The Rocky Mountain News and the Seattle Post-Intelligencer

The closures of the Seattle Post-Intelligencer (PI) and the Rocky Mountain News (Rocky) were prompted by a simple factor: insufficient revenue. ¹ From 2006 to 2009, annual nationwide newspaper advertising revenue dropped from about $49 billion to $27 billion (Project for Excellence in Journalism [PEJ], 2012). The migration of audiences to new media gave advertisers many new placement options; meanwhile, newspapers failed to monetize their developing digital audiences as their print circulation revenue eroded. During 2008, the PI lost about $14 million, and the Rocky lost about $16 million (DeBruin & Ryckman, 2009; Richman & James, 2009). Both newspapers were alternately competing and collaborating—via a joint operating agreement (JOA)—with a second daily newspaper in their home market: the Denver Post and the Seattle Times. Though the PI and Rocky had both been community fixtures for 150 years, their large, national corporate owners—E. W. Scripps (Rocky) and Hearst (PI)—no longer saw a future in publishing the second paper in a two-paper market.

The sudden collapse of advertising revenue immediately precipitated the newspapers’ closures, but a long-term decline in circulation was the underlying culprit. By the end of 2009, daily newspaper circulation was down about 31% from its peak levels nationwide (PEJ, 2010). In Denver, circulation decreases at the two newspapers were similar to nationwide trends: down steadily during the early 2000s and then sharply in 2007–2009 (PEJ, 2010). As Table 1 shows, the circulation of the Rocky Mountain News declined nearly in lockstep with the Post’s in the years preceding its closure. By the spring of 2009, both papers had about 203,000 subscribers—but the Post had more active local ownership, favorable union contracts, and lower overhead (Diddlebock, 2009). Ultimately, the Post was the stronger of the two newspapers, and its owner was more determined to stay the course.

In Seattle, the figures depart some from the national narrative. Though overall circulation was in decline, the decrease was not as pronounced as it was in Denver or nationally. The key event in the two papers’ recent relationship was the Times’ switch in 2000 to morning publication (Bishop & Richman, 2003). This put the two papers in direct competition for the first time—and started a migration of PI subscribers to the Times. This competitive shift also soured the JOA that the two papers had, and the Times moved to end the relationship in 2003 (Bishop & Richman, 2002; Pryne, 2007). For the next 5 years, a group of citizens known as the Committee for a Two Newspaper Town fought the dissolution of the JOA in court in the hopes of keeping the PI alive. But, in the end, the JOA issue was moot because Hearst chose to close the PI in the face of mounting losses.

Theorizing Local Newspapers’ Importance

Several theoretical strands from the past 200 years suggest different ways that newspapers contribute to their communities. In Democracy in America, de Tocqueville argues that newspapers were catalysts for the formation of associations that were vital to America's early participatory democracy. Similarly, Bryce (1995) and Tarde (1903) propose that newspapers set common agendas for their readers that, in a way, define the readers’ communities. Janowitz (1968) describes the reciprocal relationship between community newspaper readership and community attachment, suggesting that, one way or another, newspaper readership is an important part of being a community member. And, as urban populations suburbanized in the latter half of the 20th century, Kaniss (1997) asserted that large daily newspapers knit patchwork municipalities together into cohesive metropolitan areas by actively cultivating one overarching civic identity. But what underpins these related, yet divergent, functions of the local news media?

Habermas, in Theory of Communicative Action (1985), posits a two-level concept of society that offers a starting point for understanding the relationship between local news media and their communities. He argues that society is structured by both lifeworld—private and public communicative interactions that characterize the daily lives of individuals—and system—Weberian economic and state bureaucracies that increasingly delimit these lives (Baxter, 1987). In early societies, these two levels were closely related but, in the 21st century, they are “uncoupling” and the influence of lifeworld upon system is less direct. As the distance between lifeworld and system grows, “media replace linguistic ‘communication’ in certain functional contexts, thereby replacing the ‘mechanism’ of linguistic understanding and consensus” (Baxter, 1987, p. 56). Or, in other words, media become an important link between lifeworld and system.

