Election Campaigns,News Consumption Gaps,and Social Media: Equalizing Political News Use When It Matters?

Abstract

We investigate how inequalities in political media use develop throughout election campaigns, and in particular whether social media use helps counterbalance traditional news consumption gaps. Using a four-wave individual-level panel survey of the Norwegian 2017 national election campaign, we run a series of latent growth models to investigate whether differences in news consumption based on gender, age, education, and political interest increase or decrease during campaigns. We find that news consumption gaps are either stable or converge throughout the campaign. Importantly, social media provides political information to those groups that use traditional media channels the least and thereby reduce overall gaps in political media consumption. In this way, election campaigns, to some extent, equalize inequalities in political news consumption when it matters the most.

Keywords

election campaigns news consumption gaps social media news consumption

Introduction

In order to make informed choices on Election Day, citizens need information about the alternatives running for election and the important issues of the day. The news media is the most vital provider of such information, and by following the news, people keep themselves updated (Holbert 2005; Barabas and Jerit 2009). In this way, the news media also link citizens to the political public sphere, connecting citizens and political elites (Norris 2000: 25–8). From a normative perspective then, it is important that citizens of representative democracies follow news about politics and current affairs in the period leading up to an election (Dahl 1998; Schudson 1998; Zaller 2003).

Political news consumption is, however, not distributed equally between citizens. Previous research demonstrates that motivational factors (i.e., political interest) generate systematic differences in media consumption. Such inequalities, in turn, are related to sociodemographic characteristics, such as gender, age, and education (Esser and Steppatt 2017; Norris 2000). We know, however, little about how election campaigns influence these news gaps. A vast body of research on electoral campaigns provides vital knowledge about outcomes such as political participation (Holt et al. 2013), vote choice (Bartels 2006; Claassen 2011), political knowledge (Holbrook 2002; Hansen and Pedersen 2014; van der Meer et al. 2016), and political trust (Hooghe and Stiers 2016). Although political media use is sometimes included in this body of research as an explanatory variable, it is rarely the object of study itself. In this article, we switch the focus from studying the effects of political media consumption to examining how and why inequalities in political news consumption change during election campaigns. As news consumption is essential for citizens’ abilities to make informed political choices and for their involvement in politics as such, we need to know more about how these differences develop in the period leading up the election. By studying political media use as a dependent variable in a campaign setting, this study offers crucial insights into citizens’ linkages to the public sphere during the period when most citizens have the greatest opportunity to influence public policy.

The existing theoretical and empirical work, we argue, enables the formulation of two competing arguments. On the one hand, the exceptional level of media choice available amid contemporary electoral campaigns may facilitate the reinforcement of pre-existing differences in political media consumption (Tichenor et al. 1970; Prior 2005), suggesting a divergence pattern throughout election campaigns. On the other hand, groups that usually show little interest in politics may pay more attention to the news in periods when political alternatives are clearly articulated, information is easily accessible, and the political stakes are high. In other words, the political involvement of the “monitorial citizen” (Schudson 1998; Zaller 2003) amid electoral campaigns may level out pre-existing differences in media consumption, suggesting a convergence pattern during campaigns.

Importantly, the divergence and convergence trajectories described above might differ between traditional media and social media. Social media is distinct in that news exposure is curated by user behavior, other individuals in their networks, and algorithms (Thorson and Wells 2016). Some contend that these characteristics reduce the amount of selective news exposure, and increase the amount of incidental exposure to political news (Fletcher and Nielsen 2018; Wojcieszak et al. 2021). If this is the case, news use on social media may balance out the inequalities in news consumption typically found on traditional platforms (e.g., Holt et al. 2013; Stier et al. 2022; Newman et al. 2018). Building on these insights, we investigate to what extent social media news consumption acts as a leveler, counterbalancing traditional news consumption gaps at election times.

To study these matters we utilize the Norwegian Election Campaign Study of 2017 (NECS). The NECS is a four-wave individual-level panel survey that includes questions about news consumption on a broad range of media outlets. We take advantage of the panel structure of the data by running a series of latent growth models. These are panel-data models used to describe and explain changes over time in a given phenomenon of interest, such as individuals’ media use during election campaigns (Singer and Willett 2003).

The results suggest that election campaigns mostly have a mitigating effect on overall gaps in political media use. Well-known differences based on age, gender, education, and political interest either narrow or remain stable throughout the campaign. As for social media use specifically, we find that the young, women, and the less educated are the most active users, and these gaps increase throughout the campaign. Social media thus seems to act in a countervailing manner, providing the most political information to those who use traditional media news channels the least. In the concluding section, we critically discuss these results in terms of the quality of knowledge and informational gains obtained on social media platforms.

Theory

News Consumption Gaps in Campaigns—Divergence or Convergence?

A long line of research attributes variability in citizens’ media use to a combination of opportunity, motivation, and ability (e.g., Luskin 1990). Opportunity refers, in this context, to the availability of media content, motivation to people’s interest in seeking out this content, and ability to how skilled people are in retrieving and comprehending information. This trifold model of political behavior is known as the OMA framework. Motivation—political interest—is commonly viewed as the most important factor in explaining differences in news consumption at the individual level (Luskin 1990; Prior 2019). Numerous studies confirm that the politically interested consume more news than those who are less politically interested (Lecheler and de Vreese 2017; Shehata and Strömbäck 2011; Strömbäck et al. 2013).

During election campaigns, the opportunity structure for political media consumption changes dramatically (Strömbäck and Kaid 2008; van Aelst et al. 2012). Modern election campaigns are intense communication environments in which voters must filter through an ever-increasing stream of political information on both traditional outlets and social media (Owen 2017; Ørmen 2019). Figure 1 visualizes this changing political information environment in the case of the Norwegian 2017 parliamentary election. The graphs track the absolute number of news articles about the party leaders (black line) and the prime minister and opposition leader (dotted line) immediately before and during the campaign in four major news outlets. Evidently, the amount of content available about politics increases almost exponentially throughout the campaign and peaks on Election Day.

Figure 1.

Political news coverage during the 2017 Norwegian election campaign. The number of news stories mentioning all the party leaders and the prime minister and opposition leader in four major news outlets (Aftenposten, NRK, TV2, VG). Source: https://www.retriever.no. Analysis by the authors.

