Public and Commercial Broadcasters Compared

The possibility of knowledge gaps resulting from exposure to primarily commercial or public newscasts presupposes that the newscasts are in fact markedly different. Where PSB news varies significantly from commercial news, the potential for knowledge effects is strong, all else being equal. If PSB and commercial news programmes are quite similar, however, differential effects on knowledge are less plausible.Footnote ⁸ When content is similar, evidence for more (or less) informed PSB audiences is quite likely to be an artefact of the audience being more (or less) informed to begin with. That said, the probability of selection effects should increase where public and commercial newscasts are substantively different –real (or perceived) differences between news providers is precisely what should lead well-informed citizens to consistently select into particular streams of news. In short, teasing out the impact of PSBs versus private broadcasters, controlling for selection effects, is difficult.

Several recent studies involving PSB and commercial media content have focused on coverage patterns related to the supply of current affairs information.Footnote ⁹ Generally, results are consistent with respect to the issue of hard versus soft news content: PSBs tend to report more hard news on average than the commercial newscasts in their markets.Footnote ¹⁰ Numerous case studies highlighted the well-known mantra of ‘informing, enlightening and entertaining’ that is widely associated with the institution of public broadcasting;Footnote ¹¹ and entertainment is typically cast third of the three objectives. What this means is that, on a regular news day, people are likely to encounter a greater proportion of foreign news, domestic and international politics, and public-policy oriented reports if they tune into a PSB newscast rather than a commercial newscast. Conversely, watching commercial newscasts increases the likelihood of exposure to what is typically regarded as soft news – such as crime, human-interest, celebrity or entertainment-oriented stories.

In fact, research in this vein has pointed to heterogeneity not just across newscasts in terms of soft versus hard news, but also with respect to the placement and volume of current affairs programming airing on a typical day.Footnote ¹² Practically speaking, the literature suggests that during prime time, when television audiences spike, citizens residing in the most commercialized media systems are least likely to encounter (hard) news programming on any free-to-air television channel. Countries where public broadcasting plays a leading role in the media system are where (hard) news programming tends to flourish during prime time. In short, the aggregate provision of news versus entertainment programming varies according to the media system.

None of this necessarily means that living in highly commercial media contexts leaves citizens with less news to select from and promotes lower levels of news consumption, of course. It is entirely possible that other media – online, radio or print – compensate for relatively low levels of prime time television news in commercial media systems. Online media content is not restricted geographically, and thus there is little to prevent citizens in more commercialized systems looking elsewhere for news if they choose to do so.

The important point is this: the more commercialized the media system is, the more proactive citizens may need to be about seeking out meaningful news on a day-to-day basis. Inadvertent or incidental exposure to the news during prime time news programmes occurs more frequently in countries where public broadcasting is a strong component of the national media system.Footnote ¹³ And where PSB and commercial newscasts are concerned, there is evidence that hard news tends to receive shorter shrift on commercial compared to public television. In a nutshell, there is evidence that the quantity and quality of news varies systematically across public and private broadcasters. The potential exists, then, for a link between PSB and knowledge (driven by some combination of exposure and selection bias).

Current Affairs Knowledge and News Exposure

Much of the extant literature about news exposure and knowledge acquisition has questioned whether different media produce different effects. Until recently newspapers and television tended to dominate the debate, though a growing body of work now focuses on online news consumption. While many have found positive effects of newspaper consumption on knowledge and particularly factual information gain,Footnote ¹⁴ others have pointed to comprehension benefits and longer-term memory retention resulting from television news exposure,Footnote ¹⁵ and others still have cast a positive light on online media where integrated knowledge gain, political interest and participation is concerned.Footnote ¹⁶ So, clearly all media matter in one way or another. That said, most studies suggest that one's preferred news media form, or modality, is significantly related to their performance on a battery of current affairs questions.Footnote ¹⁷

As noted above, previous research suggests that PSB news programming varies significantly from commercial television newscasts. At the very least, PSBs cultivate a reputation (deservedly or not) for reporting current affairs in distinct ways from their commercial rivals. Thus, compared to people who regularly watch commercial news programming, it seems reasonable to expect that those who routinely consume PSB newscasts have higher levels of hard news knowledge; conversely, PSB viewers may be less knowledgeable about soft news.