Communities come into the equation because they are “the complex social structure[s] in which system and lifeworld meet” (Friedland & McLeod, 1999, p. 198). Though communities may not all be defined by geography and proximity today, local communities are still central to our governance and society (Macedo & Karpowitz, 2006). But what delineates these communities? Gerrymandered political borders? Mutable incorporation lines? Natural boundaries like rivers or ridges? To these obvious factors, both Tarde (cited in Katz, 1999) and Kaniss (1997) would add local newspapers. They argue that community springs forth from a common agenda that is cultivated by readership of the same newspaper. Friedland and McLeod (1999) move beyond readership and argue that “local media are themselves central institutions connected to other community institutions and operate to link the networks and bonds, in general the set of relations, that tie differing associations, subcommunities, and cultures together” (p. 208).

Whether through information dissemination or the work of producing the newspaper, the relations and ties that newspapers contribute to are the basis for public action—which, in modern “uncoupling” societies, is the mechanism through which citizens may influence their system. So, if local media institutions are strong and are binding individuals and groups together, then citizens should be participating in more community groups, contacting their government more frequently, and circulating more petitions because they are more aware of shared problems, interests, and opportunities. ² Readership of the same metropolitan newspaper—or other interactions with it—may spur many forms of engagement across districts and boundaries as groups of citizens respond to the same local (or regional) news in different manners. In short, civic engagement and other indicators of an active citizenry should be higher when local news media institutions are numerous and vibrant and members of the public are consuming the content that they produce.

Empirical Evidence of Local Newspapers’ Importance

Researchers who try to isolate the discrete contributions that newspapers make to their communities typically situate their work somewhere within this broad theoretical foundation. In the process, they have depicted correlations between newspaper readership and civic engagement (Kang & Kwak, 2003; McLeod, Scheufele, & Moy, 1999; Moy, McCluskey, McCoy, & Spratt, 2004; Scheufele, Shanahan, & Kim, 2002), community attachment (Jeffres, Atkin, & Neuendorf, 2002; McLeod et al., 1996; Stamm, 1988; Stamm, Emig, & Hesse, 1997; Stamm & Weis, 1986), and relevant political outcomes such as voters’ local electoral choices (Becker & Dunwoody, 1982; McLeod et al., 1999). But most prior research in this vein is limited by a reliance upon cross-sectional, single-city data sets that preclude causal tests and limit generalizability (Hoffman & Eveland, 2010; Kim & Ball-Rokeach, 2006). Inconsistent, unreliable, and often self-reported media exposure variables, which plague media research in general (Prior, 2009), are an additional obstacle that undermines researchers working to show a causal relationship between newspaper readership and community-related outcomes.

Hoffman and Eveland (2010) address some of these shortfalls by using a national panel study to assess the importance of local news media use. Their findings replicate the well-established correlational relationship between local newspaper readership and community attachment, but they do not locate evidence of a significant relationship over time. This prevents any conclusions regarding the causal ordering of the relationship between local news media use and community attachment. In explaining their null findings, Hoffman and Eveland assert that levels of both local news media use and community attachment are entrenched and unlikely to vary much over the course of a year (the time period between the waves of their survey) unless a dramatic event disrupts the status quo. As they write, “Only panel studies that track individuals over time, beginning when they enter communities or when communities are significantly disrupted by outside events, are likely to be capable of sorting out the causal ambiguities inherent in this area of great theoretical interest” (Hoffman & Eveland, 2010, p. 193).

It is a pair of such disruptions—the deaths of the PI and Rocky—that prompts the current article. And there is a small tradition of similar research that informs this work. Berelson (1948) conducted an interview-based examination of the effects of a newspaper delivery strike in New York City that took place in 1945 and lasted for 17 days. In this pre-television era, most of the respondents said they missed the hard news that their newspaper provided—but in Berelson's estimation, only about a third were able to engage in enough discussion of the news to indicate that they ever actually read it. More recently, a pair of studies replicated Berelson's work in small towns where the local newspaper was not delivered on a given day (Bentley, 2001) or was permanently shuttered (Smethers, Bressers, Willard, Harvey, & Freeland, 2007). Like Berelson, both projects found that many of the interviewees primarily missed their newspaper for reasons related to habit or entertainment—but not because of a lack of access to hard news.

If individual-level media exposure variables derived from surveys are unreliable, perhaps a better approach to assessing the importance of newspapers is to use environmental differences to, in effect, establish a natural experiment at the community level. For example, Mondak (1995) compared residents of Pittsburgh during an 8-month-long newspaper strike in 1992 and 1993 with residents of Cleveland who did not experience a disruption in their newspapers. He found that there was evidence that the lack of newspapers had a deleterious effect upon Pittsburghers’ local political knowledge and talk, but not their national or international political knowledge and talk. So, even without individual-level media exposure variables, it may be possible to measure the effects of access to different media within and across communities.

