This file is also available in Adobe Acrobat PDF format
LEGISLATORS, JOURNALISTS, AND CITIZENS
THE MASS MEDIA perform a vital function in democratic systems by reporting what elected officials are doing in office. The media convey not only factual accounts of officials' activities and decisions; they also transmit evaluations of officials' performance, including assessments by other politicians, interest group leaders, pundits, and ordinary citizens. Although the media are not the only source of information about officials' performance, they are by far the most important. Indeed, it is difficult to imagine how large-scale democracy would be possible without a free and independent press to report the actions of governmental officials. Robert Dahl, the democratic theorist, argues that the existence of alternative and independent sources of information is one of seven necessary conditions for the existence of democratic government.1
Information about elected officials' performance serves two important purposes. First, it allows citizens to evaluate the desirability of retaining or replacing officials when they run for reelection. Candidates promise all sorts of things when they first run for office. When they run for reelection, however, there is no better guide to their future performance than what they have already done. Second, a regular flow of information about governmental decision making helps keep officials on their toes when they first make decisions. Officials who expect their actions to be featured on the evening news and on the front pages of newspapers may make decisions different from officials who expect their decisions to remain forever hidden from public scrutiny.
How extensively and how effectively do media outlets in the United States cover elected officials? Do they report the kinds of information that citizens need to hold officials accountable for their actions in office? Or is coverage so spotty and incomplete that even the most diligent citizens cannot learn much about who is responsible for governmental decisions? These important questions are central to the performance of democratic government. Unfortunately, they are not questions to which we know the answers.
Most citizens are exposed to a regular diet of information about what the president is doing in office. The mass media cover presidential activities on an almost daily basis, reporting where the president travels, what he says, what he proposes, how his proposals fare in Congress, what he is doing about various crises, and what innumerable pundits, legislators, politicians, and foreign officials think of his performance in office. Although one can surely raise questions about the adequacy and fairness of the media's coverage of presidential activities and about the depth of citizens' knowledge of presidential performance, two things seem clear. First, presidents know that their deeds and misdeeds will be covered by the press and noticed by the public, so they work hard to produce pleasing records. Second, when pollsters come knocking at their doors, it is reasonable to believe that most citizens have some evidentiary basis for determining whether they "approve or disapprove of the way the president is handling his job as president."
Can one make similar arguments about the way journalists cover members of Congress? Do legislators expect that their individual activities and decisions in Washington will be covered by the press and reported to their constituents? Are citizens exposed to regular information about what their senators or representatives are doing in office? Do citizens have any evidentiary basis for determining whether they approve or disapprove of the way their representatives are performing in office? Here the issues become more complicated, in part because there are 535 legislators to cover. Journalists do not cover all senators and representatives equally well. Citizens in different states and different districts are not exposed to identical flows of information.
The so-called national press--the networks, newsmagazines, and national newspapers--could not possibly cover the individual activities of every senator and representative. The national press can cover Congress as an institution and report what it is doing about a whole range of problems. It can focus on some colorful or consequential legislators, making Ted Kennedy and Newt Gingrich into household names. The national press could not possibly make 535 legislators into household names. It has neither the time nor the space for such intensive coverage. The typical representative does not appear even once a year in Newsweek, USA Today, or on the CBS evening news.2
Local media outlets are better suited to cover individual members of Congress than the national media. Newspapers, television stations, and radio outlets serve geographically defined media markets, and most of these market areas are represented in Congress by only a few legislators. Even local media outlets, however, have constraints on their coverage. The congruence between congressional districts and media markets is far from perfect. A television station in New York City, which has thirty-five representatives within its broadcast area, has no more time in its broadcast day than a television station in Portland Maine, where the broadcast area is essentially congruent with a single congressional district.3 A small weekly newspaper has to decide whether to devote its meager resources to covering the politics of the several towns it serves or the activities of its representative in Washington. Each media outlet decides what kinds of news it wants to present. No law compels them to cover what representatives are doing in office.
Although local media outlets are better suited to covering individual representatives than are the national media, they are much more difficult to study. The difficulty involves both numbers and access. Four national networks, three newsmagazines, and a handful of major newspapers have been the focus of most previous studies of politics and the press. Once early studies established that coverage patterns were similar among national media outlets, subsequent studies often focused on a single network, news-magazine, and newspaper. In contrast, 23,000 local media outlets blanket the country in a hodgepodge of overlapping territories.4 No one knows anything about the similarity of coverage patterns across these thousands of outlets, so one cannot discover much about the universe of outlets by studying only a handful. Studying local outlets requires sampling, but how should one draw a sample that is representative of what citizens see, hear, and read?
The problem of access is even more serious. A research center at Vanderbilt University has recorded and archived network television newscasts since 1968. No one has recorded and archived local television newscasts. Radio newscasts are similarly unavailable. Although most research libraries contain archives of newsmagazines and major newspapers, most local newspapers are found only in the communities where they are published. Given the problems of sampling and access, it is not surprising that few scholars have attempted to study how local media outlets cover members of Congress.
This book is the first large-scale study of how local media outlets cover members of Congress. The focus is on local newspapers because it is for them that I have solved the twin problems of sampling and access. No cost-effective solution is in sight for studying local radio or television newscasts.5 Unlike previous studies, which largely focused on the campaign period, this book explores how local newspapers covered representatives during an entire congressional session, from the first day of 1993 to election day 1994. The longer period is essential for studying political accountability.
