Expanding the Model

INTRODUCTION | LITERATURE REVIEW | METHODOLOGY | INITIAL FINDINGS
EXPANDING THE MODEL | CONCLUSIONS | REFERENCES | PDF VERSION

Introduction

An editorial in the Chattanooga, Tennessee Times Free Press (June 8, 2011) stated one of the fundamental principles of local government budgeting.

“Tennessee lawmakers are fortunately required by our state Constitution to balance the state’s budget each year. That leaves only a couple of options when the cost of the things that the state needs or wants exceeds the funds available: Cut spending or raise taxes. Budget cuts are rarely easy, but they are far preferable to job-killing tax increases.”

Chattanooga was one of the cities in this study with substantial declines in budgeted funds between 2008 and 2011. California cities such as San Luis Obispo, also one of the cities included in this study, were not immune from the budget cuts either. “Roads are rougher, sidewalks riddled with more cracks and grass at parks left fallow as local governments continue to trim from already lean budgets,” a December 31, 2011 San Luis Obispo Tribune article by AnnMarie Cornejo read. “When the recession struck in 2008 and cities began to face steep declines in revenue, government employees faced a new level of scrutiny nationwide and a growing awareness of public salaries became a focal point of heated discussions.”

During the last decade, spending cuts have been on the minds not only of bureaucrats and politicians but also of reporters and citizens alike as the Great Recession forced towns to cut services, everything from street lights to trash pickup to summer baseball. This research examines the impact that local media outlets have on policy changes, the budget specifically. The analysis in the previous chapter showed a weak relationship between newspaper coverage and policy change, a relationship that was stronger over a longer time period than a shorter one. Focusing on the five-year analysis shown to be most relevant in the last chapter, the additional examination begins with a more complex regression analysis used to evaluate the relationship between the newspaper coverage and changes in the budget, including the impact of geography and the impact of media coverage on various local government departments. Geography does matter when studying local government budgets and not all departments are influenced equally by media coverage. The chapter also includes discussion of the variables for staff size, market saturation (circulation / population), online presence and local ownership conceived as part of the original set of hypotheses.

Coverage and Policy Change

H1: Local coverage and policy change will be positively correlated.

Of the hundreds of studies on agenda setting, mostly at the national level and mostly of public opinion, researchers often found extremely high correlations, as high as 1.0 (Maher, 1996). Over time, the concept of agenda setting, that media coverage influences public opinion became accepted. Subsequent studies took this concept one step further by examining the relationship between media coverage, public opinion and policy outcomes. Mortensen and Serritzlew (2006), for example, examined in the relationship between media coverage and changes in the budgets of 191 Danish municipalities over a 13-year period to examine the outer boundaries of media influence. “If public spending,” they concluded, “is not affected by the media, then this is an empirical finding that tells us something important about the outer limits of media effects.” They reported that the media may affect political discussions and certain political decisions, but the budgets and broader policy priorities remain largely unaffected.

Similarly, this study examined the impact of media coverage not on public opinion but on policy change. In findings similar to those of Mortensen and Serritzlew, this study finds that media have almost no impact on budget changes in the short-term (one year) and minimal impact over a five-year period — identifying more clearly an outer boundary for media effects. Even when the media coverage does impact policy change, it does not do so equally across all areas of the budget.

While media reports had an impact on changes in the budget at least in some departments, external factors, such as the economic downturn of 2008, influenced changes in policy as well. By 2012, the budgets of the towns in this study started showing signs of declining as part of the Great Recession that began in late 2007 (National Bureau of Economic Research, 2008). While it appeared that federal aid to local governments, incremental tax/fee increases and use of reserve monies helped many town avoid cutting their budgets significantly early on, by 2011, the federal aid and reserve monies had dried up and citizens were not willing to be taxed at a high enough rate to keep government services functioning at the same level as they had prior to the economic downturn. In the period between 2003 and 2008, the budgets of the towns in this study grew by an average of 33 percent. However, between 2005 and 2010, they grew by only 21 percent.