A recent study of Cincinnati and its suburbs (Schulhofer-Wohl & Garrido, 2011) that evaluated the effects of the 2007 closure of the Cincinnati Post also used this kind of community-level approach. The authors suggested that a number of negative political outcomes could be observed from 2007–2010 in the northern Kentucky communities that the Post traditionally served. These same outcomes, however, were not visible in the communities served by the Cincinnati Enquirer, which did not close. In the Kentucky communities, “Fewer people voted in elections for city council, city commission, and school board; fewer candidates sought those seats; the remaining candidates spent less money on their campaigns; and, for councils and commissions, incumbents’ chances of retaining office improved” (Schulhofer-Wohl & Garrido, 2011, p. 2). Even though the Enquirer increased coverage of northern Kentucky to compensate for the Post’s closure, the communities suffered significant, negative local political effects.

Three conclusions emerge from the body of previous empirical scholarship concerning the importance of local newspapers. First, there is an array of evidence that depicts a correlation between local newspaper readership and a variety of important civic, political, and community outcomes. These correlations offer general support for the theoretical arguments that claim newspapers are important to citizens and their communities. Second, because of data and methodological limitations, the causality and generalizability of these relationships is unclear: Does newspaper readership spur civic engagement, political action, or community attachment—or vice versa? Third, prior research suggests that it may be possible to gain some leverage on both the causality and generalizability of newspapers’ importance by strategically using community-level data.

Hypotheses

In response to the methodological limitations of much of the prior research regarding the importance of local news media—and following the innovative structure of several other projects—this analysis focuses on a community-level comparison of the change in civic engagement in two dozen cities across the United States over a 12-month period. Given prior findings (Hoffman & Eveland, 2010; Kang & Kwak, 2003; Moy et al., 2004), the first hypothesis that this article tests is as follows.

If this is true, then there is reason to investigate the change in year-over-year civic engagement in various American metropolitan areas, with a particular focus on Denver and Seattle. Though cross-sectional data cannot alone establish the causal direction of the relationship between newspaper readership and civic engagement, the tradition of correlational findings suggests that the loss of a newspaper will negatively impact civic engagement.

Meanwhile, in cities that did not experience a significant disruption in their local media environments, at first blush there is no reason to expect a significant year-over-year change in civic engagement.

At the national level, the demise of the Post-Intelligencer and Rocky Mountain News is relatively insignificant. But, because of the timing of the CPS—the 2008 iteration immediately followed the surge of political activity tied to the 2008 presidential campaign and the 2009 iteration was not preceded by a similar catalyst—it is conceivable that civic engagement would be broadly depressed in any year-over-year comparison. Accordingly, Hypothesis 4 is as follows.

Finally, even if there is evidence of a national decrease in civic engagement in 2009, an off-election year during an economic downturn, the common factors that contribute to this general decline will be present in Denver and Seattle—and compounded by the loss of newspapers in these cities.

Method

The primary data set in this article merges measures drawn from the 2008 and 2009 November supplements to the CPS. Starting in 2008, the U.S. Census Bureau added a battery on civic engagement to the CPS that initially included 25 items and was asked of nearly 68,000 respondents nationwide (U.S. Census Bureau, 2008). The next year, the battery was repeated, but 12 questions were dropped (including a set of news media usage questions) and the number of respondents was reduced to about 21,000 (U.S. Census Bureau, 2009). In addition to the civic engagement battery, the CPS data sets both provide access to basic demographic information describing the respondents and a group of items focused on social capital. ³

The specific respondents to the CPS vary from year to year, but because the survey population is derived from a probability sample designed to be representative of all households, comparisons of aggregate year-over-year data should be valid. Conducting demographic comparisons of the 2008 and 2009 samples does reveal some significant differences in the metropolitan area populations year over year. The differences that exist are generally muted and may feasibly be attributed to the natural evolution of communities or the nation's macroeconomic environment rather than systematic sampling bias. Table 2. contains a comparison of the 2008 and 2009 sample demographics in Denver and Seattle. There are no significant differences in the sample composition year over year in Seattle, but in Denver there is a significant change in the racial composition of the sample.

