The game frame and political efficacy: Beyond the spiral of cynicism

Abstract

Several observational and experimental studies have confirmed the ‘spiral of cynicism’ hypothesis: the tendency of the news media to cover politics through a game frame, which focuses on political strategy instead of political issues, leads to cynicism about election campaigns and politicians among the electorate. However, such cynicism may in itself be somewhat inconsequential, and so this article suggests that we move beyond cynicism regarding specific electoral campaigns or politicians, and that we turn our attention towards political efficacy. This is done in an empirical study, which is based on a survey among the electorate and a content analysis of political coverage in newspapers. The study shows that exposure to the game frame is indeed associated with lower levels of internal efficacy, even when controlling for potentially confounding variables.

Keywords

Cynicism efficacy framing game frame media effects

Introduction

Modern democracy is mediated democracy. The mass media are a vital source of information and vehicle of communication between the governors and the governed, and news media coverage of politics largely determines what information the citizens receive about these issues (Cook, 1998; Strömbäck, 2008). This article looks into the news media’s tendency to frame politics as a game. When applying this game frame, the media focus on the strategy, popularity and power of political parties and politicians. Numerous studies have shown that the game frame and similar frames are widely used by the media (e.g. Benoit et al., 2005; Binderkrantz and Green-Pedersen, 2009; Strömbäck and Aalberg, 2008; Strömbäck and Dimitrova, 2006; Strömbäck and Shehata, 2007). Furthermore, Cappella and Jamieson (1997), and the studies that have built on and elaborated upon this study, have shown that this frame often leads to political cynicism among the public (e.g. de Vreese, 2004; de Vreese and Elenbaas, 2008; Valentino et al., 2001). For clarity, it should be noted that some studies have used the term ‘game frame’ more narrowly – denoting only coverage focusing on the electoral race – and used terms such as ‘strategic coverage’ or ‘strategy frame’, when describing media coverage. However, the conceptual overlap between the two frames is great and the terms have been used interchangeably (e.g. Lawrence, 2000).

The purpose of this study is to investigate the effects of the game frame outside an experimental setting and in a non-electoral context. This is done through a content analysis of Danish newspapers and a survey measuring media consumption and political attitudes in the Danish electorate. While this is not the first study to investigate the effects of the game frame outside the laboratory (e.g. de Vreese, 2005a; de Vreese and Semetko, 2002; Elenbaas and de Vreese, 2008), it adds to the field by replacing cynicism with political efficacy as the dependent variable. Additionally, the study revisits the concept of the game frame and our measurement of exposure to this frame.

The spiral of cynicism

Cappella and Jamieson (1997) has been a seminal contribution to research on the media game frame. Through experiments, the study established a convincing causal link between exposure to the game frame (termed ‘strategic coverage’) and increased cynicism. These results substantiated the previously levelled criticisms of the media’s propensity to use a game frame (Jamieson, 1992; Patterson, 1993).

Following Cappella and Jamieson (1997), several studies have largely corroborated their findings. The experiments done by Valentino et al. (2001) confirmed the link between the game frame and cynicism, although the impact of the game frame was found to be contingent on partisanship and education. Similarly, when exposing individuals to different versions of a news story about EU enlargement, de Vreese (2004) found that individuals exposed to a story with a game frame had a significantly higher level of political cynicism than individuals exposed to a story with an issue frame. Likewise, experiments investigating the effects of strategic metacoverage, a subcategory of game framing focusing on the interaction between politicians and reporters, have also shown that this type of coverage leads to political cynicism (de Vreese and Elenbaas, 2008).

Observational studies have yielded similar results. Using content analysis and a survey, de Vreese and Semetko (2002) showed that exposure to news with a game frame was associated with an increased level of cynicism during the 2000 Danish euro referendum campaign. Using the same methodology in a later cross-country study, de Vreese (2005a) also found exposure to the highly strategically framed news coverage of EU politics in the Netherlands to be associated with increased levels of cynicism in the Dutch population, while exposure to the more issue-oriented news coverage in Denmark was associated with decreasing levels of cynicism among the Danish population. Similarly, Elenbaas and de Vreese (2008) also found a relationship between exposure to the strategy frame and political cynicism among the young Dutch electorate, when studying campaign coverage of the referendum on the European Constitutional Treaty in 2005. Finally, although a recent study by Adriaansen et al. (2010) did not find an effect of strategic news on cynicism, this study did find that a ‘substantive’ frame had the opposite effect of the game frame on cynicism. In sum, the observational studies confirm the experimental findings: there is a causal link between the game frame and cynicism.

Theory: Is cynicism really that important?

It is easy to take for given that the spiral of cynicism literature uses cynicism as the dependent variable. Nevertheless, the following section argues that cynicism may not be the best or theoretically most relevant measure, when analysing the impact of the game frame.