Even if it were not the case that studying local newspapers is easier than studying local television, good reasons exist for beginning with newspapers. First, local newspapers have much larger newsholes than do local television stations. Local newscasts are usually fixed at thirty or sixty minutes, so after deducting for weather, sports, and advertisements, the time available for news is quite limited. Newspapers, by comparison, can cover many more subjects and in much greater detail. Second, the constraints in large metropolitan areas, where there are dozens of representatives to cover, are particularly severe for television, whereas large metropolitan newspapers can use regional editions and regional sections to cover representatives. The news hour is fixed; the newspaper is expandable. Third, in many localities, newspapers set the local news agenda and broadcast journalists follow their lead. Jeffrey Mondak found this to be especially true for House campaigns.6 Finally, two studies of how local media outlets cover Congress found much less coverage on local television stations than in local newspapers (Hess 1991; Vinson 2003).7
A full understanding of how the mass media cover representatives requires examination of all types of media--radio, television, cable, daily newspapers, weekly newspapers, and the Web. The arguments for beginning with local newspapers are two. First, local newspapers appear to cover local representatives more intensively than do other media outlets. Beginning with local newspapers allows one to establish a baseline for comparing other types of coverage. Second, the problems of sampling and access are more easily solved for newspapers, thus allowing for much larger sample sizes. It is worth emphasizing, however, that this study is essentially measuring the high-water mark for media coverage of representatives. When local newspapers fail to cover some aspect of a legislator's behavior, it is unlikely that local radio and television outlets are somehow filling the void.
The American system does not make it easy for citizens to hold elected officials accountable for governmental decisions. Holding officials accountable is easiest when power is concentrated on a single individual or party team. In a parliamentary system, for example, where two parties compete in regular elections and where the winning party gains complete control over policy making until the next election, citizens need not monitor who is doing what within government. With the executive and legislative functions united, it is reasonable for citizens to assume the in-party is responsible for everything that government does. If citizens don't like what government has been doing, they can throw the rascals out. The incentives for the in-party to produce pleasing outcomes are especially strong when power is concentrated. The rewards are for action and results, not words and excuses.
The American system is one of dispersed power and scattered responsibility. Federalism, separation of powers, and bicameralism make it difficult for citizens to know who is responsible for improving or deteriorating conditions, and therefore whom they should reward or punish. If candidates ran for office and governed as members of strong party teams, citizens could reward and punish the team that controlled the legislative and executive branches, just as they do in parliamentary systems. But candidates run more as individuals than as members of strong party teams, citizens often split their votes among the parties' candidates in separate House, Senate, and presidential elections, and parties do not govern as unified teams. Although the norm was once for a single party to control the House, Senate, and White House, the norm now is for divided control of these three institutions. With weak parties and divided party control, citizens need to know more about what particular officials have been doing if they are to reward and punish the right officials. They need to know who the rascals are before they can throw them out.
A system of dispersed power and scattered responsibility also affects the incentives of elected officials. Whereas members of strong party teams work for the good of the team, often giving up individual glory for team success, elected officials in the American system have a stronger interest in individual glory. Achieving any kind of coordinated action among officials so motivated is difficult. Officials also have strong incentives to blame others for inaction. A president lashes out against a do-nothing Congress; senators complain about the lack of presidential leadership; House members blame a senatorial filibuster. Every participant has a favorite explanation for legislative gridlock.
If citizens are to hold legislators accountable, they need information about what their representatives are doing in office. Where might they find appropriate information? One thing is certain: Most citizens do not have an incentive to search diligently for information about representatives' actions in office. Anthony Downs made the case long ago that information is costly and that few citizens choose to incur substantial costs to become informed voters. Most citizens rely on whatever information comes their way in the course of daily life (Downs 1957, 207-37). Fortunately, many individuals and groups have incentives to inform citizens about representatives' actions in office. They willingly bear the costs so that citizens receive information with little effort. Citizens have at least four sources for information.
Incumbent representatives have the strongest incentives for informing constituents about their legislative activities. If they can shape citizens' views of their accomplishments, they gain an electoral advantage. To that end, they regularly visit their constituencies, speak before labor, business, and civic groups, and attend gatherings of every type. They use their free mailing privileges to shower constituents with newsletters and to target individuals with special mailings. They issue press releases to highlight their positions and accomplishments; they court local reporters and editors. Representatives are assisted in these tasks by press secretaries, legislative correspondents, and caseworkers, some residing in Washington, some in district offices.
Politicians who seek to remove representatives from office are another source for information about representatives' actions in office. Quite naturally, these politicians emphasize different aspects of legislators' records. They may publicize unpopular roll-call votes, complain about the lack of any real accomplishments, or place a different spin on activities that legislators consider to be accomplishments. These politicians include active challengers in primary and general election campaigns, individuals who are considering challenging incumbents, and leaders of the opposite party.
Individuals and groups who care intensely about specific policy problems are a third source for information about representatives' actions in office. Interest group leaders usually monitor what representatives are doing to help or hurt their members' interests and inform either group members or citizens more generally when they observe unfriendly actions. Individual citizens who are very interested in particular problems, policies, or programs--hereafter referred to as opinion leaders--often do the same thing. For example, much of the monitoring of what representatives say and do about abortion is performed by local opinion leaders who care intensely about this issue.
Local newspapers play several roles in conveying information to citizens about what representatives are doing. First, journalists are independent monitors of governmental decision making who actively seek and report information about what elected officials are doing in office. Most journalists consider that reporting the actions of elected officials is one of their central responsibilities. Most media outlets make politics and public affairs an important part of their news coverage. Second, journalists are conveyors of information from all sorts of interested parties. Representatives, challengers, and others who have an interest in publicizing information about representatives' actions do so by encouraging journalists to write stories in ways that further their own goals. This is an efficient way to reach citizens, since stimulating a single reporter to write a single story can reach thousands of citizens. It also increases the credibility of the message because a story published under a reporter's byline seems less promotional than an advertisement. Third, newspapers provide a forum for local opinion leaders to share their views about a representative by encouraging and publishing opinion columns and letters to the editor. Newspapers can make their editorial pages a place for public deliberation about a representative's performance in office.
