Reinforcing Spirals of Negative Affects and Selective Attention to Advertising in a Political Campaign

Abstract

The present study investigates self-reinforcing spirals processes between the negative affect toward asylum seekers and the attention to political advertising in a campaign dealing with the issue of the asylum law restriction. Based on data from a three-wave panel survey, the study found evidence of such reciprocal influences. Specifically, the initial attention to political advertising elicited negative affective reactions, such as fear or anxiety toward asylum seekers in the course of the campaign. At the same time, these affective reactions enhanced people’s attention to subsequent political ads. Not only do the findings indicate the presence of reinforcing spirals processes but also do the data suggest that the spirals process is mainly fueled by cues emanating from the political campaign.

Keywords

political advertising intergroup emotions reinforcing spirals processes research methods ethnic minorities

Political communication strategies do aim not only at voters’ minds but also at their hearts (cf. Kinder, 1994). Numerous studies have demonstrated that political ads are likely to elicit emotional reactions of voters (Brader, 2005; Chang, 2001; Lang, 1991; Thorson, Christ, & Caywood, 1991). For example, Chang (2001) showed that attack ads aroused voters’ negative emotions thereby influencing their candidate evaluations. Racial cues in campaigns also increase negative emotions (Brader, Valentino, & Suhay, 2008; Esses, Veenlivet, Hodson, & Mihic, 2008). In addition, extant research suggests that negative affective states increase people’s involvement and attention in campaigns. Specifically, fear or anxiety enhances information seeking (Boyle et al., 2004; Cho et al., 2003; Valentino, Hutchings, Banks, & Davis, 2008), political involvement (Marcus & MacKuen, 1993), and attention to campaign stimuli (Bradley, Angelini, & Lee, 2007; Lang, 1991).

Put differently, on one hand affective reactions are precursors to the attention to political advertising. On the other hand, the attention to such ads elicits affective reactions. Obviously, the relationship between advertising attention and affective reactions is reciprocal. However, previous studies have not explicitly investigated these reciprocal influences. There are some hints in research demonstrating that the relationship between exposure to TV news and negative affect in the context of 9/11 is bidirectional (Cho et al., 2003). However, this study did not deal with emotions elicited by strategic political communication but with affective reactions stemming from an exceptional event that caused traumatic responses and stress disorders in the audience (Ahern et al., 2002).

The following study builds on previous research and extends it in several ways: First, it investigates the reciprocal relationships between negative emotions in response to strategic political communication and the attention to campaign cues that is due to campaign-induced affective reactions. Specifically, the study concentrates on negative affective voter reactions to portrayals of ethnic minorities. The use of racial cues is a common strategy of campaigners to elicit negative emotions of voters thereby affecting voting intentions or attitudes toward policy programs (Brader et al., 2008; Esses et al., 2008; Peffley, Shields, & Williams, 1996).

Second, the focus is on how these processes evolve over time. That is, the study examines how attention to political ads increases negative affective reactions toward ethnic minorities and how these affective reactions elicit additional attention to the campaign. The reinforcing spirals model (RSM; Slater, 2007) suggests that media effects and selectivity can reinforce each other producing cumulative effects over time.¹ Accordingly, campaign cues should elicit negative affective reactions toward ethnic minorities. As a consequence of the campaign-induced negative affect, people’s attention to the campaign is likely to increase. Thus, in contrast to simple effects models in communication research that ignore the dynamic and reciprocal processes of audience selection and effects over time, the present research explicitly acknowledges and investigates the reciprocity of the audience selectivity and effects in a single framework. Specifically, we applied the RSM to a campaign context to study the reciprocal relationships of audience attention to campaign advertising and negative affect toward ethnic minorities over time. However, the occurrence of reinforcing spirals processes requires that such influences occur repeatedly in successive periods of a campaign. On one hand, the growing campaign volume, increasing until the hot phase of a campaign is reached, is more likely to elicit stronger emotional responses and enhance attention in voters at the end than at the beginning. On the other hand, at the peak of a campaign, advertising effects may wear out because voters become habituated or inoculated. Given the inconclusiveness of the extant literature, the present research investigates the evolution of these processes over time more thoroughly.

A third issue refers to the relative strength of both processes. When the temporal development of such a process is considered, one may wonder whether there are differences in the strength of the impact of political ads on negative affect on one hand and the influence of negative affective states on the attention to campaign cues on the other. This question is not only of theoretical interest but also normatively important. When advertising effects on affective reactions drive the reinforcing spirals process, such that the attention that voters devote to the campaign is by and large determined by campaign cues, then this implies that voters can be manipulated by campaign strategies. However, when campaign attention is the result of people’s initial interest, then this would limit the potential of voter manipulation.

Clearly, the relative impact of these processes is always specific to a certain campaign context in which it is tested and different contexts may produce different patterns of results. Nevertheless, by considering more cases and different campaign contexts, we may gain a more thorough understanding of whether campaign effects are the result of people’s voluntary attention devoted to a given campaign or whether campaigns produce cumulative effects by raising people’s attention to the campaign that then influence their affective reactions.

In sum, the aim is to test whether the influence of political ads on negative affective reactions toward ethnic minorities and the increase in attention to political advertising that is due to these negative emotions produce cumulative effects over time. In addition, the study investigates the relative strength of both processes, that is, whether the affective advertising effect is stronger than the influence of negative affect on the attention to advertising.

The Affective Impact of Political Communication

Previous research indicates that political campaign communication clearly aims at voters’ emotions (for an overview, see Kinder, 1994). For example, issue ads relative to image ads for political candidates enhance positive affective reactions such as pleasure in recipients (Thorson et al., 1991). In addition, support ads increase positive emotions while attack ads decrease pleasure (see also, Chang, 2001). Furthermore, the use of symbols, the depiction of negative events or issues like violence, drug use, crime, or riots elicit negative emotions (Brader, 2005; Brader et al., 2008; Sears, 2001). Survey evidence also indicates that the exposure to politics on TV has an affective impact on the audience. For instance, Pan and Kosicki (1994) found that exposure to TV news enhanced patriotic feelings and intensified affective responses to the Persian Gulf War. Furthermore, Holbert, Shah, and Kwak (2004) found that the use of TV news and crime dramas induces fear of crime in audience members.

