Reconsidering the role of the media in Japanese politics

In his work, Flanagan reveals a paradox: Japanese are widely saturated by media and generally retain a high level of trust in it. Yet, in spite of huge volumes of information, wide variations lay between consumers and non-consumers. One of the most significant is with regard to medium: newspaper readers are urban, male, cosmopolitan, and elite – integrated and networked in their communities; frequent TV viewers are instead primarily women: less well off and socially isolated. Newspaper readers, particularly those reading national dailies, are exposed directly to election information and are more psychologically engaged in politics. Those who rely solely or partially on TV are exposed to far less that is political and, as a result, less engaged.Footnote ¹ Flanagan finds a modest correlation between overall media exposure and knowledge – the latter measured as the respondent's ability to identify party positions and willingness to comment on parties themselves. Awareness of the Lockheed incident (where several high ranking figures, including Prime Minister Tanaka Kakuei, had taken substantial bribes from the Lockheed Corporation in exchange for lucrative aircraft purchase contracts) appeared to correlate with newspaper consumption, knowledge, and engagement, but Flanagan finds, to some surprise, that disproportionate reliance on TV for information about the scandal correlated with higher voter turnout. Even if its effects are ‘rather limited overall’, and ‘do not exert dramatic, decisive influences on Japanese voting behavior’ (1991: 330–1), television, he contends, helps to shape agendas and candidate images.

Flanagan's retake on the JABISS data in his 1996 piece, by contrast, places media exposure at the center of individual political involvement (p. 297). Where media exposure is the strongest correlate of political knowledge, it exerts stronger effects directly on the attitudinal factors underlying participation. Echoing Watanuki, social networks remain the most powerful predictor of Japanese participation but, within the complex path of politicization, the media appears to have a more prominent role than previously thought. ‘Media exposure’, he writes,

Where the Japanese media system may have fit closer to Hallin and Mancini's (Reference Hallin and Mancini2004) conception of ‘democratic corporatist’ through the 1980s, it was clear that it was shifting toward a more Anglo-American, ‘liberal’ style in the 1990s. A critical turning point, detailed by Altman (Reference Altman, Pharr and Krauss1996: 177–84; see also Krauss, Reference Krauss2000a, Reference Krauss, Gunther and Mughan2000b), came in 1993: the first election to break the LDP's uninterrupted 38-year rule and the first where television was credited with a decisive role. A series of forces converged to unseat the Miyazawa Kiichi government – Recruit and taxes, but most significantly, the emergence of ‘new media:’ programs such as Kume Hiroshi's Hōdō Station and Tahara Soichiro's Sunday Project, both on Asahi, that were giving voters unprecedented critical analysis and ‘raw political talk’ and putting LDP politicians in the unfamiliar position of having to appear unscripted and natural. Feldman's pathbreaking work had earlier demonstrated the cozy relationship between establishment Diet members and NHK reporters (1993: 43–8); Hōdō Station and Sun Pro were now rewriting the rules, so much so that the disparate left coalition that displaced the LDP was labelled the ‘Kume-Tahara Government’. Altman describes Miyazawa's fateful appearance on Sun Pro: the first time a prime minister was interviewed without pre-negotiated restrictions. Tahara pinned him to a promise of reform; Miyazawa, facing inter-party turmoil, failed to fulfill it. The leaders of the new opposition parties – eventual Prime Minister Hosokawa Morihiro, among them – took to the airwaves throughout the summer to label Miyazawa a liar. The LDP, skilled in behind-the-scenes, pork barrel politics but lacking in new media skills, stammered. ‘Without television’, Altman concludes, ‘many of the newly elected politicians would not have been successful’ (pp. 174–5).