What citizens know about politics and public affairs is largely determined by what the press chooses to cover. Citizens are more likely to learn something if the press covers it intensively than if coverage is sparse. Intensive coverage does not guarantee an informed public. Citizens must be interested in a subject to notice and process the copious information that journalists provide (Zaller 1992). When the press ignores a subject, however, most citizens remain completely uninformed about it. The most interested and attentive citizens may still acquire information through specialized informational networks, but the general public remains in the dark.
Representatives, challengers, interest group leaders, opinion leaders, and journalists play their various roles in creating and transmitting information about representatives' actions in office. How much of this information do ordinary citizens actually receive? How much does whatever information citizens receive affect their evaluations of representatives? Scholars do not have satisfactory answers to these questions.
Determining how much information citizens receive is a difficult task. There are no covert recordings of the messages that citizens actually see or hear. All we have are their own fallible memories. Most citizens do recall receiving communications from or about their representatives. In a survey of citizens who voted in the 1994 House election, 65 percent reported reading about their representative in a newspaper, 61 percent seeing her on television, 33 percent hearing her on the radio, 63 percent receiving mail from her, 14 percent seeing her at a meeting, and 15 percent meeting her personally. Nine of every ten voters recalled at least one of these forms of communications (Jacobson 1997, 100).
We know far less about the content of the information that citizens receive in these communications. Most voters apparently learn their representative's name. In 1994, 51 percent of voters recalled their representative's name from memory, and 93 percent recognized it when offered a list of names (Jacobson 1997, 96). When asked specific questions about their representative, citizens display more modest levels of knowledge. Asked in 1991 how their representative voted on the Persian Gulf Resolution, 19 percent answered correctly and 39 percent guessed correctly. Asked in 1994 how their representative voted on the recent crime bill, 23 percent answered correctly and 28 percent guessed correctly (Alvarez and Gronke 1996; Wilson and Gronke 2000).
Asking citizens to recall specific bits of information about a representative may not be the best way to determine what information citizens actually receive or how the information received affects how they evaluate their representative. Recall of information is most relevant if citizens' decision making is memory based. If citizens make decisions about whether to support or oppose a representative first by recalling all relevant information stored in memory, then evaluating the individual bits of information, and finally combining the individual evaluations into an overall evaluation, then knowing what kinds of information citizens remember would be useful. On the other hand, if citizens process information on-line as they receive it and store only summary evaluations in memory, then knowing what kinds of information citizens remember would not be as helpful.
The jury is still out as to whether citizens' decision making about politics is better captured by memory-based or on-line models. Memory-based models are better at explaining how citizens make decisions about things that they are not expecting to evaluate. John Zaller's account of how citizens answer survey questions about policy alternatives is persuasive (Zaller 1992). In his model, citizens canvass considerations at the "top of their heads" and answer according to the net value of the considerations that come to mind. Since things at the top of the head are often matters that were recently activated, perhaps by recent media stories or perhaps by the survey itself, Zaller can account for how citizens express opinions about a wide range of policy alternatives. Memory-based models seem less satisfactory for explaining how citizens evaluate things that they expect to evaluate.8 Knowing that I need to assign grades to students at the end of the semester, I constantly update my evaluations of each student, rather than storing in memory for later evaluation everything they say in class or write in their papers. Knowing that they need to evaluate regularly their senators and representatives, some citizens operate in similar fashion (Just et al. 1996, 21-22).
Milton Lodge and his colleagues offer as an alternative to memory-based models an impression-driven or on-line model of decision making in which citizens react to information as they are exposed to it, storing in memory only summary evaluations.9 In experimental settings, they show that their on-line model outperforms memory-based models. They conclude that campaign information strongly affects citizens' evaluations of candidates, even though most people cannot later recall the original information (Lodge, McGraw, Stroh 1989; Lodge, Steenbergen, Brau 1995).
If citizens use on-line information processing for evaluating representatives and quickly forget most information they receive, then measures of information recall are poor indicators of citizens' exposure to and reception of politically relevant information.10 We need more direct measures of the informational environment in which citizens operate. Knowledge about the informational environment is also helpful for understanding what citizens do happen to remember. Observers are often surprised that most citizens cannot recall how representatives voted on specific roll-call votes. It is never clear, however, whether the press featured these votes prominently and citizens failed to notice or remember them or whether the press never spotlighted the votes in the first place. Put differently, are citizens largely to blame for how uninformed they seem about politics and public affairs, or is the press more at fault for failing to report frequently and prominently basic facts about representatives' behavior in office?
Some citizens acquire information about politics and public affairs directly from the mass media. They read newspapers, watch television, or listen to radio newscasts. Many others acquire information indirectly (Huckfeldt and Sprague 1995). They learn from a spouse, friend, coworker, or union leader that their representative voted wrong on the North American Free Trade Agreement. Even when citizens do not acquire information directly from the mass media, the media are generally involved in disseminating political information at earlier stages--for example, to a spouse, friend, coworker, or union leader (Mondak 1995, 101-24). Knowledge about the informational environment is helpful for understanding citizens' decision making no matter whether citizens acquire information directly or indirectly from the mass media. A rich informational environment is more likely to produce an informed citizenry than is an informational wasteland.
Journalists need to be selective in what they report about a representative's actions. They could not possibly report everything that a representative did in office--every bill introduced, speech made, position taken, meeting attended, lobbyist met, compromise offered, and contribution solicited. These are the raw materials for good stories, but journalists must select whatever actions seem most newsworthy and write stories that summarize and interpret these actions in interesting and appealing ways.
What kinds of information would be most helpful to citizens? What facts and opinions are most relevant to citizens holding their representatives accountable? At least four kinds of information are especially useful. First, citizens profit by knowing what positions legislators have taken on the important issues of the day. How have their representatives voted on bills that reached the House floor? Where do they stand on various presidential proposals and on bills still in committee? How have representatives explained their positions, especially those that seem contrary to their campaign promises or to citizens' expressed preferences? Position taking is a major part of what legislators do. Knowing what positions representatives have taken helps citizens apportion responsibility for what Congress has done.