A closer look at previous research substantiates the notion that political media content is more likely to induce negative affective responses than positive affect (Chang, 2001; Cho et al., 2003). One reason for this negativity bias may be due to the negativism in political news or advertising (Ansolabehere & Iyengar, 1997; Cappella & Jamieson, 1997). Irrespective of the specific causes of negativity, prior research suggests that the content and form of political communication exert a strong impact on viewers’ emotions and that this communication elicits more negative than positive affective reactions.

Ethnic Minorities in Political Campaigns

Political ads also include racial cues or depict ethnic minorities to influence the public (Kinder & Sanders, 1996; Mendelberg, 1997; Valentino, Traugott, & Hutchings, 2002). Even if they are subtly embedded in campaign content, racial cues are likely to elicit strong negative affective reactions in recipients (Brader et al., 2008; Esses et al., 2008; Peffley et al., 1996). Specifically, the portrayal of members of ethnic minorities as immoral individuals who aim at cheating the welfare system arouses contempt and lack of admiration (Esses et al., 2008). In addition, anxiety toward ethnic minorities results “when negative consequences of immigration are paired with a stigmatized outgroup” (Brader et al., 2008). Finally, the use of racial cues in the 1988 U.S. election campaign elicited racial resentment which in turn increased the approval of Bush (Kinder & Sanders, 1996, p. 241).

Intergroup emotion theory offers an explanation for why people experience negative emotions toward the negative portrayals of ethnic minorities (Mackie, Devos, & Smith, 2000). Accordingly, ethnic minorities can be considered as an outgroup that threatens the ingroup. Such threats can be realistic or symbolic in nature. Realistic threats refer to material resources, such as money or jobs that the outgroup may receive at the expense of ingroup members. Symbolic threats refer to cultural values of the ingroup to which the outgroup does not subscribe. The more individuals perceive ethnic minorities as a threat, the more they will perceive negative affective reactions (Cottrell & Neuberg, 2005; Mackie et al., 2000). Specifically, threats to physical safety and the cultural values of the ingroup are likely to increase anxiety and fear, and threats to material resources (e.g., the welfare system) are likely to elicit anger in members of the ingroup (Cottrell & Neuberg, 2005; Cottrell, Richards, & Nichols, 2010). Taken together, political actors often play the race card in campaigns to elicit voters’ negative emotions.

Affect-Induced Selective Exposure and Attention to Political Communication

Previous studies also found that negative affect can increase the attention and exposure to campaign stimuli. Experiments substantiate the notion that fear-arousing communication is likely to enhance people’s attention to political cues (Nabi, 2003; Valentino et al., 2008) and the processing and recall of such information (Bradley et al., 2007; Lang, 1991; Reeves, Newhagen, Maibach, Basil, & Kurz, 1991). The allocation of attention and processing capacities is an automatic process that is triggered by formal features or the emotional tone of an advertising stimulus (Lang, 1990, 2000). When individuals experience negative emotions such as fear or anger in response to mediated stimuli, then their attention automatically increases and focuses on the stressor that elicited the affective state. This allocation of attention enables individuals to deal quickly with environmental stimuli (Frijda, 1986).²

In media consumption settings, this means that the arousal of negative emotions is associated with increased attention to mediated stimuli related to that emotion. This has also been shown in the context of political ads (Bradley et al., 2007; Lang, 1991). In addition, the fear-appeal literature suggests that fear predisposes people to attend to information that is relevant to the threat that elicited the current fear (Witte & Allen, 2000). A similar perspective is adopted by the affective intelligence framework (Marcus & MacKuen, 1993). This model assumes that emotions have important adaptive functions. This framework distinguishes between enthusiasm as a positive emotion and anxiety as a negative emotion. These emotions function as markers of interdependent emotion systems. Positive affect as a reaction of the disposition system indicates that the environment is safe and that the individual does not have to take any action. As a consequence, individuals can rely on routines and learned behavior. In contrast, negative affect or anxiety as a reaction of the surveillance system indicates a possible threat in the environment. In this situation, negative affect shifts the attention of individuals to the threatening event and prepares them for possible action. According to the affective intelligence framework, people who experience anxiety in a political campaign do not rely on routine judgment formation anymore. In contrast, anxious voters increase their attention to campaign information in order to learn more about the source of the threat. Consistent with their hypotheses, the authors found that anxiety enhanced involvement and learning in the U.S. election campaign.

Taken together, previous research clearly suggests that negative affective responses to campaign cues increase the attention to additional information that is relevant to the affective reaction. The use of racial cues in campaigns, for example, the negative portrayal of ethnic minorities will elicit negative affective reactions toward members of this group. Theses affective reactions, in turn, can increase the attention to additional information about ethnic minorities. These assumptions imply reciprocal dynamics of communication effects and increased attention that has not been previously investigated in the realm of political campaigns.

Reciprocity of Affective Campaign Effects and Selectivity in Attention

The previous sections indicate that there might be reciprocal relationships between the effect of political ads on negative emotions and emotion-induced increased attention to political ads. Previous research provides partial evidence of such dynamics in the context of 9/11 (Cho et al., 2003). In a two-wave panel survey, they found that TV news use elicited negative emotions like fear and anger. In an additional step, they demonstrated that these negative emotions predicted the use of TV news. Based on their findings the authors concluded “that the relationship between television news use and respondents’ negative emotions is reciprocal” (Cho et al., 2003, p. 323). However, it is almost questionable whether this effect can be extrapolated to political campaign contexts. On one hand, 9/11 was an exceptional event, even causing traumatic experiences in the audience (Ahern et al., 2002). Thus, these affective reactions may be more intense than emotions elicited by campaign cues. On the other hand, the news coverage of 9/11 was certainly not motivated by a political strategy like political ads in a campaign context. Therefore, these caveats preclude a generalization of the effects to political campaign contexts.

Second, the two-wave panel survey by Cho et al. (2003) was not designed as a direct test of self-reinforcing spirals processes. Therefore, no conclusion can be drawn as to whether the reciprocal effects that they observed would have occurred in subsequent periods. In other words, it is not clear whether the media-induced negative affective reactions in the second panel wave would have elicited additional attention to TV news, which, in turn, may have produced additional negative affective reactions in viewers. The minimum requirement for the study of mutually reinforcing processes of campaign effects and attention is a three-wave panel. This design provides the researcher with two periods in which media effects and selective attention can be observed.