The presidential election a year earlier, in 1992, was a similar turning point in American elections, as voters not only saw the 12-year Republican regime come to an end after a faulty promise (‘read my lips. . .’), but began to witness, for the first time, candidates' use of non-traditional outlets to convey a softer, humanized image to the public – Bill Clinton's saxophone performances on MTV and Ross Perot's chart shuffling on CNN being the quintessential examples of this ‘new news’ strategy. Research that emerged from the campaign showed a mixed impact on political learning: Chaffee et al. (Reference Chaffee, Zhao and Leshner1994) find a positive impact from talk shows, but find television only boosts ‘people knowledge’; newspapers have a stronger impact on understanding of parties and policy differences. Hollander (Reference Hollander1995) confirms the impact of talk shows, but finds the softest programs, such as MTV and late night comedy, have either a negative or no effect on knowledge. McLeod et al. (Reference McLeod, Guo, Daily, Steele, Huang, Horowitz and Chen1996) finds non-traditional media boosting interest without increasing knowledge of candidate positions.

Koizumi and the ‘Wide Show Cabinet’

By the time Koizumi Junichiro emerged as the head of the Liberal Democratic Party (LDP) in 2001, the shift toward a more personalized, commercial media-driven environment – ‘presidentialization’, in Krauss and Nyblade's (Reference Krauss and Nyblade2005) terms – was already well underway in national politics. As Kabashima and Steel (Reference Kabashima and Steel2007, Reference Kabashima and Steel2010) explain, Koizumi was well suited for the role of in-party ‘outsider’: the man who would eventually serve longer than any modern prime minister (and preside over one of the LDP's biggest landslide victories) recognized, above all, the benefits of going above what he termed ‘the forces of resistance’ to speak directly to the public. The strategy focused on the then new commercial programs –wide shows, in particular – to ‘present [Koizumi] as a beleaguered champion of change and reform, fighting the 'old guard' of his party [and] engaged in a 'civil war' with the anti-reformists’. So prolific was the strategy that Koizumi's government was referred to as 'the Wide Show Cabinet’. But where some denounced the approach as a ‘“dumbing down” of politics’, Kabashima and Steel contend that Koizumi's ‘popularizing could equally be seen as a democratizing of politics. The cozy, mutually beneficial relations among bureaucrats, politicians and political journalists were ruptured as Koizumi sought – and gained – media collaboration in taking his politics directly to the public and asking for their approval’ (2010: 82).

It is difficult to find analogues to wide shows in Western media, given their unique ability to intersperse politics with daily life, health, food, crime, sports, and entertainment news – all within the context of a diverse panel discussion format where style and gossip may take a comfortable place alongside facts and analysis. In the US, a mélange of Good Morning America, The View, and Fox and Friends might come to mind, joined by an eclectic panel that might be akin to something on cable programs Real Time with Bill Maher or Iron Chef America (a US adaptation of a Fuji network program). On Japanese television, elements of the chatty formula extend beyond the typical morning or afternoon program and into evening and weekend programming, where shows may offer ‘hard’ political subjects with a light touch (e.g., TBS’ Sunday Morning) or magazine-style news reports with wide show ‘corners’ (e.g., TBS’ Broadcaster). Hōdō Station or SunPro are credited by many for ushering in a new media landscape by presenting politics from an ‘authentic’ or critical viewpoint, and thus challenging NHK's traditional hegemony over broadcast political information. But programs like Asahi's Super J Channel – or, at the extreme end, Fuji TVs Super News, where nightly features may alternate between undercover probes of convenience store theft to reports on cheap sushi, from ‘house husbands’ to hydrangeas – have offered a different challenge to the information environment with their arrays of sensational, ‘human interest’, and banal content. Even viewers of what Krauss (Reference Krauss2000a: 211) referred to as the ‘bland, neutral, but somehow didactic’ NHK news shows now witness more colorful kyasutas and softened program sets – no doubt the result of growing competition for younger and more demanding segments of news seekers (see Yonekura and Hirotani, Reference Yonekura and Hirotani2011).

What is ‘soft’ – and what is not?