Second, citizens benefit by knowing how representatives have contributed to policy making beyond supporting or opposing other legislators' proposals. Nothing happens in Congress unless someone plans for it and works for it. How have individual representatives contributed to legislative action in areas that are important to their constituents or that are part of their committee responsibilities? Are they introducing bills, mobilizing support, and working to solve problems, or are they waiting for other legislators to do the heavy lifting? On any given bill, most legislators are position takers. That is the reality of a legislature with 535 members. But a legislature full of nothing but position takers is an institution that accomplishes little.
Third, citizens benefit by knowing how other people evaluate a representative's performance in office. Ultimately citizens can decide for themselves whether a representative deserves to be reelected or removed. But they are assisted in that task if they first hear a broad range of opinions about a representative's performance. Citizens may find these opinions expressed in news stories, where journalists often seek evaluative comments from representatives, their supporters, and their critics, or on the editorial and op-ed pages, where columnists, editorial writers, politicians, interest group leaders, and opinion leaders debate the accomplishments and failings of Congress and its members.
Finally, citizens gain by hearing about the various candidates running for Congress, including those running in primaries and the general election. Who are the candidates that are challenging the incumbent representative and what are their messages? Do journalists focus on the candidates' past accomplishments and policy differences, or do they feature horse race coverage--who is ahead and who is behind? Do journalists give balanced coverage to incumbents and challengers?
Journalists who report all four types of information increase the probability of citizens acquiring the kinds of information they need to hold representatives accountable for their actions. Journalists who focus on only one or two types of news deprive citizens of the full range of knowledge that contributes to an informed citizenry.
Political accountability in the American system is achieved not only by citizens removing from office legislators with disagreeable records but by legislators anticipating what citizens might do and working to forestall unfavorable evaluations. To be sure, the system would not work well if citizens never removed representatives for cause. All that is required, however, is that some representatives fail their reelection examinations some of the time. These failures remind the survivors and instruct the newcomers that anticipating citizens' preferences is the best way to avoid electoral trouble. Defeat at the polls is common enough among career-minded politicians that most representatives have a healthy fear of electoral retribution. They modify some of their behaviors in Washington to remain popular at home (Mayhew 1974; Arnold 1990, 1993).
Representatives may also modify their behavior based on how journalists report news about their Washington activities. Representatives who observe that newspapers regularly feature their positions and actions may make different electoral calculations when they are deciding how to vote or what activities to pursue than do representatives who seldom see news coverage of their Washington activities. When journalists cover a single story assiduously, they increase the probability that citizens will notice what a representative is doing. If a representative believes that the whole constituency is watching, he may adjust his behavior to make it more pleasing. When journalists ignore a story, however, a representative may be less concerned that citizens will ever learn of his actions. In short, the volume and content of press coverage may affect the way in which representatives anticipate and respond to citizens' preferences when they make legislative decisions. Extensive coverage may make representatives more responsive to citizens' policy preferences.
How journalists report the news can also affect the very activities that representatives undertake. If legislators observe that journalists convey little information about legislative activity beyond what legislators reveal in their press releases, they may focus their creative talents on writing press releases rather than writing laws. If legislators notice that journalists monitor carefully what legislators are doing to solve national problems and reform governmental institutions, however, legislators may decide that doing these things well is the best way to attract favorable coverage and impress their constituents. We know that representatives monitor what journalists write about them (Cook 1989, 75, 201). It does not require an enormous leap to imagine that representatives tailor their activities to attract the best coverage.
How journalists practice their profession, then, can affect the behavior of both citizens and legislators. If the press reports nothing about legislators' actions in office, citizens may have insufficient information for determining whether they should renew or terminate representatives' contracts. If legislators know in advance that their actions will go unreported, they may have less reason to make pleasing decisions in the first place. Studying how journalists report news about local representatives, therefore, allows one to make inferences both about how citizens evaluate representatives and about how representatives make legislative decisions.
The richer the informational environment, the better the two accountability mechanisms work. A rich informational environment increases the chances that citizens will have an evidentiary basis for determining whether they approve or disapprove of a representative's performance in office. A rich informational environment increases the chances that representatives will choose their positions and actions with great care. Of course, local media outlets are only a part of the informational environment in which citizens and representatives operate. Representatives, challengers, and interest groups have other ways to communicate with citizens. Representatives use newsletters and community meetings to communicate directly with citizens. During campaign season, representatives, challengers, and interest groups sponsor events, use direct mail, and purchase advertisements from local media outlets.
An environment in which incumbent representatives and their supporters emphasize their accomplishments while challengers and other critics emphasize representatives' shortcomings can be an informative one for citizens. Just as the adversarial system in trial courts, where attorneys on opposing sides make the best possible cases for their clients, can be an effective way to uncover the truth, so too can an adversarial system in politics demonstrate how well or how poorly incumbents have performed. How informative such a system is for citizens, however, depends on how likely it is that citizens hear both sides. An adversarial system in politics is not necessarily informative for citizens, any more than it is necessarily the best way to uncover the truth in court. If one litigant is represented by the best attorneys that money can buy while the other is represented by a rookie lawyer, jurors may have trouble uncovering the truth. Similarly, if incumbent legislators have ample opportunities to publicize their accomplishments, while challengers and other critics have few opportunities to publicize their criticisms, citizens will be less able to evaluate legislators' fitness for office than they would be if the flow of information were more balanced.
The system works best when lots of citizens notice and read what newspapers publish about representatives. But the system does not break down simply because most citizens are not the ideal citizens that populate democratic theory. Not every citizen needs to be a front-line sentry to keep representatives on their toes. As long as a cadre of individuals and organizations monitor what representatives are doing in office and stand ready to inform other citizens when they see something out of line, representatives know that they are being watched. Much more important is that information regularly flows to those who act as watchdogs, that these watchdogs reflect the diversity of interests in a constituency, and that they have easy ways to communicate with other citizens when they discover representatives doing disagreeable things.
