worth noting, however, that the focus of previous effect studies has been on the influence of media use on attitudes, while there are hardly any studies testing if there, in line with the reinforcing spirals model (Slater, 2015), might be reciprocal relationships between media use and immigration attitudes (although see Jacobs, Meeusen, & d’Haenens, 2016).

Another gap in the literature is that it has not been tested whether media use might stimulate different effects toward immigration from different regions. This is reasonable to expect for at least three reasons. First, research indicates that people tend to be more negative toward immigration from regions that are perceived to be more remote due to differences from the host society on a range of different dimensions, such as culture, ethnicity, language, religion and socio-economic status (Ford, 2011; Hagendoorn, 1995). Second, content analyzes suggest that immigrant groups that, in general, are perceived as more remote are more salient, as well as more negatively framed, in news media (e.g., Bleich, Stonebraker, Nisar, & Abdelhamid, 2015; Jacobs, 2016). Third, experimental studies suggest that the short-term effect of exposure to different frames about immigration diverge depending on the origin of the immigrant group (Brader, Valentino, & Suhay, 2008; Igartua, Moral-Toranzo, & Fernández, 2011). Thus, it might be that media use principally reinforces negative attitudes toward perceived remote immigration, especially when people use media types that have stronger tendencies to focus on crime, conflict, and negative aspects of immigration. In the aftermath of the refugee situation in 2015, when many people from remote regions fled to Europe, this is a highly relevant issue to investigate.

Against this backdrop, the aim of this study is to test the effects of media use on attitudes toward immigration from regions that are perceived to be remote versus close. Furthermore, the study will test if there are reciprocal relationships between the use of specific media types and negative attitudes toward immigration, and how strong the potential influence of media use on immigration attitudes (i.e., the media effects) is compared to the potential influence of immigration attitudes on media use (i.e., the selection effects).

To investigate the effects of media use on attitudes toward perceived remote versus close immigration, and to test if there are reciprocal relationships, three-wave panel data from Sweden will be used.

ATTITUDES TOWARD PERCEIVED REMOTE VERSUS PERCEIVED CLOSE IMMIGRATION

Intergroup threat theory (ITT) suggests that opposition toward immigration is grounded in different kinds of threat perceptions. An intergroup threat is experienced when members of an in-group (such as native-born people) perceive that an out-group (such as immigrants) is likely to cause them harm (Stephan, Ybarra, & Rios Morrison, 2009).

ITT distinguishes between so called symbolic and realistic threats (Stephan et al., 2009). The concept of symbolic threats is based on social identity theory, suggesting that opposition toward immigration can be explained by perceived threats to group characteristics such as cultural values, religion, and ethnicity (Stephan et al., 2009; Tajfel, 1981; Tajfel & Turner, 1979). Realistic threats, in turn, stem from realistic group conflict theory, which claims that immigration might pose a threat to the economy and welfare as well as the physical wellbeing of the majority population (Stephan & Stephan, 2000).

Importantly, however, research also suggests that people generally view immigration from some regions more negatively than others (Brader et al., 2008; Ford, 2011; Valentino et al., 2017). Specifically, people are generally less positive toward immigration from regions that are perceived to be more remote due to differences from the host society on different dimensions, such as culture, ethnicity, language, religion and socio-economic status (Ford, 2011; Hagendoorn, 1995). This has been explained by a general perception that immigration from these regions (perceived remote immigration) poses a greater threat to the host society, compared to immigration from more similar regions (perceived close immigration) (Dustmann & Preston, 2007; Ford, 2011).

It is reasonable to expect that symbolic as well as realistic threats play a role in shaping attitudes toward perceived remote and perceived close immigration. In line with social identity theory, research suggests that people in Western societies are more hostile toward immigrants from more culturally distinct regions such as Africa and the Middle East compared to immigrants with greater cultural proximity (Dustmann & Preston, 2007; Ford, 2011; Valentino et al., 2017), and evidence from Sweden indicates that people consider immigrants from the Middle East to pose a greater threat to the national culture compared to other immigrant groups (Ahmadi & Palm, 2018).

Furthermore, perceived remote immigration might be viewed as a greater realistic threat — for example, by burdening the welfare system and by pushing down wages — as many who immigrate from these regions are lower educated refugees with less economic resources (De Coninck et al., 2018). It is worth noting that research also indicates that people are less positive toward so called low-skilled immigration (e.g., Valentino et al., 2017).

MEDIA OUTLETS AS DRIVING FORCES IN MAINTAINING ETHNIC HIERARCHIES?