The mixing and mingling of politics and entertainment one sees on Japanese television not only mirrors that which is seen in the US, but is reflective of the murky problem of defining ‘news’ in the contemporary media environment, where scholars have not always been able to pin down clear distinctions between what is meant by ‘hard’ and ‘soft’ political information. Thomas Patterson (Reference Patterson2000: 3–4) contends that ‘soft’ news is essentially that which is not otherwise topically ‘hard’ (‘more sensational, more personality centered, less time-bound, more practical and more incident-based than other news’), but also that it captures a movement that represents a ‘change in the vocabulary of news’, that is ‘more personal and familiar in its form and presentation’. Reinemann et al. (Reference Reinemann, Stanyer, Scherr and Legnante2012: 233) combines these ideas into an operationalized definition, which we adhere to in this paper:

It could be argued that this raises as many questions as it clarifies – what, for example, constitutes ‘relevance’? – but what is useful here is that it suggests that softness and hardness in news may be better conceived of as a range of topical and stylistic elements rather than a clear conceptual dichotomy. One of the first efforts to categorize Japanese news programs, by Kobayashi (Reference Kobayashi1982), follows this by presenting his content analysis findings in the form of a ‘hard -> soft’ spectrum, with NHK at the extreme ‘hard’ end and TBS at the extreme ‘soft’ end and others in between. Krauss broadens this into two dimensions to include the ‘style’ and ‘content’ of networks. NHK, for Krauss, is ‘hard’/’factual’, while TBS and Nippon Television (NTV) are in the opposite corner: the ‘soft’/‘opinionated’. Asahi is ‘hard’/‘opinionated’ and Fuji is ‘soft’/‘factual’. This offers three take-away points. First, NHK's general adherence to an objective and disaffective presentation of ‘hard’ information remains, relative to the rest of the Japanese information marketplace, unique; second, there are distinctions to be made between soft and opinionated political information; third, a number of commerical networks – as typified by the wide show format – conflate personality, opinion, and news on soft topics.

Krauss’ matrix is a general representation of the media landscape; it does not account for the variations that occur across programs within networks. As Inamasu and Ikeda (Reference Inamasu and Ikeda2009) find in a rigorous content analysis of political ‘corners’ during the 2007 elections, discussions of politics on Japanese media not only vary from network to network, but vary in substance from program to program, resulting in ‘soft’ talk on otherwise ‘hard’ outlets and vice versa. In the analysis that follows, we attempt to reconcile these inconsistencies, by accounting for programs that mix elements along with those that hew more consistently to ‘harder’ or ‘softer’ topics and formats.

The (alleged) impacts of a softening news environment

More prominent than the conceputal questions about what ‘soft’ is, however, are the instrumental questions about what it is (allegedly) doing. The literature on this topic is too vast to review here (see Reinemann et al., Reference Reinemann, Stanyer, Scherr and Legnante2012), but we can simplify the findings into two sides: optimists and pessimists. Speaking for the optimists, Baum (Reference Baum2002, Reference Baum2005) demonstrates that soft news may increase political attentiveness among the unengaged and make candidates more likeable; Baum and Jamison (Reference Baum and Jamison2006) argue further for an ‘Oprah effect’, where those who watch daytime talk shows practice more consistent voting patterns. On the pessimistic side, most notably, are Prior (Reference Prior2003, Reference Prior2005) and Patterson (Reference Patterson2000). Patterson makes the case that the rise of soft and opinionated news is linked to broader declines in trust, in both government and media. Prior (Reference Prior2003), instead, finds little evidence that it promotes learning. In a subsequent paper (Prior, Reference Prior2005), he suggests that what new information outlets do is foster a type of political segregation. ‘New media do indeed increase political knowledge and involvement among some people’, he writes. But ‘other people take advantage of greater choice and tune out of politics completely. Those with a preference for entertainment, once they gain access to new media, become less knowledgeable about politics and less likely to vote.’ Understanding content preferences, he argues, is ‘the key to understanding the political implications of the new media’ (p. 587).

What of Japan? While much has been made of the changes brought about by commercial competition to NHK News, as well as Koizumi's (and others' – see Taniguchi, Reference Taniguchi2011) increasing use of these new news outlets, scholarship on Japanese politics has only begun to engage the questions surrounding the impact of its increasingly liberal media environment. And there, too, we find some disagreement. Taniguchi (Reference Taniguchi2002) finds Hōdō Station having an impact on cynicism toward politics; Inamasu and Ikeda (Reference Inamasu and Ikeda2009), similar to Baum, find instead that soft news exposure may increase interest in politics. The questions, therefore, beg: does the onset of softer news programs indeed represent a ‘dumbing down’ of politics? How, if at all, are these changes in style and political communication impacting on what the average Japanese knows about politics?