We are interested in how inequalities in political media consumption develop in this changing information environment. Based on previous research, we formulate two competing arguments. First, a divergence argument building on the seminal work of Tichenor et al. (1970) predicting that inequalities in media consumption will increase throughout the campaign. Second, a convergence argument based on Schudson’s (1998: 311–12) idea of the monitorial citizen predicting that gaps in media consumption will narrow throughout the campaign. As we will see, although the two theoretical arguments have contrasting empirical implications, political interest is central to both perspectives.

Starting with the divergence argument, Tichenor et al. (1970) argued that, when mass media information in society increases

… segments of the population with higher socioeconomic status tend to acquire this information at a faster rate than the lower status segments, so that the gap in knowledge between these segments tends to increase rather than decrease. (159–60)

Tichenor et al. (1970: 161–62) point to several factors that explain this development, but they highlight individuals’ motivation to seek out relevant information through selective exposure as the key factor (see also Stroud 2017). The role of individual motivation is also prominent in more recent work on media consumption in “high-choice environments” (Prior 2005, 2007). The transition to high-choice media environments has provided users with endless opportunities to find content matching their individual preferences. Anyone interested in politics can consume almost limitless amounts of news, while those more interested in entertainment, it is claimed, can devote their time entirely to non-political content (see Haugsgjerd et al. 2021 and van Aelst et al. 2012 for a discussion). Following this line of thought, during campaigns, the politically interested have even more opportunities to follow the news cycle, while the uninterested can easily avoid content about the election. Collectively, these perspectives lead us to expect greater differences in political media consumption throughout campaigns.

The convergence argument builds on Schudson’s (1998: 311–12) idea of the monitorial citizen, which is based on the notion that people pay more attention to political news when the circumstances require it. That is, less interested groups that usually pay little attention to politics may devote more of their attention to the news in periods when political alternatives are clearly articulated, the political stakes are high, media coverage is intense, and information is easily accessible (Zaller 2003). Those who are more interested in news and politics, who already consume large amounts of political news on a regular basis, may have fewer opportunities during campaigns to increase their already intense media consumption. Granted, Schudson’s original theory does not pertain to elections specifically but rather to extraordinary circumstances that attract citizens’ attention. Because national elections constitute extraordinary informational environments with high political stakes, we may expect that those who normally pay little attention to politics boost their news consumption during these periods. Indeed, some studies give credence to this assertion. For instance, Jerit et al. (2006) find that increased coverage of a topic on television benefits all educational groups equally (see also Hansen and Pedersen 2014; Holbrooke 2002; Ytre-Arne and Moe 2018).

Based on the above discussion, our first research question is whether inequalities in political news consumption based on political interest diverge or converge during election campaigns (RQ1).

Sociodemographic Differences in News Consumption

Previous research documents systematic differences in political media use between sociodemographic groups (see Esser and Steppat 2017 for a review). In regard to gender, the pattern is clear: women tend to consume less news than men do (Benesch 2012; Cohen 2013). Regarding age, older cohorts typically consume more news than younger cohorts do, but this is mostly restricted to the use of traditional media outlets (i.e., newspapers and television) (see Elvestad and Phillips 2018). Education is also a central predictor of media use, but this relationship is less straightforward than is often assumed (Esser and Steppat 2017). The highly educated tend to consume most news through newspapers, while the evidence is more mixed regarding TV news (Shehata and Strömbäck 2011).

As far as we know, there are no previous studies on how these differences in news consumption between sociodemographic groups develop throughout election campaigns. We know from existing research, however, that the highly educated, men, and older cohorts tend to be more politically interested than groups with lower education, women, and younger cohorts (Prior 2019). We also know that political interest is important in order to understand the abovementioned differences in political media use between socio-demographic groups (Benesch 2012; Webster 2014). The question, however, is—as discussed in the previous section—whether the high-information environment of contemporary election campaigns mostly boosts the media appetite of politically interested groups (i.e., the highly educated) or triggers news consumption motivations among those who were less interested to begin with.

Our second and third research questions are therefore whether news consumption gaps between sociodemographic groups increase or decrease throughout election campaigns (RQ2), and whether these developments can be attributed to variability in political interest between the groups (RQ3).

Social media as a Leveller?

In recent years, social media has emerged as an important news source (Fletcher and Nielsen 2018; Newman et al. 2018). More specifically, social media act as intermediaries between news outlets and citizens, as political information on these platforms is diffused and flows through active nodes in networks (Karlsen 2015; Stier et al. 2022; Wojcieszak et al. 2021). Consequently, news exposure on social media is curated by users’ behavior, other individuals in their networks, algorithms, and the interaction between these factors (Thorson and Wells 2016; Thorson et al. 2021). One scholarly tradition has focused on how social media therefore might increase individualized media diets and news avoidance, as algorithms potentially strengthen the effect that predispositions has on media consumption (Prior 2005, 2007; Kümpel 2020; Möller et al. 2020). Following this line of thought, social media will likely contribute to increased gaps in news consumption during election campaigns, strengthening the divergence argument we developed above.

Another perspective, however, argues that social media curation contributes to more diverse content exposure (Thorson and Wells 2016) and increases incidental news consumption (Fletcher and Nielsen 2018; Bergstrøm and Belfrage 2018)¹ that often leads to intentional news use (Nanz & Matthes 2022). From this perspective, social media will likely increase the levels of news consumption, also for groups less motivated to consume news (see e.g., Stier et al. 2022; Wojcieszak et al. 2021). Consequently, when the amount of news diffused on social media increases amid election campaigns (de Leon et al. 2021), people who use these platforms the most, will be more exposed to news. Following this line of thought, social media could strengthen the convergence argument and reduce traditional news consumption gaps during campaigns.²

Empirical studies indeed suggest that social media, to some extent, balance out differences in participation in civic and political life (Boulianne 2015), including news consumption gaps typically found in traditional news outlets (Holt et al. 2013).³ General usage patterns also support this notion: younger cohorts use social media for news more frequently than older cohorts (Shehata and Strömbäck, 2021), while the legacy news audience is aging (Westlund and Weibull 2013; see e.g. Esser and Steppatt, 2017 for a discussion). Studies have even found a reverse gender-gap on social media because women use this platform for news more often than men do (Bergstrøm and Belfrage 2018).