As people tend to be more negative toward perceived remote immigration than close immigration, so called ethnic hierarchies based on the “degree of the remoteness” are likely to exist (Ford, 2011; Lubbers, Scheepers, & Wester, 1998; Tajfel & Turner, 1979). In addition, different kinds of media might play a role in maintaining these ethnic hierarchies through the use of different frames. Following the emphasis framing tradition, here frames refer to the presentation of immigration related issues from a specific perspective to the exclusion of other perspectives (see e.g., Boydstun & Glazier, 2013; Entman, 1993).

Consistent with the emphasis framing literature, evidence indicates that news media, by linking immigration to problems and threats, might reinforce negative immigration attitudes (Eberl et al., 2018; Igartua & Cheng, 2009; Jacobs & van der Linden, 2018). Yet, findings from experiments indicate that negative frames about immigrants from perceived remote regions lead to more opposition than negative frames about immigration from perceived close regions (Brader et al., 2008; Igartua et al., 2011). For example, a study from Spain revealed that participants felt more negative emotions of anger, shame, and fear when the news story had a negative frame and involved Moroccan immigrants compared to when it was about Latin American immigrants (Igartua et al., 2011). The authors argue that the reason why the origin of immigrants’ influence media effects is that the negative attitudes and stereotypes that native born people already hold toward immigrants from perceived remote regions are activated, confirmed, and reinforced when the media frame these groups in negative ways (Igartua et al., 2011). In other words, the same type of media content might lead to different effects since people tend to interpret information by using their previous understandings. This might especially be the case on a salient issue such as immigration when people are more likely to have established strong opinions (Zaller, 1992).

While experiments indicate that negative media frames principally trigger and reinforce negative attitudes toward perceived remote immigration, panel data are needed to analyze how potential effects develop over time and if media use in the real world (rather than exposure to artificial experimental stimulus) influences attitudes. While the possibility that short-term effects disappear after a while cannot be excluded, I expect to find effects in line with previous experimental studies since (1) research suggests that immigrants from perceived remote regions are more salient, as well as more negatively framed, in news media (e.g., Bleich et al., 2015; Jacobs, 2016), and (2) negative attitudes toward perceived remote immigration might be more easily reinforced, as attitudes generally are more negative toward perceived remote than toward perceived close immigration. Thus, I will first test if the findings from previous research holds true when using panel data:

H1: The use of traditional news media will generate stronger and more negative attitudes toward perceived remote immigration than toward perceived close immigration.

DIFFERENT EFFECTS DEPENDING ON THE MEDIA TYPE USED?

Beyond the effects of using traditional news media on perceived remote versus close immigration, a key question is whether the use of different media generates different effects on attitudes toward perceived remote versus close immigration. Considering that different media report and frame immigration differently (e.g., Cheregi, 2015; Jacobs et al., 2016), and that people are exposed to differently framed immigration news, this is likely the case.

More specifically, the literature suggests that commercial TV news and tabloids have a stronger tendency to focus on crime and negative aspects of immigration than public service TV news and broadsheets (Cheregi, 2015; Eberl et al., 2018; Jacobs et al., 2016). Sweden scholars have found that the commercial news show TV4 Nyheterna, compared to the two public service news shows Aktuellt and Rapport, reports slightly more about crime and conflict (Jönsson & Strömbäck, 2007; Pollack, 2001).¹ Moreover, Strömbäck (2008) found that violence, crime, and similar topics that are likely to evoke threat perceptions and feelings of fear, were substantially more common on the frontpages of the tabloids Aftonbladet and Expressen compared to the broadsheets Dagens Nyheter and Svenska Dagbladet. Since research shows that immigration is often linked to crime, violence, and terrorism (Eberl et al., 2018), these content differences might affect peoples’ attitudes toward immigration. This should especially be the case when it comes to perceived remote immigration, which is particularly likely to be linked to threats such as crime and terrorism (Bleich et al., 2015; Eberl et al., 2018; von Sikorski, Schmuck, Matthes, & Binder, 2017).

Another reason to expect different effects depending on media type is that there are some, although quite limited and few, indications that different media types in Sweden cover immigration in different ways. For instance, Strömbäck and colleagues found that framing immigration as a threat to the social cohesion is slightly more common in tabloids than in broadsheets (Strömbäck, Andersson, & Nedlund, 2017). This leads to the second hypothesis:

H2: The use of commercial TV news and tabloids will generate more negative attitudes toward perceived remote immigration than the use of public service TV news and broadsheets.

