Microsoft Word - journal commonwealth 2007 FINAL 0307.doc 45 What’s Local about Local Television News? An Analysis of 2004 Election Coverage by Philadelphia’s Local Television Stations Matthew Hale Seton Hall University Because capturing and analyzing local television news broadcasts is difficult, most research on campaign coverage by local television stations centers on a single candidate race and a limited number of news broadcasts. This article expands existing research by examining all prime time election coverage by Philadelphia’s four network affiliates in the final 30 days of the 2004 campaign. These stations virtually abandoned coverage of local and statewide races in favor of the presidential contest. Most coverage focused on campaign strategy rather than issues, but there is little evidence of direct bias favoring either Democrats or Republicans. Although stations generally ignored nonpresidential elections, they gave a fair amount of coverage to the mechanics of voting, which may provide voters with valuable locally based election information. Political communication scholars are divided, often bitterly, over the potential effects of television news on citizen engagement, knowledge, turnout, and voting behavior. For many years, the groundbreaking work of Patterson and McClure (1976) held sway and the conventional wisdom was that television news had virtually no effect on voters. More recently, the strength of this minimal affects position has declined and a host of empirical studies have suggested that the media can influence citizen attitudes and behavior (Ansolabehere and Iyengar, 1995; Entman, 1992; Finkel and Geer 1998; Gilliam and Iyengar, 2000; Goldstein and Freedman 2002; Graber 1997 and 2001; Iyengar and Kinder 1987; Just, Crigler, Alger, Cook, Kern, and West 1996; Shaw 1999; Valentino 1999; Wattenberg and Brians 1999). Much of this work is based on studies of network news broadcasts. This focus is primarily one of expedience. Since network news airs once a night and can be captured in any part of the country, it is simply easier for scholars to analyze network news instead of local TV news. 46 Scholars are beginning to recognize the importance of local TV news. For example, Althaus, Nardulli, and Shaw (2002, 2) bluntly state, “Studies of the elite national news media ignore the simple fact that local news broadcasts are now the average citizen’s primary source of information about presidential campaigns.” This contention is supported by several surveys from the Pew Center indicating the centrality of local news as a trusted information source.1 Snider (2000) highlights the importance of local news with his contention that an archive of local news would be a public good and, as such, Congress should mandate the creation of a national archive of local news programs similar to Vanderbilt University’s archive of national news programs. Since Congress has not done so, researchers interested in local news and campaigns are forced to limit what local news content they capture and analyze. This happens in at least three ways. First, researchers concentrate on a single, often presidential, election (Bartels 1988; Beck, Dalton, Greene, and Huckfeldt 2002; Finkel and Geer 1998; Iyengar and Kinder 1987; Shaw 1999, Just et al. 1996). Second, the few researchers (Carter, Fico, and McCabe, 2002; Just et al. 1999) who attempt to examine multiple races are forced to limit their research to single news broadcasts (e.g. the 5:00 p.m. broadcast but not the 11:00 p.m. broadcast). Third, researchers completely ignore stories that focus on the process of the election (e.g. voter registration deadlines and locations of polling places). These “voter information” stories are comparatively prevalent on local news broadcasts. All these strategies present methodological problems. Over 40 years ago, Bernard Cohen argued that the media do not tell us what to think; they tell us what to think about. Following Cohen, it is reasonable to suggest that comparative balance or imbalance in the quantity of coverage about different elections signals voters about the comparative worth of those elections. For example, a lack of coverage about one U.S. House race tells voters that they do not need to think about it, just as blanket coverage of another U.S. House race signals voters to pay attention to this one. Research focusing on a single race cannot speak to this fundamental point. In addition, focusing on a single race assumes that the qualitative aspects of news coverage, such as bias, are uniform across elections. Yet there is no reason to believe that a television 47 news organization will act with this type of consistency. It is clearly possible that a local television station may favor a Democrat in a U.S. Senate race and favor a Republican in the Presidential race. Again, research focusing on a single race can not address this point. It is not surprising that research on local television news centers on candidates; that is, after all, what research on network news does. This focus, however, ignores stories about the process and mechanics of the election itself, stories that are in fact more likely to occur on local television stations than on network news (Kaplan, Goldstein, and Hale 2004). In addition to being more prevalent on local news than national news, these “voter information” stories often provide voters with information directly relevant to the situation they will face when they go to the polls. For example, local television stations can report on local voter registration deadlines, polling place locations, and the time and place of local events related to the upcoming election—stories that network news does not report. In addition, local news can report on specific and local instances of voting irregularities or potential difficulties in the local voting process in a way than network news cannot. As a result, there is no reason to believe that these voter information stories are irrelevant and without influence on voters’ perceptions of the election. In fact, it seems reasonable to suggest that a story highlighting a local nonpartisan voter information forum might leave voters with a generally positive view of the election and their ability to participate in it. Conversely, a story highlighting acrimonious lawsuits over the implementation of electronic voting machines might leave voters with a more negative impression. Just as the balance in the quantity of coverage about individual elections may signal voters about which elections are important, the balance in noncandidate stories may help signal voters about the value of participating in the election at all. The first goal of this article is to provide a more complete picture of election news coverage by examining news coverage of all races and all election related stories across multiple news broadcasts. The key research questions include: what is the quantitative balance of coverage across multiple races? Are there qualitative differences in coverage of different races? What types of noncandidate election stories exist, and how might these stories influence voter’s opinions of the election? 