Building on these insights, we assess the possibility that traditionally low-consumption groups increase their social media consumption the most during the period leading up to an election. Hence, our final research question is whether social media use during the course of a campaign acts as a leveler, counterbalancing traditional news consumption gaps (RQ4).

Methods

Case Selection and Data

We use the Norwegian Election Campaign Study (NECS), which consists of four rounds of individual-level panel data from the 2017 Norwegian parliamentary election campaign. In Norway, parliamentary elections are fixed and occur on the second Monday in September every fourth year. The election campaign begins immediately after the end of the summer break, in mid-August (see, e.g., Hopmann and Karlsen 2021; Skogerbø and Karlsen 2021). Election campaigns receive massive media attention, and the party leaders participate in televised debates regularly. Norway has a media system characterized by a dominant public service broadcaster, and relatively high levels of news consumption, including news consumption on social media (Skogerbø and Karlsen 2021).

The NECS were collected through four web surveys in collaboration with Statistics Norway. The surveys were conducted before the election campaign, during the election campaign (twice), and immediately after the election. Each round of fieldwork lasted approximately one week.⁴ The NECS is based on a representative probability sample drawn from the official Norwegian electoral roll by Statistics Norway. Out of a gross sample of 10,000 respondents, 4,033 (40% response rate) participated in the first round. Only those respondents who participated in the first round were contacted again in the subsequent rounds. In the second round, 2,274 (56% response rate) respondents participated, and 2,157 (54% response rate) respondents participated in the third round. Ultimately, 1,836 (46% response rate) respondents participated in the final round. 1,555 respondents participated in all four rounds.⁵ Since attrition makes the sample less representative for each wave, we run our analysis on both the full sample (i.e., an unbalanced panel), allowing us to use as much as possible of the information in the data (see Singer and Willett 2003: 146–59 for a discussion), and on a balanced panel including only individuals present in all four waves (see Table A10 and Figure A2 in the supplementary information file).

Research Design

We analyze the NECS data by means of latent growth models (Singer and Willett 2003). These are panel-data models used to describe and explain changes over time in a given phenomenon of interest, such as individuals’ media use during election campaigns.⁶ We use a multilevel approach with random intercepts and random time trends. In this approach, each individual is conceived of as the upper level in the data hierarchy (level two), while observations of each individual are conceived of as the lower level (level one). A basic growth model can be represented by the equation

Y_{i t} = β_{0 i} + β_{1 i} T i m e + ε_{i t}

(1)

where the outcome variable Y (frequency of political media use) for a particular unit (i) at a particular point in time (_t) is modeled as a function of an intercept and time. In latent growth models, each individual has their own intercept, as well as their own slope (or rate of change) over time. This allows us to separate two types of variation in media use: variation in individuals’ media use before the campaign started, located at level two (individuals’ starting level), and within-individual changes in media use throughout the campaign, located at level one (individuals’ growth rate).

Individuals’ starting level of media use can be represented by the equation

β_{0 i} = γ_{00} + γ_{01} Z + μ_{0 i}

(2)

where β_0i is the intercept from equation (1), γ₀₀ is an intercept, Z is a variable that affects initial media use levels, γ₀₁ is the effect of Z on initial media use, and μ_0i is the unexplained variance in person i’s starting level. This part of the growth model facilitates the study of whether people in different groups (based on gender, age, education, and political interest) used the same amount (and types) of media to seek political information before the parliamentary election campaign began in 2017.

Individuals’ rate of change over time (their growth) is represented by the equation

β_{1 i} = γ_{10} + γ_{11} Z + μ_{1 i}

(3)

where β_1i is the slope parameter from equation (1), γ₁₀ is an intercept, Z is a predictor of the rate of change in media use, γ₁₁ is the effect of Z on the rate of change in media use, and μ_0i is the unexplained variance in person i’s growth rate. This part of the growth model facilitates the study of whether different groups’ (based on gender, age, education, and political interest) changes in media use throughout the campaign differ from one another.⁷

Measurements

We measure our dependent variables—the frequency of political news consumption—by constructing news consumption indexes based on the following question: “How often have you heard/seen/or read news stories about the election on [channel] this last week?” The respondents were presented with five response categories. In total, we asked about the respondents’ use of eleven outlets (see Table A1 in the supplementary information file for an overview). We used these questions to construct three indexes capturing the respondents’ self-reported (i) overall consumption of political news, (ii) consumption of political news on social media, and (iii) consumption of political news in traditional news outlets (TV, radio, and newspapers).

We compute two different versions of our three dependent variables. First, as we have an unequal number of outlets within each channel, our main approach is based on taking the maximum value of the outlets within a given channel (TV, radio, newspapers, and social media). Using television consumption as an example, if a respondent watched the main commercial alternative (TV2) news several times a day but the public broadcaster (NRK) news only once a week, the television variable takes the value from the main commercial alternative item (several times a day). We then compute a mean index of the four constructed channel variables. Second, as this first approach may be susceptible to ceiling effects, we also compute standard additive indexes based on all eleven outlets in our questionnaire. The likelihood of ceiling effects is smaller using this second approach as the distribution on this index has flatter tales than the distributions on the single items (fewer have extreme values at T1). Nevertheless, we cannot rule out the possibility that the results to some extent are affected by ceiling effects. We run all analyses on both versions of our dependent variables and report additional results in the supplementary information file (see Table A11 and Figure A3).

Regarding our explanatory variables, political interest is measured by the standard item in the literature (Prior 2019).⁸ Gender is a binary variable taken from the Norwegian official registry. The respondents’ levels of education are divided into three categories—primary school, secondary school, and university—and this information is also based on the official registry. Age is categorized into three groups: 19–29, 30–59, and 60 + .⁹

Analyses

We begin by investigating our first and second research question, that is, whether news consumption gaps based on political interest (RQ1) and sociodemographic characteristics (RQ2) diverge or converge during electoral campaigns. More precisely, we use the NECS data to run a series of bivariate latent growth models using age, education, gender, and political interest as explanatory variables. These four explanatory variables are each regressed on the total news consumption index. The results from these four models are presented in Table 1.¹⁰

Table 1.

Explaining Overall Political Media Consumption (1–5). Latent Growth Models.