A division of labor is central to any large political system. Yet many observers of American politics fail to appreciate the division of labor between those who actively monitor political actors and those who reward and punish them for their actions. For example, many political scientists once concluded that representatives had little influence over bureaucrats' behavior because they did not systematically monitor what bureaucrats were doing. Mathew McCubbins and Thomas Schwartz changed that view by differentiating between two ways in which principals monitor agents' behavior. They argued that "police-patrol oversight," where a principal actively monitors everything that an agent does, is relatively rare in politics, while "fire-alarm oversight," where other actors do the monitoring and then inform a principal when an agent steps out of line, is more common. Fire-alarm oversight is efficient because it takes advantage of a division of labor (McCubbins and Schwartz 1984).
The discussion in this chapter recognizes a division of labor between professional watchdogs, amateur watchdogs, journalists, and ordinary citizens. Professional watchdogs, including challengers, potential challengers, party leaders, and interest group leaders, have the incentives and resources to ferret out information about a representative's actions even if local journalists ignore what a representative is doing. Amateur watchdogs are local citizens who are intensely interested in politics and publics affairs or in particular problems, policies, or programs. These local opinion leaders rely on journalists for most of their information about a representative's actions. Ordinary citizens may notice regular news coverage about a representative or they may simply wait for professional or amateur watchdogs to sound the alarm when a representative steps out of line.
What would be the consequence if local media outlets ignored what representatives were doing in office? Most representatives would probably be advantaged. Their ability to communicate with citizens using newsletters, meetings, and campaign advertisements would be undiminished, so they could continue to tout their accomplishments. Whether or not challengers and interest groups were able to publicize representatives' shortcomings would depend partly on the seriousness of representatives' shortcomings and partly on their access to financial resources to publicize representatives' transgressions. If a representative made lots of careless choices on major issues, if she repeatedly annoyed powerful interest groups, or if she was indicted for a felony, she would provide powerful ammunition for her opponents. As long as a representative avoided doing these things, however, it would be difficult for another candidate to mount a serious challenge. Accountability would be achieved largely by a representative anticipating what major issues might be used against her and adjusting her behavior accordingly.
When local newspapers cover representatives frequently and thoroughly, however, both accountability mechanisms work better. Extensive coverage allows local opinion leaders and ordinary citizens to monitor contemporaneously what a representative is doing. Citizens need not wait for an angry interest group or a well-funded challenger to inform them during campaign season about a representative's shortcomings; they can read about a representative's positions and actions in the morning paper and judge for themselves. Contemporaneous coverage also encourages a dialogue between a representative and her constituents. Citizens raise challenging questions when a representative attends local meetings; they write letters to the editor complaining about a representative's positions and actions; they send protest letters directly to their representative; they work to mobilize their friends, neighbors, and colleagues to do all these things. Moreover, citizens can monitor their representative for the price of a daily newspaper. They don't need to be represented in Washington by interest groups in order to discover what their elected representative is doing. A final consequence--and a crucial one--is that a representative knows that her actions in Washington are reported regularly to her constituents. Accordingly, she works hard to look good every day of the week. She takes great care when she acts on scores of issues, not just the few issues that would make good campaign issues a year hence.
Standards of News Quality
Extensive coverage of a representative's positions and actions increases the chances for accountable government. But how much is enough? Is there some minimum level of coverage that is necessary to allow citizens to monitor a representative's actions? Can one establish standards for deciding whether newspapers have created an informational environment sufficiently rich to keep representatives on their toes?
John Zaller offers two standards for evaluating the quality of news coverage. According to the Full News Standard, newspapers "should provide citizens with the basic information necessary to form and update opinions on all of the major issues of the day, including the performance of top public officials" (Zaller 2003, 110). In contrast, the Burglar Alarm Standard suggests that newspapers should focus on just a few issues that others have identified as particularly important.11 Those who might sound the alarm to attract journalists' attention to particular issues include interest group leaders, party officials, the president, and electoral challengers--essentially those I have labeled professional watchdogs. For the case of local newspapers and local representatives, Zaller argues that newspapers should focus on how representatives have voted on bills that were top presidential priorities, on representatives' ethical transgressions, and on any reelection races that appear to be close.
Zaller's distinction between the Full News Standard, which he believes has dominated media studies, and the Burglar Alarm Standard, which he argues is a more appropriate standard, is important and useful. Indeed, I use this distinction throughout the book. But I disagree with Zaller on one point. I believe it is important for local newspapers to meet both the Full News and the Burglar Alarm Standards. Zaller recognizes the need for the elite news media at the national level--the New York Times, National Public Radio, the Newshour with Jim Lehrer--to meet the Full News Standard, so that politically attentive citizens can form and update their opinions on the important issues of the day and monitor what public officials are doing. He argues that the Burglar Alarm Standard is the more appropriate benchmark for thousands of other media outlets because most citizens are not sufficiently interested in politics and public affairs to want exhaustive coverage. The basic notion is that opinion leaders use the elite media outlets to keep informed, and once they notice something disturbing, they sound the alarm through mass media outlets. In local markets, however, newspapers are simultaneously elite and mass media outlets. Local newspapers are the only media outlets capable of informing local opinion leaders about what representatives are doing. Local newspapers are also well positioned to broadcast alarms to ordinary citizens when professional watchdogs or local opinion leaders sound the alarm about disagreeable things that representatives have been doing.