When measuring cynicism, Cappella and Jamieson’s (1997: 266) primary scale consisted of 13 items of which a few were close to typical measures of political external efficacy, for example: ‘It makes a difference who is elected mayor in a large city’. However, most of the items measured attitudes towards a specific election campaign with statements such as ‘What [the candidates] said depended on who was listening’ and ‘The candidates were willing to do whatever it took to win’. These measures of cynicism have clearly impacted the field, as most of the later studies have explicitly used measures of cynicism derived from Cappella and Jamieson (de Vreese, 2004, 2005a; de Vreese and Elenbaas, 2008; de Vreese and Semetko, 2002; Elenbaas and de Vreese, 2008). There are, however, at least two drawbacks to this use of the concept of cynicism.

First, from a purely terminological perspective, one could argue that cynicism is normally understood as something far more serious and problematic than what these studies have actually investigated. Cynicism is typically defined as lack of trust (Agger et al., 1961), and measuring cynicism with evaluative statements about specific election campaigns and individual politicians arguably leads to a somewhat attenuated understanding of cynicism. This perspective on cynicism leads to a second, more serious, objection: cynicism might be of limited theoretical interest. As noted by de Vreese (2005a), there is little empirical evidence supporting the assumption that cynicism in itself is detrimental to political participation (see also Craig and Maggiotto, 1981; Hollander, 1997; Pollock, 1983). It has even been argued that cynicism can be seen as a reflection of an interested and critical public stance, which is ultimately healthy for democracy (Elenbaas and de Vreese, 2008). As already noted by Valentino et al. (2001), the theoretical breadth and practical impact of the game frame would be enhanced if the frame was shown to influence overall confidence in government and perhaps even turnout.

This study, therefore, investigates the effect on political efficacy, which has been shown to have a significant association with political participation (Pinkleton et al., 1998; Valentino et al., 2009). Political efficacy consists of two separate components, internal efficacy and external efficacy. Internal efficacy refers to beliefs about one’s own competence to understand and participate effectively in politics, while external political efficacy refers to beliefs about the responsiveness of governmental authorities and institutions to citizen demands (Niemi et al., 1991: 1407–1408).

Theoretically, there is good reason to expect that the game frame impacts external efficacy. As shown by Cappella and Jamieson, there is a significant correlation between cynicism and external efficacy (1997: 272). This is not surprising, given that external cynicism is conceptually very close to cynicism, and external efficacy can arguably be considered a form of generalised cynicism: if the game frame makes an individual cynical towards a specific campaign or politicians, this specific cynicism may also impact this individual’s general trust towards the political systems, i.e. external efficacy. The game frame might, for example, induce cynicism in the electorate by interpreting policy suggestions from a campaigning politician as guided solely by self-interest, and not as a response to the public interest (Adriaansen et al., 2010: 436). This campaign-specific cynicism will probably also by extension influence the electorate’s general perception of the political system as a field that is unresponsive to citizen interests. Therefore, the first hypothesis to be tested is:

H1: Exposure to high levels of the media game frame is associated with a lower level of external efficacy.

The theoretical link between the game frame and internal efficacy is less obvious. There are, however, at least two theoretical perspectives, which might explain such a link. First, a subjective feeling of low internal efficacy might simply be a result of having a low level of political knowledge: politically knowledgeable persons tend to have relatively high internal efficacy (Bennett, 1997), and the game frame might lead to lower levels of internal efficacy, because the focus on strategy, popularity, etc. crowds out substantial knowledge about politics. In other words, the game frame may take up time and attention that would otherwise have been used to consume news containing factual knowledge about substantive politics and thereby lead to lower levels of political knowledge, which again lead to lower levels of internal efficacy.

A second possible explanation for a link between the game frame and internal efficacy is based on the power of the media to determine what is deemed politically relevant. According to the ‘importance change’ model of framing, frames work ‘by affecting the perceived relative importance of different already accessible considerations’ (Slothuus, 2008: 5). Therefore, people exposed to a frame which highlights certain facts or values will tend to give those facts and values greater weight when forming an attitude or perception towards a certain topic (Chong and Druckman, 2007; Slothuus, 2008). By highlighting, for example, the results of an opinion poll, the game frame does not only convey facts about the popularity of specific parties. The game frame also stresses the importance of such facts and thereby conveys the impression that politics is mainly a strategic game for popularity and power, which again may decrease subjective belief of political competence in the following way: the majority of the electorate readily form opinions on salient political issues, e.g. for or against tax cuts (Zaller, 1992). Hence, it is relatively easy to achieve a subjective feeling of political competence insofar as political competence is the perceived ability to understand political issues well enough to form political opinions. However, according to the perspective of the game frame, politics is about reasons behind current fluctuations in opinion polls and interpretations of the hidden strategic reasons behind the actions of politicians, areas where the competence of the ordinary citizen is normally smaller than that of professional journalists and pundits. Not being able to participate in these analyses, the electorate is thereby, to use the terminology of Jamieson (1992) and Patterson (1993), turned into ‘spectators’ of politics. Hence, from this spectator perspective, the game frame lowers internal efficacy by creating a perception of politics as a field where the ordinary citizen has limited competence.