		Total media use
		Gender	Education	Age	Political interest
Fixed effects
Initial status	Intercept	2.20**	2.09**	1.94**	1.33**
		(0.02)	(0.03)	(0.03)	(0.06)
	Male (ref: female)	0.18**
		(0.03)
	Secondary education (ref: primary)		0.22**
	Secondary education (ref: primary)		(0.04)
	Higher education (ref: primary)		0.28**
	Higher education (ref: primary)		(0.04)
	Age: 30–59 (ref: 18–29)			0.30**
				(0.04)
	Age: 60 + (ref: 18–29)			0.64**
				(0.04)
	Little politically interested (ref: not)				0.49**
	Little politically interested (ref: not)				(0.06)
	Somewhat politically interested (ref: not)				1.17**
	Somewhat politically interested (ref: not)				(0.06)
	Very politically interested (ref: not)				1.71**
	Very politically interested (ref: not)				(0.07)
Change rates	Early campaign (ref: before campaign)	0.46**	0.42**	0.42**	0.35**
	Early campaign (ref: before campaign)	(0.02)	(0.04)	(0.04)	(0.09)
	Late campaign (ref: before campaign)	0.57**	0.48**	0.51**	0.43**
	Late campaign (ref: before campaign)	(0.02)	(0.04)	(0.04)	(0.10)
	After election (ref: before campaign)	0.70**	0.67**	0.73**	0.40**
	After election (ref: before campaign)	(0.03)	(0.05)	(0.05)	(0.13)
	Male * early campaign	−0.09**
		(0.03)
	Male * late campaign	−0.15**
		(0.03)
	Male * after election	−0.14**
		(0.04)
	Secondary education * early campaign		0.02
	Secondary education * early campaign		(0.04)
	Secondary education * late campaign		−0.02
	Secondary education * late campaign		(0.05)
	Secondary education * after election		−0.12*
	Secondary education * after election		(0.06)
	Higher education * early campaign		−0.02
	Higher education * early campaign		(0.04)
	Higher education * late campaign		0.02
	Higher education * late campaign		(0.05)
	Higher education * after election		0.01
	Higher education * after election		(0.06)
	30–59 * early campaign			0.06
				(0.04)
	30–59 * late campaign			0.06
				(0.05)
	30–59 * after election			−0.02
				(0.06)
	60 + * early campaign			−0.12**
				(0.04)
	60 + * late campaign			−0.16**
				(0.05)
	60 + * after election			−0.26**
				(0.06)
	Little interested * early campaign				0.19*
	Little interested * early campaign				(0.10)
	Little interested * late campaign				0.17
	Little interested * late campaign				(0.11)
	Little interested * after campaign				0.40**
	Little interested * after campaign				(0.14)
	Somewhat interested * early campaign				0.03
	Somewhat interested * early campaign				(0.09)
	Somewhat interested * late campaign				0.02
	Somewhat interested * late campaign				(0.11)
	Somewhat interested * after campaign				0.18
	Somewhat interested * after campaign				(0.13)
	Very interested * early campaign				−0.17⁺
	Very interested * early campaign				(0.10)
	Very interested * late campaign				−0.21⁺
	Very interested * late campaign				(0.11)
	Very interested * after campaign				−0.03
	Very interested * after campaign				(0.14)
Variance components
Level 1	Within person	0.03**	0.03**	0.03**	0.03**
		(0.00)	(0.00)	(0.00)	(0.00)
Level 2	In initial status	0.53**	0.53**	0.49**	0.36**
		(0.02)	(0.02)	(0.02)	(0.01)
	In rate of change	0.21**	0.21**	0.21**	0.21**
		(0.00)	(0.00)	(0.00)	(0.00)
	Covariance	−0.29**	−0.30**	−0.25**	−0.19**
		(0.04)	(0.04)	(0.04)	(0.04)
Goodness-of-fit	Log lik.	−10,692	−10,623	−10,567	−10,075
N	Observations	10,165	10,126	10,165	10,154
	Number of individuals	3,948	3,926	3,948	3,939
	Waves	4	4	4	4

Standard errors in parentheses.

⁺

p < 0.10, * p < 0.05, ** p < 0.01.

As discussed in the methodology section, latent growth models partition the variation in the dependent variable into two parts, providing two sets of coefficients. The “initial status” coefficients reported in the upper half of Table 1 explain inter-individual variation. Specifically, they show the frequency of political news consumption in the different groups before the 2017 campaign began (the first time point in the data). The “change rate” coefficients reported in the lower half of the table show intra-individual variation. Specifically, they indicate at what rate and in what direction political media consumption changed for each group for each time point throughout the panel (one survey wave is one time point).

For ease of interpretation, we plot the predicted levels of political media consumption by gender, age, education, and political interest with 95 percent confidence intervals in Figure 2. Thus, the figures illustrate each group’s level of media consumption before the campaign started and their subsequent development until Election Day.

Figure 2.

Total news consumption before and during the campaign by gender, education, age, and political interest (based on Table 1).

Overall, Figure 2 shows that gaps in news consumption either narrow or remain stable during the course of the campaign. In the period leading up to the campaign, we see a social stratification pattern similar to what is often found in previous research (Esser and Steppat 2017). Women consume less political news than men, those with lower education levels consume less news than those with higher education levels, young people consume less news than the elderly, and the politically interested consume more news than the less interested. These initial gaps in media use, however, vary considerably in size. While the initial status gender coefficient is relatively small (0.18, p < 0.01), initial differences based on political interest amount to almost two entire scale points (1.71, p < 0.01) on the five points scale.

Then, during the campaign, all groups increase their news consumption, but not to the same degree. First, the gender gap closes. All three change rate coefficients capturing changes in gender differences are negative and significant, (t²-0.09, p < 0.01; t³-0.15, p < 0.01; t⁴-0.14, p < 0.01), and as Election Day approaches, women therefore consume the same amount of news as men. Second, the educational gap between those with high levels of education and those with low levels of education remains stable. However, the group with low levels of education catches up with the group with medium education levels towards the end of the campaign (t⁴-0.12, p = 0.04). Third, news consumption gaps based on age narrow. Specifically, the elderly (60 + ), who consume the most news before the campaign starts, have a less steep curve (t²-0.12, p = 0.01; t³-0.16, p < 0.01; t⁴-0.26, p < 0.01) than the young adults (19–29), who consumes the least news to begin with. Fourth, differences between the least and the most politically interested remain stable throughout the campaign (p > 0.05). However, even those who are not politically interested increases their absolute news consumption in the period leading up to the election (t²-0.35, p < 0.01; t³-0.43, p < 0.01; t⁴-0.40, p < 0.01).