48 Obviously, the answers to these questions are largely descriptive. Even so, they are important because they further our understanding of how voters within a media market actually experience election news coverage. For example, it seems reasonable to suggest that voters watch entire (if not multiple) newscasts every night and not (as researchers generally do) single stories about one election. As a result, voters experience one part of election coverage (e.g. presidential election stories) not in a vacuum but in connection with other parts of election coverage (e.g. stories about other races and about the electoral process generally). By describing how the different “parts” of an election are covered in relation to each other, we are better able to understand how voters experience (and perhaps understand) both the parts of an election and the election as a whole. The Philadelphia Media Market: Size and Competition The Philadelphia media market provides an interesting case study because in 2004 it was an example of a large media market with many competitive elections. This combination is significant because the size of the Philadelphia market would suggest that presidential coverage would be much more prevalent than coverage of local elections. Yet, the number of competitive down-ballot races in the Philadelphia market would suggest that coverage between presidential and nonpresidential elections would be more balanced. Examining election coverage in Philadelphia helps inform our understanding of the relationship between these two variables. The Size of the Philadelphia Media Market The Philadelphia media market is the fourth largest media market in the country. It reaches across 18 counties and three states. Eight of the 18 counties are in New Jersey (Atlantic, Burlington, Camden, Cape May, Cumberland, Gloucester, Mercer, and Salem). Eight of the counties are in Pennsylvania (Berks, Bucks, Chester, Delaware, Lehigh, Montgomery, Northampton, and Philadelphia). Two of the counties are in Delaware (Kent and New Castle). Because of its size and cross state reach, the TV stations in Philadelphia generally have more elections to cover than almost any 49 market in the country.3 In addition to the 2004 presidential race, voters in the Philadelphia DMA participated in more than 155 nonpresidential races. These included one U.S. Senate races, six other statewide offices, 15 Congressional races, 19 state senate races, and 113 state assembly races.4 This total does not include the hundreds of local races for city councils, mayors, and law enforcement officers in the 18 counties. Researchers have been unable to make comparisons between local news reports in multiple markets. Nonetheless, it is reasonable to speculate that larger media markets may provide less coverage of local elections than do smaller markets inasmuch as the percentage of a station’s viewers that can vote in a particular race is higher in a small market than in a large market. However, ownership and operation of all the affiliate stations in large markets by the major media companies (ABC, CBS, FOX and NBC) may also contribute to a lack of coverage of local elections because reporters at these stations are more directly linked to, or perhaps identify more with, national journalists. Consequently, coverage of the presidential race and perhaps the U.S. Senate race can be expected to dominate election news coverage on Philadelphia’s television stations. Electoral Competitiveness in Philadelphia Clearly market size is not the only factor in determining what gets covered. A second variable is the presence of competitive elections. Philadelphia is an interesting case study in 2004 because in comparison to most places in the country, a number of races in the Philadelphia area were thought to be competitive or at least interesting. Pennsylvania was considered to be a crucial swing state in the presidential election, and towards the end of the election there was indication that New Jersey too might be competitive in the presidential race. The Pennsylvania Senate race was characterized by the Cook Political Report5 as “leaning Republican,” indicating some degree of competitiveness. Voters in the Philadelphia DMA cast ballots in six other statewide races.6 Of these, four were open seat races, which are generally characterized as less certain than elections with incumbents. In addition, three of the 15 U.S. House races7 held in the DMA were thought to be at least somewhat competitive.8 Given that a total of just 24 U.S. House races were characterized as even marginally competitive in 2004,9 the 50 Philadelphia DMA accounted for 13% of all U.S. House races characterized as leaning Democratic or Republican in the country. While this may be a comment on the lack of competitiveness in U.S. House races in general, it does suggest that in comparison with other media markets, Philadelphia has its share of interesting U.S. House races. Given that local television station managers and news directors routinely argue that elections would receive more coverage if more of them were competitive, it would seem reasonable to expect that Philadelphia might provide a significant amount of coverage to nonpresidential elections. In short, the Philadelphia media market is interesting because we have two competing hypotheses about how the city’s stations might balance their election coverage between presidential and nonpresidential elections. Discovering which one is correct is the second goal of this article. Methodology This article is based on a larger study of 44 stations in 11 media markets2 conducted by the Norman Lear Center at the University of Southern California and the NewsLab at the University of Wisconsin. While the 11 media markets in the full study were not randomly selected, they account for 23% of all television viewers in the country. In addition, the 11 markets are geographically diverse and somewhat politically diverse. There is also variation in the level of electoral competitiveness within the 11 markets at both the presidential and down-ballot levels. The full study examined over 8,000 hours of news coverage and almost 7,000 news stories, making it one of the largest studies of local television news ever conducted. The ability to capture and analyze election coverage across multiple races and broadcasts is made possible by the unique media capture and management system created at Wisconsin’s NewsLab. The process is divided into four distinct phases: capture, clipping, coding, and archiving. In each market, computer servers capture entire news broadcasts and transmit the content electronically to the University of Wisconsin. Once they arrive in Wisconsin, broadcasts are clipped into individual news stories and coded for primary focus (elections, crime, health, foreign policy, etc.). Election stories are then sent to individual computer work stations for coding. Highly 51 trained coders watch and code each election story directly on a computer screen. Specially designed software prevents errors in logical consistency. For example, if a coder says the story concerns only a gubernatorial race but enters the name of a U.S. Senate candidate into a soundbite field, the story is automatically returned to a supervisor to check and correct any errors. When coding is complete, the stories are automatically sent to a digital archive available at www.localnewsarchive.org. Users of the archive can search the video database on a host of items including keywords, story subject, station, market, and date aired. This article is based on a detailed analysis of the 580 election news stories aired by the Philadelphia affiliates of the four major networks (CBS, NBC, ABC, and Fox). As in the larger study, all stories aired during prime time (5:00 p.m. to 11:30 p.m.) between October 4 and November 1, 2004. All stories in Philadelphia were examined by three trained coders. Intercoder reliability was measured using Cronbach’s alpha (Cronbach 1951) and was greater than .83 on all but one variable; and in the case of primary story, focus was above .90. The one exception was the variable called the “average campaign manager score,” which is described in a later section. The initial reliability measure indicated an intercoder agreement of .77. Although some scholars find this acceptable (Nunnally 1978), other argue that intercoder agreement should be greater than .80 (Krippendorff 1978, Tinsley and Weiss 2000). To err on the side of caution, a project supervisor reviewed all stories where the three coders were less than unanimous and made a final determination, thus insuring adequate reliability on all measures. Basic descriptive elements of each story—such as date aired, station, and story length—are automatically determined during the clipping process described above, so reliability measures are not necessary on these variables. Comparing Philadelphia The first step is to compare the Philadelphia media market with the 10 other markets in the larger Lear Center Study. The results show in most respects that the stations in Philadelphia covered the 2004 election somewhat differently than the other stations in the larger study. For example, the Philadelphia stations covered the election a bit more frequently than did the other 10 markets. Seven 52 out of 10 broadcasts captured in Philadelphia contained at least one election related story