When local newspapers do not meet the Burglar Alarm Standard, they seriously impede citizens' ability to hold representatives accountable for their actions. Although well-funded interest groups and well-funded challengers can still communicate directly with citizens about representatives' shortcomings, the resources required for direct communications are so substantial that only the most serious shortcomings are likely to be publicized. When local newspapers meet the Burglar Alarm Standard, they increase the prospects for accountable government. By providing challengers, interest group leaders, and other professional watchdogs with opportunities to sound alarms about representatives' transgressions, without them having to purchase direct coverage, these papers increase the number and variety of transgressions that can be publicized. When local newspapers meet both the Burglar Alarm and Full News Standards, they further increase the prospects for accountable government because they inform and empower local opinion leaders. Representatives are likely to be especially vigilant when they know that a wide range of local opinion leaders, and not just Washington-based interest groups, are monitoring their behavior.
The literature on how the mass media cover politics is large and growing. Most of the literature focuses on how the press covers wars, presidents, election campaigns, and policy issues. Scholars have largely ignored press coverage of Congress and its members. Only ten studies have examined how local newspapers and television stations covered individual members of the House or Senate. These ten studies are best described as pilot studies, rather than systematic studies, because the authors typically chose nonrandom samples of media outlets and studied those outlets for only a few weeks, typically during the campaign period. Most studies focused more on the volume of coverage than on its content.12
This book differs from previous studies by exploring four sets of questions about the volume, content, causes, and consequences of newspaper coverage. First, it seeks to establish how frequently local media outlets cover members of Congress. Do media outlets regularly report information about representatives' actions in office, and do they display their coverage in prominent ways? Or is coverage of representatives infrequent, spotty, or buried in the back pages of newspapers? It is important to determine something about the volume and prominence of political information because both factors affect whether citizens are likely to notice and digest the information.
Second, it examines the content of press coverage of individual legislators. Do the media report the kinds of information that citizens would need to hold representatives accountable for their actions in office, or do they focus on more peripheral matters that entertain, amuse, or enrage citizens without conveying much information about legislators' actual performance? Do they feature bill introductions, roll-call votes, leadership activities, and constituency service? Are the media evenhanded in their stories, or do they offer more extensive or more positive coverage to incumbents than to challengers, or to Democrats than to Republicans?
Third, it seeks to explain why news outlets differ in their coverage of Congress and its members. Why do some media outlets provide exemplary coverage of local representatives while others largely ignore representatives' activities? Do large, well-financed urban newspapers provide better coverage of representatives, or do these papers avoid extensive coverage of local representatives because their primary circulation areas include so many congressional districts? Does press coverage depend on what representatives do in Congress? Do local media outlets cover more extensively legislators who are important participants in congressional policy making--the workhorses--or do representatives attract local press attention by constituency-oriented activities? Does it matter whether media outlets have Washington correspondents?
Finally, it attempts to discover whether differential coverage of local representatives affects citizens' political knowledge. Are citizens who live in areas where media outlets carefully cover representatives more likely to recall or recognize their representatives than citizens who live in areas where media attention is sparse? Does media attention affect the chances that citizens will know something about representatives' records? When the media report extensive information about roll-call votes, are citizens more likely to know where their representative stands on the issues?
Unlike previous studies, which largely focused on the campaign period, this project explores how local newspapers covered representatives during an entire congressional session, from the first day of 1993 to election day 1994. The longer period is essential for studying political accountability. In order to determine what kinds of information newspapers make available to citizens, one needs to collect newspaper articles from a reasonable number of papers, for a reasonable number of representatives, and over a sufficiently long time period. Focusing on how a few newspapers cover a few representatives over a few weeks does not allow one to discover how coverage patterns vary over the cycle of governing, campaigning, and elections, or to generalize with any degree of certainty to the universe of all newspapers and all representatives. Attempting to balance these competing needs, I have selected three samples of newspaper coverage, each sample designed to reveal a different aspect of press coverage.
The first data set is a sample of 25 local newspapers and 25 representatives. It contains every news story, editorial, opinion column, letter, and list that mentioned the local representative between January 1, 1993, and November 8, 1994. My sampling strategy involved first selecting as representative a set of newspapers as possible, and then selecting randomly one House member from each newspaper's primary circulation area.
Selecting the newspaper sample was the greater challenge. At the time the sample was drawn, there were 1,567 daily newspapers in the country with combined circulations of 57 million copies.13 Eighty-eight of these newspapers had publicly available electronic archives for all of 1993 and 1994.14 The problem was to draw a sample of these 88 newspapers that was a reasonable approximation of the universe of all daily papers. The good news was that the 88 papers included 38 percent of the total daily circulation in the country, despite the fact that they represented only 6 percent of all daily papers. This followed from the fact that a majority of citizens read a newspaper with a daily circulation of more than 100,000 copies,15 and large newspapers were overrepresented among the 88 papers. The bad news was that smaller newspapers were underrepresented in the electronic archives, and smaller newspapers tend to serve small cities and rural areas.
In order to draw a sample of newspapers that is representative of what the average citizen reads, I rank-ordered the 1,567 papers according to circulation, and then grouped the papers into approximate sextiles so that each group represented about one-sixth of the total daily circulation in the country. I then highlighted the 88 archived papers within the various sextiles.16 Given that the two lowest sextiles contained only seven of the 88 papers, I combined these two sextiles into a single group. I then randomly selected five papers from each of the five groups. The sample of 25 newspapers includes large national papers like the Los Angeles Times and the Boston Globe, midsized papers like the Hartford Courant and the Tulsa World, and small-city papers like the Rock Hill Herald (South Carolina) and the Lewiston Morning Tribune (Idaho). From each newspaper's primary circulation area I randomly selected one representative for study.17 Table 1.1 lists the newspapers and representatives in the first data set.
