The Structural Determinants of Local Congressional News Coverage

Transcription

1 Political Communication, 20:41 57, 2003 Copyright 2003 Taylor & Francis /03 $ DOI: / The Structural Determinants of Local Congressional News Coverage BRIAN F. SCHAFFNER and PATRICK J. SELLERS Political observers have extensively documented the national media s focus on committee and party leaders. Legislators local coverage, in contrast, remains largely unexplored and unexplained. This article examines 40 local newspapers to explore how these factors may influence legislators local newspaper coverage. We find that local newspapers do not provide more coverage to congressional leaders and that independent papers write more frequently than chain-owned competitors about the local House delegation. Additionally, the extent to which a legislator s district geographically overlaps with the newspaper s market has a strong effect on legislators mentions. Keywords Congress, House of Representatives, markets, media, newspapers, press Since the late 1970s, national news outlets have reduced their coverage of Congress (Cook, 1989; Underwood, 1998). In the scant congressional coverage that remains, national correspondents focus on congressional leaders, largely ignoring rank-and-file members (Kuklinski & Sigelman, 1992; Squire, 1988; Hess, 1986). Political observers have extensively documented the national media s focus on committee and party leaders. Legislators local coverage, in contrast, remains largely unexplored and unexplained. 1 It is unclear, for example, whether local outlets mirror the national media s focus on the congressional leadership or broaden their coverage to include local rank-and-file members of Congress. The diversity of local news outlets also presents new questions about congressional news coverage that do not arise on the national level. Do local papers owned by a national chain provide different levels of news coverage than independent papers? Do newspapers in state capitols focus on state-level politics, to the exclusion of congressional news? How does the overlap between a congressional district and a newspaper s market affect that paper s congressional news coverage? This article explores how these factors may influence legislators local newspaper coverage. This local news coverage and its determinants are crucial to the interaction between legislator and constituent. While members of Congress build support among voters by developing unique homestyles (Fenno, 1978), no legislator can directly interact with Brian F. Schaffner is Assistant Professor of Political Science at Western Michigan University. Patrick J. Sellers is Associate Professor of Political Science at Davidson College. This research received funding from the National Science Foundation (SBR # ) and the Dirksen Congressional Center. Address correspondence to Brian F. Schaffner, Department of Political Science, 3302 Friedmann Hall, Western Michigan University, Kalamazoo, MI 49008, USA. brian.schaffner@wmich.edu 41

2 42 Brian F. Schaffner and Patrick J. Sellers every constituent on a regular basis. The local news media help many legislators overcome this constraint by transmitting political events and other displays of homestyle throughout the electorate. The transmission of information helps the voters hold their elected representative accountable. By publicizing the legislators statements and records, local papers can help these voters learn about the representatives and make informed choices in elections. Not all local papers provide the same levels of congressional news coverage. To document and explain this variation, we focus on 42 local newspapers and their coverage of members of the U.S. House of Representatives during the first eight months of The next section of the article explains how structural factors may influence the representatives news coverage. The second part of the article explains how we measure the news coverage and the influences driving it. We then present a brief analysis of legislators national news coverage and compare this baseline to local levels of news coverage. Determinants of Congressional News Coverage Two groups of legislators dominate the national news coverage of Congress. The first are congressional leaders. Full committee chairs or party leaders won nearly half of the network news coverage of the House between 1969 and 1986 (Cook, 1989). Senate leaders enjoy a similar dominance (Kuklinski & Sigelman, 1992; Squire, 1988). This dominance results from the national media s focus on the overall legislative process. When finding sources for their stories, national correspondents prefer members with great influence over that process, that is, congressional leaders. The resulting national news coverage may in turn enhance the leaders legislative influence, thus creating a self-perpetuating cycle of coverage and influence. Rank-and-file members find it difficult to break into this cycle. A second group of legislators winning more national coverage are members of the out party, that is, the party not controlling the presidency. The news routine of objectivity leads reporters to present competing sides of any conflictual issue (Ansolabehere, Behr, & Iyengar, 1993). National political issues often simplify into a disagreement between the two major parties. Reporters turn to the president as a spokesperson for his party, thereby excluding the president s fellow partisans in Congress. For the opposing side on the issue, the journalists rely upon legislators from the opposing party. In contrast to the national media, state and district outlets may base their congressional news coverage on factors other than whether a legislator is a leader or member of the out party. Unlike national journalists focus on the overall legislative process, local reporters often work to add a local angle to this process. The latter reporters select spokespeople from the local congressional delegation (Vinson, 2003). But do all state and district news outlets work equally hard at localizing their congressional coverage? And when adding this local angle, how might a reporter choose which of the local congressional delegation to cover? Localizing Congressional News Coverage We argue that two types of local newspapers are less likely to add a local angle to their congressional coverage. The first are papers owned by a major national chain. Critics have attacked these papers as overly concerned with profit and as offering less diversity of coverage, compared to independently owned papers (Bagdikian, 1987). While chain-