Figure 2 demonstrates that political interest is key to understanding variation in news consumption in an election campaign setting. This finding raises the question of the extent to which the sociodemographic gaps in political media use also evident in Figure 2 are shaped by differences in political interest (RQ 3). In order to assess this question, we show in Figure 3 the gaps in news consumption by gender, education, and age after controlling for political interest.¹¹

Figure 3.

Total news consumption by gender, education, and age, controlling for political interest (based on Table A8 in the supplementary information file).

The results (Figure 3) indicate that most of the sociodemographic differences in news use evident in Figure 2 are strongly related to political interest (see Table A8 in the supplementary information file for complete results). First, we see that the gender gap in news consumption disappears entirely: the confidence intervals for men and women overlap at every time point. Second, the differences between educational groups are also clearly related to political interest. The gap between the three groups disappears both before and during the campaign when political interest is controlled for (p > 0.05). Political interest does not, however, explain age differences to the same extent. Here, the gap between the youngest and the oldest age group, though smaller, remains intact (the confidence intervals do not overlap) and resemble the pattern we see in Figure 2. We return to this latter result in the discussion.

Thus far we have analyzed people's overall news consumption. However, previous research made us suspect that developments in social media and traditional media may differ from one another. In order to find out, we compare usage patterns on social media and traditional media platforms (RQ4) and present the results in Figure 4 (see Table A6 and A7 in the supplementary information file for complete results).

Figure 4.

Comparing social and traditional media by gender, education, age, and political interest (based on Table A6 and A7 in the supplementary information file).

Figure 4 reveals major differences between the social profile of news use on social media and traditional media platforms. In fact, with respect to the sociodemographic characteristics, the profiles are almost opposite. While men, the highly educated and the elderly use traditional news media the most, women, the less educated, and the youngest age group are the most active users of social media and these gaps widen throughout the campaign. Women increase their news consumption to a greater extent than men on both traditional media (t²-0.08, p = 0.01; t³-0.15, p < 0.01; t⁴-0.12, p < 0.01) and social media (t²-0.10, p = 0.01; t³-0.15, p < 0.01; t⁴-0.19, p < 0.01). In traditional media, women almost close the initial gap to men, but on social media, women move away from men creating a reversed gender gap. There are small differences between educational groups on social media, but the group with low education has a steeper rise than the group with middle education towards the end of the campaign (t⁴-0.21, p = 0.01). As for age differences, the youngest age group narrows the gap to the oldest group somewhat on traditional media (t²-0.10, p = 0.04; t³-0.14, p = 0.01), while on social media the youngest group are the most active news users, and increase their consumption compared to the elderly (t²-0.32, p < 0.01; t³-0.47, p < 0.01; t⁴-0.42, p < 0.01). Finally, for political interest, the pattern before the campaign started is similar to what we observed for overall news consumption, and the gaps remain stable during the campaign for both traditional and social media.

While it is no big surprise that the youngest age group is the most active users of social media, the gender pattern on these platforms is more difficult to explain. We can only speculate, but previous studies suggest that the traditional gender news gap is partly due to women having less time to prioritize news (Benesch 2012). As discussed, news consumption on social media is curated (Thorson and Wells 2016) and incidental (Fletcher and Nielsen 2018). Thus, on these platforms, users do not necessarily need to set aside time to actively consume news (it will appear in their feed). The reversed gender gap evident on social media might therefore reflect general usage patterns on these platforms.

Discussion and Conclusion

We have investigated citizens’ linkages to the public sphere during a parliamentary election campaign, asking whether pre-existing political inequalities in media use are strengthened or reduced. Our results show that systematic differences in political media use indeed exist and, to some extent, persist during campaigns. Overall, the young, women, the less educated, and those that show little interest in politics consume less news than the elderly, men, the highly educated, and the politically interested. However, all groups speed up their news consumption as Election Day approaches. Importantly, high consuming groups do not increase their news use more strongly than low-consuming groups. To the extent that we do see changing patterns, news consumption differences become smaller: women increase their overall news use to the extent that the gender gap disappears. The gap based on age also becomes smaller during the course of the campaign. This type of convergence dynamic suggests that politically less-attentive citizens are not permanently detached from politics. Rather, echoing Schudson’s (1998) idea of the monitorial citizen: citizens pay attention when political alternatives are clearly articulated, the political stakes are high, and the political media coverage is intense (Zaller 2003).

The results also show that political interest is crucial to understanding sociodemographic differences in news consumption. When we control for political interest, the gender gap disappears, and educational gaps narrow considerably. This finding speaks to the general importance of motivation as a driver of news consumption (e.g., Luskin 1990; Benesch 2012; Webster 2014) and it suggests that political interest is particularly important for news use during electoral campaigns. In her study from the United States, Benesch (2012) found that political interest did not reduce the gender gap in news consumption. Rather, the dual burden of job and children was the most important driver. We can only speculate, but the campaign context of our study can likely help us understand these different results. Moreover, the comparatively generous Norwegian welfare state context might also reduce the importance of the job–children burden crucial in the U.S. context. That being said, motivational factors do not account for every social demographic difference in news use in our study. Age differences are to a lesser extent based on political interest. This latter finding echoes the literature on political participation, which shows that participation increases gradually as voters become more integrated into society and take on various adult obligations through, for instance, marriage, home ownership, and permanent employment (Highton and Wolfinger 2001).

Another essential finding is that developments on social media are distinct from traditional media. On social media, the sociodemographic profile is almost the opposite of what we find in traditional news outlets: the young, women, and to some extent, the less educated are the most active users. These gaps are small or insignificant before the campaign starts, but increase as Election Day approaches. Social media thus—to some extent—counterbalances the inequalities found in traditional news outlets by providing most political information to those groups who use traditional media channels the least. This finding is interesting as our study is conducted more than a decade after the introduction of social media in general. Our results therefore suggest that Holt et al.’s (2013) findings, that social media functions as a leveler with respect to news consumption inequalities in election campaigns, were not just due to social media being in its infancy in 2010 characterized by early adopters usage patterns (Rogers 1995). Our study shows that these patterns endure, which suggests that they are related to inherent qualities of the media technology, and specifically, how the network structure of social media enables curated flows (Thorson and Wells 2016).