In sum: the knowledge perspective posits that the game frame leads to lower levels of internal efficacy through lower levels of factual political knowledge The spectator perspective claims that the game frame leads to lower levels of internal efficacy by creating a perception of politics where the factual knowledge that the electorate may hold is irrelevant and thereby insufficient to make the electorate competent. Both theoretical perspectives lead to the following hypothesis:

H2: Exposure to high levels of the media game frame is associated with a lower level of internal efficacy.

Data and methods

The study drew on a content analysis of five major national Danish newspapers and a survey with a probability sample of the Danish electorate.

Content analysis: Identifying the game frame

Determining the prevalence of the game frame in newspapers of course requires a way of indentifying this frame. Cappella and Jamieson (1997) never formally defined the frame, but instead described it by reference to the characterisation of the ‘game schema’ in Patterson (1993) and particularly to the characterisation of ‘strategy coverage’ in Jamieson (1992):

Jamieson argues that strategy coverage is marked by several features: (1) winning and losing as the central concern; (2) the language of wars, games, and competition; (3) a story with performers, critics, and audience (voters); (4) centrality of performance, style, and perception of the candidate; (5) heavy weighing of polls and the candidates standing in them. (Cappella and Jamieson, 1997: 33)

It is noteworthy that the news material used in the experiments by Cappella and Jamieson (1997) did not adhere strictly to these indicators, as the examples of news coverage with an issue frame actually contained indicators of the game frame, e.g. references to polls, the language of games and focus on perception of the candidate:

Democrat Rendell has the chance tonight to cement voter approval on this issue. Polls show [. . .] But Rendell has seized the upper hand [. . .] may play to Rendell’s advantage and he seeks to impress voters with his determination [. . .] Dennis Wesley, a former Republican, says he can win big if he scores in the Black Community [. . .] The finish line for the race is almost here. (Cappella and Jamieson, 1997: 247–253)

The characterisation of the game frame made by Jamieson (1992) and Cappella and Jamieson (1997) has, nevertheless, impacted the literature significantly. The characterisation has often been treated as an actual definition, and observational studies have measured the prevalence of the game frame by coding on a presence–absence basis for these, or slightly modified, indicators (de Vreese, 2005a, 2005b; de Vreese and Semetko, 2002; Elenbaas and de Vreese, 2008).

Describing a frame through the presence of such indicators is rather straightforward in experimental studies, where it is sufficient to identify or construct news messages that are highly typical examples of the game frame.¹ However, the use of these presence–absence indicators in observational studies has the drawback that they might not cover all articles that apply a game frame. Conversely, articles might contain these indicators but not really focus on the game. Another approach when measuring the game frame is to code the dominant frame in each article, also known as the metaframe. When coding this way, articles are either coded as having a game frame, an issue frame or, sometimes, a mixed frame (Lawrence, 2000; Strömbäck and Aalberg, 2008; Strömbäck and Dimitrova, 2006). As shown by Strömbäck and Van Aelst (2010), the choice between these two coding approaches can have a strong impact on the results: articles which are clearly using a meta game frame do not always contain indicators such as those mentioned by Cappella and Jamieson (1997). And conversely, articles containing an indicator of the game frame do not always apply the meta game frame. Therefore, this study measured the metaframe of newspaper articles, employing the following definition:

The issue frame refers to political news that focuses on policy, policy issues, the policy positions on these issues, real life conditions with relevance for issue positions and the effects of enacted policies and initiatives on conditions outside the political sphere. In contrast, the game frame refers to political news that frames politics as a game in which the political parties and politicians compete. The game frame focuses on the strategy and motives behind policy positions, and the popularity and power of political parties and politicians. The game frame focuses on the effects of policies and initiatives inside the political sphere, for example whether a policy or policy position will improve the popularity or power of a politician or a political party.

The definition is a modified and elaborated version of the definition found in Strömbäck and Aalberg (2008). The definition explicates that the game frame is conceptually juxtaposed with issue framing, as framing in itself is ‘inescapable’ (Strömbäck and Dimitrova, 2006; although see Van Gorp, 2010). It is worth stressing that both frames can contain criticism and conflict, so conflict does not define any of the two frames (Lawrence, 2000).

The analysis measured the prevalence of the game frame in articles from five major national Danish newspapers: JyllandsPosten (JP), Berlingske Tidende (Be), Ekstra Bladet (EB), BT and MetroXpress (MX). Selecting these five newspapers had two advantages. First, they are among the most read newspapers in Denmark. Second, this selection of newspapers includes both traditional morning papers (JP and Be), tabloids (EB and BT) and a free daily (MX). The analysis was based on 301 articles about politics in the newspapers’ main sections, covering a month leading up to the conducted survey.²