3 Determinants of Local News Coverage 43 owned newspapers may produce fewer articles (Litman & Bridges, 1986), it remains unclear whether the quality of their overall news coverage differs from that of independent papers (Drew & Wilhoit, 1976; Demers, 1999). Despite this inconclusive evidence, chain-owned papers have a strong incentive to provide less localized coverage of Congress. These newspapers may jointly employ a single correspondent on Capitol Hill, who writes more general stories (with few local angles) that can run in all of the chain s outlets. In contrast, an independent paper may hire its own Washington correspondent. Writing exclusively for the paper s local audience, this reporter can focus congressional stories on the elected representatives of that audience. The second type of newspaper providing less congressional news coverage may be those based in a state s capitol city. In these newspapers, U.S. representatives and senators must share column inches with their counterparts in state government. This tension is weaker in papers based outside state capitols, where coverage of local members of Congress may crowd out that of state and local politicians. Thus, chain ownership and location in a state capitol may reduce the number of stories that a paper writes about the local congressional delegation. In addition to the number of such stories, each paper must also choose their content. Which member(s) of the local delegation should be the focus of the localized stories? Representatives and Reporters Distinct Incentives In local papers stories on Congress, the mention of particular representatives hinges, at least in part, on the overlap between each legislator s congressional district and each paper s market. Different aspects of this overlap may influence individual representatives incentives to work for coverage and individual papers incentives to provide that coverage. 2 In fact, House members may win more coverage, and more favorable coverage, when their districts are highly congruent with a media market (Vinson, 2003). Newspapers markets and representatives districts may overlap in diverse ways. At one extreme is the Second Congressional District of Missouri, which lies completely within the market of the St. Louis Post Dispatch (see Figure 1). At the other extreme, the market of the Bend Bulletin lies completely within the Second Congressional District of Oregon (see Figure 2). Most combinations of markets and districts lie between these two extremes. The Ninth District of Missouri, for example, only partially covers the St. Louis Post Dispatch s market (see Figure 1). To capture this overlap, we look at market-district dyads. Each dyad is a pairing of a newspaper s market and a representative s district that overlap. A single market may be in more than one dyad, since that market may geographically cover parts of multiple congressional districts. Similarly, a single district may be in more than one dyad, since that district may geographically cover parts of several newspapers markets. Our examination of dyads is unusual in research on politics and the media. Most analysis of politicians focuses on congressional districts and individual politicians overall pursuit of media coverage (Cook, 1989). At the same time, many studies of the media examine the overall coverage of individual papers or other outlets (Lasorsa & Reese, 1990). These analytical frameworks are less appropriate for examining how individual politicians receive coverage in individual newspapers, because they cloud the complicated interaction between market and district boundaries. The United States includes 435 congressional districts and 210 media markets. Accordingly, the boundaries of districts and markets rarely match. Many House members represent districts that are

4 44 Brian F. Schaffner and Patrick J. Sellers IL #17 IL #18 MO #9 IL #20 IL #19 MO #4 IL #12 MO #8 MO #2 MO #3 MO #1 Figure 1. St. Louis media market. The shaded portion represents market area. split among a number of different newspaper markets. Each member may work harder to win coverage in certain newspapers than others. In addition, newspaper markets are even more likely to be split among multiple congressional districts. Accordingly, a reporter for a local paper may find certain representatives more newsworthy than others. Unlike the earlier analytical frameworks, the market-district dyads help us begin to examine these variations in representatives and reporters incentives. We argue that these two sets of incentives are linked to the market-district dyad in which representatives and reporters interact. Different dyads encourage different behavior. We consider how three distinct aspects of each dyad may influence politicians and journalists. The first aspect is the extent to which a market geographically covers a district in a particular dyad. An example of extremely high coverage is the dyad of Missouri s Second District and the St. Louis Post Dispatch, whose market geographically covers 100% of the district (see Figure 1). The level of coverage is lower in other market-district dyads. The market of the Bend Bulletin, for example, covers only 4% of Oregon s Second District (see Figure 2). A newspaper market s geographic coverage of a congressional district can influence a representative s incentive to work to win that paper s attention, and thus affect the actual number of news stories about the representative. When a newspaper s market geographically covers only a small part of a district, relatively few of the representative s constituents lie within that market. Winning news coverage in that newspaper would provide limited benefits to the politician. The legislator may therefore send few press