Supporting the individual-variable analysis in the previous chapter, none of the additional independent variables had any impact on the one-year analysis, as table 9 indicates. Only the relationship between local ownership and staff size was found to be significant, and then with a barely moderate effect size. One year did not give the media outlets enough time to have a perceptible impact on budgetary changes.

Table 9: Correlations of Newspaper and Town Variables
on 2009-2010 City Budget Totals
 
 

Mean

Standard deviation

Local ownership

Market saturation

Staff size

Web coverage

Content analysis

Budget 728.55 8424.08 -0.06 -0.04 0.10 0.05 -0.03
Local ownership 0.31 0.46   0.10 0.29** -0.09 0.01
Market saturation 0.58 0.50     0.10 -0.06 -0.07
Staff size 21.14 12.45       -0.05 -0.11
Web coverage 3.61 0.75         0.03
 

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

Similarly, in a one-year model from 2009-2010 (table 10), none of the individual policy areas (economic development, parks, recreation and tourism, public safety and public works) revealed statistically significant results. The one-year models added little to the discussion except that media effects do not occur over the short-term. Instead, as this analysis showed and the analysis in the previous chapter supported, examination over a longer time period provides more insight. Other research by Walker (1977), Baumgartner and Jones (1993) and Kingdon (2003) and basic decision-making theory supports this insight. It simply takes time to establish objectives, alternatives and make decisions. “Time is an ever present and prominent dimension in all human decision making. Decisions are oriented towards future time, they take time to make, their consequences develop over time, and they are sometimes thought about for a long time afterwards” (Ranyard, Crozier and Svenson, 1997; Ariely, Zakay, 2001).

Table 10: Correlation of Newspaper and Town Variables on Various City Budget Line Items
 
  2009-2010 Model
Economic development 0.01
N 137
Adjusted R2 -0.02
   
Parks, recreation and tourism 0.00
N 137
Adjusted R2 -0.02
   
Public safety 0.04
N 137
Adjusted R2 -0.01
   
Public works 0.00
N 137
Adjusted R2 -0.02
   
Total -0.03
N 133
Adjusted R2 -0.02

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

Initially, an evaluation of the assumptions of the independent variables led to transformation of the independent variables several of which had a skewness value exceeding 1 and a higher than expected kurtosis value which might have indicated the presence of outliers. To examine the possible presence of outliers and skewness, each of the groups for analysis (economic development, parks, recreation and tourism, public safety and public works) were run first with no transformations, then with a logarithmic transformation on market saturation (circulation divided by population) and finally with a logarithmic transformation on web coverage, staff size, budget change and market saturation. While there was no theoretical basis for the transformations, the high values for kurtosis indicated this transformation might improve the normal distribution of the variables. However, while it did decrease the values for kurtosis and skewness, it did so only marginally and neither improved the significance nor the effect size of the equations.

Because market saturation includes population and because there is a theoretical foundation for transformation of variables that are spread over several orders of magnitude, including population, the transformation of this variable was retained. A transformation is validated in literature and the logarithm transformation is favored to improve normality and to reduce positive skewness (Hopkins, 2002; Allison, 1999; Tabachnick and Fidell, 2001). Market saturation had skewness of 3.055 initially and 0.321 after the logarithmic transformation. Initially, the variable also had a kurtosis value of 12.674, which was reduced to 0.083 after the transformation. In addition, as part of the analysis of the equation, to reduce the impact of missing values absent in the five-year model due to the inability to obtain five consecutive years of budget or content analysis data, the missing values were replaced with the mean.

Table 11 displays the correlations between the variables. In all cases, the values of r (and hence the variance explained) improved with the five independent variables included when compared to the simple correlations of budget change and change in newspaper coverage over the time period studied.