compared with slightly more than six out of 10 in the other markets studied.10 One possible explanation for this difference is that stories in Philadelphia were on average 13 seconds shorter than stories in the other markets. The Philadelphia stations aired more stories about the presidential race and fewer stories about local races than did the other markets. In addition, Philadelphia stations gave significantly more attention to campaign strategy and the “horserace” and less attention to campaign issues than did the other markets. The stories on Philadelphia stations included a candidate soundbite somewhat less frequently than did the stations in the other markets, but the difference was not statistically significant. There was no difference between Philadelphia and the other markets in the average length of a candidate soundbite. Table 1 contains these results. Table 1 Overview of 2004 Election Coverage Study Sample (all newscasts at 5:00 p.m. and 11:30 p.m. aired between October 4 and November 1, 2004) 10 Markets* Philadelphia Number of local television stations 40 4 Total hours of news programming 8,070 596 Total number of local news broadcasts 4,035 298 Total number of campaign stories 6,441 580 53 *The 10 markets include New York, Los Angeles, Dallas, Miami, Tampa, Orlando, Denver, Seattle, Des Moines, and Dayton (Ohio) Comparing Elections: Quantitative Disparities in Race Coverage Given the importance of the office and the overall competitiveness of the race, it is not surprising that presidential coverage trumped coverage of local and even statewide races in all 11 markets studied in 2004. That Pennsylvania was considered a crucial swing state caused both presidential candidates to spend a great deal of time in the state, and specifically in Philadelphia, so it is not surprising that the presidential race received a great deal of coverage by Philadelphia’s stations. These facts, however, do not make the dominance of the presidential race in Philadelphia any less striking, especially given the presence of other competitive races in the market. For example, not only did the Philadelphia stations devote significantly more of their stories to the presidential race than did stations in the other 10 markets, but they also devoted significantly more of their election coverage to the presidential race than did stations in other presidential battleground states (Dayton, Ohio; Des Moines, Iowa; and Miami, Orlando, and Tampa in Florida). Presidential stories made up 59% of the stories in these five markets compared to 76% of the stories in Philadelphia.11 Overall, the Philadelphia stations aired a total of just 30 stories focused on nonpresidential candidates. Of these, 16 focused on the U.S. Senate race,12 and just 11 stories focused exclusively on U.S. Variables 10 Markets* Philadelphia t test Percentage of broadcasts with at least one campaign story 63% 70% T= 2.45 (347), p=.014 Average length of a campaign story 87 seconds 75 seconds T= 4.48 (702), p=.000 Percentage of stories about the presidential race 60% 76% T= 8.59 (722), p=.000 54 House races.13 The remaining stories focusing on nonpresidential candidates included a single story about the Attorney General race, one story about the Delaware gubernatorial race, and one story featuring a state senate candidate who was only marginally related to the election. One way to put these results in perspective is to compare the total amount of news time captured with how much of it focused on these races. Out of almost 600 hours of news time captured, a total of slightly more than 13 minutes of coverage focused on the U.S. Senate race and just 12 and a half minutes focused on any U.S. House race. The total amount of air time devoted to all nonpresidential candidates was just 28 minutes and 24 seconds. In comparison, Philadelphia’s stations devoted a total of nine hours and 45 minutes of election coverage to the presidential race. This means that presidential candidates received 95% of all the candidate centered air time on Philadelphia’s stations. In addition, presidential candidates received 90% of all candidate soundbite time aired by Philadelphia’s stations. While this is still dominant, it is slightly less than the overall percentage of air time devoted to presidential candidates, suggesting a slight tendency on the part of Philadelphia’s stations to show nonpresidential candidates speaking more often than presidential candidates. Even so, the difference in soundbite time remains staggering. Presidential candidates were given a total of one hour and 47 seconds of speaking time, compared to a total of seven minutes and nine seconds of speaking time for all nonpresidential candidates. Overall, these results suggest that a typical voter would have to be fairly vigilant to see, let alone learn, anything about a nonpresidential election by watching Philadelphia’s local television news. Clearly, one of the strongest signals sent to voters by Philadelphia television stations was that the presidential election was the only thing that mattered, or to paraphrase Cohen, the only candidates to “think about” were those running for president. Table 2A contains these results. 55 Table 2A Quantity of Presidential and Nonpresidential Coverage in Philadelphia Presidential election stories (n=440) Nonpresidential election stories (n=30) Quantity of Coverage Total air time 9 hours, 45 minutes 28 minutes, 24 seconds Percentage of candidate centered air time 95% 5% Candidate Soundbites Total soundbite time 1 hour, 47 seconds 7 minutes, 9 seconds Percentage of all candidate speaking time 90% 10% Average length of a candidate soundbite 12 seconds 11 seconds Comparing Elections: Qualitative Similarities in Race Coverage It is fairly clear from the previous section that Philadelphia stations all but ignored nonpresidential races. The next section compares how the stations covered presidential and nonpresidential races. This is important in part because in our 11 market analysis we found that stories about local races were longer, contained longer soundbites, and were more likely to focus on issues than on stories about presidential candidates.14 Given that many media reform advocates15 argue that longer news stories featuring candidates talking about issues are the qualitative improvements necessary in news coverage, it is reasonable to suggest that stories about nonpresidential races were qualitatively superior to stories about presidential races. In Philadelphia the comparison is at best mixed. The average length of a presidential story in Philadelphia was 80 seconds while the average length of a nonpresidential story was just 57 seconds. While this appears to be a large difference, it was not statistically significant (p=.060). The results, however, show no significant differences between Philadelphia’s presidential and nonpresidential stories in terms of average soundbite length or percentage of stories 56 focusing on strategy or issues. These results, therefore, show that unlike other markets in the larger study, there is not a clear pattern of providing qualitatively superior coverage of nonpresidential elections in Philadelphia. Table 2B contains these results. Table 2B “Quality” of Presidential and Nonpresidential Coverage in Philadelphia “Quality” of coverage Presidential election stories (n=440) Nonpresidential election stories (n=30) Significance Average length of a story 80 seconds 57 seconds t= 1.88 (468), p=.060 Average length of a candidate soundbite 12 seconds 11 seconds t= 1.06(348), p=.287 Percentage of stories about strategy or horserace 74% 70% t= .435 (468), p=.664 Percentage of stories about issues 18% 23% t=.668 (468), p=.505 Comparing Candidates: Visibility and Treatment of the Presidential Candidates This section explores the data to see if there is an overall pattern of bias toward one party or individual candidate. We look at these bias questions from two perspectives: visibility and treatment of the candidates. These are explained below. Before turning to the results, however, some clarifications are important. First, because the presidential race dominated coverage, the focus is on comparing the major presidential and vice presidential candidates. The results generally hold for the U.S. Senate and