It is also important to account for right-wing alternative media, which have become an increasingly important part of the Swedish media environment (Newman, Fletcher, Kalogeropoulos, Lewy, & Nielsen, 2018). The content in these outlets is often characterized by polarization, emotional agitation, anti-immigration, and racism (Atkinson & Berg, 2012; Holt, 2016a, 2016b). Additionally, many working for these outlets often hold hostile attitudes toward Islam and are especially critical toward the impact of immigration from Muslim countries (Holt, 2016a). Hence, the use of right-wing alternative media outlets might especially reinforce negative attitudes toward perceived remote immigration. Based on this, it is hypothesized that:

H3: (a) The use of right-wing alternative media will generate more negative attitudes toward perceived remote immigration compared to the use of traditional media.

(b) The use of right-wing alternative media will generate smaller, and less negative, effects on attitudes toward perceived close immigration than perceived remote immigration.

CLARIFYING THE CAUSAL CHAIN BY TESTING THE REINFORCING SPIRALS MODEL

Even if the empirical findings would support the hypotheses outlined above, it could still be that media use and attitudes influence each other mutually. At present, however, there is a lack of studies testing if there might be so called reinforcing spiral processes between media use and immigration attitudes. The reinforcing spirals model (RSM) aims to understand the role of media in helping create and maintain attitudes (Slater, 2015). According to the RSM, the process of media selection and effects of using selected media is dynamic. In other words, the influence does not merely flow from media use to immigration attitudes or from immigration attitudes to media use. Rather, the effect of using specific media types should influence the strength and accessibility of immigration attitudes — which in turn should influence people’s media selection, leading to reinforced immigration attitudes over time (Feldman, Myers, Hmielowski, & Leiserowitz, 2014; Slater, 2015). These types of reinforcing spiral processes have been observed in several studies (e.g., Eveland, Shah, & Kwak, 2003; Kruikemeier & Shehata, 2017; Schemer, 2012). To mention one example, Schemer (2012) found reciprocal influences between the attention to political advertising about asylum law restriction and the negative affect toward asylum seekers.

The RSM is closely linked to social identity theory and suggests that media use is often used as a tool for maintaining social identities (Slater, 2015). I find it reasonable to expect that this is particularly likely to be the case in the context of immigration, since immigration —above all, perceived remote immigration — might be viewed as a threat to people’s cultural habits, religion, and social identity. Thus, media use might be especially likely to serve as a means for attitude-maintenance or attitude-reinforcement in this context. Another reason why it may be fruitful to explore reinforcing spirals in the context of immigration is that previous research suggests that specific media types tend to diverge in their coverage of immigration-related topics. Hence, people are provided with the opportunity to select a media diet that matches their immigration attitudes, which is not necessarily the case for all political issues. It should also be noted that findings from selective exposure literature suggest that people are likely to select media content that matches their immigration attitudes (Trilling, van Klingeren, & Tsfati, 2017).

However, while it might appear obvious why a person who holds hostile opinions toward immigration might select right-wing alternative media, it may not be as clear why the same person would prefer to watch commercial TV news or read tabloids instead of watching public service news and reading broadsheets. Overall, the differences in how traditional media outlets in Sweden cover immigration appear to be quite modest. Yet, there are, as mentioned, other content-differences between media types in Sweden that arguably might be relevant. Specifically, commercial TV news and tabloids tend to cover issues such as crime and conflict to a larger degree than public service news and broadsheets. Since research shows that immigration often is linked to these issues, people who already hold negative attitudes toward immigration might therefore be more attracted to this type of content, as it is likely to match the attitudes better. In other words, a person who believe that immigration contributes to more crime and insecurity might be more likely to find support for this belief when consuming commercial TV news or tabloids, since these outlets focus more on these issues and since these issues are often linked to immigration.

As effect studies suggest that media use influences immigration attitudes on the one hand, and evidence from the selective exposure literature indicates that immigration attitudes influence media use on the other, the relationship appear to be reciprocal. This, together with the theoretical argument about why the RSM is likely to be supported in the immigration context — and above all, for remote immigration — leads to the following hypotheses:

H4: There is a reciprocal influence between the use of commercial TV news and tabloids and negative attitudes toward immigration. (b) The reciprocal influence is stronger for perceived remote immigration than perceived close immigration.

H5: There is a reciprocal influence between the use of right-wing alternative media and negative attitudes toward immigration. (b) The reciprocal influence is stronger for perceived remote immigration than perceived close immigration.