Table 11: Correlations of Newspaper and City Variables
on 2005-2010 City Budget Totals Including All Regions
 
  Mean Standard deviation Content analysis Local ownership Market saturation (log) Web coverage Staff size
Budget change (DV)

117.97

20.21

-0.03

0.07

-0.10

-0.07

-0.05

Content analysis

635.57

583.57

 

-0.10

-0.08

0.03

-0.01

Local ownership

0.31

0.46

   

0.09

-0.07

0.28***

Market saturation (log)

-0.78

0.71

     

-0.05

0.14**

Web coverage

3.60

0.74

       

-0.06

Staff
size

21.09

12.04

         

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

The first examination of the data involved the aggregated change in total city budget during the five-year period and change in total number of terms in the database during the same time period. The only two variables that showed any significance at all were local ownership / staff size and the log of market saturation / staff size. Papers with higher market saturation utilize a larger staff. Because the direction of the relationship cannot be determined from this analysis, it is possible that larger staffs generate papers with higher market saturation. Regardless of which direction that relationship developed, it was not as interesting as the significant relationship between local ownership and staff size. In terms of direction, it seems much more plausible that locally owned papers have larger staffs than non-locally owned papers than that larger staffs generate local ownership.

Despite validation of some ideas presented in the literature, the entire equation was not statistically significant possibly due to the aggregation of the data. Though one department’s budget, or a line item within that budget, might change significantly resulting in major changes in service, the money for that increased service may come from another area or the money for decreased service might go to another area resulting in little net change in the budget. Similarly, newspapers may choose to cover public safety, for example, to the exclusion of economic development, public works or parks/recreation/tourism due to some issue in the area of public safety. Due to staff or space limitations, the total amount of coverage in the paper may remain the same. The amount of coverage (sheer number of stories) is probably more related to advertising income than any desire to cover more stories.

Geographic Significance

With the anecdotal evidence piling up that some states were hit harder than others for reasons that were beyond the scope of this study, one area worth examining was geographic though it was not a part of the original study design. Indeed, geography does seem to matter when it comes to budget stability as model 3 in table 12 demonstrates in comparison to model 2 with subregion 4.9 excluded. The U.S. Department of the Census divides the country into four major regions: 1 — northeast, 2 — central, 3 — southeast and 4 — west. A univariate analysis of variance of both the one-year data and the five-year data resulted in a significant Levene’s Test, showing that the groups formed by region vary significantly in their variations on change in the budget. Between 2009 and 2010, region 4 saw an average budget decrease of 2.59 percent while region 1 saw an average increase of 1.81 percent. Over both the short-term and the long-term, the budget shortfalls hit states in region 4 harder.

The trends continued to be apparent when the subregions were included in another univariate analysis of variance and regression analysis. The U.S. Department of the Census further breaks down the country into nine subregions. This analysis too was significant. The Partial Eta Squared value for the one-year change for subregion 4.9 was 0.10 (p<0.01) and 0.71 for the five-year change (p<0.01). Between 2009 and 2010, the budgets of the states in subregion 4.9 (California, Oregon and Washington) declined by 5.8 percent. Between 2005 and 2010, the budgets of the towns in subregion 4.9 grew by only 15.5 percent compared to an overall average of 21.3 percent. The towns in subregion 4.8 grew by 31.1 percent over the same time period.[1] The population in subregion 4.9 increased by 8.1 percent while other subregions saw an average population increase of 4.1 percent. All other regions during the five-year period showed an increase in total budgeted allocations of between 12 percent and 31 percent. Subregion 4.8 saw the largest increase, 31 percent. In short, the budgets of the Western states (California, Oregon and Washington) increased at a rate slower than the rest of the country. Further, as the literature indicated, California, due to decades of over-spending and poor fiscal management, bore the brunt of the economic downturn.

The hardships in California, in particular, are nothing new, brought on initially and immediately with the passage of Proposition 13 in 1978. After the passage of that proposition, governments went into deep fiscal crisis in the recession of 1980-82 and the state had to bail them out by transferring money from the general fund (Schrag, 1998). To make matters worse, the Reagan administration in Washington cut federal revenue sharing and aid to cities (Davis, 1993) and the Deukmejian administration in Sacramento put its money into the greatest prison-building splurge in U.S. history (Gilmore, 2007). In the downturn of 1990-93 after the tech stock bubble, the state took a $11 billion nosedive into the red on a $50 billion budget (Walker, 1995). To keep things running on reduced tax revenues, the government began to issue more bonds, pushing the state further and further into debt. The result has been horrendous slashing of the budgets for schools, higher education, health and welfare, and local government functions (Schrag, 2009) and an economic system that is unresponsive to policy changes, including changes in the budget. This stalemated society simply cannot respond to the issues it faces. Hence, excluding California, Oregon and Washington (subregion 4.9) shows us what influence media has on policy where the system allows for change. For the media or other external factors to have an impact on California, will require some type of revolution that results in a change in the constitutional structure at the foundation of that state’s government.