This assumption is also put forward by Slater (2007). The RSM proposes a synthesis of “the process of media selection and media effects into a more comprehensive model” (Slater, 2007, p. 281). Accordingly, attention to mediated information is seen as an endogenous variable that affects an audience characteristic, for example, negative emotions, but that is also affected by this audience characteristic. These reciprocal influences between attention to information and the audience characteristic evolve over time. As a consequence of this reciprocity, media effects and selectivity can produce cumulative effects over time. Self-reinforcing spirals processes producing cumulative effects over time are most likely when self-regulating processes stemming from external forces do not intervene (Slater, 2007). In the present context, communication cues that undermine the impact of campaign ads on negative affect can represent such an external force (e.g., a countercampaign). Such a model has been tested in other domains of media effects research, for example, sexual satisfaction or behavior (Peter & Valkenburg, 2009), social risk judgments (Slater & Rasinski, 2005), or aggressiveness (Slater, Henry, Swaim, & Anderson, 2003). However, to date there is no direct test of the RSM in a political campaign context.

Therefore, the aim of the present research is to test whether the influence of political ads on negative affective reactions toward ethnic minorities and the increase in attention to political ads that is due to these negative emotions produce cumulative effects over time. There are two important reasons for an integration of the processes of media selection and media effects into a single framework. First, a test of reinforcing spirals processes can shed light on the relative strength of media effects and the effects of selectivity. More to the point, results from a RSM can reveal whether people’s attention to political information is driven by negative affect that is induced by campaign cues. In this vein, both media selection and effects are largely determined by political campaign cues. Conversely, it may be the case that people’s initial selectivity determines the extent to which they are emotionally affected by a political campaign. Thus, the RSM does not only allow the modeling of cumulative effects but also allow an assessment to what extent certain phenomena are initialized or determined by the cues emanating from the campaign or by internal motivations of audience members. This question is still one of the most pressing questions in research on media effects and gratifications (Slater, 2007).

Second, the substantive conclusion that can be drawn from such a model is whether the interplay of campaign-induced negative affect toward ethnic minorities and the affect-induced attention to political ads produce cumulative effects over time. In this specific case, the result of such cumulative effects would be an increase of negative emotions toward ethnic minorities. These negative emotions, in turn, have been shown to function as an important predictor of the discrimination of ethnic minorities (Talaska, Fiske, & Chaiken, 2008). Thus, applying the RSM to a political campaign in which racial cues are embedded will contribute to the understanding of an important social phenomenon, that is, the perpetuation of negative affective reactions toward and the discrimination of ethnic minorities.

Thus, the previous sections suggest that the relationship between negative affective reactions toward ethnic minorities and attention to political cues is reciprocal in nature. In addition, in the course of a political campaign this reciprocity is likely to produce cumulative campaign effects that can have huge consequences with respect to how ethnic minorities are evaluated or treated. The RSM offers an integrative framework for synthesizing such processes of campaign effects and attention.

Hypotheses

The study aims at testing the reciprocal influences between the effects of political ads on affective reactions toward minorities and the impact of negative affectivity toward that group on the attention to political ads. As previously shown, the negative portrayal of ethnic minorities elicits negative affective reactions toward the members of these groups. Therefore, the first hypothesis states as follows:

Hypothesis 1 (H1): Political ads increase negative affective reactions toward ethnic minorities.

In addition, research also shows that such negative affective reactions can heighten attention to and information seeking in campaigns. Therefore, the second hypothesis states as follows:

Hypothesis 2 (H2): Negative affective reactions toward ethnic minorities enhance the attention to campaign information.

The combination of these hypotheses indicates a reinforcing spirals process and suggests that negative affective reactions and the attention to political ads reinforce each other in the course of a political campaign.

A reinforcing spirals process depends on the fact that the influences assumed before can be observed more than once in the course of a campaign (Slater, 2007). An interesting question that has not been addressed so far is whether the influence of negative affect on attention to political ads and the affective influence of political ads change in strength over time. For example, it may be that cues in a campaign elicit stronger emotions as the campaign progresses to a decisive or “hot” phase. Thus, as the volume of ads increases, affective reactions are more likely to be elicited (Pan & Kosicki, 1997). Similarly, it can be expected that a growing campaign volume captures the attention of people that did not follow the political campaign before. According to this reasoning, the effects hypothesized in H1 and H2 should increase as the campaign progresses.

The alternative hypothesis suggests that the effects are stronger at the beginning than at the end because the advertising effects simply wear out over time (Craig, Sternthal, & Leavitt, 1976). Political messages at the beginning of a campaign are likely to be more powerful because people are not aware of competing arguments and are not certain in their decision. Therefore, they may be more susceptible to campaign effects. In addition, recipients of a political campaign get accustomed to campaign messages. This habituation is likely to reduce the power of campaigns to influence the public. Finally, given that people have not paid attention to the campaign before, the attention-grabbing power of campaign cues may be stronger at the beginning than at the end of the campaign. Given that people show some interest in a given campaign it may be more difficult to elicit additional attention to campaign cues later on. As previous research does not offer straightforward rationale with respect to the temporal development of the effects, we formulate the following research questions:

Research Question 1 (RQ1): Is the effect of attention to political advertising on negative affect stronger at the end of the campaign than at the beginning?

Research Question 2 (RQ2): Additionally, it is asked whether the affect-induced attention to political ads increases over time.

The previous reasoning referred to the absolute strength of media effects and selectivity and to the temporal development. An additional issue is whether the influence of people’s campaign attention that is due to negative affects is stronger or weaker than the influence of political ads on voters’ negative affective reactions. Ultimately, this amounts to the question of whether voter reactions are determined by external influences such as campaign ads or whether campaign effects are limited by internal forces such as individual selectivity in attention. This is an important question because, when voter reactions are determined by campaign cues, then this is a potential gateway to manipulation. However, selectivity in the use of campaign cues would shield voters from manipulative campaign strategies. Given that previous research is also not conclusive about this question, we formulate the following research question:

Research Question 3 (RQ3): What is the relative strength of the attention to political ads on negative affect compared with the influence of negative affect on the attention to political ads?

Method

Context of the Study

In order to test the RSM, we rely on data from a three-wave panel study conducted in the context of a campaign dealing with the asylum law restriction in Switzerland. As in many other immigrant-receiving countries, asylum seekers and minorities from non-European countries have a negative image (Schneider, 2008). Unsurprisingly, immigration policy is one of the most important issues on the agenda in Switzerland (Lavenex, 2007). To date, several referendums have been held that consistently aimed at enhancing migration control, reducing immigration, and deterring asylum seekers. The public approved most of these restrictions.