Coding instructions required that each article was classified in one of four categories: (1) ‘Article is exclusively or primarily using an issue frame’; (2) ‘Article is using a mixed frame, equally using the issue frame and the game frame’; (3) ‘Article is exclusively or primarily using a game frame’; or (4) ‘Cannot be coded – Article is neither game frame nor issue frame’.³ The coding was done manually by reading each article and subsequently classifying it according to the definitions of game frame and issue frame, respectively. To assist the coder, the coding manual gave examples of questions which are often asked or answered by the two frames (as suggested by Van Gorp, 2010). For example, writing about policy actors, news with an issue frame will often answer the (implicit) question: ‘what is their policy position (ideology) and specific policy proposals for the future?’, while the game frame will answer questions such as ‘how popular are they?’⁴

To test for intercoder reliability, a subsample of 48 randomly selected articles was coded by two independent coders. The test yielded a Krippendorff’s α = .82 (Krippendorff, 2004), with a 95% confidence interval of .72 ≤ α ≤ .92 (Hayes and Krippendorff, 2007). Kappa was .74 (with linear weighting).

Survey data

The study utilised data generated in a recruitment experiment for a web panel. While the primary aim of this experiment was to investigate recruitment techniques, the results of the survey turned out to be appropriate for this study. A sample of 2,500 individuals were drawn from the Danish Civil Registration System, and were subsequently invited, through letter, email, phone and text messages (sms) to participate in a web survey. A total of 394 respondents completed the survey, yielding a response rate of 16% (for further details, see Hansen and Pedersen, 2012).

The dependent variables were measured on two additive scales, each ranging from 0 to 1. External efficacy was measured using these items: (1) ‘Politicians do not really care what the voters think’; (2) ‘Usually you can trust the political leaders to do what is best for the country’; (3) ‘Citizens like me don’t have any influence on the decisions of the parliament and government’; and (4) ‘The politicians waste a lot of taxpayers’ money’. Internal efficacy was based on these items: (1) ‘Sometimes politics is so complicated that a person like me cannot really understand what is going on’; (2) ‘When politicians debate economic policy, I only understand a small part of what they are talking about’; (3) ‘Citizens like me are qualified to participate in political discussions’; and (4) ‘Citizens like me have opinions on politics that are worth listening to’. Reliabilities of the two scales were acceptable, with Cronbach’s α = .62 for external efficacy and α = .69 for internal efficacy. Some of the items were answered ‘don’t know’ by one to seven respondents. The values of these items were imputed by maximum likelihood estimation on the other items. Newspaper reading habits were measured by asking respondents how often they generally read specific newspapers. Furthermore, the survey contained standard questions on gender, age, education, employment and political attitudes.

Exposure to the game frame: Combining content analysis and survey data

Having analysed media content and surveyed media consumption, every respondent’s exposure to the game frame was calculated by merging the results: an individual’s exposure to the game frame was calculated as the sum of total exposure to the game frame from the five newspapers, divided by total exposure to the five newspapers. Formally, the measure was calculated with the following formula:

E x p o s u r e_{i} = \frac{\sum_{j} (R_{i j} \times G_{j} \times A_{j})}{\sum_{j} (R_{i j} \times A_{j})}, j \in {B e, B T, E B, J P, M X}

Exposure_i is an individual’s exposure to the game frame, R is the individual’s readership of the specific newspapers (reading the newspaper every day, for example, results in a value of 7/7 = 1.0), G is the prevalence of game framed articles in the specific newspapers and A is the number of articles about politics in the newspaper. Exposure to the game framed articles was measured as a proportion of total media consumption because exposure to other frames can mitigate the effects of exposure to the game frame (Cappella and Jamieson, 1997). This approach is in line with the perspective that the effect of news is not just a function of the amount of news, but rather a function of news characteristics (Kleinnijenhuis et al., 2006).

Using this aggregate measure instead of measures for each newspaper has the advantage that it does not assume that a specific newspaper has the same effect on all consumers: if exposure to individual newspapers was used, a model would have to show a newspaper to have a negative, positive or insignificant correlation with efficacy for all respondents. Whether exposure to a specific newspaper increases or decreases an individual’s efficacy depends, however, on whether the newspaper increases this person’s relative exposure to the game frame or not, which cannot be known without taking into account the prevalence of the game frame in other newspapers read by this individual. As the focus on this analysis is not exposure to the newspapers per se, but rather exposure to a certain news frame, this aggregate measure of exposure is preferable.

Naturally, the measure can only be calculated for individuals reading at least one of the five newspapers. Ninety-three respondents did not read any of the newspapers and were excluded from the analysis. Furthermore, although we are mainly interested in the share of game framed articles, we should still expect this measure of game frame exposure to have the highest explanatory power for individuals reading one or more of the newspapers on a regular basis. For example, a person reading one of the newspapers on a daily basis would be influenced more by this newspaper than an individual only reading the newspaper once a week, as this latter person would probably receive the bulk of political news from other sources. To account for this, the respondents were divided into two equally sized groups of infrequent and frequent readers of the five newspapers. Median readership of the five newspapers was 0.5. (i.e. reading one of the newspapers every second day). Hence, infrequent readers were defined as readers with a readership of less than this value, while frequent readers had a readership above this limit. It is important to note that the term infrequent readers does not necessarily imply that all respondents in this group rarely read newspapers in general, but rather that they were infrequent readers of the five newspapers used in this analysis. Descriptive data for the two subsamples of infrequent and frequent readers are summarised in Appendix 1. The two subsamples differed significantly in voting intention, but otherwise resembled each other reasonably well. When compared to the general population, the respondents were somewhat higher educated and slightly older, but otherwise reasonably representative of the Danish population.