What happens in California as a state, with the largest state and local government budget in the country, has an impact on municipalities within the state and also on other parts of the country simply due to California’s size. The state is the country’s largest sub-economy, accounting for roughly 13 percent of national output and the largest budget after the federal government, about $100 billion per annum in the 2000s. California also has the largest budget deficit of any state today (McNichol and Johnson, 2010). California’s budget deficits peaked at $45.5 billion in 2010, now down to $8.4 billion projected for 2013, about one-tenth of the state’s budget. However, California is not the only Western state with budget challenges. Washington and Oregon (the other two states that make up subregion 4.9) too have faced their challenges. By 2013, Oregon faced a $1.7 billion projected deficit, about one-fourth of its budget. Washington faced a $3.5 billion deficit (down from $4.8 billion in 2010), one-fifth of that state’s budget (McNichol, Oliff and Johnson, 2012).

The downsizing of the manufacturing industry, the bursting of the tech stock bubble and the rapid increase in home sales and prices that made for unaffordable housing made California’s problems multifaceted, complex and difficult to solve. Over time, all of these factors, and others, led to increasing unemployment. “The state’s new unemployment rate — 12.2 percent, according to the Bureau of Labor Statistics — is far above the national average of 9.7 percent and places California, the national’s most-populous state, fourth behind Michigan, Nevada and Rhode Island” (Steinhauer, 2009). “Many of the cities with the longest road to recovery are California cities, where home prices rocketed out of control and entire economies were supported largely by a real estate bubble. Fresno, Modesto, Salinas, Bakersfield, Stockton and Los Angeles all saw home prices soar to unsustainable levels and then begin their inevitable plunge. The collapse of the housing markets pushed unemployment rates in these cities above 10 percent” (Zumbrun, 2009). In turn, the high unemployment led to declining tax revenues at all levels of government and to deeper cuts in public services, everything from street lights to trash collection to public transportation.

According to researchers, California is not simply another state falling in the shadow of the economic changes — it was the cause. “[I]f Wall Street was the eye of the financial hurricane of the last decade, then California was the equivalent of the tropical oceans that provide the heat to feed such raging storms. More than any other place, California was the source of mass mortgage lending, ballooning home values and dubious subprime operations. In short, California needs to be recognized as the pivotal site of the bubble of the 2000s, the bursting of the financial markets and the Great Recession that followed” (Bardhan and Walker, 2010).

Following four years of what some have termed draconian budget cuts, some indicators by 2012, an election year, show that the state of California as well as the cities and towns in that state, are poised for a slow recovery. While the state’s fiscal conditions are improving along with the broader economy, the state is coming out of a deep hole. Further, states face major obstacles slowing their fiscal recovery, including the shifting role of federal government, which allowed much of the emergency aid to states to expire in 2011 — a decision that left states with fewer options to address their still substantial budget shortfalls in fiscal year 2012 and beyond (McNichol, Oliff and Johnson, 2012). If revenues continue to grow at the 2011 rate, it would take seven years to get some state budgets back on a normal track The short-term outlook for California and the other Western states primarily depends on U.S. trends. (Levy, 2011) Cities with robust technology sectors, including some in the Western states, are poised for stronger recoveries than manufacturing or finance centers. Cities with high-tech capabilities such as Seattle, Huntsville, Ala., or Boulder, Colo., could see quick recovery in coming months (Zumbrun, 2009).