The present referendum dealt with the restriction of the asylum law. In the debate, the political left (Social Democrats and Green Party) attacked a proposal supported by liberal and right-wing parties. Especially, the populist right argued that most refugees were bogus asylum seekers and criminals. Accordingly, stricter laws should be adopted to deter the abuse by foreign scroungers and to deal with crime (e.g., the reduction of social security benefits for refugees whose applications were dismissed). Especially, the ads of the pro camp have the potential to arouse negative emotions. The issue of crime is likely to be perceived as a threat to the physical safety of the ingroup. As a consequence of this safety threat, people are most likely to experience fear and anxiety. The issue of abuse is a cue that can stir people’s anger, because it refers to financial resources that are in danger when ethnic minorities try to obtain welfare benefits by fraud (Cottrell et al., 2010). In contrast, the political left along with humanitarian organizations, churches, or labor unions argued that the restriction of the asylum law was inefficient and contradicted the humanitarian tradition of Switzerland.³ The referendum took place on September 24, 2006, and 68% of the people approved of the asylum law restriction.

Panel Survey

The sample for the three-wave panel survey was recruited by random digit dialing. The response rate (RR3) was .59.⁴ The questionnaire was programmed for an application of computer-assisted telephone interview (CATI). For the first wave (W1), respondents (N = 1,725) were contacted between July 4 and July 20, 2006. The second wave (W2) followed between August 28 and September 2 (N = 1,415). The final wave took place just after the referendum, that is after September 24 (N = 1,094, 52% women, mean age = 48.18, SD = 17.11; 59.8% of our respondents held lower educational qualifications, i.e., elementary school and subsequent vocational education but no academic education in high schools, colleges, or universities).⁵

Measurement

Given that the campaign deals with the restriction of the asylum law, group-related emotions are measured. Specifically, negative affective reactions that respondents experienced toward asylum seekers are assessed in all panel waves with a three-item scale (fear, anxiety, and anger; Cronbach’s alpha = .73 in the first, .76 in the second, and .78 in the third waves). The interviewees report the strength of their feelings toward asylum seekers on a scale ranging from 1 (not very strong) to 5 (very strong). In addition, attention to political advertising is assessed with two items. The respondents are asked to rate the importance of advertising posters and ads in the newspaper to get campaign information (1 = not very important to 5 = very important; Pearson’s r = .57 in the first, .64 in the second, and .65 in the third waves). In Switzerland, political TV ads are not allowed. Therefore, no questions ask about that. As control variables, demographics (age, sex [0 = male, 1 = female], and education [0 = low education to 1 = highest education]) and ideology (left–right self-positioning, 1 = left to 10 = right) are measured in the first panel wave.

Panel Attrition

Those respondents who could not be reinterviewed do not differ significantly in terms of negative affect or attention to political ads from respondents who remained in the panel (see Table 1). However, the proportion of men who dropped out is higher than the proportion of women. In addition, the reinterview rate of younger people is lower than the proportion of older people. There are no differences in education or ideology of the sample in the first compared with the last panel wave. Thus, panel attrition does not affect the crucial variables, that is, negative affective reactions toward asylum seekers and attention to political ads.

Table 1.

Panel Attrition: Comparison of Respondents Missing in Last Wave and Respondents Who Dropped Out

	Not missing in W3	Missing in W3
	M (SD)/Percentage	M (SD)/Percentage
Negative affect W1	2.44 (1.00)	2.39 (1.00)
Negative affect W2	2.48 (1.02)	2.49 (1.05)
Attention to political ads W1	1.96 (0.96)	2.01 (0.94)
Attention to political ads W2	2.09 (0.96)	2.10 (0.99)
Sex (percentage female)	54.7^a	48.0^b
Age	50.44 (17.13)^a	45.17 (16.58)^b
Ideology (left–right self-positioning)	6.06 (2.05)	5.96 (2.05)
Education (percentage lowest degree)	60.2	59.3

Note: Numbers in a row with a different superscript differ significantly, p < .05.

Data Analysis

In order to test self-reinforcing spirals of negative emotions and attention to ads a parallel-process model using structural equation modeling is appropriate (Bollen & Curran, 2006; see also Slater & Hayes, 2010 for an application of the RSM using parallel-process modeling). Negative affect toward asylum seekers and attention to political ads are treated as latent variables. Then, a linear growth process for each construct is modeled to investigate the changes in negative affect and in attention to ads over time. Latent growth curve models involve the modeling of two growth factors, a latent intercept and a latent slope. The former represents the initial level of a construct, that is, the initial level of attention and negative affect. The latter captures the intraindividual change in these constructs. By convention, the paths from this intercept to the latent variables (negative affect and attention) are constrained to 1 (see Figure 1). The means of both intercepts are constrained to zero for identification purposes. A linear intraindividual growth process for attention to political ads and negative affect is achieved by fixing the paths from the slope to the latent variables to 0, 1, and 2, respectively. The growth factors are allowed to covary.

Figure 1.

A reinforcing spirals model of attention to political advertising and negative affect toward asylum seekers

Results

In general, the model showed a good fit to the data (χ² = 128.57, df = 114, p = .17, comparative fit index [CFI] = .99, root mean square error of approximation [RMSEA] = .02, 90% confidence interval [CI] [.01, .03]). In order to facilitate the understanding of the findings, descriptive results related to the changes in negative affects and advertising attention are reported first. Then, the effects of the covariates are presented in brief. The final section is dedicated to hypotheses tests.

The Change of Negative Affects and Attention to Political Ads Over Time

First of all, the mean of the negative affect slope is significantly greater than zero (M = .054, p < .05) indicating that negative emotions toward asylum seekers became more negative over time. Obviously, the campaign increased people’s fear and anger toward asylum seekers. An inspection of the latent means confirmed this. The negative affect increased continuously from the first (M = 2.64, SD = .03) over the second (M = 2.67, SD = .03) to the last panel wave (M = 2.71, SD = .04). In addition, the variance of the negative affect slope that is significantly greater than zero (σ² = .041, p < .05) means that there are interindividual differences in the increase of negative affective reactions toward asylum seekers. In other words, the change in the emotions of some individuals deviates significantly from the assumed linear growth process.