Few studies have empirically addressed this possibility of knowledge gaps resulting from exposure to public and commercial television, and within this small literature results are rather mixed. Consider some recent results from two panel studies. Using Norwegian election study participants, Jenssen finds no evidence that exposure to NRK news – from Norway's main (public) broadcaster – was more informative than following commercial news.Footnote ¹⁸ Another panel study in the Netherlands and Denmark involving knowledge of European Union events and leaders finds positive effects for public broadcasting in certain contexts.Footnote ¹⁹ Using cross-sectional Dutch Election Study data, Aarts and Semetko find strong evidence that not just knowledge but also attitudes and engagement vary according to public versus private media consumption.Footnote ²⁰

While panel studies provide some real strength in terms of causality, existing work is limited in its generalizability. To our knowledge, only three (non-panel) studies have addressed the knowledge question with data from more than two countries.Footnote ²¹ All suggest that there are modest knowledge gaps based on PSB viewership. Holtz-Bacha and Norris find significant knowledge effects for public television preferences in ten out of fourteen countries.Footnote ²² Jenssen et al. find positive effects of PBS news exposure and mainly negative effects for commercial news – though most effects, but not all, are not significant when controlling for background characteristics and political interest. Popescu and Toka also report positive effects across thirty-five country cases, but only among those least informed to begin with, and the effect varies according to the broadcasting model.Footnote ²³ In each case, results are limited to European countries, and to specific knowledge of European political issues and/or party placement on ideological scales. Generally, results point to some important differences between public and commercial broadcast news on current affairs knowledge. And the Popsescu and Toka paper is one of several recent studies to focus on the influence not just of specific media outlets, but media systems on various forms of citizen engagement in public life.Footnote ²⁴

In a similar vein, Aalberg and colleagues’ exhaustive thirty-year, six-country study suggests that the daily diet of news programming varies according to the (system-level) degree of media commercialization.Footnote ²⁵ Overall, the supply of prime time news programming is greatest in countries where public broadcasters dominate in terms of audience ratings (such as Norway and Sweden) and lowest in countries where commercial media control the market (for instance, the United States).Footnote ²⁶ Dual broadcasting models (for example, the United Kingdom) – in which public and private channels coexist on a relatively equal footing – fall somewhere in between when it comes to prime time news availability.

The recurring message, in short, is that simply living in more commercially-oriented (or ‘liberal’) media systems makes an individual less likely to be politically engaged and broadly less aware of the events occurring in the world around them.Footnote ²⁷ As Aalberg and Curran report, for instance, citizens who are not interested in politics and choose not to pay attention to domestic news, still manage to be relatively well informed in European countries with strong PSBs. Yet within the United States, those with little interest in politics tend to be largely uninformed about current affairs.Footnote ²⁸ Put another way, where private companies dominate the news media landscape, citizens tend to be less aware of what and who is making news nationally, and also in others parts of the world. Moreover, it appears that knowledge gaps between traditionally advantaged and disadvantaged groups in society widen in more commercialized media settings.Footnote ²⁹ What this all suggests is that people who are motivated to get informed are able to do so in spite of the media system in their country, but that current affairs awareness for those with fewer resources and/or weaker motivation is strongly tied to their national media context.

Ultimately, what remains unclear is not so much whether the media system plays a role in knowledge transmission (it clearly does), but to what degree PSB matters in each case. The fact is that PSBs exist in one form or another in virtually all developed countries, and they operate in varying contexts of commercialization. Yet PSBs differ widely in terms of audience share, levels and types of public funding, and institutional independence from the political process. This raises the possibility that knowledge effects resulting from exposure to commercial or public news are not always consistent. Is exposure to public broadcasting related to increases in knowledge of current affairs? Under what conditions do public broadcasters best perform the educative and enlightenment function they all profess to fulfil? We turn now turn to these questions.