5 Determinants of Local News Coverage 45 WA #3 WA #4 OR #3 OR #5 OR #4 OR #2 ID #1 Figure 2. Bend, Oregon media market. The shaded portion represents market area. releases to that paper and devote little attention to its reporter(s), which results in little news coverage. The benefits of news coverage grow dramatically if the paper s market geographically covers much or all of the district. Newspaper appearances can communicate the legislator s name and message to many more constituents. In this context, the legislator may work harder for this paper s news coverage and receive more of it. Constituents in this latter type of district tend to know more about their representative, compared to individuals living in districts that overlap little with media markets (Campbell, Alford, & Henry, 1984; Niemi, Powell, & Bicknell, 1986; Levy & Squire, 2000). Alternatively, House members may work equally hard to gain coverage in any market that intersects any part of their district, no matter how small. We will examine whether market coverage of a district has an effect on local newspaper coverage. A second and distinct aspect of the market-district overlap is the extent to which a district geographically covers a market in a particular dyad. An example of high coverage of this type occurs in the aforementioned dyad of the Bend Bulletin s market and Oregon s Second District. Here, the district geographically covers 100% of the newspaper s market. The coverage is lower in other dyads, such as the 35 involving the New York Times and its market. In each of these 35 dyads, the particular congressional district geographically covers only a small portion of the paper s market (the district coverage of the market ranges from 0.1% to 3.1% across the 35 dyads). In a particular dyad, the district coverage of the paper s market affects that paper s incentive to write articles about the representative, which in turn influences the actual number of articles. If a district geographically covers only a small portion of a newspaper s market, the paper s reporters may devote little attention to the district and its

6 46 Brian F. Schaffner and Patrick J. Sellers representative. The district s citizens make up only a small portion of the paper s readership. The journalists may instead focus their stories on a legislator from a district that geographically covers a high percentage of the market. This legislator represents many of the paper s readers, who would be interested in his or her work in Congress. The preceding paragraphs have described two important aspects of the overlap between market and district. The labels for the two characteristics sound similar (market coverage of a district and district coverage of a market). But the labels capture the intuitive concept of one geographic entity physically covering part or all of another such entity. For each characteristic, the extent of geographic coverage suggests different incentives for legislators and journalists. In turn, these varying incentives may affect the actual number of column inches that each representative receives. A third aspect of the market-district overlap may also influence both incentives and actual news stories: whether the newspaper of a dyad is based in the same state as the corresponding congressional district. Many newspaper markets cross state lines, particularly in urban areas. Of the 35 congressional districts overlapping the New York Times local market, 20 are in the state of New York. Legislators representing the remaining 15 districts may receive less attention from the New York Times than their congressional colleagues from the New York districts. News reporters and editors may be most concerned about efforts to help their own district and state. Legislators share this focus on their district and state. As a result, a newspaper may print more stories on legislators from its state and fewer on representatives from nearby states. These paragraphs outline our arguments about the diverse origins of local congressional news coverage. We summarize these arguments in six specific hypotheses: 1. Unlike the national media, local newspapers do not devote more news coverage to congressional leaders and to legislators not sharing the president s party. 2. A member of Congress receives less local news coverage in chain-owned papers than in independent ones. 3. A member of Congress receives less local news coverage in papers in state capitols than in other papers. 4. As a paper s market geographically covers more of a congressional district, the member of Congress representing that district works harder to earn that paper s attention and receives more news coverage. 5. As a congressional district geographically covers more of a paper s market, reporters for that paper find the member of Congress representing that district more newsworthy and allot the legislator more news coverage. 6. A member of Congress receives more local news coverage in a paper based in the same state as the legislator s district than in a paper based in another state. Measuring Markets, Districts, and Coverage To test these hypotheses, we first created market-district dyads for all possible media markets and congressional districts in the United States. The pairing resulted in a total of 920 dyads. Twenty three have a market geographically covering a single district. Seventy dyads use a market covering two districts, and 827 dyads are based on a market covering more than two districts. Similar variation arises from a congressional district geographically covering more than one media market. Of the actual dyads, 214 use a district covering part or all of one market. One hundred sixty-six dyads are based on a district covering two markets, with the remaining 540 dyads using districts that cover