Five core civic engagement items are available in both the 2008 and 2009 CPS data sets. These five questions are analyzed individually, recoded into yes (1) and no (0) dummy variables, and also as a cumulative index of civic engagement that averages all five together (0–1, M = .12, SD = .20). Together, they are the key dependent variables in the analyses below. They are as follows:

Additionally, of the 13 substantive items that are repeated from year to year in the CPS civic engagement supplement, several target indicators of social capital that have, at most, an indirect conceptual connection to newspaper readership. Three of these 5-point ordinal items tracked the frequency of social activities—having a household dinner, talking with neighbors, and doing favors for neighbors—from 1 (not at all) to 5 (basically every day). These questions are averaged into a social capital index (1–5, M = 3.22, SD = .94), used below as a point of comparison for the civic engagement analyses.

The initial regression analyses below establish the basic relationship between newspaper readership and civic engagement at the individual level. For these, the 2008 CPS data set is used alone because the news media use questions were eliminated from the 2009 survey. ⁴ The newspaper readership question on the 2008 survey is:

Next, the 2008 and 2009 data sets are merged to allow for community-level analyses that seek to isolate the effect of the newspaper closures in Denver and Seattle. ⁶ Since the 2009 data do not offer news media use variables, the critical independent variable in the year-over-year analyses is implied: the change (or lack thereof) in the media environment within each community. Simple t tests are initially used to assess the significance of the observed changes.

In the combined data set, the populations of various metropolitan areas are identified according to the Census CBSA FIPS code. The Census Bureau's definitions of metropolitan areas are often relatively broad, encompassing multiple contiguous cities and sometimes even spanning across state lines in the process (U.S. Census Bureau, 2008). As scholars have argued (Kaniss, 1997), newspapers also transcend municipal borders, so defining the community of study according to these metropolitan areas is intuitive. In addition, this broad definition of communities yields sample sizes that allow for localized year-over-year comparisons to be made with reasonable statistical power. Accordingly, eight comparison cities were chosen with both a purposive and practical reason in mind. First, in terms of culture and demographics, some cities are more comparable in nature (Seattle and Portland, for example) than others (Seattle and Houston). Second, the CPS data, though abundant, limit local-level comparisons to only the largest metropolitan areas because of the reduced sample in 2009.

The final analyses are conducted on a novel data set built from several sources that allows for a more systematic analysis of the importance of newspaper closures. The primary dependent variable in these models comes from the merged 2008/2009 CPS data set and is the year-over-year change in civic engagement for each of the top 20 U.S. metropolitan areas—including Denver (number 12) and Seattle (number 20)—by population according to the 2008 CPS. For each city, the index of civic engagement (described above) in 2009 is subtracted from the index of civic engagement in 2008. The same process is completed for the social capital index outlined above, and the difference in social capital is used as an alternate dependent variable.

The year that passed between November 2008 and November 2009 was eventful, and three key control variables are included in this data set to address alternate causal explanations for observed changes in civic engagement. A measure of total 2008 political campaign spending on TV advertising for all campaigns combined, by media market, is drawn from data provided by the Wisconsin Advertising Project (Goldstein, Niebler, Neiheisel, & Holleque, 2011). Newspaper circulation reports from the Audit Bureau of Circulations are used to create a variable that tracks the percentage change in the total circulation of all major dailies present in each of the top 20 cities from the final quarter of 2008 to the final quarter of 2009. Finally, the percentage change in unemployment levels from November 2008 to November 2009 by metropolitan area is gleaned from reports made available by the U.S. Bureau of Labor Statistics.

Results

Table 3. contains the results of three OLS regressions that predict cumulative civic engagement in three populations: the United States, Seattle, and Denver. At the national level, the sample is over 50,000 and every independent variable has a significant relationship with civic engagement, including newspaper readership (β = .014, p ≤ .001). In both Denver and Seattle, the positive relationship between newspaper readership and civic engagement holds even with a sample of just over 500 (β = .015, p ≤ .05). Thus, the data confirm H1: There is a positive relationship between newspaper readership and civic engagement.

Table 4. contains the results of year-over-year comparisons of the five civic engagement indicators described earlier. Data for 10 cities, including Denver and Seattle, as well as the nation as a whole are reported. Each cell contains the difference from 2008 to 2009, in terms of percentage, in the proportion of the survey sample within each metropolitan area that responded in the affirmative to the particular engagement question at hand. Negative numbers indicate that the percentage of respondents who reported taking a specific action declined year over year; positive figures indicate the opposite. The significance of each difference was assessed with simple two-tailed t tests.