Methodology

The reference country for almost all media system and knowledge research is the United States. Widely acknowledged as the prototypical commercial (liberal) media system,Footnote ³⁰ however, PSB news plays a marginal role in the news diet for the vast majority of Americans.Footnote ³¹ We accordingly do not examine the impact of PSB in the United States below (given the very low number of regular PSB viewers, the American survey did not ask about PSB consumption), though we do look at American as well as Australian data to explore briefly the impact of television exposure versus newspaper readership on knowledge. Those results are included in the online appendix, and discussed briefly in the concluding section. Here, we focus on six countries for which we have comparable survey data, and in which the PSB receives a reasonable audience share: Canada, Italy, Japan, Norway, Britain and South Korea. Note that these six provide a good amount of variance where media system variables are concerned. We discuss media systems in more detail in the sections that follow.

Data for all countries was drawn from a unique survey on news exposure and political knowledge, fielded (nearly) simultaneously in 2010 across eleven countries.Footnote ³² Surveys in each of the six (plus two) countries used here were conducted online by YouGov Polimetrix. Details of these surveys, including survey firms, field dates and availability, are provided in the Appendix.

As noted above, a central issue in the study of media effects, particularly though not exclusively in cross-sectional data, is the problem of self-selection. PSB may well provide a greater amount of hard news information than commercial broadcasting. But it is almost certainly also true that people interested in news, and with higher levels of hard news knowledge to begin with, choose to watch PSB, while those with less interest are less likely to do so. Thus, we cannot easily tell which came first, the chicken (knowledge) or the egg (PSB exposure).

The result where statistical analysis is concerned is that the importance of media to political knowledge is, as a consequence of endogeneity, fairly easy to overestimate. Indeed, several studies, by controlling for self-selection, have rendered ostensibly positive effects for newspaper consumption on knowledge spurious because newspapers readers were knowledgeable, motivated or better educated to begin with.Footnote ³³ Finding an appropriate way to control for the cognitive and demographic differences between media audiences is essential.

One potentially useful approach is propensity score analysis. The approach is designed to allow for more reliable causal inferences in those observational studies for which randomization was not possible.Footnote ³⁴ The crux of the method is as follows: based on a number of background characteristics, captured in the ‘propensity score’, those in a treatment group are ‘matched’ to a group of similar others in a non-treatment group. The idea is to approximate randomization in treatment – to produce two relatively similar groups, one exposed to a treatment and the other not. The difference in outcomes between these two groups is then the critical test of the impact of treatment.

Comparing means across two groups is of course relatively easy; producing the groups themselves is rather more difficult. Doing so relies on a propensity score, capturing similarities in individuals across n dimensions. That score is typically produced in a regression model of the binary treatment variable. So, in the current case, we first use a probit regression model to estimate the likelihood that different individuals are exposed to public television. The results (predicted likelihoods) from that regression are the propensity score. ‘Like’ individuals, who were exposed to public television or not, where likeness is determined by their propensity score, are then selected into the treatment and non-treatment groups.Footnote ³⁵ And the differences between the two groups in terms of outcomes (here, political knowledge) is the critical test of the impact of treatment (here, public broadcasting).

The central difficulties in employing matching methods lie in the production of the propensity score, and then the matching. In short, the estimation of a propensity score is subject to all the same problems as a regular ordinary least squares (OLS) model. Using the correct set of covariates to produce propensity scores is critical. As may happen with regular regression models, the exclusion of an important covariate opens up the possibility that estimated effects are a consequence not of treatment alone, but something else. That is, an inadequate propensity score model leaves open the possibility that the matching methods approach does not get around the problem of endogeneity.Footnote ³⁶ To be clear, then: we do not claim here that matching necessarily removes the difficulties associated with endogeneity in the estimation of media effects. But a properly-specified propensity score model can yield somewhat more accurate (and often more conservative) estimates of treatment effects.Footnote ³⁷ In this case, we view propensity score matching as a potentially useful way to remove some, though probably not all, of the problems associated with self-selection. It is in this way not unlike a similarly-specified regression model, though with some additional statistical advantages.Footnote ³⁸