Results

The content analysis of the newspapers revealed a substantial amount of game framed articles and large variation among the newspapers. As shown in Table 1, 25% of the articles used a game frame, 58% of the articles used an issue frame and the remaining articles contained a mixed frame.

Table 1.

Frames in the press.

Frame	Be (n = 106)	JP (n = 78)	BT (n = 18)	EB (n = 57)	MX (n = 42)	Total (n = 301)
Issue	61%	67%	39%	42%	67%	58%
Mixed	21%	12%	11%	23%	10%	17%
Game	18%	22%	50%	35%	24%	25%
Measure of game frame	.28	.28	.56	.46	.29	.33

Note: Percentages denote share of articles with frame.

The two tabloids, BT and EB, used the game frame substantially more than the two morning dailies, Be and JP. This finding is in line with previous studies, showing tabloids as using the game frame relatively frequently (Strömbäck and Van Aelst, 2010). The prevalence of game framing in the free daily (MX) was comparable to the morning dailies. Summing the prevalence of the game frame into a single measure for each newspaper was done by adding half the number of mixed articles to the number of game framed articles, and dividing this result by the total number of articles. This yields values between .28 and .56.

Internal efficacy

A bivariate regression showed a highly significant negative relationship between exposure to the game frame and internal efficacy among the frequent newspaper readers (B = –.78, p < .001, adjusted R² = .08). Among the infrequent readers, this relationship was not significant (B = –.37, p = .13, adjusted R² = .01). To account for the correlation between efficacy and other demographic and attitudinal variables, the relationship was tested in two multivariate models (Table 2).

Table 2.

Explaining internal efficacy.

	Model 1		Model 2
	Infrequent readers	Frequent readers	Infrequent readers	Frequent readers
Intercept	.45*** (.07)	.50*** (.08)	.54*** (.12)	.81*** (.14)
Basic demographics
Male	.10** (.03)	.09** (.03)	.10** (.03)	.09** (.03)
Age	.00** (.00)	.00 (.00)	.00* (.00)	.00 (.00)
Education
Primary, secondary and short-cycle tertiary	.06 (.04)	.04 (.05)	.05 (.04)	.02 (.05)
Medium-cycle tertiary	.06 (.04)	.06 (.05)	.05 (.04)	.03 (.05)
Long-cycle tertiary	.16** (.05)	.10† (.05)	.15** (.05)	.05 (.05)
Occupation
Lower white collar	−.06 (.05)	.11* (.05)	−.06 (.05)	.12* (.05)
Higher white collar	.01 (.04)	.10* (.05)	.00 (.05)	.09* (.05)
No employment	−.02 (.04)	.04 (.04)	−.03 (.04)	.04 (.04)
Vote
Government or supporting party	−.01 (.03)	.03 (.03)	−.01 (.03)	.02 (.03)
Exposure to game frame			−.23 (.25)	−.64** (.24)
R² (adjusted)	.13	.10	.13	.13
R² change (model 1→model 2)			.00	.04**
N	158	143	158	143

Note: OLS regressions, unstandardised betas (standard errors in parentheses).

***

p < .001, **p < .01, *p < .05, †p < .1.

Reference categories are: Gender: female, Education: vocational education, Occupation: worker or self-employed, Vote: opposition or not voting.

Model 1 includes gender, age, education, occupation and voting intention, which clearly explain a part of the variance in internal efficacy. There is some variance in the estimates for infrequent and frequent readers, but none of the coefficients differs significantly from each other, when comparing the two subgroups. Model 2 adds exposure to the game frame. Among the infrequent readers, the impact of this variable is not significant, which is not surprising, given the results of the bivariate analysis. Exposure to the game frame is, however, still significantly correlated with internal efficacy among the frequent readers, and including this variable increases adjusted R² significantly (excluding six respondents with imputed values on the efficacy items has no significant effect on the results). Hence, exposure to the game frame does seem to have an independent, significant and substantial effect on internal efficacy for frequent readers, and hypothesis 1 is not rejected.

Two additional models (not shown) can be considered. First, political interest is highly correlated with internal efficacy, and adding this variable to model 2 for frequent readers has marked effects on the model: adjusted R² increases to .35, and the effect of exposure to the game frame is attenuated (B = –.43), although still significant (p = .043). However, such a model is likely to be misspecified, as interest might be more of a result than a cause of internal efficacy. Although political interest has been included in models explaining efficacy (e.g. Kenski and Stroud, 2006), the main causal effect may run in the opposite direction. Using structural equation modelling, Brussino et al. (2011) find efficacy to be a cause rather than an effect of political interest. Hence, without a convincing theoretical explanation of why political interest should be a determinant rather than a result of internal efficacy, and without any empirical studies pointing in this direction, interest is not an appropriate variable in these models.