Interestingly, the attention to political ads decreased from the first to third wave (M = –.079, p < .05). Accordingly, voters became less attentive to political ads as the campaign unfolded. An in-depth analysis indicated that the advertising attention developed in an inverted U-shape fashion with attention to ads increasing between the first (M = 2.00, SD = .03) and the second wave (M = 2.14, SD = .03) and then decreasing to a level that was lower in the last wave (M = 1.88, SD = .03) compared with the first one. In addition, there are individual differences in the attention growth process (σ² = .018, p < .05). The significant variance parameters of both intercepts indicate that there are individual differences in the initial level of the attention to ads (σ² = .412, p < .01) and negative affect (σ² = .583, p < .01).

The correlations among the growth factors (intercepts and slopes) shed light on the associations among the initial status and the growth trajectories. The negative correlation between the slope and the intercept of the attention to political ads (r = –.19, p < .05) indicates that people with a high initial attention to political ads increased their attention significantly less in the course of the campaign compared with people starting out with a low attention. Furthermore, the intercept of advertising attention is positively correlated with the slope of negative affect (r = .86, p < .01). In other words, people with a high initial attention to political ads showed a steeper increase in negative affective reactions toward asylum seekers than people with a low initial advertising attention. No other correlations among the growth factors achieved statistical significance.

Before presenting the tests of hypotheses, results referring to the effects of the covariates on the growth factors are reported (see Table 2). In short, female respondents experience more negative emotions toward asylum seekers and are more attentive to political ads in beginning of the campaign. Age exerts only an effect on the negative affect intercept, indicating that older respondents experience more negative emotions toward asylum seekers. More highly educated respondents show fewer initial negative emotions, a less steep increase in negative affects over time, and lower attention to political ads compared with respondents with lower educational qualifications. Finally, right-wing respondents feel more negative about asylum seekers and are more attentive to political ads at the beginning. However, they show a smaller decrease in attention to political ads.

Table 2.

Effects of Control Variables on the Growth Factors

	Intercept negative affect	Slope negative affect	Intercept attention to political advertising	Slope attention to political advertising
Sex	.13**	−.11	.11**	.03
Age	.10**	−.05	−.05	−.05
Education	−.24**	−.20**	−.12**	.03
Ideology	.33**	.11	.10**	−.11**

Note: Entries are standardized effects of control variables.

p < .05. **p < .01.

Self-Reinforcing Spirals Effects of Attention to Political Advertising and Negative Affect

The first hypothesis stated that the attention to political ads should elicit negative affective reactions. The findings are consistent with this hypothesis (see Table 3). Specifically, attention to political ads in the first wave produces negative affective reactions toward asylum seekers in the second wave. Similarly, attention to political advertising in the second panel wave enhances negative emotions in the final panel wave. This result confirms that campaign ads increase people’s negative feelings toward the minority group over time.

Table 3.

Reciprocal Influences of Attention to Political Advertising and Negative Affect Toward Asylum Seekers

Parameters	Estimates (SE)	Parameters	Estimates (SE)	z difference
Political advertising_t1 → Negative affect_t2	0.15 (.03)**	Political advertising_t2 → Negative affect_t3	0.15 (.03)**	0.33
Negative affect_t1 → Political advertising_t2	0.11 (.02)**	Negative affect_t2 → Political advertising_t3	0.07 (.03)**	1.72
z difference	1.23	z difference	2.7*

Note: Entries are unstandardized estimates; z difference represents the result of significance tests of the difference between parameters in a row and in a line, respectively.

p < .05. **p < .01.

In the second hypothesis, it was assumed that negative affective reactions should enhance attention to political ads. The findings indicate that this effect occurs in both periods: Negative affects toward asylum seekers in the first wave increase advertising attention in the second wave. In addition, negative affect in the second wave enhances the attention that people pay to political ads in the last wave. Thus, the findings confirm Hypotheses 1 and 2. In combination, the results show evidence of reinforcing spiral processes. More to the point, attention to political ads enhances negative affective reactions toward asylum seekers that, in turn, increase the attention to campaign ads. Then, these campaign ads induce additional negative affective reactions toward asylum seekers. These reinforcing processes produce cumulative media effects over time with the result of more negative affective reactions toward asylum seekers at the end of the campaign than at the beginning.

RQ1 and RQ2 addressed the temporal development of advertising effects and attention. More specifically, RQ1 asks whether the effect of attention to political ads on negative affect is stronger at the end of the campaign than at the beginning. The findings in Table 3 show that the affective impact of political ads is exactly the same in both periods. The second research question speculated about differences in the effect of affect-induced selectivity between the first and the second phase of the asylum law campaign. However, there is no difference in the influence that negative affect toward asylum seekers has on advertising attention in the first compared with the second period. In both periods, negative affective reactions enhance the attention to political ads in a similar fashion. The opposite effect, that is, the influence of negative affect on the attention to advertising also remains constant throughout the campaign.

Finally, the third research question asked about the relative strength of the influence of people’s selective attention to campaign ads that is due to negative affect compared with the impact of political ads on voters’ negative affective reactions. Results for the first phase of the campaign show that there is no difference between the advertising effect and selective attention effect. However, in the second phase the impact of attention to political ads on negative affect is stronger than the influence of negative affect on the attention to ads. This finding indicates that in the course of the political campaign the self-reinforcing spiral process is fueled or kept alive to a stronger extent by campaign cues, that is, political ads, than by internal forces, that is, enhanced attention to these ads.

Discussion

The present study demonstrated the reciprocal relation between attention to political ads and negative affective reactions in political campaigns. Specifically, attention to political ads depicting asylum seekers in a negative light directly elicited negative affective reactions toward that social group. This result corroborates prior research on the impact of racial cues on affective reactions of the public (Brader et al., 2008; Cottrell & Neuberg, 2005). In addition, the reverse dynamics were also shown, that is, that negative affective reactions, such as fear and anxiety, enhanced attention to campaign cues. This conforms to the results of extant research on affect-induced increase in attention to political information (Boyle et al., 2004; Cho et al., 2003; Valentino et al., 2008).

