House Members on the News: Local Television News Coverage of Incumbents

Abstract

The accountability relationship between voters and elected members of Congress (MCs) hinges on the potential for citizens to learn about legislator behaviour. In an era of declining local newspapers, local television coverage of MCs potentially fulfils this important role. However, few studies have comprehensively examined the determinants of contemporary MC coverage by local television news broadcasts. In this paper, we leverage a vast database of local television news broadcast transcripts spanning two years to identify which factors explain MC coverage. We find that MCs receive little coverage outside the general election campaign season. Media market and campaign-specific factors are associated with more exposure when coverage occurs. Finally, we find that within competitive elections, incumbents receive only a marginal advantage in coverage. These findings provide a springboard to explore further questions regarding Congress, local media, and political accountability.

For democratic accountability mechanisms to function, the voting public must be informed about their elected representatives’ actions (Prior 2014). Yet, research on contemporary information environments regarding legislator performance is scarce. In particular, can voters rely on the media to learn about their members of Congress (MCs) and candidates for that office? Prior work has focused mainly on local newspaper coverage of MCs. While local newspapers' circulations continue to decline and cable television and national broadcasts remain salient news outlets, local television news remains the most common local news source for most Americans.¹ Can citizens learn about incumbent MCs from television news coverage? Given how news is produced, do incumbents dominate coverage or can credible challengers garner sufficient coverage to become known to voters? Furthermore, what factors shape coverage patterns across candidates, districts, and stations?

We provide what we believe is the first near-comprehensive descriptive analysis of how frequently MCs are featured on contemporary local news broadcasts. While previous studies have addressed how changes in local media ownership structure influence news reporting patterns (for example, Martin and McCrain 2019), few have specifically examined incumbent coverage on local television in and outside of a campaign (although, see Dunaway 2008; Fowler et al. 2007; Hale, Goldstein, and Kaplan 2005; Schaffner and Gadson 2004). Using a novel data set of local television news broadcasts, we demonstrate that locally elected incumbent MCs receive little coverage from their local television press. During election season, however, we observe sharp increases in airtime devoted to incumbents. Focusing on this period, we investigate how geographic congruence between the boundaries of a legislator's district and the district's media market, the competitiveness of elections, and corporate ownership structure affect the frequency of coverage of incumbents and challengers.

Our approach and findings are important in several respects. First, we find little evidence that incumbents are advantaged large-scale vis-à-vis credible challengers. This lack of coverage has implications for citizens' knowledge of political actors. Limiting coverage to campaign season and ‘horse race’ dynamics could disincentivize attentiveness to local constituencies if roll-call votes and casework remain unreported, perhaps further nationalizing elections (Hopkins 2018). Second, MCs in districts that overlap substantially with a media market's boundaries are more likely to be covered than those in metropolitan markets that extend across many districts. Unlike newspapers that publish subregion-specific content (see Arnold 2004), television broadcasts are largely ‘one size fits all’, exacerbating these pressures. Finally, our approach encourages a more textual analysis of coverage. Unlike previous work that relied on limited broadcast recordings, the tools we use can capture nearly all stations on a continuing basis. This study serves as a foundation to understand the electorate's political information environment better and lay the groundwork to answer essential questions regarding Congress and the media.

Local media Coverage of MCs

Previous studies of how local press cover MCs primarily analyzed print newspapers (Arnold 2004; Dunaway 2008; Dunaway 2013; Fogarty 2008; Fogarty 2011; Gershon 2012a; Gershon 2012b; Hall and Lim 2018; Hayes and Lawless 2015; Schaffner and Sellers 2003; Snyder and Stromberg 2010; Vinson 2003). Many television studies focus on MCs' appearances on national television rather than local news (for example, Dietrich, Schultz, and Jaquith 2019; Kuklinski and Sigelman 1992). Earlier studies of local television occurred in a vastly different media landscape; however, they were novel in collecting recordings from local broadcast stations (thirty-three and eight stations, respectively) (Hess 1991; Vinson 2003). More recent work uses broadcast news transcripts to circumvent recording challenges (Ansolabehere, Snowberg, and Snyder 2006; Moskowitz 2021; Schaffner 2006). Overall, past work highlights television's potential as a news source about MCs, but most evidence is dated or limited in the number of stations considered. Nor has analysis been undertaken to understand the relative advantage in coverage that incumbents enjoy vis-à-vis challengers, both on average or across districts and races with different characteristics. While studies outside of the US have addressed legislator appearances on television, they are typically limited to national broadcasts (Amsalem et al. 2020; Tsfati, Elfassi, and Waismel-Manor 2010) or based upon perceptions of media coverage by survey respondents (Elmelund-Praestekaer, Hopman, and Norgaard 2011; Maier and Nai 2020).