The 25 newspapers are a diverse lot. Although very small newspapers are necessarily absent, the inclusion of a few small papers should allow one to determine if small newspapers cover House members differently than large newspapers. The sample also contains various types of papers and not just the most celebrated newspapers in the country. It includes tabloids, such as the Chicago Sun-Times and Newsday, rather than their highbrow competitors, the Chicago Tribune and the New York Times; it includes the upstart Washington Times rather than the Washington Post. The sample is also geographically diverse, with newspapers from eighteen states and the nation's capital. The newspapers sell nearly 7 million copies daily--about 12 percent of the nation's total daily circulation. The sample is not as diverse as one would have obtained if one selected a stratified random sample from the complete list of 1,567 newspapers. Taking that route, however, would have required searching most newspapers manually, which would have necessitated a much smaller sample and a much shorter time frame.
The 25 House members are reasonably representative of the whole House.18 The match in party and seniority was especially good. Fourteen representatives in the sample were Democrats (56 percent), just shy of their actual percentage in the House (60 percent). The median representative in the sample was elected in 1986, as was the median member of the House. Two representatives in the sample, James Bilbray and Larry LaRocco, ran for reelection and lost, exactly matching the percentage for the whole House. Retiring members were underrepresented (only Romano Mazzoli), women were underrepresented (only Barbara Kennelly), and black members were overrepresented (Ronald Dellums, Louis Stokes, and Albert Wynn). Two of the 25 richest districts in the country made the list (CA24 and NY3); none of the 25 poorest districts did. Although none of these differences are out of line for a sample this small, they are worth noting.
After choosing a sample of newspapers and representatives, I used computerized routines for searching the text of the 16,950 daily newspapers (25 newspapers times 678 days). This search identified and retrieved 8,003 news stories, editorials, opinion columns, letters, and lists that mentioned the 25 local representatives. Three research assistants read the material, coded the articles for their objective content, and summarized the tone and valence of each article. They used 68 variables to code a variety of information, ranging from the size, location, and prominence of each article, to whether an article mentioned a representative's policy positions, roll-call votes, or leadership activities. They also tracked the appearance of 214 policy issues in order to see how journalists portrayed representatives' connections to highly visible issues, such as NAFTA, the budget, crime, and gun control, as well as to less visible issues that Congress handles every year.19
One limitation of the first data set is that one cannot determine what accounts for large differences in coverage. Why, for example, did some newspapers cover their representatives more heavily than other papers covered theirs? Did these differences in coverage reflect differences in the newsworthiness of representatives or in the editorial practices of newspapers? The question is unanswerable with a data set in which there is a one-to-one correspondence between each newspaper and each representative. The second and third data sets are designed to overcome this limitation.
The second data set parallels the first. I simply paired six newspapers from the first data set with six newspapers that are published in the same cities. The paired newspapers are from Boston, Chicago, San Francisco, Seattle, Tucson, and Washington.20 The aim is to determine how pairs of competing newspapers covered the same legislators. My assistants coded the news stories, editorials, columns, letters, and lists in the additional papers according to the same procedures used for the first data set. The second data set contains 2,175 articles--1,053 from the original six papers and 1,122 from the six comparison papers.
The third data set includes information about the volume and timing of coverage for a much larger sample of newspapers and representatives. This data set shows how 67 local newspapers covered 187 representatives during 1993 and 1994, with a total of 242 representative/newspaper dyads. The 61,084 citations--headline, date, section, page, and byline, but not full text--allow one to analyze how the amount and timing of coverage depend on the newsworthiness of individual representatives, the competitiveness of elections, and the resources and constraints of individual newspapers. The third data set is not a random sample of all newspapers; it is closer to the universe of all newspapers that were available for electronic searches in 1993 and 1994.21 But imbedded in this data set are the 25 randomly selected newspapers from the first data set. By analyzing separately how these 25 newspapers covered the 91 legislators within their primary circulation areas (22,175 citations), I can determine if the larger but less representative sample differs significantly from the smaller but more representative sample.
The fourth data set is designed to determine whether the volume of newspaper coverage affected what citizens knew about their representatives. This data set was constructed by linking information about how extensively the 67 newspapers in the third data set covered particular representatives with information about citizens' knowledge of their local representatives, as recorded in the autumn 1994 survey conducted by the National Election Studies. The unit of analysis is the individual citizen. Added to the usual attitudinal data about each citizen is information about how a local newspaper covered that citizen's representative during 1993 and 1994. The original 1994 NES data set had 1,795 respondents. I have information about local newspaper coverage for 675 of these respondents. Although the fourth data set is not ideal, it is the best that can be assembled, given the original NES survey. The survey contained information about how many times a week a citizen claimed to read a newspaper but not the name of the newspaper that a citizen read. So, I have been forced to assume that the local newspaper for which I have data is the same newspaper that a citizen actually read. The result, of course, is noisy data.22
The main innovation in this study is the use of computerized text searching to locate each mention of a representative's name in a local newspaper between January 1993 andNovember 1994. Although computerized text searching has been used for studies of the national news media (Fan 1988; Fan and Norem 1992), it has not been used to study how local newspapers cover members of Congress. As with any new technology, there are advantages and disadvantages.
The principal advantages of computerized text searching are efficiency and accuracy. It would be extraordinarily expensive to acquire and read paper copies of the 45,426 daily newspapers searched for this study (67 papers times 678 days). It was much less expensive to search through the electronic archives of the same newspapers. Computerized text searching is also more accurate. Computers are very good at finding every mention of a name in millions of lines of text. Human coders are remarkably poor at this mind-numbing task. One of the earliest studies of how newspapers covered members of Congress employed a commercial clipping service (Clarke and Evans 1983). A subsequent audit revealed that the clipping service missed about two-thirds of the articles (Goldenberg and Traugott 1984, 133).