What follows is a first cut at a response, as we examine first the determinants of political knowledge and then the impact of hard, soft, and mixed programs on political learning. We juxtapose their impact vis-à-vis newspapers and place it in the broader context of the sociological, structural, attitudinal, and behavioral factors presumed to be important in the knowledge acquisition process.

The basis of Japanese political knowledge

A general image of Japanese knowledge acquisition, based on a similar path analysis by Delli Carpini and Keeter (1996: 181), is displayed in Figure 1 using SPSS Amos. The dependent variable is the balanced score of political knowledge, with the first stage independent variables (at the bottom) – gender, age, city size, and marital statusFootnote ⁴ – modeled under the assumption that these influence knowledge in subsequent stages. The second stage ‘structural’ variables include education and employment (public servant), modeled as such because they further define an individual's position in Japanese society. The last stage factors – interest, civic duty, party ID strength, and media consumption – are combined into a ‘behavioral’ category, which are those most closely related to the level of knowledge.Footnote ⁵

Figure 1.

A simplified path model of political knowledge

Our path model, overall, explains knowledge fairly well (JESIII: R² = 0.28; GLOPE2005: R² = 0.35), with model diagnostics confirming a good fit in both datasets.Footnote ⁶ As expected, behavioral variables have the largest direct impact on knowledge in both datasets. The structural variables also have some direct impact on knowledge, though it is notably smaller than behavioral factors. In the first stage, only gender has a direct effect; the others are absorbed into the structural and behavioral factors. One thing that should be noted here is the varying age effects. While age has strong negative impact on the structure of political knowledge, it retains a positive impact on the attitudinal/behavioral aspects of knowledge attainment.

Multiple regression

The results of our OLS models appear in Tables 1 (JESIII) and 2 (GLOPE2005). The first three columns indicate the results where the dependent variable is a specific dimension of knowledge. The last two columns show the results for the balanced and added ‘all knowledge’ variables.

Table 1.

The determinants of political knowledge (JESIII)

Notes: ***p<0.001, **p<0.01, *p<0.05, ^p<0.1 (OLS regression coefficients).

Every first row represents standardized coefficients and the second rows show (unstandardized coef./standard error).

Table 2.

The determinants of political knowledge (GLOPE2005)

Notes: ***p<0.001, **p<0.01, *p<0.05, ^p<0.1 (OLS regression coefficients).

Every first row represents standardized coefficients and the second rows show (unstandardized coef./standard error).

Media exposure: general effects

What the findings indicate is that regular newspaper reading and consumption of a wide range of television programs increases all types of knowledge and knowledge on the whole. Watching TV, in and of itself, has no impact. In the GLOPE2005, a one unit increase in newspaper reading offers a knowledge boost comparable to a one unit increase in party identification strength or working as a public servant. Reading a wider variety of newspapers, as the JESIII results show, does not have the same effect across-the-board as watching a wider range of TV news programs.Footnote ⁸ It does seem to provide a modest boost to knowledge of political leaders.

The impact of consuming political news on the internet is marginal, as we see in the JESII data in Table 1. While it does seem to have a modest effect on knowledge of leaders, it has no impact on policy awareness. The coefficients in the institutional knowledge and overall knowledge equations only meet a marginal p<0.10 standard.

Media exposure: specific source effects

The impact of specific media sources follows, beginning with descriptive analysis and followed by a summary of regression models that isolate the effect of the news source in a fully specified equation.

Television

Compared to newspapers, TV programs enjoy more varied consumption patterns, with each JESIII respondent watching a little more than three (out of 17 tested) shows. NHK News programs enjoy an audience that is nearly twice as large as the second and third most popular programs in the study, Asahi TV's, Hōdō Station and TBS' News 23.Footnote ¹¹ About one in four watch Sunday Project, making it the most viewed ‘hard’ program other than NHK. With NHK set aside, ‘soft’ and ‘mixed’ programs are, in total, watched more than twice as frequently as ‘hard’ programs.