Finally, I will investigate if the potential influence of the same media types on immigration attitudes (media effects) are stronger that the potential influence of immigration attitudes on media use (selection effects). Since the RSM has not previously been tested in this context, the following research questions are formulated:

RQ1: Is the effect of using commercial TV news on immigration attitudes stronger or weaker than the effect of immigration attitudes on using of commercial TV news?

RQ2: Is the effect of using tabloids on immigration attitudes stronger or weaker than the effect of immigration attitudes on using tabloids?

RQ3: Is the effect of using right-wing alternative media on immigration attitudes stronger or weaker than the effect of immigration attitudes on using right-wing alternative media?

METHOD

The Swedish Context

In order to test the hypotheses above, panel data from Sweden will be used. The panel data was collected between the years 2014–2016, which was a period when refugee immigration to Sweden increased sharply (Migrationsverket, 2017). Like citizens in many other countries, Swedes report more positive attitudes toward perceived close immigration than toward perceived remote immigration. Specifically, Swedish public opinion is most positive toward immigration from other Nordic and European countries, and least positive toward immigration from the Middle East and Africa (Ahmadi & Palm, 2018; Strömbäck & Theorin, 2018).

Furthermore, as in most Western countries, Swedish citizens’ opportunities to choose media that match their preferences have increased with the growing number of media outlets (Djerf-Pierre & Shehata, 2017), and there are relevant content differences between media types. In light of this, Sweden is a suitable case for testing the hypotheses tapping into attitudes toward remote versus close immigration, as well as the potential dynamic process between media selection and immigration attitudes.

At the same time, it should be noted that Swedish public opinion is relatively positive toward immigration compared to the public opinion in many other European countries (Heath & Richards, 2016). Hence, Swedish citizens might be quite resistant toward negative frames about immigration. This, together with the fact that there is a high level of agreement of journalistic news values and practices among Swedish journalists² (Djerf-Pierre & Shehata, 2017), makes Sweden a least likely case: if the hypotheses are supported in the Swedish context, a similar pattern will most probably also be found in other countries.

The Panel

The paper is based on a three-wave panel study with respondents recruited through stratified probability sampling from a database consisting of approximately 35,000 individuals from Novus’s pool of Web survey participants. In the first panel wave, in November 2014, 7,652 respondents were invited to join. The second wave took place in November 2015, and the final wave in November 2016. The analyses in this study are based on the respondents who either participated in wave 1 and wave 2 (N = 578) or in all three panel waves (N = 2,254). This final sample is generally representative of the Swedish population, on socio-demographic characteristics such as age, gender, and education.³

Measures

Attitudes toward Immigration

The measures on attitudes toward perceived remote and close immigration is based on a battery of items measuring attitudes toward immigration from different regions in the world. The response options ranged from 1 (very negative) to 5 (very positive). The question and items are presented in detail in the Online Appendix, Table A1.

For the analyses, one index for perceived remote immigration was constructed based on items measuring attitudes toward immigration from Africa and the Middle East (Cronbach’s alpha: .97), and one index for perceived close immigration based on items measuring attitudes toward immigration from other Nordic countries and Europe (Cronbach’s alpha: .82). The indices were rescaled to range from 0 to 1, with higher values indicating more positive attitudes.

The theoretical rationale for these categorizations is that immigrants from Africa and the Middle East differ more from Sweden on the dimensions that are likely to lead to perceptions of remoteness, such as culture, ethnicity, and religion (Ford, 2011; Hagendoorn, 1995). Also, respondents reported less positive attitudes toward immigration from Africa and the Middle East (see table A2 in the Online Appendix). Additionally, for the years 2015 and 2016, a principal component analysis distinguished two different factors, showing that the attitudes toward immigration from other Nordic and European countries loaded substantially stronger on one of the factors while attitudes toward the Middle East and Africa loaded stronger on the other factor. This suggests that attitudes toward immigration from the Middle East and Africa form one latent variable (attitudes toward perceived remote immigration), while attitudes toward immigration from other Nordic and European countries form another latent variable (attitudes toward perceived close immigration). The results are displayed in Table A3 in the Online Appendix. It is noteworthy that only one factor was identified in 2014. The fact that two factors developed over time is an indication that the distinction between perceived remote and close immigration became more important over time. It is reasonable to expect that this, at least partly, can be explained by the increased refugee immigration from the Middle East in 2015.