An additional advantage of using electronic archives is that the archives contain articles from all regional editions and neighborhood sections that a newspaper happens to publish. The Los Angeles Times, for example, publishes six regional editions, as well as special sections for various neighborhoods. Libraries acquire paper or microfilm copies of only one edition. Searching a library's copies of the Times would not be a problem if one were studying a subject that appears in all daily editions--for example, stories about the president or governor. It turns out to be a very serious problem, however, when one is studying how newspapers cover representatives because the editors of the Times intentionally place much of the coverage of individual representatives in the various regional editions and neighborhood sections. The editors target constituents with news about their own representatives without distracting them with news about every other representative in Southern California. In order to study how large metropolitan newspapers like the Times cover members of Congress, one needs either to acquire all the regional editions and neighborhood sections of these newspapers--something that no library does--or to use the electronic archives that contain material from all editions and sections.
Computerized text searching allows one to analyze very large samples. This is a direct consequence of the technology's twin advantages of efficiency and accuracy. Previous studies limited themselves to a few weeks of coverage. This book includes 97 weeks. Previous studies examined an average of 908 articles. This book examines 61,084 citations and 9,125 full-text articles. The first data set alone is roughly equivalent to the combined sample sizes of the ten previous studies.23 Large samples allow one to find subtler patterns in newspaper coverage and to generalize with greater confidence to other newspapers and representatives.
Electronic archives have several disadvantages. Some newspapers do not own the electronic rights to everything they publish. As a consequence, information about representatives that appeared in syndicated columns or wire services' reports may not be included in the electronic archives for particular newspapers (Snider and Janda 1998). In addition, some newspapers choose not to include other types of materials to which they do control the rights--for example, letters to the editor and information in tables.24 Although Nexis and Dow Jones list the types of information excluded for each newspaper, there is no way to estimate the amount of coverage that is actually excluded. My sense is that excluded information is not a serious problem for a study of how local newspapers cover members of Congress, whereas it would be a very serious problem for a study of how local newspapers cover international affairs, since most papers rely heavily on wire services for foreign coverage.
Another problem with computerized text searching is that it is difficult to construct searches that retrieve only articles mentioning the representative. A search that is excessively narrow misses actual coverage of a representative; a search that is too broad finds articles about people who share a representative's name. My solution was to conduct broad searches and then use human coders to discard articles that referred to someone other than the representative.25
Another problem is that electronic archives contain duplicate copies of some articles. Twenty-three of the 25 newspapers archived at least some duplicate articles about local representatives. The problem was most common for newspapers that published timed or regional editions. For example, the San Diego Union-Tribune, which publishes as an all-day newspaper with up to nine editions over a twenty-four-hour period, archived the most duplicate articles (26 percent).26 Others newspapers placed nearly identical materials in different regional editions, especially when a congressional district was split between two or more newspaper regions. My solution was to use computerized text searching to find all possible articles and then use human coders to discard duplicates.27 Table 1.2 shows summary information for the four data sets, including details about sampling and error correction.28
The advantages of computerized text searching far outweigh the disadvantages. Two of the disadvantages can be overcome by using human coders to discard duplicate articles and articles that refer to someone other than the representative. The other disadvantage--the exclusion of syndicated columns and wire services' reports--is probably not a serious problem when studying how local newspapers cover members of Congress. In any event, this one disadvantage is a small price to pay for obtaining very large samples and for correctly observing the coverage patterns in newspapers that publish regional editions and neighborhood sections.
Description and Analysis
Description is a necessary part of this research enterprise. We know so little about how local newspapers cover members of Congress that one cannot assess the adequacy of newspaper coverage or analyze the causes and consequences of that coverage without first describing the nature of coverage. No doubt each reader of this book begins with some sense of how local newspapers cover members of Congress, perhaps derived from reading a single local paper, perhaps derived from reading papers in several localities where the reader has lived over the years. My first task is to replace whatever impressionistic sense the reader may have about how local papers cover members of Congress with firm evidence from a sample of newspapers.
The descriptive task is complicated by the enormous disparities in newspaper coverage. Some newspapers cover representatives extensively, painting rich portraits of where they stand on the issues of the day and what they are doing in Washington. Other newspapers provide meager coverage--meager in both quantity and quality. Throughout this book I convey the nature of these disparities, both with tables that show how each newspaper covered various aspects of the representative's behavior and with occasional summaries of typical, exemplary, or superficial articles. Description of newspaper coverage appears in chapters 2 through 7.
In addition to describing coverage patterns, I assess the quality of newspaper coverage, and particularly whether the information that journalists publish helps citizens hold representatives accountable for their actions. I offer three vantage points for this assessment. Previous work on political accountability shows the importance of particular types of information about legislators' behavior (Mayhew 1974; Arnold 1990, 1993). The current book is partly an attempt to see how much of this information journalists actually provide. A second gauge is provided by Zaller's distinction between the Full News Standard and the Burglar Alarm Standard. Do newspapers provide coverage that allows individuals and groups to monitor what representatives are doing in office? Do newspapers provide opportunities for opinion leaders to communicate with citizens when they discover representatives doing disagreeable things? A third yardstick is provided by journalists themselves. Since journalists often reject outsiders' recommendations about how they should cover the news--what could scholars know about the realities of attracting readers and satisfying advertisers and stockholders?--I treat as a baseline for assessment the actual coverage in the best newspapers in the sample. The best practices of these newspapers--some small, some medium, some large--offer a standard that is more palatable to journalists. Preliminary assessments appear in each of the descriptive chapters (chapters 2 through 7). Assessment takes center stage in chapter 9.
In addition to describing and assessing newspaper coverage, I seek to identify the causal forces that produce particular coverage patterns. For example, chapter 2 first describes the volume of coverage in each newspaper, and then proceeds to test various hypotheses that seek to explain these variations. Other chapters attempt to explain variations in how newspapers cover electoral campaigns, bill introductions, and other topics. I also seek to determine whether variations in newspaper coverage have consequences for what citizens know about representatives. This is the central subject of chapter 8. In short, the book is a combination of description, assessment, and the testing of various hypotheses about the causes and consequences of newspaper coverage.
Return to Book Description
File created: 8/7/2007
Questions and comments to: email@example.com
Princeton University Press