At first glance, viewers of ‘hard’ programs appear to be more knowledgeable, viewers of soft and ‘mixed’ programs less so. T-tests reveal that Sunday Project Hōdō Station and Sunday Morning on TBS, as well as Tokyo TV's World Business Satellite, have viewers who score higher across knowledge dimensions. NHK viewers are not as knowledgeable on average as those who consume hard programs, although they score higher than the average JESIII respondent. Soft program viewers generally score lower, particularly in institutional knowledge. Consumers of Fuji TV's News Japan and Kyō no Dekigoto on Nippon TV in our ‘mixed’ category are among the least knowledgeable overall.

Table 3 presents the results when dummy variables for ‘often’ watching specific TV programs are added to our full regression model. The results of two separate sets of equations are presented: with a control for total TV news consumption (‘TVAGG’) and without.

Table 3.

TV news source effect on political knowledge (JESIII)

Notes: 1. ***p<0.001, **p<0.01, *p<0.05, ^p<0.1 (OLS regression).

2. Entries represents the standardized coefficients from OLS regression with all the variables appeared on Table 1. The coefficients of which did not come out significant are not shown.

3. The TV station names are following: ATV = TV Asahi, TBS = Tokyo Broadcasting System, TTV = TV Tokyo, NTV = Nippon TV, FTV = Fuji TV.

The most prominent finding in the analysis is the effect of NHK. Ceteris paribus, we see that watching NHK news significantly boosts policy and people knowledge and knowledge overall. The effect of watching NHK ‘often’ is comparable to a one unit increase in educational attainment and exceeds the impact of most attitudinal variables in our knowledge model, except interest. Several Asahi TV programs show positive coefficients, but the effects are confined to models without a control for overall news program consumption. World Business Satellite, in our ‘mixed’ category, has an impact on institutional knowledge.

The second noteworthy finding is the apparent depressive impact of softer programs. As the table shows, the coefficients for Super News are negative and significant in two of the three knowledge domains and overall. Those for News Plus One on Nippon TV are negative and borderline at a p<0.10 cutoff standard. The depressive effects also extend into several programs in the ‘mixed’ category, as News 23 appears to diminish policy knowledge. Fuji TV programs appear more consistently with negative coefficients. Among commercial networks, Asahi programs appear consistently with positive coefficients.

When viewers of ‘hard’. ‘mixed’, and ‘soft’ programs are combined, recoded and reinserted in the full model, the effects of format become clearer. Table 4 displays two results: when hard/mixed/soft is recoded as a ‘dummy’ variable (reflecting that the respondent watches at least one such show ‘often’) and as an ‘additive’ scale (reflecting the total number of shows in each category that the respondent watches.) We see that the ‘hard’ category shows positive and significant (at p<0.001) impacts on the three domains and overall knowledge, although the effects diminish when the amount of TV news watching is controlled.

Table 4.

Hard and soft news source effect on political knowledge (JESIII)

Notes: ***p<0.001, **p<0.01, *p<0.05, ^p<0.1 (OLS Regression standardized coefficients).

Hard news: Wake Up! (NTV), Sunday Project (ATV), Hodo 2001 (FTV), Hodo Tokusyū (TBS), Sunday Morning (TBS).

Mixed news: Hodo Station (ATV), Kyo no Dekigoto (NTV), News 23 (TBS), World Business Satellite (TTV), News Japan (FTV).

Soft news: Super J Channel (ATV), Broadcaster (TBS), Super News (FTV), News Plus One (NTV), The Sunday (NTV), News no Mori (TBS).

The soft program coefficients for people and overall knowledge, by contrast, turn negative when the TVAGG variable is in the equation. Their diminishing effect is equivalent to being one year younger or living in a smaller city and appears to mitigate, at a similar rate, whatever knowledge may be gained from hard news. Noteworthy are that the effects appear when soft program viewing is coded as an additive scale. Put another way, merely watching a particular soft program ‘often’ may diminish institutional knowledge, but the cumulative impact of watching additional soft programs appears to diminish overall knowledge as well. In the ‘mixed’ category, we only witness depressive effects on policy knowledge; in other knowledge domains, such programs appear to have a positive impact, but only in the absence of a control for overall consumption.