all U.S. House races too. Appendix A contains an overview of the results for these down-ballot races. Second, the results do not include stories about third party presidential 57 candidate Ralph Nader because coverage of Nader was almost nonexistent. Appendix B contains comparisons between coverage of Nader and his major party counterparts, and they show significant differences in how Nader was covered by Philadelphia stations. Third, because the findings are remarkably consistent across individual stations, the results are presented at the market rather than the station level. Appendix C contains the results for the individual stations. Comparing Candidates: Visibility of Candidates on the News The first set of measures focus on the visibility of each candidate on the news in comparison to his counterpart(s). The rationale for these measures is similar to one used with paid campaign advertising, namely more is better. If, for example, John Kerry appears on the news much more often than George Bush, voters, at the very least, receive more exposure to Kerry than they do to Bush. These “visibility” results are presented in two ways. First, Table 3 reports the total amount of air time and soundbite time each candidate received over the 30 day study period. The results show very few differences between the major party candidates on these aggregated totals. For example, out of more than eight hours of coverage focused on the two presidential candidates, the difference in air time was less than five minutes. Table 3A Quantity of Coverage by Candidate Candidate Total number of stories featuring candidat e Total time of stories Total number of candidate soundbite s Total amount of soundbite time George Bush (R) 357 4 hrs, 13 mins 135 27 mins, 23 secs John Kerry (D) 351 4 hrs, 18 mins 111 22 mins, 12 secs Dick Cheney (R) 60 28 mins 30 4 mins, 33 secs John Edwards (D) 60 30 mins 31 6 mins, 15 secs 58 Next, the aggregated results are broken down to the story level, and we calculate the average percentage of each story that was devoted to each candidate and the average number of seconds each candidate appears in a story. As shown in Table 3B, there were no significant differences between major party candidates on either variable. Table 3B Comparing Candidate Visibility in Stories Average Percentage of a Story Devoted to Candidate N Mean S D Significance George Bush 357 .53 .239 John Kerry 351 .51 .238 t=1.35 (706), p=.177 Dick Cheney 60 .33 .235 John Edwards 60 .29 .210 t=.846 (118), p=.399 Average Number of Seconds Candidate Appeared in a Story N Mean SD Significance George Bush 357 42.6 33.11 John Kerry 351 44.1 39.61 t= .550 (706), p=.583 Average Number of Seconds Candidate Appeared in a Story N Mean SD Significance Dick Cheney 60 28.4 25.91 John Edwards 60 29.5 31.04 t= .223(118), p=.824 Comparing Candidates: Treatment of Candidates The previous tables indicate that the major party presidential candidates were given virtually the same amount of air time by Philadelphia stations. While the quantity of coverage given each candidate is one indication of bias, it does not address the possibility of other qualitative differences in candidate coverage. For example, it is possible that a candidate embroiled in a scandal might actually receive more coverage than his opponent even 59 though the quantitative advantage may actually hurt the candidate’s chances. Two different measures are used to examine the “treatment” of the presidential candidates by the news in Philadelphia. First, coders recorded how often a candidate was criticized by someone else in each story. For each story, coders recorded instances of a candidate being critiqued by a reporter, another candidate, a campaign staffer, an ordinary citizen, or some other person. By aggregating each criticism category, it is possible to discover the total number of times each candidate faced some form of criticism and the percentage of stories in which each candidate received some form of criticism. If one candidate faced significantly more criticism than another, it is possible some partisan bias exists. Second, coders were asked to play the role of a campaign manager by saying for each story whether or not they would be unsatisfied, neutral or satisfied with how their candidate was portrayed. A score of one equals unsatisfied, two equals neutral, and three equals satisfied. Averaging these scores creates a composite measure of story satisfaction where the higher the score the more satisfied a hypothetical campaign manager would be with the story. If one candidate receives a higher overall satisfaction score than his opponent, it is possible some bias exists. Although these two measures are related, they are actually designed to detect different aspects of candidate treatment. The campaign manager question looks at the story as a whole, while the criticism question looks for instances within a story where a candidate receives criticism. This means that it is possible for a story to be coded as favorable toward one candidate by a “campaign manager” even if it includes some criticism of the candidate in the story. For example, a story might focus almost entirely on the enthusiastic support George Bush received at a campaign rally, then at the very end the story it may show a single soundbite from someone representing the small group of protesters that also attended the rally. This story would be coded as favorable toward Bush, but it also would be counted as an instance of him receiving criticism. The reverse is also true; a story can be coded as unfavorable toward a candidate but not necessarily contain direct criticism. For instance, a story might simply report that John Kerry was slipping in the polls but do so without any direct criticism of 60 him or his campaign by a third party. This story would be coded as unfavorable toward Kerry but it would not be counted as an example of the story containing criticism. The results of both measures are reported in Table 4. Once again they show little evidence of direct favoritism among the major party candidates. None of the comparisons indicate a statistically significant difference between the major party candidates on the campaign manager satisfaction question. The results do suggest, however, that Vice President Dick Cheney received more criticism than his rival, John Edwards. Vice President Cheney received some form of criticism in 15% of the stories he appeared in compared to just 5% of the stories that Edwards appeared in, although the difference was not statistically significant at the .05 level (p=.069). Table 4: Comparing Candidate Treatment in Stories Percentage of stories where candidate faced criticism N Mean SD Significance George Bush 357 28% .448 t=.133 (706), p=.910 John Kerry 351 27% .446 Dick Cheney 60 15% .360 t=1.836 (118), p=.069 John Edwards 60 5% .219 Average Campaign Manager Score N Mean SD Significance George Bush 357 2.10 .555 t= .774 (706), p=.439 John Kerry 351 2.13 .581 Dick Cheney 60 2.08 .497 t=1.094 (118),p=.276 John Edwards 60 2.18 .504 Voting Issue Stories To this point the focus has been on stories about candidates, particularly presidential candidates. Election coverage on local 61 television news, however, is not limited to candidate focused stories. For example, stories report on voter registration deadlines and provide information on various ways of voting, such as by absentee ballot. Other stories focus on the process of setting up the election and the mechanics of the voting process. Still others report on efforts by celebrities to get people to the polls. Many stories also focus on allegations of voting irregularities, fraud, or election scams. In Philadelphia, a total of 99 stories (17% of all stories) were these types of “voter information” stories. This is actually a sizable amount, more than three times the number of stories about all of Philadelphia’s nonpresidential candidates combined. Even so, Philadelphia actually aired these stories somewhat less frequently than the other 10 markets in the study.16 That these stories appear to be quite prevalent is, by itself, an interesting finding. The key question, however, is how might these stories influence voters? Since there is no literature on this topic, the following section