We estimate matching below using ‘pscore’ in STATA.Footnote ³⁹ There are a number of different matching algorithms available.Footnote ⁴⁰ Here, we rely on radius matching;Footnote ⁴¹ though our results do not change fundamentally when other matching algorithms are used.Footnote ⁴² The models for propensity scores are relatively simple, but also thorough. The ‘treatments’ investigated below are as follows: (a) public television viewing, and (b) commercial television viewing. In each case, we produce a binary treatment variable that divides the sample roughly in half, where one group watches four days or less and the other watches five days or more.

Each of these treatment variables is regressed on a series of demographic and other variables likely to affect media exposure. We include the following:

• • Age: two dummy variables for 35–54 years and 55 and over, with 18–34 years as the residual category.Footnote ⁴³

• • Education: an ordinal variable where 1 is primary education, 2 secondary education, and 3 tertiary education.

• • Political Interest: an index based on self-reported interest in national news, international news and local news. In all countries, respondents are asked to rate their interest in each on a five-point scale. In each country, the index is used to produce an interest scale, recoded here to produce four categories (roughly, quartiles), and then split into dummy variables for the second, third and fourth categories, where the first tercile is the residual category.

• • Exposure to Other Media: a continuous index, recoded into terciles by country, based on self-reported (5-point) exposure to newspapers, radio, and online news.

Propensity scores are generated country by country, rather than on a pooled basis, in order to allow for the fact that the correlates of media exposure matter differently in each country.Footnote ⁴⁴ The magnitude of individual coefficients varies, of course; so too does the proportion of variance explained by our model. Pseudo R ² for the binary probit estimations ranges from roughly 0.05 to 0.25, for instance, with an average of about 0.12.Footnote ⁴⁵ And to review: the idea is to match respondents based on basic demographics, interest in politics and exposure to other media; compare differences in political knowledge across two groups with similar propensities to watch public (or private) television, but where only one of those groups has actually been regularly exposed to public (private) television; and then use the estimated difference in means as the test of the impact of the treatment variables.

Political knowledge is measured using an index of between ten and sixteen knowledge questions, capturing knowledge of hard and soft news, both national and international. These knowledge questions are included in the online appendix; suffice it so say here that the questions tap a combination of hard and soft news knowledge, both domestic and international. The use of both hard and soft news items is relatively unique; as is the use of knowledge questions focused in part on events in the news at the time of the survey. International news knowledge questions were common across all surveys; domestic news stories obviously varied from country to country, but were designed to be similar in theme and difficulty. In each country, regardless of the total number of questions, the resulting cumulative knowledge index was rescaled to range from 0 to 1, where 1 reflects a correct score across all available questions. In order to account for the possibility that, despite our best efforts, the difficulty of questions varies somewhat from one country to the other, we use knowledge measures below rescaled to standard units (standard deviations from the mean, within each country).

Discussion and Conclusions

What do our results suggest about the relative success of public over commercial broadcasters in producing an informed citizenry? Public broadcasting, in some countries at least, clearly matters. Controlling for self-selection as best we can, it appears as though those exposed to public television news learn more about hard news (and in some cases soft news as well) than those exposed to private television news for the same amount of time per week. This effect is markedly greater, however, in countries where the public broadcasters are funded mainly if not exclusively from public money, and where they also have de jure independence from government. Essentially, freedom from interference by market forces and government seems to lead to a form of public broadcasting that is markedly ‘better’ than its commercial rivals.