A consistent finding across media studies is that the frequency of MC's coverage varies across legislators, markets, and electoral contexts (Arnold 2004; Vinson 2003). When a congressional district and a media market's borders are more congruent, the market's press has more incentive to cover the incumbent since they represent more of the outlet's potential audience (Ansolabehere, Snowberg, and Snyder 2006; Arnold 2004; Levy and Squire 2000; Manheim 1974; Moskowitz 2021; Schaffner 2006; Schaffner and Sellers 2003; Snyder and Stromberg 2010; Vinson 2003). Additionally, representing more of a market translates to fewer rival MCs competing for coverage.

Local media ownership structure may also influence the coverage of MCs. Profit drives corporate media ownership, while political ideology may motivate independently owned outlets (Hamilton 2004). Corporately owned media may cut the resources necessary to regularly cover incumbent MCs, relying on national wire services instead. Empirically, congressional reporting from corporately owned newspapers is less substantive, focusing more on the horse race aspects of elections (Arnold 2004; Dunaway 2008; Dunaway 2013) and mentions incumbents less frequently (Schaffner and Sellers 2003). This relationship is particularly interesting considering that between 2004 and 2020, the number of local television affiliates owned by the five businesses that owned the most local affiliates (Sinclair, Nexstar, Gray, Tegna, and Tribune) has more than tripled. While previous work has not taken up this question directly, using a large data set of local television news transcripts from 2017 through 2018, Martin and McCrain (2019) demonstrated that those affiliates purchased by Sinclair Broadcasting in 2017 provided significantly less coverage to local politics following the ownership change.

District competitiveness is also associated with greater newspaper coverage. Incumbents facing no challenger or only a token challenger typically receive less coverage. Conversely, more viable candidates who raise and spend more money gain greater visibility from the local press (Gershon 2012a; Gershon 2012b; Snyder and Stromberg 2010). Incumbents or challengers who raise large sums of money can signal to the local press that their campaign is intense and worth covering (Gershon 2012b). Additionally, they can afford to hold more campaign events that the local press may attend. Swing districts are associated with more coverage for incumbents (Arnold 2004; Gershon 2012a; Hayes and Lawless 2015). Local newspapers are drawn to campaigns' horse race aspects (Kahn 1991). As a result, they are more likely to cover incumbents during the final months of a general election campaign. For example, Arnold (2004) found that while the average MC received between 14 and 15 mentions per month in his sample, the average number of newspaper mentions jumped to 19.6 and 24.4 in September and October 1994, respectively.

Data and Methods

Our primary dependent variable is the frequency of MC mentions during local news broadcasts in their home district. This measure is constructed from local news transcripts collected by a private company, TVEyes. TVEyes monitors radio and television broadcasts in every local American market. For their project monitoring local news broadcasts, Martin and McCrain (2019) obtained from TVEyes records of each local news broadcast from television stations in each media market. Martin and McCrain generously supplied us with their raw data for this project. Their study resulted in a database containing the transcripts for local news broadcasts in six different periods throughout 2017 and 2018. The periods cover 2017 (3/10/17–9/5/17, 9/8/17–10/10/17, and 10/25/17–12/8/17), the pre-general election campaign in 2018 (1/10/18–2/9/18 and 2/21/18–5/8/18), and the 2018 general election campaign season (9/7/18–10/1/18). While off-year elections are typically less salient in the eyes of the public than presidential elections, research suggests that congressional races receive more news coverage during the midterms (for example, Kahn 1991).