Media Use

The media use variables are based on items measuring respondents’ use of specific media outlets in different formats (see the details of the measures in Table A1 and descriptive information of media use among respondents in Table A4 in the Online Appendix). The individual media use variables were first recoded so that each respondent for each individual media outlet was assigned a value that was the most common for the particular outlet. For example, a respondent who reported that s/he read the tabloid Aftonbladet on print 3–4 days a week and on the phone or tablet 5–6 days a week, was assigned the value corresponding to the use of Aftonbladet on the phone or tablet. To construct the variables measuring the use of different media types, one index for the use of broadsheets (Dagens Nyheter and Svenska Dagbladet⁴ ), one for use of tabloids (Aftonbladet and Expressen⁵ ), and one for use of public service TV news (Aktuellt and Rapport⁶ ) were created. The rationale for computing indices is that research from Sweden as well as other countries suggests that commercial TV news and tabloid newspapers have a stronger tendency to focus on crime and negative aspects of immigration than public service news and broadsheets (Eberl et al., 2018; Cheregi, 2015; Jacobs et al., 2016; Jönsson & Strömbäck, 2007; Strömbäck, 2008).

For commercial TV news and for alternative right-wing media, the original measures were used since the respondents were only asked about one commercial TV news-show⁷ and one right-wing alternative media outlet.

It is worth noting that the indices for tabloids, broadsheets, and public service were rescaled to range from 1–6, to make them comparable with the variables for commercial TV and right-wing alternative media.

Control Variables

Age, gender, and education are all factors playing a role in predicting immigration attitudes as well as media use and processing (Bondad-Brown, Rice, & Pearce, 2012; Demker, 2017; Eveland & Scheufele, 2000). Hence, these factors were controlled.

DATA ANALYSIS

To test the hypotheses about the influence of traditional media in general (hypothesis 1) and the use of specific media types (hypotheses 2 and 3) on perceived remote versus close immigration, a fixed effects model was estimated. To investigate reciprocal influences between immigration attitudes and media use (hypotheses 4 and 5), and to explore the relative strength of the media effects and the selection effects (research questions 1, 2, and 3), cross-lagged structural equation models were conducted. The cross-lagged structural equation models were also complemented with cross-lagged generalized structural equation models (as a robustness check), since the media use measures are ordinal variables.

While fixed effects models offer the opportunity to rule out potential spurious relationships due to time-invariant omitted variables, the cross-lagged structural equation models provide the possibility to explore potential reciprocal influences between the use of certain media types and immigration attitudes over time. Thus, combining these analytic methods provide good opportunities for making claims about causal inference (see e.g., Kruikemeier & Shehata, 2017).

It should be noted that fixed effects models are designed to study the cause of change within a person, by using within variation over time exclusively. By demanding the data to arrive at fixed effects estimates we lose degrees of freedom, which results in less precise estimates (Andreß, Golsch, & Schmidt, 2013). Thus, in the present study, an effect that is significant in the cross-lagged models might be insignificant in the fixed effects model.

RESULTS

Table A6 in the Online Appendix gives a first indication that specific media types are correlated with immigration attitudes, and that the association in many cases appears to be stronger and more negative for perceived remote immigration than for close immigration. However, to understand the effects of media use on perceived remote and close immigration we need to look at Table 1 below. Here, it should be clarified that both perceived remote and close immigration and all media types were included in the same fixed effects model displayed in Table 1. This decision was made after assessing collinearity and finding that all VIF⁸ -values were below 2, suggesting that there were no serious problems with collinearity (see Table A7 in the Online Appendix).

Table 1 Media Effects on Attitudes toward Perceived Remote and Close Immigration (Fixed Effects Model)

	Perceived Remote Immigration	Perceived Close Immigration
Commercial TV news	−.002 (.002)	−.001 (.002)
Public service TV news	.005 (.002)	−.001 (.002)
Tabloids	.001 (.003)	−.003 (.002)
Broadsheets	.005 (.003)	−.003 (.003)
Right-wing alternative media	−.018*** (.004)	.008* (.004)
R Square (within)	.184	.182
N (individuals/observations)	2,832/8,947	2,832/8,947

Note: *p< .05. **p< .01. ***p< .001. Standard errors in parentheses. The dependent variables are measured on a scale from 0 to 1 (0 = Negative attitude toward immigration, 1 = Positive attitudes toward immigration).

Turning to the results, it can be noted that no significant effects of using traditional media types could be identified. This holds true for perceived remote immigration as well as close immigration. Hence, the first and second hypotheses are not supported.