The broader democratic implications of our findings, however, are not necessarily positive. Several studies show that social media are less effective than traditional news in informing citizens about politics and current affairs (Cacciatore et al. 2018; Shehata and Strömbäck 2021; Lee et al. 2022: 47). This relates to how news consumed on social media consists of a mix of content made by professional news providers and alternative information sources providing low-quality news (Neudert et al. 2019). Moreover, the news encountered online is not curated by news organizations, but rather personalized and tailored to consumers’ specific desires (Thorson and Wells 2016). People also tend not to read full news stories online due to the overload of information (van Erkel and van Aelst 2020). Finally, and importantly, a growing body of research suggests that misinformation and disinformation are widespread on social media (Allcott et al. 2019, Tandoc jr. et al. 2020), potentially misleading voters about key policy issues (Bode and Vraga 2015). Taken together, although social media create new linkages to the public sphere and include traditionally less active groups in democratic politics, the informational consequences of news exposure on social media in campaigns are more problematic in terms of knowledge gains and uninformed political participation (Lee et al. 2022). This is an essential topic for future research.

Although this study offers important insights into campaign dynamics, it has limitations worth noting. For one thing, it relies on self-reports to measure people's exposure to news. Previous research shows that people tend to overreport their exposure to both traditional news and news on social media (Prior 2009). As the focus of this article is on relative differences between groups (and not absolute levels), overreporting is less of a problem. However, it is possible that overreporting is not randomly distributed between social groups. Still, although several demographic factors have been shown to affect reporting accuracy, there is less agreement about the nature of these associations (Vraga and Tully 2020: 553). We, therefore, encourage future research to assess campaign dynamics also with behavioral measures of news exposure.

Finally, similar studies from other countries are necessary to determine the extent to which our findings are context-specific. In Norway, Internet penetration, social media use, and online news subscription are comparatively high (Newman et al. 2018). Moreover, the country is characterized by comparatively low levels of political and economic inequality. These factors, together with a media system dominated by a quality public broadcaster, providing citizens with quality news content in prime time, may constitute a mix of factors preventing diverging patterns in political media use. An important task for future research is to comparatively investigate the political dynamics of contemporary electoral campaigns.

To conclude, the results show that all groups increase their political media use as Election Day approaches and that we do not see increasing but rather somewhat reduced gaps in news consumption. Elections campaigns, then, to some extent, equalize inequalities in political news consumption when it matters the most.

Supplemental Material

sj-docx-1-hij-10.1177_19401612221112014 - Supplemental material for Election Campaigns, News Consumption Gaps, and Social Media: Equalizing Political News Use When It Matters?

Supplemental material, sj-docx-1-hij-10.1177_19401612221112014 for Election Campaigns, News Consumption Gaps, and Social Media: Equalizing Political News Use When It Matters? by Atle Haugsgjerd and Rune Karlsen in The International Journal of Press/Politics

Footnotes

ORCID iD

Rune Karlsen

Declaration of Conflicting Interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This work was supported by the Research Council of Norway (grant number 249687).

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Atle Haugsgjerd (PhD in Political Science, University of Oslo) is a Senior Research Fellow at the Institute of Social Research, Oslo, Norway. He is currently also a Postdoctoral Fellow at the University of Amsterdam in connection to the research project Challenges to Democratic Representation. His research interests include political communication, studies of voter behavior, and political trust and welfare politics, and he has published in journals like Electoral Studies, Political Communication, and West European Politics.

Rune Karlsen is a Professor at the Department of Media and Communication, University of Oslo, and a Research Professor at the Institute of Social Research, Oslo, Norway. His research interests include political communication, elections, election campaigns, media effects, and political elites. He has edited the book Parliamentary Candidates between Voter and Parties. A Comparative Perspective (with Lieven De Winter and Hermann Schmitt) and authored the book After the Mass Party (with Elin Allern and Knut Heidar). He has published in journals like the British Journal of Political Science, Political Communication, and West European Politics.

References

Allcott

Gentzkow

. 2019. “Trends in the Diffusion of Misinformation on Social media.” Research and Politics 6(2): 1–8. https://doi.org/10.1177/2053168019848554.

Barabas

Jerit

. 2009. “Estimating the Causal Effects of Media Coverage on Policy-Specific Knowledge.” American Journal of Political Science 53(1): 73–89. http://dx.doi.org/10.1111/ajps.2008.53.issue-1.

Bartels

L. M.

2006. “Priming and Persuasion in Presidential Campaigns.” In Capturing Campaign Effects, ed. Brady

H. E.

Richard

, 78–114. Ann Arbor: University of Michigan Press.

Benesch

2012. “An Empirical Analysis of the Gender Gap in News Consumption.” Journal of Media Economics 25(3):147–67. https://doi.org/10.1080/08997764.2012.700976.

Bergstrøm

Belfrage

J. M.

. 2018. “News in Social Media. Incidental Consumption and the Role of Opinion Leaders.” Digital Journalism 6(5):583–98. https://doi.org/10.1080/21670811.2018.1423625.

Bode

Vraga

E. K.

. 2015. “In Related News, That was Wrong: The Correction of Misinformation Through Related Stories Functionality in Social Media.” Journal of Communication 65(4):619–38. https://doi.org/10.1111/jcom.12166.

Boulianne

2015. “Social Media Use and Participation: A Metaanalysis of Current Research, Information.” Communication & Society 18(5):524–38. https://doi.org/10.1080/1369118X.2015.1008542.

Cacciatore

M. A.

Yeo

S. K.

Scheufele

D. A.

Xenos

M. A.

Brossard

Corley

E. A.

. 2018. “Is Facebook Making Us Dumber? Exploring Social Media Use as a Predictor of Political Knowledge.” Journalism & Mass Communication Quarterly 95(2):404–24. https://doi.org/10.1177%2F1077699018770447.

Claassen

R. L.

2011. “Political Awareness and Electoral Campaigns: Maximum Effects for Minimum Citizens?” Political Behavior 33:203–23. https://doi.org/10.1007/s11109-010-9129-6.

10.

Cohen

A. A.

, ed. 2013. Foreign News on Television: Where in the World is the Global Village? New York, NY: Peter Lang.