Might something else fill the void if PSBs were to disappear from the national media landscapes? We cannot rule out the possibility that in the absence of a strong public broadcaster, commercial stations would produce news more like PSBs. This seems doubtful, however. It is costly to assemble hard news; and news gathering in foreign countries is particularly expensive. This is perhaps why one of the main commercial news broadcasts in the United States, called ‘World News’, offers less than 2 minutes of international affairs coverage per day.Footnote ⁵⁴ Broadcasting requires good (highly paid) journalists to produce current affairs programming, and for the most part advertisers are not very interested in hard news.Footnote ⁵⁵ There is evidence that the quantity of public affairs news supplied by commercial broadcasters is not much affected by market conditions;Footnote ⁵⁶ and there is also a growing sense that part-time citizen journalists, PR specialists and bloggers simply cannot replace full-time paid professional journalists.Footnote ⁵⁷ Editing, fact-checking and job security are important parts of news production. So, independent, well-funded public broadcasting may really make a difference.

Another possibility is that news reading is more educative than news watching. If this is true, the absence of a strong public broadcaster might make little difference to citizens’ knowledge of current affairs; people would either learn (or not) primarily from newspapers and news websites, and PSBs would not be missed either way. Our evidence for this proposition is scant, however. As presented in the online appendix, newspaper reading is not consistently more strongly associated with knowledge (hard or soft) than is television viewing. In short, our analyses do not support the view that newspaper reading has a stronger effect on citizen knowledge than viewing news on television.Footnote ⁵⁸

None of this is to say that PSB is a magical elixir for countries with low levels of political knowledge. In some countries, including Japan and the United States, the public service broadcaster is regarded (perhaps justifiably) as bureaucratic and intellectually aloof. Existing work also points to real problems with reliance on public broadcasters in post-authoritarian countries.Footnote ⁵⁹ And work on media systems (rather than individual broadcasters) suggests that variation in political knowledge might be seen as a product of the larger media system, rather than the product of individual news outlets.

We have focused here, however, on the potential differences between public and private broadcasters in producing political knowledge, across a range of six developed countries. Analyses suggest that the more PSBs come to resemble their commercial counterparts in terms of dependence on advertising revenue, the less distinguishable their effect on citizens becomes. Similarly, and perhaps most importantly, political independence appears to be a key prerequisite for PSBs ability to perform the educative task they are expected to fulfil.Footnote ⁶⁰ Note that these patterns are based on trends from a limited number of countries – trends which cannot be confirmed statistically, and may not be generalizable given our relatively small sample. That said, this is among the most detailed individual-level cross-national studies of the impact of public versus private broadcasters to date. We are inclined to see the results in Figure 3 as, at least, strongly implying a link between the funding and content of PSBs.

Thinking more broadly, this study clearly has implications for evaluating the role of public service broadcasting. It has often been pointed out that the audience share of public broadcasters is declining as a consequence of the growth of channel competition.Footnote ⁶¹ Intensified competitive pressure has led, it is suggested, to a weakening of public purpose in some public channels’ output.Footnote ⁶² This tradition records in effect ‘the decline and fall of public broadcasting’ – the title of a well-known book.Footnote ⁶³

One response to this decline literature has been to question the speed and extent of the decline. Thus, it has been argued in relation to some European and English-speaking countries that public broadcasting's audience share has stabilized or increased where there are two or more public broadcasters; that survey evidence indicates a public willingness to pay more for public broadcasting (with the exception of Canada); and that some public broadcasters have launched successful websites.Footnote ⁶⁴ The implication is that public broadcasting is worth supporting because it remains popular.

Another response has been to reformulate the case for public service broadcasting within the framework of neo-liberal thought. Public broadcasting, it has been argued, is worth preserving because it constitutes a still valid way of compensating for continuing market failure (a tendency towards oligopoly) or as a way of reaping positive externalities that cannot be realized fully through the market process.Footnote ⁶⁵ Public broadcasting, in other words, has a legitimate place within a market system.

The implication of this study, in contrast, is that public broadcasting has an important role in supporting full citizenship. It suggests that the functioning and performance of public broadcasting should be evaluated not only in terms of customer satisfaction, or within the horizon of market thought, but in terms of what television can contribute to the functioning of democracy. In short, given that public affairs knowledge appears to be significantly improved through the publicly-funded provision of news (here, on television, but potentially online as well), governments’ decisions about funding for public broadcasters seem in many cases to be very much like decisions about just how well informed their citizens will be.