Turning to the influence of right-wing alternative media, Table 1 displays support for the prediction that it will generate more negative attitudes toward perceived remote immigration compared to use of traditional media (hypothesis 3a). Further, the expectation that use of right-wing alternative media will generate smaller, and less negative, effects on attitudes toward perceived close immigration than perceived remote immigration is also supported (hypothesis 3b). In fact, the use of right-wing alternative media had a positive effect on perceived close immigration. This is an interesting finding, demonstrating that specific types of media use are not only more likely to trigger negative attitudes toward perceived remote immigration but might also lead to more positive attitudes toward perceived close immigration.

Turning to the cross-lagged models, hypothesis 4 predicted (a) that there is a reciprocal influence between the use of commercial TV news and tabloids and negative attitudes toward immigration, and (b) that the reciprocal influence is stronger for perceived remote compared to perceived close immigration. The results, however, show that no reciprocal effects could be identified either for commercial TV news or for tabloids (see Figures 1, 2, 3 and 4 below). Hence, hypothesis 4 was not supported. However, the use of commercial tv-news in 2014 led to slightly more negative attitudes toward perceived remote immigration which was also confirmed in the generalized cross-lagged model (see Figure A1 in the Online Appendix). However, Table 2 below shows that this media effect was not significantly different from the effect of immigration attitudes on the use of commercial tv-news⁹ (research question 1).

Table 2 Statistical Differences in Coefficients for Media Effects and Selection Effects

	Commercial and Remote	Tabloids and Close	Right-Wing and Remote	Right-Wing and Close
W1-W2	ME > SE Not significant (p = .56)	ME > SE Not significant (p = .95)	SE > ME Significant (p = .03)	ME > SE Not significant (p = .12)
W2-W3	No significant effect identified (see Figure 1)	No significant effect identified (see Figure 4)	SE > ME Not significant (p = .4)	ME > SE Significant (p = .00)

Note: The p values are from Wald tests of the equality of two parameters, indicating if two coefficients are significantly different from each other. ME = Media effects. SE = Selection effects.

Figure 1 Commercial TV news and perceived remote immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived close immigration and the other media types were controlled. Model fit: Chi-square (2) = 25.43, RMSEA = 0.022, CFI = 0.999. *p < .05. **p < .01. ***p < .001.

Figure 2 Commercial TV news and perceived close immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived remote immigration and the other media types were controlled. Model fit: Chi-square (2) = 35.81, RMSEA = 0.028, CFI = 0.997. *p < .05. **p < .01. ***p < .001.

Figure 3 Tabloids and perceived remote immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived close immigration and the other media types were controlled. Model fit: Chi-square (2) = 16.62, RMSEA = 0.015, CFI = 0.999 *p < .05. **p < .01. ***p < .001.

Figure 4 Tabloids and perceived close immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived remote immigration and the other media types were controlled. Model fit: Chi-square (2) = 23.23, RMSEA = 0.020, CFI = 0.998. *p < .05. **p < .01. ***p < .001.

Further, Figure 4 suggests that the use of tabloids in 2014 led to somewhat more positive attitudes toward perceived close immigration in 2015. However, the effect became insignificant when estimating a generalized cross-lagged model (Figure A4 in the Online Appendix), and Table 2 displays that the media effect was not statistically different from the selection effect.

Thus, the main conclusion from these models is that the media effects as well as the selection effects appears to be weak and few.

Hypothesis 5 predicted that (a) there is a reciprocal influence between the use of right-wing alternative media and negative attitudes toward immigration, and (b) that the reciprocal influence is stronger for perceived remote than for perceived close immigration. The results pertaining to this hypothesis are displayed in Figures 5 and 6. Figure 5 demonstrates that there is a reciprocal influence between the use of right-wing alternative media and negative attitudes toward perceived remote immigration, which was also supported in the generalized cross-lagged model (Figure A5 in the Online Appendix). It is worth noting that no significant media effect was identified between 2015 and 2016 in the main model (Figure 5). At the same time, however, Table 2 suggests that the media effect and the selection effect were not statistically different from each other between 2015 and 2016. Between 2014 and 2015, the result suggests that the media effects were weaker than the selection effect (Table 2).

Figure 5 Right wing alternative media and perceived remote immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived close immigration and the other media types were controlled. Model fit: Chi-square (2) = 14.04, RMSEA = 0.008, CFI = 1.000. *p < .05. **p < .01. ***p < .001.

Figure 6 Right wing alternative media and perceived close immigration (structural equation model).

Note. N = 2,832. After assessment of model fit, all previous values of the dependent variable were added in each equation. Perceived remote immigration and the other media types were controlled. Model fit: Chi-square (2) = 51.12, RMSEA = 0.031, CFI = 0.995. *p < .05. **p < .01. ***p < .001.