Second, instead of creating separate models for infrequent and frequent readers, model 2 could include all readers and add a term for the interaction between exposure to game frame and frequency of readership (not shown). To avoid misspecification, the second constitutive term, frequency of readership, should also be added to this model (Brambor et al., 2006). A probing of this interaction model with the Johnson–Neyman technique (Hayes and Matthes, 2009) confirms that the effect of exposure to the game frame is significantly negative (p < .05) for readership just above .50 (the effect becomes insignificant again for readership values above 1.5 because of large standard errors on these estimates). However, applying the interaction to the data does not significantly improve the model (adjusted R² = .12), which also seems to have severe problems with multicollinearity (VIF = 41.7, Condition Index = 41.4), resulting in highly inefficient parameter estimates.

External efficacy

A bivariate regression also shows a highly significant negative relationship between exposure to the game frame and external efficacy among the frequent newspaper readers (B = –.69, p < .001, adjusted R² = .07), while the relationship was non-existent among infrequent readers (B = –.06, p = .80, adjusted R² = –.01) (Table 3).

Table 3.

Explaining external efficacy.

	Model 1		Model 2
	Infrequent readers	Frequent readers	Infrequent readers	Frequent readers
Intercept	.51*** (.07)	.44*** (.07)	.51*** (.13)	.68*** (.13)
Basic demographics
Male	−.01 (.04)	−.01 (.03)	−.01 (.04)	−.01 (.03)
Age	.00 (.00)	.00 (.00)	.00 (.00)	.00* (.00)
Education
Primary, secondary and short-cycle tertiary	−.02 (.05)	.08† (.04)	−.02 (.05)	.06 (.04)
Medium-cycle tertiary	−.01 (.05)	.08† (.04)	−.01 (.05)	.05 (.04)
Long-cycle tertiary	.05 (.06)	.05 (.05)	.05 (.06)	.01 (.05)
Occupation
Lower white collar	.06 (.05)	.08† (.04)	.06 (.05)	.08† (.04)
Higher white collar	.08 (.05)	.11** (.04)	.08 (.05)	.11* (.04)
No employment	.02 (.04)	−.01 (.04)	.02 (.04)	−.01 (.04)
Vote
Government or supporting party	.05 (.04)	.10*** (.03)	.05 (.04)	.09** (.03)
Exposure to game frame			.00 (.27)	−.51 * (.22)
R² (adjusted)	−.00	.17	−.01	.20
R² change (model 1→model 2)			.00	.03*
N	158	143	158	143

Note: OLS regressions, unstandardised betas (standard errors in parentheses).

***

p < .001, **p < .01, *p < .05, †p < .1.

Reference categories are: Gender: female, Education: vocational education, Occupation: worker or self-employed, Vote: opposition or not voting.

Turning to the multivariate models, model 1 shows a noticeable difference in the explanatory power between infrequent and frequent readers, with adjusted R² of –.01 and .20, respectively. The main reason behind this difference is the fact that voting intention does not reach significance among the infrequent readers, as this sample has relatively few respondents voting for government or supporting party. Introducing exposure to the game frame in model 2 increases adjusted R² from .17 to .20 among the frequent readers. Exposure to the game frame is still significant (p = .02), supporting hypothesis 2. Adding interest to the model does not significantly alter the results, and the interaction model confirms a significantly negative effect (p < .05) for levels of readership above .87 (adjusted R² = .09, again with severe multicollinearity). However, the results for external efficacy are not very robust: excluding four respondents with imputed values on the efficacy items makes exposure to the game frame insignificant (p = .08), so the results for external efficacy should be treated with some caution.

Conclusion and discussion

This article has argued that we look beyond cynicism and instead focus on efficacy when investigating the effects of the game frame. The empirical analysis confirmed that there is a significant negative correlation between the game frame and internal efficacy, even when controlling for potentially confounding variables. The relationship between exposure to game frame and external efficacy was also significant, although not robust. The lack of robustness could, however, easily be because of a relatively small number of observations and attenuation bias from measurement error, so further studies on the game frame and external efficacy are clearly warranted.

These results do of course come with the caveat that the direction of causality is assumed, not investigated. The question of selective exposure is central in almost all observational media studies (see e.g. Bennett and Iyengar, 2010; Holbert et al., 2010), and the significant correlation between exposure to the game frame and internal efficacy could be a consequence of individuals with low efficacy seeking out certain news outlets. However, in a panel study, Gastil and Xenos (2010) found, contrary to their expectations, that media use had a significant effect on internal efficacy, whereas the reverse causal effect was insignificant. Furthermore, this article has offered two theoretical accounts of possible causal mechanisms from game frame exposure to lowered internal efficacy: first, from a knowledge perspective, the game frame may crowd out substantial information on policy in the news, meaning that the low levels of internal efficacy could be interpreted as the result of low levels of actual knowledge on policy issues. Second, from the spectator perspective, the game frame might lower internal efficacy by creating a perception of politics as a field where the ordinary citizen has limited competence.