However, the present study goes beyond previous research by simultaneously modeling advertising effects and selection dynamics over time. The findings indicate that mutually reinforcing influences were at work. In other words, attention to political ads that was due to negative affect further enhanced these affective reactions toward asylum seekers. At the same time, initial advertising attention produced additional negative affective reactions that, in turn, increased attention to subsequent campaign advertising. Thus, the relationship between attention to political ads and negative emotions toward asylum seekers is bidirectional in nature. This is an important finding because it suggests that people’s attention to the campaign is a product of the affective reactions elicited by the campaign itself. By capturing people’s attention to subsequent ads campaign cues reinforced their effects on negative affective reactions toward asylum seekers. Thus, campaign effects perpetuated themselves over time and increased the negativity in affective reactions toward ethnic minorities. This is a major contribution to the field because it also shows that by continuously portraying minorities as outgroups political ads are a driving force in maintaining negative sentiments toward and the derogation of ethnic groups.

Consequently, the increase in affective negativity toward asylum seekers may have also contributed to the adoption of the asylum law restriction.

Once a negative affective reaction is activated by negative portrayals of asylum seekers, people exhibit functionally congruent behavior (Cottrell et al., 2010). Thus, when the portrayal of outgroups induces fear and anger, then voters are likely to act accordingly and vote in favor of policies that aim at a limitation of minority rights (Wirth, Schemer, & Matthes, in press). Previous research has convincingly shown that affective reactions toward minority groups influence policy opinions (Brader et al., 2008; Esses et al., 2008), the evaluation of political candidates (Kinder & Sanders, 1996; Mendelberg, 1997), and discrimination (Talaska et al., 2008). Thus, the arousal of negative emotions toward outgroups can have far-reaching political consequences.

Although the attention to ads continually increased negative emotions over time, this effect itself did not grow in size. Similarly, the impact of affect-induced attention to political ads remained fairly constant. These findings indicate that—in terms of systems theory—no positive feedback loops occurred that steadily increased attention or negative affects to extreme values. This is also the reason for the failure to find a significant covariance between the negative affect slope and the advertising attention slope. A significant covariance would have meant that both developmental processes go hand in hand. However, the present analysis can only demonstrate a shock-like effect, that is, that attention to political ads increased negative affective reaction at a later point in time. Put differently, political ads acted as a situational disturbance of the affect system. By eliciting negative affective reactions these cues destabilized the affect system temporarily and made people more attentive to campaign cues that continued to arouse negative affective reactions. However, there was no steady increase of negative affective reactions or a continuous increase of attention over time.

Two forces may have prevented a spinning out of negative affects: self-regulation or external forces (von Bertalanffy, 1968). Campaign cues from the opponents of the asylum law restriction may have functioned as external forces that reduced negative affects. Given that people paid more attention to the campaign due to the increase of negative affects they were also more likely to encounter ads from opponents of the asylum law restriction. Thus, heightened attention to political ads not only did produce more negative affect but also may have lead to exposure to campaign information that was likely to undo such effects.⁶

Self-regulating processes are those influences that the individuals can use to adjust and tune their amount of attention or their affective reactions. Even if negative affective reactions automatically increased attention to political advertising, the decision to act accordingly (e.g., search of and exposure to additional campaign information) is effortful and under the individual’s control. Individuals may have reduced their voluntary attention which, in turn, may have reduced the potential of negative ads to infinitely elicit negative affective reactions.

Another self-regulating process is the active management of the negative emotion. Although negative emotions may be a spontaneous response to political ads in the campaign individuals are able to regulate their affective reactions actively, for example, by avoiding cues that aroused the emotion, by suppressing emotional reactions, or by reappraising situations that elicited the affective reaction (Gross, 2007). This may have also prevented negative affective reactions from increasing to extreme values.

The impact of such internal and external constraints indicates that individuals in a campaign are open to a multitude of influences that may interfere in their effects. In addition, voters must be considered as more or less active message recipients that are able to regulate their attention to environmental stimuli and their affective reactions in response to these cues. Under these circumstances mutually reinforcing feedback loops are unlikely to occur. Positive feedback loops are most likely to be observed in closed systems that are more reactive in nature (von Bertalanffy, 1968).

However, people may vary in their openness to competing communicative influences or in their activity to regulate internal states. Thus, for some people who selectively avoid information that challenges their views or who do not actively engage in emotion regulation positive feedback loops of negative affect and advertising attention may occur. In the end, these assumptions must remain speculative because there is no means of testing them with the present data. Future research can directly put these hypotheses to an empirical test and shed more light on the nature of mutual influences of media effects and selectivity.

Another account for why the impact of affect-induced attention did not grow over time is due to the fact that the last panel wave took place when the attention to the campaign had already diminished to the initial level. Thus, attention to political ads did not increase steadily but in a curvilinear fashion. Even if negative affects had raised people’s level of attention to political ads, there would not have been any campaign stimuli available because the campaign was over. Obviously, three waves of panel data are necessary to demonstrate such reinforcing spirals process. However, more than three waves are desirable because the higher the temporal resolution, the higher the probability to detect an increase or decrease of advertising effects or affect-induced attention to the campaign over time.

A final important finding refers to the relative strength of both processes. Specifically, the impact of political advertising in the second panel wave on negative affective reaction toward asylum seekers in the third wave was stronger than the affect-induced attention to ads in the same period. Thus, obviously the spirals process is fueled to a stronger extent by racial cues in political ads than by people’s enhanced attention. Although the results show what the driving force of the spirals process is when the process is running, the findings do not shed light on what initialized the spirals process. In the first phase, the effect of advertising on negative affect was similar in size compared with the influence of negative affect on the attention to political ads. In other words, the survey is inconclusive as to whether campaign cues or initial attentiveness got the ball rolling at the beginning of the spirals process.

This difference in relative strength of the effects has also normative implications. As the findings suggest, the increase in attention of voters in the campaign is partly determined by the strategic communication that elicited negative affective reactions toward asylum seekers. Thus, by arousing negative emotions toward a minority group that was in the focus of the campaign, the political actors increased attention for the cues that elicited further negative emotions. This result implies that campaigners have powerful weapons in their arsenal to persuade and manipulate voters in political contests.

The present finding that the effect of advertising attention on affective reactions was stronger than the affect-induced selectivity in attention to ads may also be due to the lag between the panel waves. As the descriptive results have demonstrated, the attention to political advertising first increased and then decreased again below the initial level. This decrease in the final panel wave is mainly due to the choice of the last measurement occasion that took place after the referendum was held. After this date, no additional attention to political advertising was possible because the campaign was over.