7 Determinants of Local News Coverage 47 more than two markets. We have already suggested how these two types of variation can shape representatives and reporters incentives and, by extension, the local news coverage. Using dyads as our unit of observation allows us to incorporate both types of variation at the same time. After we created the dyads, our next step was to find a newspaper in each media market making up the dyads. We turned to the Lexis-Nexis electronic archive, from which we downloaded all articles for an eight-month period (January 1 to August 31, 1997) from one purely national paper (USA Today) and one paper providing local news coverage in each of 40 states. 3 The archive did not contain a paper from seven states. The 40 papers with local news coverage allowed us to measure this coverage in 294 dyads, or 32% of all possible dyads. These 294 dyads are the units of observation for our analysis. We excluded the remaining 626 dyads for which we have no measure of local media coverage. The useable and excluded dyads resemble each other in terms of political orientation but not market size. Among the useable dyads, 46% of the representatives are Democrats. The corresponding percentage for the omitted dyads is 38%. In terms of market size, the useable dyads have an average population of 5,631,525; the equivalent mean for excluded dyads is 1,546,865. The difference in market size reflects the few small-state papers that are available electronically through Lexis-Nexis. This unavoidable constraint makes our analysis more relevant for markets and districts with high populations. We return to this point when discussing the results of our analysis below. For each market-district dyad, the dependent variable is the number of mentions of the district s representative s name in the market s paper during the eight-month period. To obtain these counts, we applied the Context content analysis program (produced by Oracle) to the newspaper articles for each dyad. The program searched the articles for mentions of the district representative for that dyad. Each search included variations of the legislator s name (e.g., United States Representative Barney Frank, Congressman Barney Frank, etc.). We applied the same program and search parameters to the articles from USA Today, which produced measures of each representative s national coverage. Among the 294 dyads, the highest number of mentions was 1,797, won by Representative Billy Tauzin (R-LA) in the New Orleans Times-Picayune. Fifteen House members received no mentions at all. These representatives came from 14 different states and 10 different markets. The measure of local news coverage had a standard deviation of , suggesting that the count varied widely across the dyads. To explain this variation, we required additional data on individual legislators and newspapers. To capture the diverse types of leaders in the House, we created dummy variables for whether the representative was a party leader, 4 a committee chair, or the ranking minority member of a committee. Another dummy variable indicated whether each representative was a member of the out party in 1997 (0 if Democrat, 1 if Republican). 5 Our analysis also incorporates characteristics of individual newspapers. We created a dummy variable indicating whether a chain owns a paper (1 if chain-owned, 0 if independent). 6 Another variable measures whether a newspaper s market includes a state capitol (1 if market includes capitol, 0 if not). In our final three independent variables, we capture how different aspects of marketdistrict overlap may affect local congressional news coverage. In the first aspect of the market-district overlap, the newspaper market in a particular dyad geographically covers a certain percentage of the district in that dyad. This percentage is the market coverage of the district. The second aspect of the overlap is similar: The congressional district in a dyad geographically covers a percentage of the dyad s market. This percentage is another independent variable: the district coverage of the market. 7 Finally, the third

8 48 Brian F. Schaffner and Patrick J. Sellers aspect of the overlap is whether the newspaper of a dyad is based in the same state as the dyad s congressional district. We operationalize this aspect with another dummy variable (1 if same state, 0 if not). We use regression analysis to investigate whether these 10 independent variables influence local press coverage. The dependent variable is the number of stories that mentioned a particular House member during the first eight months of The mean number of stories that mention a particular legislator is 66.77, with a low value of 0 for 19 legislators and a high of 1,797 for Billy Tauzin (R-LA). This variable is a count and shows evidence of overdispersion. In other words, the likelihood that a House member receives coverage in one article is not independent of the likelihood that he or she wins coverage in a second article. Once the first piece of coverage appears, the second becomes more likely. Accordingly, we use a negative binomial model to estimate the first model (King, 1988). The 40 different local newspapers allow us to examine variation in local news coverage across newspapers. Unfortunately, this approach may also create an estimation problem, since the news routines of congressional coverage may vary across papers in ways that we have not anticipated. For example, the Houston Chronicle provides local news coverage for all representatives in the Houston area. The Indianapolis Star does the same for much of the Indiana House delegation. We theorize that the Indianapolis paper provides less congressional coverage because Indianapolis (and not Houston) contains the state capitol. Our model attempts to incorporate such differences in newspapers that affect their news coverage of Congress. But we may not have considered all relevant differences in papers. If we fail to incorporate all such differences in papers, our estimated coefficients may be biased. Legislators would cluster into 40 groups (for each local newspaper) receiving different levels of coverage, for reasons beyond the independent variables in our model. We address this problem by estimating our model of local coverage in two ways. The first approach directly captures our expected differences in newspapers, using the measures of newspaper characteristics. But since our independent variables may not incorporate all relevant differences in newspapers, the second estimation approach assumes that each paper has its own unique level of coverage. In this second approach, we incorporate a dummy variable for each paper. These additional independent variables effectively group the observations (dyads) by paper, thus capturing every possible difference across the papers both the ones we suggest and any others influencing news coverage. This second estimation approach thus serves as a check on the completeness of our explanations. If our independent variables produce similar results with both estimations, we can be confident that we have accurately captured important influences on local news coverage and that the clustering problem described above does not bias our results. 8 Explaining Local Congressional News Coverage Our analysis of local congressional news coverage proceeds in two steps. We first estimate and briefly discuss a baseline model of national coverage. Then we estimate the two models of local news coverage. National News Coverage as a Baseline The first column of Table 1 presents the results of the model of USA Today mentions of House members. 9 These results fit our expectations about legislators national news