11.

Dahl

R. A.

1998. On Democracy. New Haven: Yale University Press.

12.

de Leon

Ernesto

Vermeer

Trilling

. 2021. “Electoral News Sharing: A Study of Changes in News Coverage and Facebook Sharing Behaviour During the 2018 Mexican Elections.” Information, Communication & Society: 1–17. https://doi.org/10.1080/1369118X.2021.1994629.

13.

Elvestad

Phillips

. 2018. Misunderstanding News Audiences. Seven Myths of the Social Media Era. London: Routledge.

14.

Esser

Steppat

. 2017. “News Media Use: International Comparative Research.” In The International Encyclopedia of Media Effects, ed. Rössler

Hoffner

C. A.

Zoonen

, 1–17. Chichester, Malden, MA: Wiley and Sons.

15.

Fletcher

Nielsen

R. K.

. 2018. “Are People Incidentally Exposed to News on Social Media? A Comparative Analysis.” New Media & Society 20(7):2450–68. https://doi.org/10.1177/1461444817724170.

16.

Hansen

K. M.

Pedersen

R. T.

. 2014. “Campaigns Matter: How Voters Become Knowledgeable and Efficacious During Election Campaigns.” Political Communication 31(2):303–24. https://doi.org/10.1080/10584609.2013.815296.

17.

Haugsgjerd

2019. “Moderation or Radicalisation? How Executive Power Affects Right-Wing Populists’ Satisfaction with Democracy.” Electoral Studies 57:31–45. https://doi.org/10.1016/j.electstud.2018.09.008.

18.

Haugsgjerd

Hesstvedt

Karlsen

. 2021. “Increased Media Choice and Political Knowledge Gaps: A Comparative Longitudinal Study of 18 Established Democracies 1995–2015.” Political Communication 38(6):731–50. https://doi.org/10.1080/10584609.2020.1868633.

19.

Heiss

Matthes

. 2019. “Does Incidental Exposure on Social Media Equalize or Reinforce Participatory Gaps? Evidence from a Panel Study.” New Media & Society 21:11–2. https://doi.org/10.1177/1461444819850755.

20.

Highton

Wolfinger

R. E.

. 2001. “The First Seven Years of the Political Life Cycle.” American Journal of Political Science 45(1):202–9. https://doi.org/10.2307/2669367.

21.

Holbert

R. L.

2005. “Back to Basics: Revisiting, Resolving, and Expanding Some of the Fundamental Issues of Political Communication Research.” Political Communication 22(2):511–14. doi:10.1080/10584600500311436.

22.

Holbrook

T. M.

2002. “Presidential Campaigns and the Knowledge Gap.” Political Communication 19(4):437–54.

23.

Holt

Shehata

Strömbäck

Ljungberg

. 2013. “Age and the Effects of News Media Attention and Social Media use on Political Interest and Participation: Do Social media Function as Leveller?” European Journal of Communication 28(1):19–34.

24.

Hooghe

Stiers

. 2016. “Elections as a Democratic Linkage Mechanism: How Elections Boost Political Trust in a Proportional System.” Electoral Studies 44:46–55. https://doi.org/10.1016/j.electstud.2016.08.002.

25.

Hopmann

D. N.

Karlsen

. 2021. “Elections and Political Communication in the Nordic Countries.” In Power, Communication, and Politics in the Nordic Countries, ed. Skogerbø

Ihlen

Kristensen

N. N.

Nord

, 219–40. Gothenburg: Nordicom.

26.

Jerit

Barabas

Bolsen

. 2006. “Citizens, Knowledge, and the Information Environment.” American Journal of Political Science 50:266–82. https://doi.org/10.1111/j.1540-5907.2006.00183.x.

27.

Karlsen

2015. “Followers are Opinion Leaders. The Role of People in the Flow of Political Communication on and Beyond Social Networking Sites.” European Journal of Communication 30(3):301–18. https://doi.org/10.1080/1369118X.2021.1994629.

28.

Kümpel

A. S.

2020. “The Matthew Effect in Social Media News Use: Assessing Inequalities in News Exposure and News Engagement on Social Network Sites (SNS).” Journalism 21(8):1083–98. https://doi.org/10.1177%2F1464884920915374.

29.

Lecheler

de Vreese

C. H.

. 2017. “News Media, Knowledge, and Political Interest: Evidence of a Dual Role from a Field Experiment.” Journal of Communication 67(4):545–64. https://doi.org/10.1111/jcom.12314.

30.

Lee

Yamamoto

Tandoc

E. C.

. 2022. “Why People Who Know Less Think They Know About COVID-19: Evidence from US and Singapore.” Journalism & Mass Communication Quarterly 99(1):44–68. https://doi.org/10.1177%2F10776990211049460.

31.

Luskin

R. C.

1990. “Explaining Political Sophistication.” Political Behavior 12(4):331–61. https://doi.org/10.1007/BF00992793.

32.

Möller

Van de Melde

Merten

Psuchman

. 2020. “Explaining Online News Engagement Based on Browsing Behavior: Creatures of Habit?” Social Science Computer Review 38(5):616–32. https://doi.org/10.1177%2F0894439319828012.

33.

Nanz

Matthes

. 2022. “Democratic Consequences of Incidental Exposure to Political Information: A Meta-Analysis.” Journal of Communication 72(3): 345–73. https://doi.org/10.1093/joc/jqac008.

34.

Neudert

Howard

Kollanyi

. 2019. “Sourcing and Automation of Political News and Information During Three European Elections.” Social Media + Society 5(3):1–13. https://doi.org/10.1177%2F2056305119863147.

35.

Newman

Fletcher

Kalogeropoulos

Levy

D. A. L.

Nielsen

R. K.

. 2018. Reuters Institute Digital News Report 2018. Oxford: Reuters Institute for the Study of Journalism.

36.

Norris

2000. A Virtuous Circle: Political Communications in Postindustrial Societies. Cambridge: Cambridge University Press.

37.

Ørmen

2019. “From Consumer Demand to User Engagement: Comparing the Popularity and Virality of Election Coverage on the Internet.” The International Journal of Press/Politics 24(1):49–68. https://doi.org/10.1177%2F1940161218809160.

38.