Table 12: Regression of Total Change in Newspaper Coverage and Total Change in Budget, 2005-2010 (Unstandardized Coefficients)
 
  Model 1 Model 2 Model 3
Content analysis 0.00
(0.00)
0.00
(0.00)
0.00
(0.00)
Local ownership   3.71
(3.86)
1.72
(2.86)
Market saturation (log)   -2.93
(2.44)
-3.88**
(1.90)
Web coverage   -2.00
(2.31)
-3.83**
(1.71)
Staff size   -0.11
(0.15)
-0.11
(0.11)
Constant 121.47***
(2.21)
124.67***
(9.50)
134.74***
(7.02)
       
R2 0.01 0.03 0.08*
Adjusted R2 -0.01 -0.01 0.04
N 75 142 121
Standard error of estimate 14.76 20.31 14.21
       

NOTE: Standard errors are in parenthesis

Model 1: No control variables
Model 2: Control variables with all regions
Model 3: Control variables excluding region 4.9

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

Comparing Models

Taking into account the independent variables for local ownership, market saturation, online coverage and staff size and eliminating subregion 4.9 produces the best results when examining the relationship between newspaper coverage and changes in the budget during a five-year period as the comparison of the three models in table 12 demonstrates. While the model is only significant at p<0.1 (unacceptable for a determination of significance in most situations), given the small sample size, it is worth noting. The third model presents not only an increased level of significance, it represents a larger effect size. While still weak, accounting for only 4 percent of the variance (adjusted r2=0.04), it improves on the basic correlation between coverage and changes in the budget over a five-year period and over the model with subregion 4.9 included. The relatively large value for the probability, however, could indicate that there is no detectable difference between budget changes in relation to coverage and budget changes without coverage.

As the model indicates, however, two independent variables, the log of market saturation (circulation / population) and web coverage are statistically significant, p<0.05. They also contribute to the model with large coefficients, -0.19 (1.90 standardized) for the log of market saturation and -0.20 (1.712 standardized) for Web coverage. For every unit increase in the log of market saturation, the town’s budget decreases by 0.19 units holding all other variables constant. Similarly, for every unit increase in Web coverage, the town’s budget decreases by 0.20 holding all other variables constant. The interpretation of why a town’s budget would decrease simply because the market saturation of a newspaper goes up or online coverage increases is challenging. Clearly more factors influence the changes in a town’s budget. The coefficients for coverage, local ownership and staff size are not significant.

Part of the challenge for this model, as noted any time the total budget was used, is that the town may have the same total budget (and therefore no change over a given time period) but may have changed the allocations for line items drastically. Such changes would not be indicated when the aggregated budget is used, hence explaining why various line items were used for the bulk of this study. While the aggregated budget and coverage equations gave little insight into the relationship between coverage and policy outcomes, the individual budget lines provided more insight.

Departmental Significance

When examining the relationship between change in media coverage and change in departmental budgets, the first step was to look at the relationship between each of the independent variables. For economic development (table 13), none of the independent variables was related to the change in the budget. However, some independent variables were statistically related to one another. For example, change in coverage (content analysis) was statistically highly correlated with local ownership. As the literature confirms, locally-owned newspapers cover municipal issues more than papers that are not locally owned. Similarly, (log of) market saturation were statistically related to coverage. That is, newspapers with a higher market saturation covered local issues more than papers with a lower market saturation. Of course, correlation does not determine causation. The relationship could mean that papers with a higher market saturation cover more local issues or because they cover local issues, they have a higher market saturation. Finally, as is also covered in the literature, newspapers that are locally owned have larger staffs. Though the observation is a correlation, not a causation, it is less likely based on prior research that papers are locally owned because they have larger staffs. While the relationship between local ownership and market saturation and the relationship between market saturation and staff size were both statistically significant, they were significant only at the lowest level (p<0.1). While both of those relationships are substantiated in the literature and seem logical, they merit little discussion as part of this research.