9 Determinants of Local News Coverage 49 Table 1 Influences on U.S. House members national and local coverage Local papers Measures of Newspaper newspaper dummy Variable USA Today characteristics variables Constant 2.86***(.12) 4.48*** (.27) 7.20*** (.29) Party leader 2.28* (.99) 1.29 (.97).28 (.33) Full committee chair.87* (.39) 1.12** (.39).14 (.24) Full committee ranking member 1.25** (.41).64 (.36).33 (.34) Out party.66***(.18).71*** (.19).43* (.21) Owned by chain 1.02*** (.22) State capitol.28 (.17) Market coverage of district.34 (.33).08 (.44) District coverage of market 4.95***(1.06) 3.88** (1.49) District in newspaper s state.15 (.26).85* (.36) α 2.91***(.18) 1.82*** (.13).98*** (.12) χ *** 71.94*** *** N Note. The third model contains 39 dummy variables not shown here. Standard deviations are in parentheses. *p <.05; **p <.01; ***p <.001. coverage. The measures of party and committee leadership all produce significant and positive coefficients. Table 2 contains the predicted number of mentions for these types of legislators. The model predicts mentions for a party leader over the eightmonth period, which is 1,562% more than the prediction for a nonleader. The committee variables affect the predicted number of mentions in a similar manner. Membership in the out party has a significant and strongly positive impact on mentions in USA Today. The average representative in President Bill Clinton s party received a predicted mentions, while the typical Republican appeared times. As expected, the national newspaper gave more coverage to members of the party not holding the presidency. These baseline results replicate the findings of earlier studies and therefore establish the validity of our content analysis program and measure of news coverage. We next apply these tools, and our full set of independent variables, to representatives local news coverage. Explanations of Local News Coverage The second and third columns of Table 1 present the results of our model of local congressional news coverage, estimated in the two ways outlined above. These results support many, but not all, of our hypotheses. First, the local papers do not give more coverage to congressional leaders. Across the two sets of results, five of the six coefficients for leadership variables are insignificant. But the local papers follow USA Today by giving more mentions to legislators not sharing the president s party. This variable is

10 Table 2 Predicted levels of coverage Local papers Measures of newspaper Newspaper dummy USA Today characteristics variables Percentage Percentage Percentage Predicted difference Predicted difference Predicted difference mentions from baseline memtions from baseline mentions from baseline Legislator characteristics House leaders Nonleader* Party leader ,562 Full committee chair Full cmte. ranking member Out party Democrat* Republican Newspaper characteristics Independent newspaper* Chain-owned paper

11 Market-district overlap District coverage of market 3%* % % % District in newspaper s state No* Yes Note: We calculated the predicted values shown here using the models in Table 1. To obtain predicted values for a particular variable, we assigned all other variables their mean values. For the variable in question, we then chose one category and calculated the predicted number of mentions for that category. We repeated the process for other categories of the variable. For district coverage of the market, we calculated predicted values for one standard deviation below the variable s mean, the mean itself, one standard deviation above the mean, and two standard deviations above. An asterisk denotes a baseline category used to calculate the percentage difference for other categories. 51