Owen

2017. “New Media and Political Campaigns.” In The Oxford Handbook of Political Communication, ed. Kenski

Jamieson

K. H.

, 823–36. Oxford: Oxford University Press.

39.

Prior

2005. “News vs. Entertainment: How Increasing Media Choice Widens Gaps in Political Knowledge and Turnout.” American Journal of Political Science 49(3):577–92. https://doi.org/10.1111/j.1540-5907.2005.00143.x.

40.

Prior

2007. Post-Broadcast Democracy: How Media Choice Increases Inequality in Political Involvement and Polarizes Elections. Cambridge: Cambridge University Press.

41.

Prior

2009. “The Immensely Inflated News Audience: Assessing Bias in Self-Reported News Exposure.” Public Opinion Quarterly 73(1):130–43. https://doi.org/10.1093/poq/nfp002.

42.

Prior

2019. Hooked. How Politics Capture People’s Interest. Cambridge: Cambridge University Press.

43.

Rogers

E. M.

1995. Diffusion of Innovations. New York: Free Press.

44.

Schudson

1998. The Good Citizen: A History of American Civic Life. New York: The Free Press.

45.

Shehata

Strömbäck

. 2011. “A Matter of Context: A Comparative Study of Media Environments and News Consumption Gaps in Europe.” Political Communication 28(1):110–34. https://doi.org/10.1080/10584609.2010.543006.

46.

Shehata

Strömbäck

. 2021. “Learning Political News from Social Media: Network Media Logic and Current Affairs News Learning in a High-Choice Media Environment.” Communication Research 48(1):125–47. https://doi.org/10.1177/0093650217749354.

47.

Singer

J. D.

Willett

J. B.

. 2003. Applied Longitudinal Data Analysis. New York: Oxford University Press.

48.

Skogerbø

Karlsen

. 2021. “Media and Politics in Norway.” In Power, Communication, and Politics in the Nordic Countries, ed. Skogerbø

Ihlen

Kristensen

N. N.

Nord

, 91–111. Gothenburg: Nordicom.

49.

Stier

Mangold

Scharkow

Breuer

. 2022. “Post Post-Broadcast Democracy? News Exposure in the Age of Online Intermediaries.” American Political Science Review 116(2): 768–74. http://dx.doi.org/10.1017/S0003055421001222.

50.

Strömbäck

Djerf-Pierre

Shehata

. 2013. “The Dynamics of Political Interest and News Media Consumption: A Longitudinal Perspective.” International Journal of Public Opinion Research 24(4):414–35. https://doi.org/10.1093/ijpor/eds018.

51.

Strömbäck

Kaid

L. L.

, eds. 2008. The Handbook of Election News Coverage Around the World. New York: Routledge.

52.

Stroud

N. J.

2017. “Selective Exposure Theories.” In The Oxford Handbook of Political Communication, ed. Kenski

Jamieson

K. H.

. Oxford: Oxford University Press.

53.

Tandoc jr

E. C.

Lim

Ling

. 2020. “Diffusion of Disinformation: How Social Media Users Respond to Fake News and Why.” Journalism 21(3):381–98. https://doi.org/10.1177%2F1464884919868325.

54.

Thorson

2020. “Attracting the News: Algorithms, Platforms, and Reframing Incidental Exposure.” Journalism 21(8):1067–82. https://doi.org/10.1177%2F1464884920915352.

55.

Thorson

Cotter

Medeiros

Pak

. 2021. “Algorithmic Inference, Political Interest, and Exposure to News and Politics on Facebook.” Information, Communication & Society 24(2):183–200. https://doi.org/10.1080/1369118X.2019.1642934.

56.

Thorson

Wells

. 2016. “Curated Flows: A Framework for Mapping Media Exposure in the Digital Age.” Communication Theory 26(3):309–28. https://doi.org/10.1111/comt.12087.

57.

Tichenor

P. J.

Donohue

G. A.

Olien

C. N.

. 1970. “Mass media Flow and Differential Growth in Knowledge.” Public Opinion Quarterly 34(2):159–70. https://doi.org/10.1086/267786.

58.

van Aelst

Thorbjørnsrud

Aalberg

. 2012. “The Political Information Environment During Election Campaigns.” In How Media Inform Democracy. A Comparative Approach, ed. Aalberg

Curran

, 34–51. London: Routledge.

59.

van der Meer

T. W. G.

Walter

Aelst

P. V.

. 2016. “The Contingency of Voter Learning: How Election Debates Influenced Voters’ Ability and Accuracy to Position Parties in the 2010 Dutch Election Campaign.” Political Communication 33(1):136–57. https://doi.org/10.1080/10584609.2015.1016639.

60.

Van Erkel

van Aelst

. 2021. “Why Don’t We Learn from Social Media? Studying Effects of and Mechanisms Behind Social Media News Use on General Surveillance Political Knowledge.” Political Communication 38(4): 407–25. https://doi.org/10.1080/10584609.2020.1784328.

61.

Vraga

E. K.

Tully

. 2020. “Who is Exposed to News? It Depends on How You Measure: Examining Self-Reported Versus Behavioral News Exposure Measures.” Social Science Computer Review 38(5):550–66. https://doi.org/10.1177/0894439318812050.

62.

Webster

J. G.

2014. The Marketplace of Attention: How Audiences Take Shape in a Digital Age. Cambridge, MA: MIT Press.

63.

Westlund

Weibull

. 2013. “Generation, Life Course and News Media Use in Sweden 1986–2011.” Northern Lights 11(1):147–73. https://doi.org/10.1386/nl.11.1.147_1.

64.

Wojcieszak

Menchen-Trevino

Goncalves

J. F. F.

Weeks

. 2021. “Avenues to News and Diverse News Exposure Online: Comparing Direct Navigation, Social Media, News Aggregators, Search Queries, and Article Hyperlinks.” The International Journal of Press/Politics: 1–27. https://doi.org/10.1177/19401612211009160.

65.

Ytre-Arne

Moe

. 2018. “Approximately Informed, Occasionally Monitorial? Reconsidering Normative Citizen Ideals.” The International Journal of Press/Politics 23(2):227–46. https://doi.org/10.1177/1940161218771903.

66.

Zaller

2003. “New Standard of News Quality: Burglar Alarms for the Monitorial Citizen.” Political Communication 20(2):109–30. https://doi.org/10.1080/10584600390211136.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.21 MB