reports the results of a closer examination of the 99 voter information stories that was conducted by the author. The goal is to propose a categorization scheme for thinking about these stories in a systematic way and begin to address the effects question posed above. For each story the three coders wrote a detailed headline about the main point of the story. The author reviewed these headlines and made an initial categorization of each coder’s headlines. The initial categories were developed based on the review of the headlines. For instance, the presence of headlines containing the words “kids, children, schools and voting,” led to the creation of a “kids and voting category.” Similarly, headlines containing the words “fraud, scam, illegal, lawsuits,” led to the creation of a “fraud category.” If the initial categorization by the author “fit” for all three coders, the story was not reviewed. This occurred in 88 of the 99 stories. In the remaining 11 stories, the initial categorization by the author did not clearly fit with all of the coders or the author was unable to make a clear categorization based on the headlines. In these 11 cases, the story was reviewed by the author. This review process led to the realization that a few of the initial categorizations were too broad because it seemed reasonable that stories within a single category might have different effects on voters. For example, stories about expectations of a record voter 62 turnout could have a positive influence on voters since they focused on the importance other people are placing on the election. By contrast, other stories about voter turnout focused on how people were likely to have to wait for hours in order to cast ballots. These might have a negative influence on voters since it is possible a voter might decide that voting was not worth the time and effort. For this reason, additional categories were created and all stories in each initial category were re-examined by the author. With this process complete, all the story categories were then grouped into positive, neutral, and negative categories based on how the story might affect voters’ perceptions of the election. For instance, positive stories generally cast the election process as valid, worthy of the viewer’s attention, helpful to viewers in understanding the process of the election, or focused on the election as a “civic” function. The neutral category includes stories likely to have little impact on voters one way or another, such as those advertising a station’s upcoming election night coverage or those about Election Day weather. The negative category contains stories that generally question the validity of the election process or focus primarily on difficulties or obstacles voters are likely to face should they attempt to vote. The neutral category also includes stories that focus on the efforts by nonprofit groups or election officials to help overcome potential problems in the election. These stories contain both positive (a group working to solve potential problems) and negative (the potential problem itself). These stories were the only consistent case of story category containing both positive and negative components. In the few cases (less than five stories) where a different story topic included both positive and negative aspects, the aspect that was presented first in the story was chosen. Table 5 shows that the majority of these stories most likely provide voters with information helpful to voting and portray the election process in a positive light. Even so, 29 of the 99 stories focused on subjects that may cast some doubt on the validity of the election process and the value of participating in it. 63 Table 5: Voting Issue Stories Type of Voting Issue Story % of all voting issue stories (n=99) “Positive” stories Voter Registration 7 Celebrities to increase voter turnout 11 Absentee ballot positive 1 Voter Turnout positive 3 Kids and voting 3 Methods and Mechanics of voting 25 Total positive stories 50 “Neutral” stories “Advertisements” about a station’s election coverage 10 Election day weather forecasts 2 Citizen group involvement with election 8 Total neutral stories 20 “Negative” Stories Absentee ballots negative 4 Voter Turnout negative 3 The potential for disruption of elections by terrorism 4 Fraud, Scams, pending lawsuits 18 Total negative stories 29 Conclusion The first goal of this study was to expand existing research concerning local television and elections by focusing on multiple races across multiple stations and news broadcasts. The results show that in Philadelphia the only type of election story that received significant coverage were those about the Presidential race. It does not seem too much of an exaggeration to suggest that a typical voter would be hard pressed to learn anything about down- ballot races from watching local news in Philadelphia. 64 Furthermore, these results show that for candidates in Philadelphia’s down-ballot races, getting publicity on local television news was not a viable option. As a result, these candidates would seem to have few alternatives but to turn to paid advertising or possibly other media venues like radio and local newspapers to garner publicity for their campaigns. Reliance on paid advertising requires candidates to raise enormous amount of money, which according to campaign finance reform advocates is a fundamental problem with American democracy. It is also important to remember that one of the cornerstones of FCC licensing of local television stations is a commitment to “localism,” which at least in terms of election coverage seems defined by coverage of down-ballot races. If Philadelphia’s stations failed the “localism” test, they clearly passed another important measure of journalistic quality. This article shows that the stations were very balanced in the amount and type of coverage they gave to the major party candidates. While the results presented here focus on the presidential race, they generally held across all elections and even across individual stations. These results can be interpreted in different ways. First, perhaps a lack of overt bias is a conscious choice by news directors and reporters who work hard to provide fair and balanced coverage. Second, perhaps the lack of bias is driven by time and money constraints faced by news directors and reporters. It is certainly easier and cheaper to attend two candidate events and “roll tape” than it is to engage in hard-hitting investigative journalism of even one candidate. It is difficult (but not impossible) for the former to be bias, while it is likely that the latter would include some overtly positive or negative framing of candidates. Third, perhaps the lack of bias actually has little to do with the stations themselves and more to do with the highly disciplined and to some extent programmed nature of political candidates today. Successful candidates often provide voters and television stations with little substantive information, preferring instead to speak in bland generalities. It is clearly more difficult for reporters to critique candidate statements that do not say anything than to critique even quasi-controversial statements from candidates. Finally, it is important to note that to some extent this article equates equality with lack of bias. The abilities or behaviors (past or 65 present) of the candidates are not factored into the bias equation, but perhaps future research should do so. Clearly this work needs to be expanded to more media markets, a process currently underway by the author using data from the larger Lear Center study. Future research should also examine off- year elections to see how the lack of a presidential race changes the quantity and nature of election coverage on local television news. A Closer Look at Election Competitiveness in Philadelphia The second goal of this article was to compare two seemingly reasonable hypotheses about how Philadelphia’s television stations might balance coverage between presidential and nonpresidential elections. One hypothesis was that the presence of competitive down-ballot races would in a sense balance the coverage Philadelphia’s television stations provided to presidential and nonpresidential races. Obviously this