12 52 Brian F. Schaffner and Patrick J. Sellers significant with both estimation approaches. As indicated in Table 2, Republican representatives won at least 53% more mentions than President Clinton s fellow Democrats in the House. This unexpected pattern may reflect an effort by journalists at all levels to present both sides of conflictual issues. Second, members of the House do receive less local news coverage in papers owned by a chain. This variable generated a strongly significant and negative coefficient. Chain ownership lowers the predicted number of mentions by 63% from the baseline number of mentions in an independent paper. Third, we found no support for the expectation that papers in state capitols provide fewer mentions of U.S. House members. The coefficient for this variable is insignificant. The papers in state capitols may provide so much political news coverage that they can extensively discuss state politics while giving the same level of congressional news coverage as other local papers. Fourth, the district coverage of the market was the only overlap variable to produce a significant coefficient in both models. 10 If a legislator s district geographically covers only a small portion (3%) of a newspaper market, the two models predict that the paper will provide relatively few mentions of the representative (43.23, 28.62). 11 This outcome is consistent with our argument. The legislator represents only a small portion of the newspaper s readers, and reporters and editors for the paper therefore have little incentive to publish articles on the representative. As the district geographically covers more of the market, the predicted number of mentions also increases, to some of the highest levels in Table 2. In the case where the district covers 48% of the paper s market, the predicted mentions of the representative are 529% to 858% higher than the previous example of low district coverage. 12 In this case of high district coverage, we argue that editors and reporters are more likely to find newsworthy those legislators who represent large portions of their readership. We found this pattern under both estimation approaches, which further increases its validity. The remaining two overlap measures produce much weaker results in the two models. We expected legislators to receive fewer mentions in local papers whose markets geographically covered small portions of the representatives districts. Instead, a representative s local news coverage does not appear linked to how much a paper s market geographically covers the legislator s district. This measure produces an insignificant coefficient under both estimation approaches. It may be that House members work equally hard to win news coverage in all local papers covering their districts. Successful members of Congress allocate their time and resources efficiently, to further their own unique set of goals (Fenno, 1992). But winning local press coverage may not require such a selective allocation. A congressional press secretary can find it just as easy to send a press release to a local paper reaching 3% of the district as to a paper delivered to half the district s constituents. The third overlap variable provides more mixed evidence. In the first set of results (the second column of Table 1), we found an insignificant coefficient for the measure of whether a dyad s paper is based in the same state as the dyad s district. This model contained measures of individual papers characteristics (whether owned by a chain and containing a state capitol in the paper s market). In our second set of results (the third column of Table 1), we replace these measures of individual characteristics with dummy variables for each paper. These dummy variables effectively account for all possible variation across individual papers. After accounting more completely for such variation, our second set of results produces a significant coefficient for the measure of whether a

13 Determinants of Local News Coverage 53 dyad s paper is based in the same state as the dyad s district. According to this set of results, having both in the same state leads to a predicted number of mentions 129% higher than if they are not. This significant effect may be hidden in the first model of local coverage, because unexplained variation across individual papers makes it harder to uncover the joint effects of each paper and the market in its dyad. Such effects emerge once we add dummy variables to control for all possible variation across papers. Overall, the models support many of our expectations about the origins of local congressional news coverage. Local newspapers do not provide more mentions of congressional leaders, compared to rank-and-file members. But House members not sharing the president s party win slightly more news coverage. Independent papers write more frequently than chain-owned competitors about the local House delegation. Finally, the extent to which a dyad s district geographically covers the corresponding market has a strong positive effect on legislators mentions. This link suggests that a local paper s editors and reporters choose whom to cover at least in part on the basis of how much each House member represents the newspaper s audience. These conclusions may be less valid for remote regions of the country with fewer residents, since our collection of dyads oversamples markets with high populations. But the dynamics underlying our conclusions are likely to apply to markets with all levels of population. The norm of objectivity likely guides reporters in any outlet, suggesting that out-party representatives may always win more news coverage. Legislators may always work to win the attention of any news outlet in their district. As a result, a market s geographic coverage of a district may not affect mentions of a representative in any of the 920 possible dyads. Unfortunately, we cannot systematically assess how these patterns apply to small markets because the Lexis-Nexis archive contains few newspapers from these markets. Conclusion Constituents often rely upon the news media to learn about their elected representative in Congress. Incumbents tend to win the bulk of this coverage during their reelection campaigns, giving them an advantage over challengers (Clarke & Evans, 1983). However, our analysis has demonstrated that the amount of local news coverage won by incumbents may vary a great deal. A newspaper in one district may provide much more extensive news coverage of the local representative than a paper in another district. Constituents in the latter district may therefore find it more difficult to hold their legislator accountable and make informed electoral choices. We have attempted to explain why local news coverage varies by focusing in particular on the structural origins of that coverage. Most notably, the extent to which a congressional district geographically covers a newspaper market exerts a strong influence on legislators local news coverage. Local editors and reporters appear to publish more stories about a House member if that legislator represents a large portion of the paper s readership. Another important structural influence is chain ownership. A chain of newspapers enjoys great economies of scale, with a single Washington reporter writing congressional stories for all of the papers. Independent papers must each hire their own congressional reporter. Our analysis points to an important drawback of chain ownership: a decline of the local angle in congressional reporting. Since chain reporters in Washington do not write for a specific locality, they are less likely to shape their coverage