Transcripts for each station's newscasts were segmented into 2.5-minute intervals. We counted the number of blocks that mentioned the incumbent or challenger in each relevant district. For example, if an incumbent is mentioned within the first and last minute of a half-hour transcript, they are coded as being mentioned twice on that station; if their name appears two times within the first minute of the transcript, they are coded as only being mentioned once. While the data are intended to capture the content of local news coverage, the intervals may include campaign advertisements. Therefore, we omit intervals mentioning the incumbent's or the challenger's name that include the phrase ‘approve this message’ because candidates are required to provide such an approval in a campaign advertisement by the Bipartisan Campaign Reform Act (BCRA).² Excluding these intervals has little effect on our counts during the months leading up to the general election campaign but decreases the average counts by 35 per cent for incumbents and 53 per cent for challengers.

We measure competitiveness in three ways. First, we use a transformation of the Cook Presidential Vote Index (PVI) as calculated following the 2016 presidential election for each district. District PVI is calculated by taking the average of the difference between the two-party presidential vote averages for the Democrats and Republicans for the two previous presidential elections. Scores are reported as being an advantage for one party (for example, R + 1 indicates a district that leans only slightly favourable to Republicans). We take the absolute value of the PVI. Higher values indicate less competitive districts.

Second, we gathered disbursements for both challengers and incumbents for the entire 2018 congressional election cycle, as reported in millions of dollars, from the Federal Election Commission (FEC). To account for the possibility that one-sided races are less newsworthy, we test for the effect of the absolute difference in spending between the Republican and Democratic candidates in a given race. One-sided spending is distinct because it likely does not indicate a competitive race worth covering.

Finally, we collected data on the types of challengers and the types of incumbents. We create an indicator variable for incumbents facing a general election challenge. We divide challengers into two groups: viable challengers, coded as those who spent at least $500,000 (mean spending is $2.21 million), and not viable challengers, those who spent less than $500,000 (mean spending is $79,025). We also identify which incumbents were unchallenged in the general election and the incumbents who did not run for re-election, subsetting the latter groups further into those who ran for higher office (that is, governor, senator, or another statewide office) and those who did not. For challengers, we create an indicator variable for open-seat elections. We also create a variable measuring primary competitiveness. We identify competitive primaries as elections where the general election candidate finished with a margin of less than 20 percentage points relative to their next closest competitor. Uncompetitive primaries have margins of 20 points or larger, including uncontested primaries.

We mapped districts to media markets using data from Daily Kos and dishuser.org. This allows us to identify which stations cover any portion of a district and identify which names we should search for in each station's transcripts. We calculate Station District Congruence for each MC-station dyad for which a district is any proportion of that station's media market. This measure is calculated as the proportion of a station's viewership population that lives in a member's district. A score of 1 means that all of a station's viewership lives in a member's district (for example, 100 per cent of the Bangor, ME media market is contained in ME-2). By contrast, scores closer to 0 indicate that the district is only a small proportion of the station's viewership, which can happen in large urban areas with many districts (for example, New York City) or if only part of a district lies in a market (see, for example, Levy and Squire 2000; Schaffner and Sellers 2003; Snyder and Stromberg 2010).

Finally, we gathered local affiliate ownership data from the Federal Communications Commission (FCC) licensing data. While previous work focused on the differences between corporately and privately owned newspapers, this approach is infeasible when analyzing local television stations because corporations currently own the overwhelming majority of stations (73 corporations own 702 local American television stations). Instead, we focus on whether stations owned by larger corporations cover incumbents and challengers differently from those stations owned by companies that own fewer stations. We created a dichotomous indicator for Large Media Corporation, which indicates whether a station is owned by a company that owns more than 5 per cent of all stations. Eleven companies owned more than 5 per cent of local affiliates, accounting for over 70 per cent of all stations (Only 4 per cent of local television stations are individually owned).

We examine coverage rates at the station/district dyad level separately for incumbents and challengers over time and by our different categorizations of race-level features. This choice allows us to understand patterns in coverage, differences between incumbents and challengers, and how these patterns vary by race (for example, incumbent versus viable challenger versus open seat races). We then turn to multivariate regression to examine the effect of multiple predictors of coverage. We focus on coverage during the fall election season.