The results show that, in Denver, four of the five civic engagement indicators declined significantly from 2008 to 2009. In Seattle, two of the five indicators decreased significantly over the same period. The other indicators in both cities were not significantly different year over year. Together, these results offer support for H2: Many measures of civic engagement in Seattle and Denver declined from 2008 to 2009. Overall, aggregate civic engagement declined significantly in Denver (−4.6%, p ≤ .05) but not in Seattle (−1.4%). As the scaling in Figure 1. shows, the declines in both cities outstripped declines (or increases) in their closest peer cities. ⁷ So, although civic engagement declined 4.6% in absolute terms in Denver, relatively speaking, there was about 30% less civic engagement in 2009 than 2008.

Dead Newspapers and Citizens’ Civic Engagement

All authors

Lee Shaker

https://doi.org/10.1080/10584609.2012.762817

Published online:

30 January 2014

Figure 1. Percentage relative change in civic engagement, 2008–2009. The difference in Denver is significant (p ≤ .05) in a two-tailed t test.

Display full size

Meanwhile, in the other eight cities, 35 of the 40 year-over-year comparisons are nonsignificant. Only one indicator across all of the cities—boycotts of products or services among Cincinnatians—significantly declined year over year. Four indicators—contacting a public official in Cleveland and Minneapolis, boycotting a product or service in Dallas, and participating in a civic group in Phoenix—increased from 2008 to 2009. Thus, H3 is not uniformly supported: There are some differences in the data across the nation. But, as H3 predicts, the vast majority of the city-level indicators were not significantly different across the 2 years.

The national data in Table 4 provide a wider view. Here each indicator is significantly different (due in part to the large sample size) in the year-over-year comparisons: Four decreased and one increased. The general trend reflects a decline in civic engagement, nationwide, which supports H4. The magnitude of these differences, however, ranges between just 1% and 2% across the board. The observed declines in Seattle and Denver, meanwhile, are roughly 5% to 8%, which supports H5.

Table 5 presents a more systematic assessment of the effect of newspaper closure upon citizens’ civic engagement. In the three models presented, the dependent variable is the aggregate year-over-year change in either social capital or civic engagement. ⁸ Though the sample size in these models is only 20, the first regression that predicts change in civic engagement is significant (F = 3.209, p ≤ .05), and there is a strong, negative relationship (β = −3.704, p ≤ .01) between the closure of a newspaper and civic engagement—even after controlling for the amount of 2008 campaign TV advertising spending in each market and the change in unemployment from 2008 to 2009. Meanwhile, a similar model that predicts change in social capital is not significant; none of the independent variables are significant. Controlling for the year-over-year change in newspaper circulation muddies the picture of civic engagement some. In this model, only the newspaper closure variable is significant (β = −3.571, p ≤ .05), but the regression itself retreats from significance (F = 2.402, p ≤ .10). Together, these models offer further support for H2 and H5. ⁹

Discussion

The results reported here suggest that the collapse of the Rocky Mountain News and Post-Intelligencer adversely affected the civic engagement of citizens in Denver and Seattle. There are many possible explanations for these results: measurement error, chance, and an array of unmeasured factors that may have intervened in the year between measurements. In isolation, it would be difficult to rule out these competing hypotheses and attribute the change to the downfall of a pair of newspapers.

Yet, the data presented here are not in isolation—there are multiple points of comparison that can be used to address alternate explanations. National data suggest that there was a slight decline in many civic engagement indicators from 2008 to 2009—not surprising considering the change in the macro-political environment. But the magnitude of the decline in Seattle and Denver outstripped the national trend. And in other, similar cities, measurement of the same indicators, collected in the same surveys and at the same times, showed little evidence of decline. Contrasting the Denver and Seattle results with those from like cities offers a pair of important benefits. First, concern that a measurement issue that systematically skewed the data toward a year-over-year decline is assuaged. Second, national-level third variables (economic turmoil, different political conditions, etc.) that could plausibly explain variation in any one city should also cause the same variation in other cities as well. The absence of uniformity across the cities suggests that differences within a city from year to year can be attributed to factors at the metropolitan level.