14 54 Brian F. Schaffner and Patrick J. Sellers around local members of Congress. As a result, legislators from areas served by chainowned papers find it harder to get local news coverage. Citizens subscribing to these papers receive substantially less information about their local representative than if they relied on an independent paper. Our aggregate-level analysis of these structural constraints can guide future investigations of individual-level behavior. The results in Tables 1 and 2 suggest that reporters are likely to perceive as less newsworthy those legislators who represent only a small portion of the reporters readership. Representatives, on the other hand, may work hard to win news coverage in every outlet in their district. Such structurally driven patterns of behavior are not the only influences on local news coverage of House members. More transient factors may also raise or lower the number of stories about a local representative. The prospect of a difficult reelection may push a politician to work harder to win local news coverage. That same prospect may make the representative more newsworthy in the eyes of local editors and reporters. Members of a local congressional delegation may compete for scarce column inches, pushing the most desperate of them to work even harder to win coverage. These transient factors must operate within the structural constraints that we have demonstrated. Observers of politics and the media must begin to consider both types of influences if we want to develop an accurate understanding of how the news media transmit information between legislators and constituents. Our study takes a step toward that understanding. Notes 1. Tidmarch and Pitney (1985) and Vinson (2003) are notable exceptions. 2. The data on congressional districts, media markets (designated market areas), and their overlap come from Congressional Districts in the 1990 s, published by Congressional Quarterly Press. We use the media market in which a newspaper is located as a proxy measure for that paper s market. See Goldenberg and Traugott (1984) for evidence in support of this assumption. 3. See the Appendix for a list of the actual papers. Lexis-Nexis includes articles from different regional editions of each paper when the content differs. The list of newspapers includes the New York Times and the Los Angeles Times, which are often considered national newspapers. But these publications cover politics in their local markets as well. We repeated the analysis below with the New York Times and the Los Angeles Times omitted; the results did not differ substantially. 4. We simply coded each party s leader for this variable Richard Gephardt and Newt Gingrich in this case. A separate variable that included additional party leaders was insignificant and did not change our results. 5. Politics in America (1998) is the source for all four independent variables. 6. We defined a newspaper chain as a company that owned more than 30 daily newspapers across the United States. This designation included Gannett, Media News Group, Thomson, Knight-Ridder, and Copley. 7. To create these measures, we used Congressional Districts in the 1990 s. This publication lists for each congressional district the percentage of that district, that overlaps with local media markets. These figures capture how much each market geographically covers the district. We also used the figures to create our second measure of overlap, district coverage of market. For each market, we added together the percentages reported for districts overlapping that market. For each of those districts, we divided the district s percentage by the sum of the districts percentages (described in the previous sentence). The result indicated how much the district geographically covered the market. 8. Clustering may also affect the precision of our estimates, since the variance in news