Table 13: Correlations of Newspaper and City Variables
on 2005-2010 Economic Development Budget Excluding Subregion 4.9
 
  Mean Standard deviation Content analysis Local ownership Market saturation (log) Web coverage Staff size
Budget change (DV) 0.54 2.72 0.08 0.08 -0.05 -0.03 -0.02
Content analysis 0.82 3.94   0.23*** 0.18** 0.12* -0.04
Local ownership 0.33 0.47     0.13* -0.06 0.25***
Market saturation (log) -0.75 0.70       -0.06 0.14*
Web coverage 3.59 0.76         -0.04
Staff size 21.46 12.55          
 

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

Like the aggregated, total budget, the equation using the economic development budget as the dependent variable was not statistically significant (table 14). The coefficients, which were not statistically significant, were all small.

Table 14: Regression of Change in Newspaper Coverage and Change in Economic Development Budget, 2005-2010 (Unstandardized Coefficients)
 
  Model 1 Model 2 Model 3
Content analysis 0.13 (0.10) 0.14 (0.11) 0.06 (0.07)
Local ownership   -0.14 (0.95) 0.42 (0.57)
Market saturation (log)   -0.66 (0.60) -0.29 (0.37)
Web coverage   0.15 (0.56) -0.16 (0.34)
Staff size   -0.05 (0.04) -0.01 (0.02)
Constant 0.88** (0.44) 0.83 (2.28) 0.82 (1.34)
       
R2 0.01 0.04 0.02
Adjusted R2 0.00 0.00 -0.03
N 138 142 122
Standard error of estimate 4.99 4.94 2.76
       

NOTE: Standard errors are in parenthesis

Model 1: No control variables
Model 2: Control variables with all regions
Model 3: Control variables excluding region 4.9

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

Of interest, however, was that the mean percentage change in the economic development budget during the five-year period was only 1.02 percent, the smallest change of any of the lines affected when controlling for the change in the total budget. Economic development budgets remained stable during this time period in comparison to the total budget. As discussed previously, economic development budgets, representing only 2.28 percent of the budgets of the towns in this study, are small when compared to the overall budget. Cities such as West Covina, Calif., an outlier when examining the economic development budget since the town devoted nearly 40 percent of its general fund budget and 19 percent of its entire budget to economic development, included items such as developing affordable housing for residents and the rehabilitation of deteriorating business centers in that city’s line item for economic development. Other towns included such items in other lines or established separate economic development authorities, with separate budgets, for such activities.

The relationships in the variables related to parks, recreation and tourism provided more insight. The evaluation showed changes in newspaper coverage were more closely related to changes in policy (table 15). First and foremost, change in the coverage of parks, recreation and tourism was highly statistically significant. In addition, at least for this budget area, the newspaper’s staff size was also highly significant in its relationship to policy changes during the five-year period. Further, the change in the coverage of parks, recreation and tourism (content analysis) was also related to local ownership of the paper, market saturation and online coverage. While it cannot be determined whether policy change resulted from the change of coverage or coverage resulted from the change in policy — even the basic model predicts that the relationship between coverage and policy change is bi-directional — they are more strongly related in this policy area than in any other. Relationships between other independent variables, local ownership and market saturation, local ownership and staff size as well as staff size and market saturation are statistically significant at least at p<0.1 and were discussed as part of the discussion above on economic development.

Table 15: Correlations of Newspaper and City Variables
on 2005-2010 Parks, Recreation and Tourism Budget Excluding Subregion 4.9
 
  Mean Standard deviation Content analysis Local ownership Market saturation (log) Web coverage Staff size
Budget change (DV) 1.97 3.70 0.52*** 0.10 0.10 -0.04 -0.21***
Content analysis 4.72 5.55   0.16** 0.18** 0.15** -0.07
Local ownership 0.33 0.47     0.13* -0.06 0.25***
Market saturation (log) -0.75 0.70       -0.06 0.14*
Web coverage 3.59 0.76         -0.04
Staff
size
21.46 12.55          

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

Whether the model was the most basic regression with one independent variable (content analysis) or a complex, yet a still incomplete, set of independent variables, the relationship between the change in newspaper coverage during a five-year period and the change in the budget for that line are strongly related (p<0.01). As table 16 indicates, the model explains about 30 percent of the variance in the budgeting for parks, recreation and tourism in those states other than California, Washington and Oregon. Of all the relationships in this study, this one was the statistically largest and most significant relationship. Both the newspaper coverage and staff size were statistically significant (p<0.05) in relation to the budget changes. Newspaper coverage had a standardized regression coefficient of 0.51 (0.34 unstandardized) indicating that an increase of one standard deviation in coverage produces an increase of 0.51 standard deviations in budget, clearly something worth noting. During the five-year period in this study, when controlling for the change in the total budget, PRT budgets increased by 2.13 percent when compared to the approximately 18 percent increase of the total budget over the same time period.