The design of this study may also explain the small effects. The effects shown in the study are influences that survived a period of 4 weeks. Given these long interpanel intervals one may assume that, originally, these effects were stronger but diminished as the time passed by. An additional explanation for the effect sizes found in the present study refers to the fact that not all recipients are equally likely to react emotionally to the political campaign. From experimental research it is know that certain boundary conditions moderate affective influences or selectivity in attention or exposure to political communication. One important moderator may be the political knowledge of respondents (Dillard & Wilson, 1993). On one hand, knowledgeable voters are likely to correct or discount affective reactions in response to political messages. On the other hand, political sophistication can enhance the selectivity in media use. The reason is that political sophisticates have more knowledge about the sources that provide the most valuable information. It may be that these people avoid political campaign advertising that contains easy-to-understand information and rely more on media content that reports on background information. The results in Table 1 tentatively corroborate this assumption: Respondents with higher educational degrees showed less attention to political advertising at the beginning of the campaign compared with people with lower educational qualifications.

Given the caveats of the present research, future studies should address such reinforcing spirals processes in more detail. This refers to both the methodology and the substance of the investigation. With regard to the methodology, research should study the time lags in spiral processes in more detail. Here, the temporal resolution is the most crucial aspect. The greater the number of measurement occasions, the more insight researchers will get into self-reinforcing spirals processes. In addition, the investigation of possible moderators would also be a valuable avenue of research. With regard to the substance, it is suggested that scholars should investigate reinforcing spirals over longer periods in order to capture the impact of political campaigns on the maintenance of social stereotypes in society. For instance, the elicitation of negative emotions toward ethnic minorities over several campaigns may contribute to the discrimination against members of such groups. Given the trend toward permanent campaigns, the use of racial appeals in political contests is likely to increase negative affective reactions and discriminatory attitudes toward ethnic minorities in the long run. The reason is that when “specific emotion is experienced repeatedly over time in response to a particular group, this feeling will become associated with the mental representation of the group through the process of classical conditioning” (Smith & Mackie, 2006, p. 182). Thus, campaigns that make use of racial cues might thwart efforts aiming at integrating ethnic minorities and facilitate the segregation of members of such groups.

The RSM should also be replicated in other campaign contexts and tested in other domains. The present study dealt with the effects of print advertisements in a political campaign where no television ads were allowed. Thus, one may wonder whether the present findings generalize to televised campaigns, for example, to U.S. presidential elections. With two or more opposing camps or candidates, most elections are similar to the present context. Given that only the campaign volume is increased in televised campaigns, it is likely that similar reciprocal influences of media selectivity and effects can be observed in televised political campaigns. Future research will have to provide empirical evidence of the general nature of such processes. The studies that have been conducted so far provide evidence that the framework proposed by Slater (2007) is more appropriate to explain cumulative communication effects than unidirectional analyses alone. Thus, the RSM encourages the integration of media effects and audience choice approaches in order to capture social phenomena more appropriately.

Future research can also make use of more diverse measures of attention to campaign information. The present study relied on a two-item measure to assess attention to ads. This measure offers a somewhat limited basis to study the multifaceted communication processes that may occur in the course of a campaign. Upcoming research should also consider additional communication channels that may increase or decrease reinforcing spirals process over time. For instance, attention to news media reporting of the campaign is an important source of information that in the present case may have decreased the effects of the advertising campaign. Furthermore, interpersonal communication may also be an important moderator of the effects that were observed. Homogeneous interpersonal networks and networks that provide people with a view that is congruent with the dominating perspective in advertisements is likely to enhance self-reinforcing spirals processes that were hypothesized in the present context. Cross-cutting interpersonal communication, however, may function as an external force that reduces cumulative effects that are due to self-reinforcing spirals processes. All these limitations could serve as directions for future research on self-reinforcing spirals processes.

Footnotes

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

This research was supported by the Swiss National Science Foundation.

Notes

References

Ahern

Galea

Resnick

Kilpatrick

Bucuvalas

Gold

Vlahov

(2002). Television images and psychological symptoms after the September 11 terrorist attacks. Psychiatry, 65, 289-300.

Ansolabehere

Iyengar

(1997). Going negative: How political advertising shrink and polarize the electorate. New York, NY: Free Press.

Bollen

K. A.

Curran

P. J.

(2006). Latent curve models. A structural equation perspective. Hoboken, NJ: Wiley.

Boyle

M. P.

Schmierbach

Armstrong

C. L.

McLeod

D. M.

Shah

D. V.

Pan

(2004). Information seeking and emotional reactions to September 11 terrorist attacks. Journalism & Mass Communication Quarterly, 81, 155-167.

Brader

(2005). Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions. American Journal of Political Science, 49, 388-405.

Brader

Valentino

N. A.

Suhay

(2008). What triggers public opposition to immigration? Anxiety, group cues, and immigration threat. American Journal of Political Science, 52, 959-978.

Bradley

S. D.

Angelini

J. R.

Lee

(2007). Psychophysiological and memory effects of negative political ads. Journal of Advertising, 36, 115-127.

Bundesamt für Statistik. (2011). Statistisches Lexikon der Schweiz [Statistical encyclopedia of Switzerland]. Neuchâtel, Switzerland: Author.

Cappella

J. N.

Jamieson

K. H.

(1997). Spiral of cynicism: The press and the public good. Oxford, UK: Oxford University Press.

10.

Chang

(2001). The impacts of emotion elicited by print political advertising on candidate evaluation. Media Psychology, 3, 91-118.

11.

Cho

Boyle

M. P.

Keum

Shevy

M. D.

McLeod

D. M.

Shah

D. V.

Pan

(2003). Media, terrorism, and emotionality: Emotional differences in media content and public reactions to the September 11th terrorist attacks. Journal of Broadcasting and Electronic Media, 47, 309-327.

12.

Cottrell

C. A.

Richards

D. A. R.

Nichols

A. L.

(2010). Predicting policy attitudes from general prejudice versus specific intergroup emotions. Journal of Experimental Social Psychology, 46, 247-254.

13.

Cottrell

C. A.

Neuberg

S. L.

(2005). Different emotional reactions to different groups: A sociofunctional threat-based approach to “prejudice.” Journal of Personality and Social Psychology, 88, 770-789.

14.

Craig

C. S.

Sternthal

Leavitt

(1976). Advertising wear out: An experimental analysis. Journal of Marketing Research, 13, 365-374.

15.