15 Determinants of Local News Coverage 55 coverage may differ across papers. We therefore use robust estimations of variance to control for the effects of clustering on our standard errors. 9. This model includes all representatives who served a complete term in the 105th Congress. 10. These measures of coverage are based on geographic area and not population. As a result, the measures could conceivably contain measurement in rural districts where the population is unevenly distributed. However, we found similar results when we controlled for the geographic makeup of a dyad s district (rural, urban, suburban, or mixed). 11. This case is similar to Districts 1 20 in New York, each of which covers about 3.1% of the New York Times market. Among these legislators, Representative Carolyn McCarthy (D) received just one mention, while Representative Charles Schumer (D) received the most coverage with 27 mentions. We found the same effect even when we controlled for the number of legislators (i.e., districts) sharing a market. 12. This case is similar to the 2nd District in Arkansas, which covers 47% of the Little Rock market. Representative Vic Snyder (D) received 114 mentions in the Arkansas Democrat, representing this district, during the first eight months of References Ansolabehere, S., Behr, R., & Iyengar, S. (1993). The media game: American politics in the television age. New York: Macmillan. Bagdikian, B. H. (1987). The media monopoly (5th ed.). Boston: Beacon Press. Campbell, J. E., Alford, J. R., & Henry, K. (1984). Television markets and congressional elections. Legislative Studies Quarterly, 9, Clarke, P., & Evans, S. H. (1983). Covering campaigns: Journalism in congressional elections. Stanford, CA: Stanford University Press. Cook, T. E. (1989). Making laws and making news: Media strategies in the U.S. House of Representatives. Washington, DC: Brookings Institution. Demers, D. (1999). Corporate newspaper bashing: Is it justified? Newspaper Research Journal, 20, Drew, D., & Wilhoit, C. G. (1976). Newshole allocation policies of American daily newspapers. Journalism Quarterly, 53, Fenno, R. (1978). Home style. New York: HarperCollins. Fenno, R. (1992). When incumbency fails. Washington, DC: Congressional Quarterly Press. Goldenberg, J. E., & Traugott, M. W. (1984). Campaigning for Congress. Washington, DC: Congressional Quarterly Press. Hess, S. (1986). The ultimate insiders: U.S. senators in the national media. Washington, DC: Brookings Institution. Hibbing, J. (1991). Contours of the modern congressional career. American Political Science Review, 85, Kuklinski, J. H., & Sigelman, L. (1992). When objectivity is not objective: Network television news coverage of U.S. senators and the paradox of objectivity. Journal of Politics, 54, Lasorsa, D. L., & Reese, S. D. (1990). News source use in the crash of 1987: A study of four national media. Journalism Quarterly, 67, Levy, D., & Squire, P. (2000). Television markets and the competitiveness of U.S. House elections. Legislative Studies Quarterly, 25, Litman, B. R., & Bridges, J. (1986). An economic analysis of daily newspaper performance. Newspaper Research Journal, 7, Niemi, R. G., Powell, L. W., & Bicknell, P. L. (1986). The effects of congruity between community and district salience of U.S. House candidates. Legislative Studies Quarterly, 9, Squire, P. (1988). Who gets national news coverage in the U.S. Senate? American Politics Quarterly, 16,

16 56 Brian F. Schaffner and Patrick J. Sellers Tidmarch, C. M., & Pitney, J. J., Jr. (1985). Covering Congress. Polity, 17, Underwood, D. (1998). Market research and the audience for political news. In D. Graber, D. McQuail, & P. Norris (Eds.), The politics of news: The news of politics. Washington, DC: Congressional Quarterly Press. Vinson, D. C. (2003). Through local eyes: Local media coverage of Congress and its members. Cresskill, NJ: Hampton Press.

17 Determinants of Local News Coverage 57 Appendix: Newspapers Used in the Analysis Newspaper Market The Montgomery Advertiser The Arizona Republic Arkansas Democrat-Gazette The Sacramento Bee Los Angeles Times The Denver Post The Hartford Courant The Florida Times-Union The Atlanta Journal and Constitution The State Journal-Register The Indianapolis Star The Des Moines Register The Kansas City Star The Louisville-Courier Journal The New Orleans Times-Picayune Bangor Daily News The Boston Globe Detroit News Minneapolis Star Tribune St. Louis-Post Dispatch Omaha World Herald Las Vegas Review-Journal The Time Union New York Times The News & Observer The Columbus Dispatch Tulsa World The Bulletin Pittsburgh Post Gazette The Providence Journal-Bulletin The Post & Courier The Tennessean The Austin American-Statesman The Salt Lake Tribune Richmond Times Dispatch Albuquerque Journal Bismarck Tribune Tacoma News-Tribune Charleston Daily Mail Capital Times USA Today Montgomery-Selma Phoenix Little Rock Sacramento-Stockton Los Angeles Denver Hartford-New Haven Jacksonville Atlanta Springfield-Decatur-Champaign Indianapolis Des Moines Kansas City Louisville New Orleans Bangor Boston Detroit Minneapolis-St. Paul St. Louis Omaha Las Vegas Albany-Schenectady-Troy New York Charlotte Columbus Tulsa Bend Pittsburgh Providence-New Bedford Charleston Nashville Austin Salt Lake City Richmond Albuquerque Minot-Bismarck-Dickinson Seattle-Tacoma Charleston-Huntington Madison National