Is it possible that the declines in civic engagement in Seattle and Denver could be the result of factors other than the newspaper closures? Yes. In the 12-month period between surveys, many important and unique events surely transpired in each city in the nation. Yet, as prior research suggests (Kang & Kwak, 2003; Mondak, 1995; Moy et al., 2004), because newspaper readership is significantly related to civic engagement, it seems very likely that the newspaper closures would bear negatively upon civic engagement in Denver and Seattle. And, even after controlling for the possibility of a spillover effect on civic engagement caused by greater political advertising in some cities compared to others, the likelihood that the economic downturn that began in 2008 had a depressing effect upon civic engagement, and the ongoing decline of newspaper circulations, the data suggest that the closure of a newspaper still had a measurable effect on citizen engagement. Finally, models that control for differences in the racial composition of the year-over-year CPS samples suggest that this factor is not related to the change in civic engagement. ¹⁰

Perhaps the observed decline in Denver and Seattle was a chimera: It is conceivable that the estimate of civic engagement in the 2008 CPS data was spuriously high. If this is the case, then the 2009 results would merely reflect regression to the mean, and the change from 2008 to 2009 would not be meaningful. Because the CPS civic engagement module began in 2008, it is difficult to address this concern. But, in perhaps the most comparable prior research, the Saguaro Seminar's multi-city social capital benchmark surveys in 2000, civic engagement in Denver and Seattle significantly outstrips civic engagement in the nation as a whole. ¹¹ This implies that the high reading in 2008 was not aberrant, and the decline from 2008 to 2009 was not random in nature.

If the year-over-year decline is real, perhaps it is only temporary. Data from the 2010 CPS lend some support to this notion. ¹² In 2010, civic engagement rebounded in Denver: It increased significantly (by 3.4%, p < .05), though it did not reach its 2008 level. In Seattle, there was a small nonsignificant decline from 2009 to 2010, continuing the trend from the year before. Additional years of CPS data will offer two advantages: more case cities (in which newspapers have closed) and more data points to track trends over time. A long-term study building on the current research is a worthy future direction. For now, this preliminary result offers some context: The negative impact of newspaper closures reported here may be only a short-term effect as communities develop new ways of producing and accessing information.

Though Denver and Seattle both experienced a similar shock to their local media environments, neither the disruption nor its effects were identical. Clearly, the data show that the magnitude of the negative effect was greater in Denver than in Seattle. Two clear explanations for this divergence come to mind. First, the relative importance of the two newspapers to their communities was different in at least one key respect: In Seattle about 8% of households received the PI, while in Denver about 20% of households received the Rocky. Accordingly, we should expect to see larger effects in Denver. Second, the PI halted its print edition but continued to produce and distribute news on the Web; meanwhile, the Rocky went completely out of business. If we think of local news as a public good—something that is broadly beneficial even if it is not specifically consumed—then this is an important distinction (Baker, 2001). In Seattle, at least some of the news produced for a print edition of the PI is now created for the Web. Even if the audience for the online edition is smaller, the positive externalities of the news it creates may still exist. In Denver, the amount of local news was simply reduced.

This raises another compelling theoretical question: Is the decline of civic engagement driven by a decrease in the production of news or a decrease in the consumption of it? Perhaps Denver and Seattle, as they lost their second newspaper, simply retreated to the level of other one-newspaper towns. If this is the case, then it would imply that decreased production of news was the critical shift. Simple analyses show that civic engagement in the one- and two-newspaper top 20 metropolitan areas is statistically indistinguishable. ¹³ Looking at the top 50 metropolitan areas, it appears that civic engagement is significantly higher in one-paper markets—but this comparison is conflated by the fact that the one-newspaper communities tend to be smaller. At least as far back as Tönnies and Loomis (1957), researchers have theorized that social ties are stronger in smaller communities. In the regressions reported above, however, overall circulation is included as a term to test the importance of readership. It is nonsignificant, which undermines the notion that readership matters. Again, this question warrants further research as the current study does not yield a satisfying answer.

In sum, the evidence reported here implies that Seattle and Denver suffered significant negative declines in civic engagement when they lost one of their daily newspapers. But perhaps the causal ordering of this relationship is reversed: It is possible that the newspapers were no longer viable because of apathy or weak civic engagement in these communities. The available contextual evidence undermines this perspective. The decline in newspaper circulation in Denver and Seattle was no worse than the nationwide trends. Levels of civic engagement in the communities were high, relative to the top 20 cities, in 2008, but this was consistent with prior measurements. And in Seattle, a community organization formed explicitly to fight the demise of the PI in court years before its eventual closure. In the end, the PI and Rocky were victims of economics in a changing media environment: Very few cities can support two local newspapers, no matter what their civic value may be.

Supplementary Material

Supplemental data for this article can be accessed on the publisher's website at http://dx.doi.org/10.1080/10584609.2012.762817.