Results

Table 1 displays the descriptive statistics for the frequency of coverage for MCs in the 115th Congress. The values reported in the table are the average number of 2.5-minute segments mentioning an incumbent per week for a station in a media market that serves their district in the three periods we present: 2017 (3/10/17–12/8/17), 2018 in the pre-general election period (1/10/18–5/8/18), and the 2018 general election period (9/7/18–10/1/18). The average station covering an incumbent mentions them in 0.38 2.5-minute segments per week in a non-election year. This frequency increased slightly in early 2018 to 0.79 but then increased to 5.78 segments in September.

Table 1. Incumbent House members: Mean number of mentions per week per station

Cells present the mean number of 2.5-minute segments in which a local television station's news programme mentioned an incumbent MC whose district was included in its media market per week. Standard deviations are in parentheses.

Table 1 also displays how these mentions are related to electoral context. First, we find minimal difference in the earliest period between incumbents who will or will not face a challenge in the general election. As time progresses, those facing challengers and running for re-election are much more likely to be discussed (5.01 mentions) compared to those unopposed (1.03 mentions). Candidates facing less well-funded challengers are discussed about 50 per cent less than those facing stronger challengers in the earlier months. In the general election, incumbents facing a viable challenger were discussed 11.61 times per week, while incumbents facing less viable challengers were discussed only 1.84. That is, in the pre-general election campaign period, incumbents facing viable challengers were discussed about twice as much as incumbents facing weak challengers. By the general election, the former were discussed about 6.3 times more than the latter. Incumbents not running for re-election are discussed frequently. We find that an incumbent running for higher office received 26.2 mentions per week in September, compared to 1.98 mentions for an incumbent not running for higher office (which is about the same as the coverage an incumbent running against a non-viable challenger attracts). See Appendix Table A5 for general election challenger descriptive statistics.³

Table 2 investigates the effect of structural, district, and candidate-level factors in explaining incumbent coverage during the general election campaign season.⁴ Greater geographic congruence between the incumbent's district and the media market is associated with significantly more coverage. Column I demonstrates that a district perfectly congruent with the media market (a value of 1) is associated with 58.77 more mentions. A one standard deviation in congruence (that is, an increase of 0.23) is associated with an increase of 13.52 mentions. This increase is roughly one-quarter of the standard deviation of mentions for incumbents (which is 49.74). The models predict a relatively more modest relationship with respect to media ownership: incumbents will receive 3.69 to 5.58 more mentions from stations owned by large media corporations, roughly 0.07 to 0.11 of the outcome variable's standard deviation. These findings are largely consistent with previous work, which suggests that more congruent legislative districts and media markets are associated with higher levels of coverage for representatives (for example, Ansolabehere, Snowberg, and Snyder 2006; Snyder and Stromberg 2010, but see also Levy and Squire 2000 for similar implications with challengers).

Table 2. Predicting coverage of incumbents in all districts (Autumn 2018)

Models were estimated using ordinary least squares. The unit of analysis is a station-incumbent pair. DV = Number of 2.5-minute segments by a given station in an incumbent's district that mention the incumbent. Standard Errors are clustered by incumbent. * = p < 0.05.

We find consistent evidence that district partisan composition predicts coverage: incumbents in less competitive districts are less likely to receive coverage. This effect may be mediated by candidate spending, per column VI. Coverage of incumbents running for higher office surges during the general election. On average, Column V's model estimates that they receive roughly 68.22 more mentions than an incumbent who faces no challengers. This effect is more than double that of an incumbent who faces a challenger who spent more than $500,000. Those incumbents facing non-viable challengers are predicted to receive coverage that is not statistically distinct from an unchallenged incumbent. Finally, in Column VI, we find that both incumbent and challenger spending is associated with greater coverage for the incumbent. We also find that the absolute difference between the two candidates is associated with less coverage. This finding indicates that although greater spending increases coverage, the value of spending decreases as one candidate spends more than the other. In Appendix Table A4, we perform the same analyses but focus on general election challengers.⁵ The results are largely consistent with Table 2.⁶

In Table 3, we limit our analysis to the general election season and those races in which an incumbent faces a challenger, including both challengers and incumbents. Our results are consistent with Table 2. Congruence, ownership, and competitiveness are associated with greater coverage of candidates. The candidate's own spending, as well as the spending of opponents, is positively associated with coverage, but lopsided spending is associated with less coverage. Columns VII and VIII subset the results to races where both candidates spent more than $500,000. Our results are consistent with the greater sample in these analyses, suggesting that the findings were not the result of races in which incumbents were assured victory. Instead, we find that the magnitude of our effects increases significantly. These results suggest that the effects of congruence and ownership are stronger in a competitive race. The effects of spending are more consistent with those of the broader sample, suggesting that spending is a strong predictor of coverage in competitive and noncompetitive elections.