Dillard

J. P.

Wilson

B. J.

(1993). Communication and affect: Thoughts, feelings, and issues for the future. Communication Research, 20, 637-646.

16.

Esses

V. M.

Veenlivet

Hodson

Mihic

(2008). Justice, morality, and the dehumanization of refugees. Social Justice Research, 21, 4-25.

17.

Frijda

N. H.

(1986). The emotions. Cambridge, UK: Cambridge University Press.

18.

Gross

J. J.

(Ed.). (2007). Handbook of emotion regulation. New York, NY: Guilford.

19.

Holbert

R. L.

Shah

D. V.

Kwak

(2004). Fear, authority, and justice: Crime-related TV viewing and endorsements of capital punishment and gun ownership. Journalism & Mass Communication Quarterly, 81, 343-363.

20.

Kinder

D. R.

Sanders

L. M.

(1996). Divided by color: Racial politics and democratic ideals. Chicago, IL: University of Chicago Press.

21.

Kinder

D. R.

(1994). Reason and emotion in American political life. In Schank

R. C.

Langer

(Eds.), Beliefs, reasoning, and decision making (pp. 277-314). Hillsdale, NJ: Erlbaum.

22.

Kriesi

(Ed.). (in press). Political communication in direct democratic campaigns: Enlightening or manipulating? New York, NY: Palgrave Macmillan.

23.

Lang

(1990). Involuntary attention and physiological arousal evoked by structural features and emotional content in TV commercials. Communication Research, 17, 275-299.

24.

Lang

(2000). The limited capacity model of mediated message processing. Journal of Communication, 50, 46-70.

25.

Lang

(1991). Emotion, formal features, and memory for televised political advertisements. In Biocca

(Ed.), Television and political advertising: Psychological processes (Vol. 1, pp. 221-243). Hillsdale, NJ: Erlbaum.

26.

Lavenex

(2007). Policies of immigration, asylum, and integration. In Klöti

Knoepfel

Kriesi

Linder

Papadopoulos

Sciarini

(Eds.), Handbook of Swiss politics (pp. 621-644). Zürich, Switzerland: Neue Zürcher Zeitung.

27.

Mackie

D. M.

Devos

Smith

E. R.

(2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602-616.

28.

Marcus

G. E.

MacKuen

M. B.

(1993). Anxiety, enthusiasm, and the vote: The emotional underpinnings of learning and involvement during presidential campaigns. American Political Science Review, 87, 672-685.

29.

Mendelberg

(1997). Executing Hortons: Racial crime in the 1988 Presidential campaign. Public Opinion Quarterly, 61, 134-157.

30.

Nabi

R. L.

(2003). Exploring the framing effects of emotion: Do discrete emotions differentially influence information accessibility, information seeking, and policy preference? Communication Research, 30, 224-247.

31.

Pan

Kosicki

G. M.

(1994). Voters’ reasoning processes and media influences during the Persian Gulf War. Political Behavior, 16, 117-156.

32.

Pan

Kosicki

G. M.

(1997). Priming and media impact on the evaluations of the president’s performance. Communication Research, 24, 3-30.

33.

Peffley

Shields

T. G.

Williams

(1996). The intersection of race and crime in television news stories: An experimental study. Political Communication, 13, 309-327.

34.

Peter

Valkenburg

P. M.

(2009). Adolescents’ exposure to sexually explicit internet material and sexual satisfaction: A longitudinal study. Human Communication Research, 35, 171-194.

35.

Reeves

Newhagen

J. E.

Maibach

Basil

Kurz

(1991). Negative and positive television messages: Effects of message type and context on attention and memory. American Behavioral Scientist, 34, 679-694.

36.

Schneider

S. L.

(2008). Anti-immigrant attitudes in Europe: Outgroup size and perceived ethnic threat. European Sociological Review, 24, 53-67.

37.

Sears

D. O.

(2001). The role of affect in symbolic politics. In Kuklinski

J. H.

(Ed.), Citizens and politics: Perspectives from political psychology. (pp. 14-40). Cambridge, UK: Cambridge University Press.

38.

Slater

M. D.

(2007). Reinforcing spirals: The mutual influences of media selectivity and media effects and their impact on individual behavior and social identity. Communication Theory, 17, 281-303.

39.

Slater

M. D.

Hayes

A. F.

(2010). The influence of music television viewership on changes in cigarette use and association with smoking peers: A social identity, reinforcing spirals perspective. Communication Research, 37, 751-773.

40.

Slater

M. D.

Henry

K. L.

Swaim

R. C.

Anderson

L. L.

(2003). Violent media content and aggressiveness in adolescents: A downward spiral model. Communication Research, 30, 713-736.

41.

Slater

M. D.

Rasinski

K. A.

(2005). Media exposure and attention as mediating variables influencing social risk judgments. Journal of Communication, 55, 810-827.

42.

Smith

E. R.

Mackie

D. M.

(2006). It’s about time: Intergroup emotions as time-dependent phenomena. In Brown

Capozza

(Eds.), Social identities: motivational, emotional, cultural influences (pp. 173-187). New York, NY: Psychology Press.

43.

Talaska

C. A.

Fiske

S. T.

Chaiken

(2008). Legitimating racial discrimination: Emotions, not beliefs, best predict discrimination in a meta-analysis. Social Justice Research, 21, 263-296.

44.

Thorson

Christ

W. G.

Caywood

(1991). Effects of issue-image strategies, attack and support appeals, music, and visual content in political commercials. Journal of Broadcasting and Electronic Media, 35, 465-486.

45.

Valentino

N. A.

Hutchings

V. L.

Banks

A. J.

Davis

A. K.

(2008). Is a worried citizen a good citizen? Emotions, political information seeking, and learning via the Internet. Political Psychology, 29, 247-273.

46.

Valentino

N. A.

Traugott

M. W.

Hutchings

V. L.

(2002). Group cues and ideological constraint: A replication of political advertising effects studies in the lab and in the field. Political Communication, 19, 29-48.

47.

von Bertalanffy

(1968). General systems theory: Foundations, development, applications. New York, NY: George Braziller.

48.

Wirth

Schemer

Matthes

(in press). The impact of positive and negative affects in direct-democratic campaigns. In Kriesi

(Ed.), Political communication in direct democratic campaigns: Enlightening or manipulating? (pp. 205-224). New York, NY: Palgrave Macmillan.

49.

Witte

Allen

(2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health Education & Behavior, 27, 591-615.