With respect to perceived close immigration, Figure 6 illustrates that there were positive effects between the use of right-wing alternative media and positive attitudes toward perceived close immigration, although the selection effect disappeared between 2014 and 2015 when estimating generalized cross-lagged models (see Figure A6 in the Online Appendix). Table 2 suggests that the media effect and the selection effect were not statistically different from each other between 2014 and 2015. Between 2015 and 2016, however, the media effect appeared stronger than the selection effect. Thus, while the results suggest that the media effect was weaker than the selection effect when studying perceived remote immigration, it was stronger than the selection effect when it comes to perceived close immigration (research question 3).

DISCUSSION

This study has gone beyond previous research in primarily two ways: First, by looking at the effect of media use on attitudes toward perceived remote versus close immigration. Second, by testing reciprocal relationships between the use of specific media types and attitudes toward perceived remote and close immigration. The results indicate that while there were no significant effects of traditional news media use on immigration attitudes in the fixed effects model, the use of right-wing alternative media appears to be more powerful in influencing attitudes. While the identified negative effect on perceived remote immigration is in line with previous research (Brader et al., 2008; Igartua et al., 2011), one might find it more surprising that the use of right-wing alternative media generated more positive attitudes toward perceived close immigration. This finding could however be interpreted in light of the intergroup threat theory, suggesting that an intergroup threat is experienced when members of an in-group, such as native born people, perceive that an out-group, such as immigrants, is likely to cause them harm (Stephan et al., 2009). Yet, it might be that some immigrant groups are considered to belong to the in-group rather than the out-group, for instance due to relatively similar culture, ethnicity, and religion. Moreover, it is reasonable to expect that the use of right-wing alternative media, where the focus is often on immigrants that are perceived to be remote (Atkinson & Berg, 2012; Holt, 2016a, 2016b), can enhance a feeling of belonging with immigrants from regions that are perceived to be closer to the host society. It might also be that negative frames about perceived remote immigration contribute to a notion that Europeans must collaborate and be united in order to increase the chances to successfully defend their (relatively similar) culture and religion as well as economic resources. The positive effect of using right-wing alternative media on perceived close immigration might also be related to positive coverage about these regions; it is a reasonable conclusion, not least since the Sweden Democrats — the right-wing and anti-immigration party in Sweden, that also had close links to the right-wing alternative media outlet Avpixlat — underline the value of close relations with other Nordic countries, and they also seem to be increasingly positive toward European collaboration (Sverigedemokraterna, 2019).

The null effect of using traditional news media, and the fact that right-wing alternative media appears to be quite powerful in influencing immigration attitudes, might be interpreted in light of research on interpersonal and mediated intergroup contact. While evidence suggests that interpersonal contact is likely to reduce prejudice (Park, 2012), intergroup contact via media tends to have the opposite effect, due to generally unfavorable media representations (Park, 2012). Thus, it might be that the positive effects of interpersonal contact and the negative effects of mediated contact cancel each other out in most cases. This may be especially likely in Sweden where one-sixth of the population consists of immigrants (SCB, 2018), which increases the likelihood of interpersonal contact. The negative effects of using right-wing alternative media on perceived remote immigration might however have trumped the positive effects of (possible) constructive personal contacts, considering the anti-immigration content of the right-wing alternative media site which is much more pronounced than the coverage in traditional media outlets.

Another explanation for not finding distinct media effects for the use of traditional media types might be that the study was done in a country where there is a high level of agreement of journalistic news values and practices among journalists (Djerf-Pierre & Shehata, 2017). Thus, the content variations in how different media types report on immigration might not be large enough to generate distinct effects. In addition, Swedish citizens, in general, are quite positive toward immigration, which should make them more resistant toward problem-oriented news coverage about immigration-related topics. It should also be noted that the fixed effects method approach might lead to an underestimation of effects due to the loss of degrees of freedom (Andreß et al., 2013).

Moving on to the cross-lagged models, few media effects and no selection effects were identified for the traditional news media types, which is consistent with the findings from the fixed effect model. However, reciprocal relations between the use of the right-wing alternative media and attitudes toward immigration were found. In line with the fixed effect model, right wing-alternative media and negative attitudes toward perceived remote immigration influenced each other mutually, while a reciprocal influence between right wing-alternative media and positive attitudes toward perceived close immigration could be identified. Further, while the results suggested that the selection effect was stronger than the media effect for perceived remote immigration, the media effect was stronger than the selection effect for perceived close immigration.