hypothesis is not supported by the data. In fact, a closer look at the competitiveness of elections within the media market actually shows how unimportant competitiveness was in what Philadelphia stations decided to cover. According to vote totals gathered from the Secretary of State in Pennsylvania and the Division of Elections in New Jersey and Delaware,17 the eight Pennsylvania counties made up 66% of all the 2004 votes cast in the Philadelphia media market. President Bush lost these eight counties by a 21 point margin and lost the entire media market by an 18 point margin. This was driven in part by President Bush’s loss of the most populous county in the market (Philadelphia) by a 60 point margin. In fact, only four counties in the market (Bucks, Chester, Lehigh, and Northampton) were decided by less than five percentage points. So while Pennsylvania as a whole was competitive in the presidential race, it is possible that Philadelphia as a media market was not. At the same time, the opposite was true for other races in the Philadelphia market. Prior to the election, the Pennsylvania Senate race was characterized by the Cook Political Report18 as “leaning Republican,” indicating at least some degree of competitiveness. While statewide the election did not turn out to be close, it was fairly close in the counties making up the Philadelphia media market. Senator Arlen Specter beat Democrat Joe Hoeffel handily 66 statewide (52% to 43%), but in the Pennsylvania counties in the Philadelphia market, Hoeffel actually won the vote by three percentage points.19 Despite the level of within market competitiveness, a total of just 16 stories focused on the U.S. Senate race. In addition, three (PA 6th, 8th and 13th) of the 15 of U.S. House races20 held within the market were thought to be at least somewhat competitive prior to the election.21 It is true that only the 6th district contest remained competitive on Election Day with Republican Jim Gerlach winning over Democrat Lois Murphy by just three percentage points. Even so, that only 11 out of almost 600 election stories focused on a U.S. House race is a telling example of how unimportant these races were viewed by Philadelphia’s local television stations. While the low number of stories about U.S. House races makes “more” or “less” a relative concept, it seems that electoral competitiveness did not drive coverage of U.S. House races. Candidates in the competitive 6th district appeared in just two stories while candidates in the 8th appeared in six stories and 13th district candidates appeared in nine stories even though those two contests were not competitive. Obviously, additional research using more markets and in off- year elections is necessary before definitive conclusion can be made. Still, the finding that within-market electoral competitiveness was essentially unimportant in driving Philadelphia’s stations to cover down-ballot races is important. This is especially true given the repeated claims by broadcasters that lack of competitive down- ballot elections is one reason stations fail to cover them.22 A Closer Look at Market Size and Ownership This article’s second hypothesis was that presidential coverage would dominate in Philadelphia because it is a large market (making coverage of local races difficult) and because all the stations in the sample are owned by one of the four major media conglomerates (making coverage of local races less interesting to nationally centered reporters). Clearly, presidential coverage dominated coverage in Philadelphia; and while far from definitive, the results suggest that perhaps market size and ownership may have played a role in this outcome. 67 We can see the pattern when we compare the balance in Philadelphia with the balance in the five markets in the larger Lear Center study that were also in what were considered to be presidential battleground states (Dayton, Ohio; Des Moines, Iowa; and Miami, Orlando, and Tampa in Florida). Fifty-nine percent of the stories in these five markets focused on the presidential race, an almost identical percentage to the breakdown in all 10 non- Philadelphia markets in the larger Lear Center study. In contrast, 76% of the stories in Philadelphia focused on the presidential race. Obviously, Philadelphia is significantly larger in size than these five markets, perhaps making coverage of any one local race more difficult. In addition, Philadelphia is the only one of these markets where all four stations are owned and operated by one of the big four media companies (ABC, CBS, FOX, and NBC). As a result, perhaps Philadelphia’s reporters and anchors feel more pressure to appear that they are covering elections as network or national journalists might. This contention is supported by the fact the two other “large” markets in the Lear Center study (New York and Los Angeles) also devoted more than 70% of their stories to the presidential election.23 If larger markets are more likely to focus on top of the ticket races, this tendency might have been reinforced in Philadelphia by the demographic characteristics of four nearby counties where the presidential race was close (Bucks, Chester, Lehigh, and Northampton). All four counties have higher median family incomes and levels of education than more populous Philadelphia.24 The demographic makeup of these counties is likely more attractive to advertisers and hence television stations than is the demographic makeup of Philadelphia. But because education and affluence are positively related to watching local television news,25 it is also possible that more local news viewers may live in the four smaller counties where the presidential race was highly competitive than in the rest of the market where the race was not competitive. In either case, it possible that television stations target their election coverage not to their entire media market but to selected segments of it. Although more sophisticated analysis beyond a single market is needed, the results from Philadelphia reveal a rich avenue for future research. 68 What’s Local about Local News? Perhaps the only answer to the question posed in the title of this article is that while Philadelphia’s local television stations ignore local candidates, they do provide a sizable number of local voter information stories. More stories aired in Philadelphia about the mechanics and procedures of voting (25) than about either the U.S. Senate race (16) or all U.S. House races combined (11). In addition, half of the voting issues stories were framed in ways that might signal voters that participating in elections has some intrinsic value. This suggests that Philadelphia’s local television stations have not completely abandoned the media’s role as facilitator or at least advertiser of civic life, even if they have discarded any interest in critiquing or assessing local candidates. The Philadelphia results, which were even more pronounced in the other markets in the larger Lear Center study, suggest that this may be a new and important area for future research. Examining these noncandidate stories across a larger number of markets is a necessary first step in determining whether other markets cover voter information stories in ways similar to Philadelphia. If the pattern recurs, the stories should be examined in more detail to discover how they may influence citizens’ beliefs about the validity of the electoral process. Appendix A: Overview of Candidate Visibility in U.S. Senate and U.S House Races The lack of stories about U.S. Senate and U.S. House candidates makes statistical comparisons problematic. As a result, Appendix A simply reports the aggregate totals for each race. They clearly show that the amount of time devoted to candidates running for the same office was virtually identical in all races. 