Future studies could use experiments and panel data to further investigate the causal direction between the game frame and efficacy. Another avenue for future studies is the exact mechanism through which the game frame impacts internal efficacy. Analyses of knowledge levels and perceptions of politics could be used to determine whether the effect is best explained by the knowledge perspective, the spectator perspective, or another perspective. Finally, as efficacy is a predictor of political participation (Pinkleton et al., 1998; Valentino et al., 2009), it would be highly relevant for future studies to move further in the causal chain and investigate the actual impact of the game frame on political participation, including voting behaviour.

From a democratic perspective, it is exactly the game frame’s potentially detrimental effects on political participation which is the main question. The results of this study suggest that the effects of the game frame may be more consequential than suggested by the previous studies within the spiral of cynicism literature. The problem with the game frame is not just that it turns people cynical, but rather that it may also turn them politically inactive. The results of this study do not, however, have to lead to a general perspective of mediamalaise (Newton, 1999), as they in no way contradict studies finding positive effects of more issue-oriented news coverage (e.g. Adriaansen et al., 2010). Some news coverage of politics may very well foster a virtuous circle, helping the electorate ‘to improve our understanding of public affairs, to increase our capacity and motivation to become active in the political process, and to thereby strengthen civic engagement’ (Norris, 2000: 317). However, the game frame does seem to be at odds with this ideal.

Footnotes

Appendix

Appendix 1:

Respondent and population characteristics (share, percentages)

	Respondents		Population (ages 18–69)
	Infrequent readers (n = 158)	Frequent readers (n = 143)
Gender
Male	44.3	55.2	50.2
Female	55.7	44.8	49.8
Age
<36	32.3b	19.6^{a b}	32.4
36–47	22.8	26.6	26.0
48–58	24.7	27.3	21.5
>58	20.3	26.6	20.1
Education
Vocational education	24.1a	25.2a	32.9
Primary, secondaryand short-cycle tertiary	26.6^a	25.9^a	45.4
Medium-cycle tertiary	33.5^a	31.5^a	14.7
Long-cycle tertiary	15.8^a	17.5^a	7.0
Geographical region
Hovedstaden	28.5	30.8	31.1
Sjælland	12.7	16.8	14.6
Syddanmark	22.2	15.4	21.3
Midtjylland	24.7	28.7	22.6
Nordjylland	12.0	8.4	10.4
Occupation
Worker or self-employed	25.3	23.8
Lower white collar	17.7	17.5
Higher white collar	25.9	30.1
No employment	31.1	28.7
Vote
Government/supporting party	28.5^b	51.7^b
Opposition /not voting	71.5^b	48.3^b
Efficacy (mean)
Internal	0.66	0.69
External	0.51	0.52

Significantly different from population (p < .05, z-test and t-test, where applicable).

Significant difference between subsamples (p < .05, z-test and t-test, where applicable).

Population data: Statistics Denmark.

Funding

This work is based on research supported by The Danish Council for Independent Research [grant number 275-07-0190].

Notes

References

Adriaansen

Van Praag

De Vreese

(2010) Substance matters: How news content can reduce political cynicism. International Journal of Public Opinion Research 22(4): 433–457.

Agger

Goldstein

Pearl

(1961) Political cynicism: Measurement and meaning. Journal of Politics 23(3): 477–506.

Bennett

(1997) Knowledge of politics and sense of subjective political competence: The ambiguous connection. American Politics Quarterly 25(2): 230–240.

Bennett

Iyengar

(2010) The shifting foundations of political communication: Responding to a defense of the media effects paradigm. Journal of Communication 60(1): 35–71.

Benoit

Stein

Hansen

(2005) New York Times coverage of presidential campaigns. Journalism and Mass Communication Quarterly 82(2): 356–376.

Binderkrantz

Green-Pedersen

(2009) Policy or processes in focus. International Journal of Press/Politics 14(2): 166–185.

Brambor

Clark

Golder

(2006) Understanding interaction models: Improving empirical analyses. Political Analysis 14(1): 63–82.

Brussino

Medrano

Sorribas

Rabbia

(2011) Young adults’ knowledge of politics: Evaluating the role of socio-cognitive variables using structural equations. The Spanish Journal of Psychology 14(1): 183–194.

Cappella

Jamieson

(1997) Spiral of Cynicism. New York: Oxford University Press.

10.

Chong

Druckman

(2007) Framing theory. Annual Review of Political Science 10: 103–126.

11.

Cook

(1998) Governing with the News: The News Media as a Political Institution. Chicago: University of Chicago Press.

12.

Craig

Maggiotto

(1981) Political discontent and political action. The Journal of Politics 43(2): 514–522.