To summarize, the findings demonstrate the relevance of distinguishing between perceived remote and close immigration, accounting for reciprocal influences and alternative media.

LIMITATIONS AND DIRECTION FOR FURTHER RESEARCH

The results above notwithstanding, some limitations of the current study need to be addressed. First, it might be argued that the measures on immigration attitudes should have been even further specified, as it is unclear what people had in mind when they answered the questions. For example, when answering the question about attitudes toward migration from European countries, did people think about immigration from Western European countries such as Germany, Southern European countries such as Spain, or Eastern European countries such as Romania?

Furthermore, it should be noted that the modest effects of using traditional media might be related to the design of the panel study, with one year between each panel wave. It might, for example, be the case that a person who watched a news show reporting on criminal immigrants would show up a more negative attitude a couple of days after, but that this effect faded away until the next panel wave. The risk that external events impacted how attitudes developed arguably increases with the relatively long gaps between the panel waves. More measuring points would, of course, provide an opportunity to get a more detailed picture over how the atypical refugee situation potentially influenced the relationship between media use and attitudes. The fact that the immigration issue was at the top of the political agenda during this period might also have limited the scope for media effects, as many people already had developed strong attitudes.

Moreover, the effects of media use on immigration attitudes and the reciprocal relationships could also be investigated in more detail by looking at individual level differences. For example, and related to the discussion above, it would be relevant to investigate the potential moderating role of interpersonal contact with immigrants.

CONCLUSION

In the aftermath of the so-called refugee crisis, many European countries are populated by a larger proportion of people who are from regions perceived to be relatively remote. This, together with the rising importance of the cultural conflict dimension — with immigration as a core issue — in many Western European countries (Bornschier, 2010), underlines the societal relevance of continuing to study attitudes toward perceived remote versus close immigration. While these attitudes are likely to be formed by several different factors, this study has demonstrated that specific types of media use might be one of them. The fact that right-wing alternative media appears to have quite strong effects on attitudes toward immigration is an important finding since these types of media outlets appear to have become more important information sources in many Western countries. The same holds true for the findings that the effects differ between perceived close and remote immigration. In that sense, the findings suggest that at least some media function to maintain ethnic hierarchies.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Notes on contributors

Nora Theorin

Nora Theorin (M.A., University of Gothenburg, 2015) is a Ph.D. student in the Department of Journalism, Media and Communication at the University of Gothenburg. Her research interests include media effects, public opinion, and migration.

Notes

¹ The factors that explain media content are many and complex (Shoemaker & Reese, 2014). Yet, it is reasonable to expect that the content differences between public service and commercial TV news can be explained by the fact that they are facing different conditions. For example, commercial TV channels compete in the advertisement market (Jönsson & Strömbäck, 2007), which might lead to increased focus on issues with high audience popularity. This might in turn stimulate coverage about crime and conflict, since people are likely to pay more attention to negative than positive information (Strömbäck, 2008).

² This is the case for traditional media. Right-wing alternative media in Sweden are guided by their political agendas rather than journalistic norms such as objectivity and impartiality (Holt, 2016a, 2016b).

³ In 2014, the mean age in the Swedish population was 41 years. The gender distribution was 50% male and 50% female. Thirteen percent had completed grammar school, 45% high school and 41% had a college or university degree (SCB, 2019). For 2%, the educational level was unknown. Among those who participated in two or three panel waves, the mean age was 46 years. The gender distribution was 49.7% male and 50.3% female; 8.2% had completed grammar school, 43.7% high school, and 48% had a college or university degree.

⁴ Cronbach’s alpha 2014 = .63, 2015 = .62 and 2016 = .66.

⁵ Cronbach’s alpha 2014 = .65, 2015 = .67 and 2016 = .67. It is noteworthy that the alpha values for broadsheets and tabloids are below the usually accepted threshold. Yet, a Principal Components Analysis (PCA) showed that broadsheets load was substantially stronger on one factor and tabloids on the other (see Table A5 in the Online Appendix). Thus, they can be considered to form two distinct latent variables.

⁶ Cronbach’s alpha 2014 = .87, 2015 = .85 and 2016 = .87. Correlation 2014 = .778**, 2015 = .755**, 2016 = .764***.

⁷ TV4 Nyheterna is the only commercial traditional television news show in Sweden.

⁸ VIF = Variance Inflation Factor.

⁹ I tested if the cross-lagged effects were statistically different from each other using the command estat stdize in Stata.