Table 3. Predicting coverage of incumbents and challengers, limited to districts with incumbents and challengers running in the general election

Models were estimated using ordinary least squares. The unit of analysis is a station-candidate pair. DV = Number of 2.5-minute segments by a given station in a candidate's district that mention the candidate. Standard Errors are clustered by candidate. * = p < 0.05. Table A3 clusters by CD.

When we account for all of these factors, incumbents still have a marginal advantage in local television coverage over their general election challengers. The incumbent can expect to receive between 5.17 and 5.73 more mentions in the general election period than a challenger, all else being equal. When we limit this sample to those races in which both candidates are viable, our estimates still suggest incumbents have an advantage in coverage over challengers, although the estimate is less precise. These results indicate that challengers still face obstacles in reaching the level of coverage for incumbents on local television during a general election campaign.

Conclusion

Local television news plays a key role in citizens' ability to hold legislators to account. Further, prior research did not fully measure the context in which the medium covered incumbents and challengers. We have provided what we believe to be the most systematic and comprehensive analysis of local television news coverage of MCs. Incumbents receive little coverage outside of a general election campaign. This coverage increases significantly during a campaign, but wide variations still exist. Most campaign season coverage of incumbents and challengers occurs in competitive contests. We also found the geographic congruence between a media market and a congressional district predicted a significant variation in coverage. In the campaign context, those races with greater spending and more competitive electoral environments were predicted to provide more coverage for incumbents and challengers. We found marginal evidence that stations owned by the larger media corporations were associated with more coverage of challengers and incumbents. Finally, we found marginal evidence that incumbents hold a slight advantage in coverage, but when we control for disbursements in a race, this effect is less clear.

These results have ambiguous implications for representative government. On the one hand, incumbents receive little coverage of their activities for most of the legislative cycle. That we observe frequent discussion of the candidates on news broadcasts only during the campaign season suggests that local television news provides little substantive coverage of MCs' behaviour in Congress. Previous studies on the closure of local newspapers suggest that media outlets not focusing on local politics reduces the incumbency advantage and decreases the frequency of split-ticket voting (Darr, Hitt, and Dunaway 2019). While we are unable to identify a relationship between local television consolidation and a lack of incumbent coverage, our results suggest that local television broadcasts do not provide a sizable amount of coverage year-round.

If local television news remains a primary source of information, then MCs may be free from adversarial coverage. They could provide their own press as a substitute for work previously performed by journalists (Grimmer 2013). At the same time, news coverage allows an incumbent MC to provide their personal brand. Without a foundation to create that personal brand through traditional media, MCs may be less able to cultivate a personal vote (Snyder and Stromberg 2010). As a result, they could become more vulnerable to the national mood and economic conditions. As American elections become more nationalized, incumbent MCs may lament the lack of local television press to provide them with an opportunity to distance themselves from their party (Hopkins 2018).

The changing media environment's effects on accountability demand more attention, particularly through a longitudinal perspective. We have demonstrated that local coverage of incumbents and challengers is affected by local markets’ political, geographic, and commercial structure. While the local media landscape has changed, many market incentives associated with varying coverage levels have remained the same. Broadcasters are still motivated by attracting larger audiences and thus tend to promote competitive elections and the legislators who represent the most viewers. Yet, as these market structures continue to change, we might expect that the relationship between the press and the political elites will follow suit. Our approach can be adapted to observe shifts in the frequency of coverage and the tone and content of such coverage. If American politics continues to move in a more nationalized direction (for example, Hopkins 2018), local news may lose its local content. This analysis serves as a foundation to answer future questions regarding the congressional information environment and potential shifts from that foundation.