13.

De Vreese

(2004) The effects of strategic news on political cynicism, issue evaluations, and policy support: A two-wave experiment. Mass Communication and Society 7(2): 191–214.

14.

De Vreese

(2005a) The spiral of cynicism reconsidered. European Journal of Communication 20(3): 283–301.

15.

De Vreese

(2005b) News framing: Theory and typology. Information Design Journal and Document Design 13(1): 51–62.

16.

De Vreese

Elenbaas

(2008) Media in the game of politics: Effects of strategic metacoverage on political cynicism. International Journal of Press/Politics 13(3): 285–309.

17.

De Vreese

Semetko

(2002) Cynical and engaged: Strategic campaign coverage, public opinion, and mobilization in a referendum. Communication Research 29(6): 615–641.

18.

Elenbaas

De Vreese

(2008) The effects of strategic news on political cynicism and vote choice among young voters. Journal of Communication 58(3): 550–567.

19.

Gastil

Xenos

(2010) Of attitudes and engagement: Clarifying the reciprocal relationship between civic attitudes and political participation. Journal of Communication 60(2): 318–343.

20.

Hansen

Pedersen

(2012) Efficiency of different recruitment strategies for web panels. International Journal of Public Opinion Research 24(2): 238–249.

21.

Hayes

Krippendorff

(2007) Answering the call for a standard reliability measure for coding data. Communication Methods and Measures 1(1): 77–89.

22.

Hayes

Matthes

(2009) Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research Methods 41(3): 924–936.

23.

Holbert

Garrett

Gleason

(2010) A new era of minimal effects? A response to Bennett and Iyengar. Journal of Communication 60(1): 15–34.

24.

Hollander

(1997) Fuel to the fire: Talk radio and the Gamson hypothesis. Political Communication 14(3): 355–369.

25.

Jamieson

(1992) Dirty Politics. New York: Oxford University Press.

26.

Kenski

Stroud

(2006) Connections between internet use and political efficacy, knowledge, and participation. Journal of Broadcasting and Electronic Media 50(2): 173–192.

27.

Kleinnijenhuis

Van Hoof

AMJ

Oegema

(2006) Negative news and the sleeper effect of distrust. Harvard International Journal of Press/Politics 11(2): 86–104.

28.

Krippendorff

(2004) Reliability in content analysis: Some common misconceptions and recommendations. Human Communication Research 30(3): 411–433.

29.

Lawrence

(2000) Game-framing the issues: Tracking the strategy frame in public policy news. Political Communication 17(2): 93–114.

30.

Newton

(1999) Mass media effects: Mobilization or media malaise? British Journal of Political Science 29: 577–599.

31.

Niemi

Craig

Mattei

(1991) Measuring internal political efficacy in the 1988 national election study. American Political Science Review 85(4): 1407–1413.

32.

Norris

(2000) A Virtuous Circle: Political Communication in Postindustrial Societies. Cambridge: Cambridge University Press.

33.

Patterson

(1993) Out of Order. New York: Alfred A Knopf, Inc.

34.

Pinkleton

Austin

Fortman

KKJ

(1998) Relationships of media use and political disaffection to political efficacy and voting behavior. Journal of Broadcasting and Electronic Media 42(1): 34–49.

35.

Pollock

(1983) The participatory consequences of internal and external political efficacy: A research note. Western Political Quarterly 36(3): 400–409.

36.

Slothuus

(2008) More than weighting cognitive importance: A dual-process model of issue framing effects. Political Psychology 29(1): 1–28.

37.

Strömbäck

(2008) Four phases of mediatization: An analysis of the mediatization of politics. International Journal of Press/Politics 13(3): 228–246.

38.

Strömbäck

Aalberg

(2008) Election news coverage in democratic corporatist countries: A comparative study of Sweden and Norway. Scandinavian Political Studies 31(1): 91–106.

39.

Strömbäck

Dimitrova

(2006) Political and media systems matter: A comparison of election news coverage in Sweden and the United States. Harvard International Journal of Press/Politics 11(4): 131–147.

40.

Strömbäck

Shehata

(2007) Structural biases in British and Swedish election news coverage. Journalism Studies 8(5): 798–812.

41.

Strömbäck

Van Aelst

(2010) Exploring some antecedents of the media’s framing of election news: A comparison of Swedish and Belgian election news. International Journal of Press/Politics 15(1): 41–59.

42.

Valentino

Beckmann

Buhr

(2001) A spiral of cynicism for some: The contingent effects of campaign news frames on participation and confidence in government. Political Communication 18(4): 347–367.

43.

Valentino

Gregorowicz

Groenendyk

(2009) Efficacy, emotions and the habit of participation. Political Behavior 31(3): 307–330.

44.

Van Gorp

(2010) Strategies to take subjectivity out of framing analysis. In: Paul

Kuypers

(eds) Doing News Framing Analysis: Empirical and Theoretical Analysis. New York: Routledge.

45.

Zaller

(1992) The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press.