69 Candidate Total stories candidate appeared in Total time stories focused on Average percentage of story that focused on candidate Total amount of soundbite time Average length of candidate soundbite U.S. Senate Arlen Specter (R) 24 7 mins, 36 secs 39% 93 secs 12 secs Joseph Hoeffel (D) 22 6 mins, 38 secs 35% 94 secs 12 secs James Clymer (I) 7 1 min 33% 20 secs 20 secs Betsy Summers (I) 5 51 secs 17% 17 secs 17 secs U.S. House (6th) Jim Gerlach (R) 2 1 min, 15 secs 63% 22 secs 7 secs Lois Murphy (D) 2 1 min, 5 secs 38% 25 secs 8 secs U.S. House (8th) Mike Fitzpatrick (R) 6 1 min, 48 secs 34% 14 secs 14 secs Virginia Schrader (D) 5 2 mins, 55 secs 41% 38 secs 10 secs U.S. House (13th) Allyson Schwartz (D) 9 1 min, 56 secs 26% 24 secs 8 secs John McDermott (Con) 4 30 secs 24% 0 0 Chuck Moulton (Lib) 4 30 secs 24% 0 0 U.S. House (15th) Charles Dent (R) 1 1 min, 3 secs 50% 16 secs 8 secs Joe Driscoll (D) 1 1 min, 3 secs 50% 16 secs 8 secs Appendix B Ralph Nader Comparisons Ralph Nader did not appear on the Pennsylvania ballot in 2004. So it is not surprising that he received less coverage than the two major party candidates. The difference, however, is quite dramatic. Nader appeared in 34 stories compared with over 350 for both Bush and Kerry. Nader received a total of 17 minutes of air time 70 compared with over four hours for both Bush and Kerry. Nader spoke in three soundbites for a total of 24 secs. In comparison, both Bush and Kerry were shown speaking over 100 times for more than 20 minutes each. In addition to these vast differences in the quantity of air time, Philadelphia’s stations covered Nader in qualitatively different ways. Most of Nader’s stories focused on his difficulties getting on the ballot. In many respects this explains the results. Stories about Nader’s attempt to get on the ballot were much more likely to focus exclusively on Nader than were “regular” stories about Bush and Kerry. This helps explain why a higher percentage of the overall story time was on average higher for Nader (.61) than it was for either Bush (.53) or Kerry (.51). In addition, these stories were more likely to be shorter, consist of straight news reporting, and because Nader failed to get on the Pennsylvania ballot to be unfavorable toward Nader. This helps explain why Nader received less air time per story and less criticism but more unfavorable coverage than the other candidates. Average Percentage of a Story Devoted to Candidate N Mean SD Significance Ralph Nader 34 .61 .398 George Bush 357 .53 .239 t=1.714(389), p=.087 Ralph Nader 34 .61 .398 John Kerry 351 .51 .239 t=2.245(383),p=.025 Average Number of Secs Each Candidate Appeared in a Story N Mean SD Significance Ralph Nader 34 28.1 29.92 George Bush 357 42.6 33.11 t= -2.470(389),p=.014 Ralph Nader 34 28.1 29.92 John Kerry 351 44.1 39.61 t= -2.302(383),p=.022 71 Percentage of Stories Where Candidate Faced Criticism N Mean SD Significance Ralph Nader 34 .09 .287 George Bush 357 .28 .448 t= -2.411 (389), p=.016 Ralph Nader 34 .09 .287 John Kerry 351 .27 .446 t= -2.371 (383), p=.018 Average Campaign Manager Score N Mean SD Significance Ralph Nader 34 1.50 .615 George Bush 357 2.10 .555 t=-5.948(389),p=.000 Ralph Nader 34 1.50 .615 John Kerry 351 2.13 .581 t=-6.020(383), p=.000 Appendix C Individual Station Comparisons The limited difference in how stations covered the presidential race is by itself an interesting finding. The data below show the percentage of presidential candidate appearances that each station devoted to each candidate. Station Bush % (n) Kerry % (n) Nader % (n) Cheney % (n) Edwards % (n) ABC (n=333) 40.2 (134) 40.2 (134) 6.0 (20) 6.9 (23) 6.6 (22) CBS (n=206) 41.7 (86) 40.3 (83) 3.4 (7) 7.8 (16) 6.8 (14) Fox (n=27) 40.7 (11) 37.0 (10) 0.0 (0) 11.1 (3) 11.1 (3) NBC (n=296) 42.6 (126) 41.9 (124) 2.4 (7) 6.1 (18) 7.1 (21) 72 On the more qualitative variables, the results again show almost no differences in how each station treated each candidate. On the ABC station in Philadelphia none of the comparisons proved significant. On the CBS station, Dick Cheney had a significantly higher percentage of story time devoted to him than John Edwards did (t=2.726 (28), p=.011). None of the other CBS comparisons were significant. Because of the small number of stories aired by the Fox station, only presidential comparisons are possible. The only comparison that proved even marginally significant statistically was that George Bush had a lower campaign manager score than did John Kerry (t= -1.832(19), p=.083) on the Fox station. Similarly, the only even marginally significant difference on the NBC station was that John Edwards received a higher campaign manager score than did Dick Cheney (t= -1.874 (37), p=.069). Because of the small number of stories featuring Ralph Nader, comparisons are made only between the major party candidates. ABC Stories Average Percentage of a Story Devoted to Candidate N Mean SD Significance George Bush 134 .51 .228 John Kerry 134 .47 .222 t=1.592 (266), p=.113 Dick Cheney 23 .29 .207 John Edwards 22 .31 .220 t=-.388 (43), p=.700 Average Number of Secs Each Candidate Appeared in a Story N Mean SD Significance George Bush 134 42.1 34.15 John Kerry 134 42.3 44.52 t= -.053 (266), p=.958 Dick Cheney 23 32.2 31.75 John Edwards 22 34.1 39.44 t= -.185 (43), p=.855 Percentage of Stories Where Candidate Faced Criticism N Mean SD Significance George Bush 134 .32 .468 John Kerry 134 .34 .474 t= -.259 (266), p=.796 Dick Cheney 23 .26 .448 John Edwards 22 .09 .294 t=1.495 (43), p=.142 73 Average Campaign Manager Score N Mean SD Significance George Bush 134 2.08 .476 John Kerry 134 2.09 .527 t=-.122 (266), p=.903 Dick Cheney 23 2.00 .522 John Edwards 22 2.14 .468 t= -.921(43), p=.362 CBS Stories Average Percentage of a Story Devoted to Candidate N Mean SD Significance George Bush 86 .53 .263 John Kerry 83 .55 .241 t= -.346 (167), p=.729 Dick Cheney 16 .48 .304 John Edwards 14 .24 .180 t=2.726 (28), p=.011 Average Number of Secs Each Candidate Appeared in a Story N Mean SD Significance George Bush 86 46.6 36.35 John Kerry 83 51.9 42.11 t= -.868(167), p=.387 Dick Cheney 16 27.3 19.15 John Edwards 14 24.7 33.93 t=.261 (28), p=.796 Percentage of Stories Where Candidate Faced Criticism N Mean SD Significance George Bush 86 .28 .451 John Kerry 83 .27 .444 t=.203 (167), p=.839 Dick Cheney 16 .06 .250 John Edwards 14 .07 .267 t= -.095 (28), p=.925 Average Campaign Manager Score N Mean SD Significance George Bush 86 2.17 .723 John Kerry 83 2.20 .712 t= -.275 (167), p=.783 Dick Cheney 16 2.38 .500 John Edwards 14 2.29 .726 t=.396 (28), p=.796 FOX Stories26 Average Percentage of a Story Devoted to Candidate N Mean SD Significance George Bush 11 .52 .260 John Kerry 10 .49 .251 t=.293 (19), p=.773 74 Average Number of Secs Each Candidate Appeared in a Story N Mean SD Significance George Bush 11 57.6 40.44 John Kerry 10 59.3 36.13 t= -.099 (19), p=.922 Percentage of Stories Where Candidate Faced Criticism N Mean SD Significance George Bush 11 .09 .301 John Kerry 10 .10 .316 t= -.067 (19), p=.947 Average Campaign Manager Score N Mean SD Significance George Bush 86 1.91 .302 John Kerry 83 2.20 .422 t= -1.832 (19), p=.083 NBC stories Average Percentage of a Story Devoted to Candidate N Mean SD Significance George Bush 126 .55 .234 John Kerry 124 .52 .249 t=.896 (248), p=.371 Dick Cheney 18 .26 .133 John Edwards 21 .33 .220 t= -1.290 (37), p=.205 Average Number of Secs Each Candidate Appeared in a Story N Mean SD Significance George Bush 126 39.19 28.45 John Kerry 124 39.70 31.00 t= -.135 (248), p=.893 Dick Cheney 18 26.1 25.45 John Edwards 21 28.5 19.92 t= -.335 (37), p=.739 Percentage of Stories Where Candidate Faced Criticism N Mean SD Significance George Bush 126 .25 .432 John Kerry 124 .23 .419 t=.375 (248), p=.708 Dick Cheney 18 .11 .323 John Edwards 21 .00 .000 t=1.578 (37), p=.123 Average Campaign Manager Score N Mean SD Significance George Bush 126 2.08 .515 John Kerry 124 2.12 .550 t= -.617 (248), p=.538 Dick Cheney 18 1.94 .416 John Edwards 21 2.19 .402 t= -1.874 (37), p=.069 75 Notes 1. Pew Center for the People and the Press Survey Reports, “News Audiences Increasingly Politicized: Online News Audience Larger, More Diverse” (June 8, 2004); “Public More Critical of Press, But Goodwill Persists” (June 26, 2005); “Voters Impressed with Campaign: But News Coverage Gets Lukewarm Ratings” (October 24, 2004). See http://people press.org. 2. The 11 markets studied were New York, Los Angeles, Philadelphia, Dallas, Miami, Tampa, Orlando, Denver, Seattle, Des Moines, and Dayton (Ohio). The results of the full study are available at www.localnewsarchive.org. 3. The New York Media market also covers three states: New York, New Jersey, and Connecticut. 4. The number of elections was calculated based on data from the Secretary of State in Pennsylvania (http://www.dos.state.pa.us/dos/site/default.asp) and the Division of Elections in New Jersey http://www.state.nj.us/lps/elections/ electionshome.html and Delaware http://www.state.de.us/election/archive/ elect04/2004_electionindex.shtml. 5. http://www.cookpolitical.com October 29, 2004. 6. Pennsylvania voters had races for Attorney General, Auditor, and State Treasurer. Delaware had races for Governor, Lieutenant Governor, and Insurance Commissioner. The Attorney General, Auditor, and State Treasurer races in Pennsylvania and the Insurance Commissioner in Delaware were all open seats. 7. Philadelphia DMA voters participated in 16 federal legislative elections, 10 in Pennsylvania, five in New Jersey, and one in Delaware. 8. According to the Cook Political Report for October 26, 2004, the 6th and 8th Pennsylvania Congressional district were also classified as “leaning” Republican. The 13th district in Pennsylvania was classified as leaning Democratic. 9. Cook Political Report, October 29, 2004. http://www.cookpolitical.com 10. Much of this pattern was driven by the NBC and ABC affiliates in Philadelphia, which were both in the top 15 of the 44 station sample in terms of total number of stories aired. In comparison, the FOX affiliate aired a total of just 21 election stories, the lowest number among the 44 stations studied. The CBS affiliate was average in terms of the number of stories aired, in part because it airs an evening news program at 4:00 p.m. outside the time period captured. 11. t= 7.57(4029), p=.000 12. An additional nine stories included a U.S. Senate candidate with other candidates, so at best the four stations aired a total of 25 stories about the U.S. senate race. Of these 25 stories, 19 (or 76%) focused on strategy or horserace and four stories (16%) focused on issues. Sixteen of the 25 stories (60%) aired during the final week and eight (32%) aired on the day before the election. 13. An additional seven stories mentioned U.S. House candidates either with candidates for other offices or while discussing noncampaign-related activities. This means at best a total of 18 stories aired featuring a U.S. House candidate. Of these 18 stories, 14 focused on strategy/horserace, one focused on issues, and three were coded as “other.” Fifty percent of these stories aired in the final week of the campaign. 14. In the larger report the comparisons described above did not include U.S. senate races or other statewide races. Because the number of nonpresidential 76 election stories in Philadelphia is so small, the U.S. Senate and statewide races are included together. 15. See http://freepress.net/conference/. Also see the recommendations of the so-called Gore Commission at http://www.ntia.doc.gov/pubintadvcom/ pubint.htm. 16. Voting issue stories made up 22% of the stories in the other markets. The difference between Philadelphia and the other 10 markets in the quantity of voting issues stories was statistically significant (t= 2.547 (7019), p=.011. 17. For Pennsylvania see http://www.electionreturns.state.pa.us/ElectionReturns. For New Jersey see http:// www.state.nj.us/lps/elections/electionshome.html. For Delaware see http://www. state.de.us/election/archive/elect04/2004_election_index.shtml. 18. http://www.cookpolitical.com October 29, 2004. 19. See http://www.electionreturns.state.pa.us/ElectionReturns. 20. Philadelphia DMA voters participated in 15 U.S. House elections, nine in Pennsylvania, five in New Jersey and one in Delaware. 21. According to the October 26, 2004, Cook Political Report, the 6th and 8th Pennsylvania Congressional district were classified as “leaning” Republican. The 13th district in Pennsylvania was classified as leaning Democratic. 22. See the National Association of Broadcaster’s “Free Air Time” newsletter available at http://www.nab.org. 23. As in Philadelphia, all the stations in New York and Los Angeles are owned and operated by one of the four major media companies. In the New York market 76% of the stories were about the presidential race, and in Los Angeles 74% of the stories were about the presidential race. 24. According to the 2000 U.S. Census, the median family income in these counties was: Bucks $68,727; Chester $76,916; Lehigh $53,147; Northampton $53,955; and Philadelphia $37,036. The percentage of people with at least a B.A. degree was: Bucks 31.2%; Chester 42.5%; Lehigh 23.3%; Northampton 21.2%, and Philadelphia 17.9%. 25. See http://www.stateofthenewsmedia.org/2006/. 26. The FOX station only aired three stories on the vice presidential candidates making statistical comparisons meaningless. The low number of stories about presidential candidates requires that even these results be viewed with caution. References Althaus, S.L.; Nardulli, P.F.; and Shaw, D.R. 2002. “Candidate Appearances in Presidential Elections, 1972–2000.” Political Communication 19: 49–72. Ansolabehere, S., and Iyengar, S. 1995. Going Negative: How Political Ads Shrink and Polarize the Electorate. New York: Free Press. Bartels, L.M. 1988. Presidential Primaries and the Dynamics of Public Choice. Princeton: Princeton University Press. Beck, P.A.; Dalton, R.J.; Greene, S.; and Huckfeldt, R. 2002. “The Social Calculus of Voting: Interpersonal, Media, and Organizational Influences on Presidential Choices.” American Political Science Review 96: 57–73. 77 Carter, S.; Fico, F.; and McCabe, J.A. 2002. “Partisan and Structural Balance in Local Television News Coverage.” Journalism and Mass Communication Quarterly 79 (1): 41–53. Cronbach, L. J. 1951. “Coefficient Alpha and the Internal Structure of Tests.” Psychometrika 16: 297–334. Entman, R.M. 1992. “How the Media Affect what People Think: An Information Processing Approach.” Journal of Politics 51: 347–370. Finkel, S. E., and Geer, J.G. 1998. “A Spot Check: Casting Doubt on the Demobilizing Effect of Attack Advertising.” American Journal of Political Science 42: 573–595. Gilliam, F.D., and Iyengar, S. 2000. “Prime Suspects: The Influence of Local Television News on the Viewing Public.” American Journal of Political Science 44: 560–573. Goldstein, K., and Freedman, P. 2002. Campaign Advertising and Voter Turnout: New Evidence for a Stimulation Effect. Journal of Politics 64: 721–740. Graber, D.A. 1997. Mass Media and American Politics. 5th ed. Washington, D.C.: CQ Press. ________. 2001. Processing Politics: Learning from Television in the Internet Age. Chicago: University of Chicago Press. Iyengar, S., and Kinder, D. 1987. News That Matters. Chicago: University of Chicago Press. Just, M.; Crigler, A.; and Buhr, T. 1999. “Voice, Substance, and Cynicism in Presidential Campaign Media.” Political Communication 16 (1): 25–44 ________; Crigler, A.; Alger, D; Cook, T.; Kern, M; and West, D. 1996. Crosstalk: Citizens, Candidates, and the Media in a Presidential Campaign. Chicago: University of Chicago Press. Kaplan, M.; Goldstein, K.; and Hale, M. (2004) “Local News Coverage of the 2004 Campaigns: An Analysis of Nightly Broadcasts in 11 Markets.” University of Southern California Norman Lear Center. www.localnewsarchive.org. Krippendorff, K. 1978. “Reliability of Binary Attribute Data.” Biometrics 34: 142– 144. Nunnally, J. C. 1978. Psychometric Theory. 2nd ed. New York: McGraw Hill. Patterson, T.E.; and McClure, R. 1976. The Unseeing Eye: The Myth of Television Power in National Politics. New York: Putnam. Shaw, D. 1999. “The Effect of TV Ads and Candidate Appearances on Statewide Presidential Votes, 1988–1996.” American Political Science Review 93: 345– 361. Snider, J. 2000. “Local Television News Archives as a Public Good.” Harvard International Journal of Press and Politics. 5 (2): 111-118. Tinsley, H. E. A., and Weiss, D. J. 2000. “Interrater Reliability and Agreement.” In Handbook of Applied Multivariate Statistics and Mathematical Modeling, eds. H. E. A. Tinsley and S. D. Brown. San Diego, CA: Academic Press. Valentino, N.A. 1999. “Crime News and the Priming of Racial Attitudes During Evaluations of the President.” Public Opinion Quarterly 60 (4): 515–541. Wattenberg, M.P., and Brians, C.L. 1999. “Negative Campaign Advertising: Demobilizer or Mobilizer?” American Political Science Review 93: 891–899.