Table 16: Regression of Change in Newspaper Coverage and Change in Parks, Recreation and Tourism Budget, 2005-2010, (Unstandardized Coefficients)
 
  Model 1 Model 2 Model 3
Content analysis 0.28*** (0.05) 0.26** (0.05) 0.34*** (0.05)
Local ownership   0.35 (0.73) 0.46 (0.64)
Market saturation (log)   0.52 (0.46) 0.08 (0.42)
Web coverage   0.014 (0.44) -0.57 (0.38)
Staff size   -0.07** (0.03) -0.06** (0.02)
Constant 0.82** (0.41) 2.67 (1.79) 3.57** (1.52)
       
R2 0.17 0.20*** 0.32***
Adjusted R2 0.16 0.17 0.29
N 138 143 122
Standard error of estimate 3.91 3.85 3.13
       

NOTE: Standard errors are in parenthesis

Model 1: No control variables
Model 2: Control variables with all regions
Model 3: Control variables excluding region 4.9

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

While newspaper coverage of economic development issues, including taxation, was difficult to find, newspaper reports seem more willing to cover parks, recreation and tourism in their communities. One of the frequent topics for coverage was a decrease in service, including the closing of various parks. If it were not a reflection of tough budget times and real cuts in services, the commentary on Denis C. Theriault’s newspaper blog about budget cuts in Portland would be humorous. “Yesterday, the Bureau of Parks and Recreation was the first city office to release its own doomsday cutback plans, and it’s not pretty,” he said on December 9, 2011. “Even in the lightest scenario, portable toilets would replace park restrooms (like at Occupy Portland!), Buckman Pool would close, and so would one community center. Layoffs appear to be in the worst-case plan.”

While Portland was not one of the cities in this study, the impact on parks, recreation and tourism in local communities was so obvious that even the New York Times reported on it (McKinley, 2011). “There are few things in life more doleful than a child looking at a closed pool on a steamy summer day, and yet that sad scene has become as common as sunburns and mosquito bites as struggling local governments make the painful choice to shut their pools to save the budget. The list of locales where public pools have been in jeopardy in recent years includes some of the sweatiest spots in the nation, including Central Florida (90s and humid on the Fourth), Atlanta (90), and Houston (97).”

One of the complications of analyzing parks, recreation and tourism as part of the budget was that, like public works, cities and towns often included only operational costs in the budget, most often as part of the general fund. Physical plant improvements, land purchases and other costs which might be directly associated with parks and recreation but over a longer time period than the typical annual budget were sometimes included in separate funds or capital improvement budgets typically not included in this study. On average, parks, recreation and tourism made up 6.12 percent of the budgets in this study, larger than economic development but less than one-third of public works and one-fifth of public safety.

The relationship between coverage of public safety and change in the line-item budget for public safety (including police, fire and EMS) (table 17) bore similarities to economic development (where there was no relationship) and parks, recreation and tourism (where there was a strong relationship). Descriptively, while the budgets did see an increase over the period studied (1.98 percent on average), newspaper coverage of public safety increased 10.01 percent on average. During the five years between 2005-2010, while public safety budgets increased marginally, less than the 4.7 percent increase in population, news coverage increased five times as much. For every unit increase in newspaper coverage, the town’s budget increases by 0.15 (p<0.1) holding all other variables constant. Local ownership of the newspaper also seemed related to changes in the public safety budget. Local newspaper owners, like all residents of a community, have a vested interest in the safety and security of that community. In addition, the change in newspaper coverage was also related to the online presence of the newspaper. At least in the area of public safety, for every unit increase in newspaper coverage, Web coverage increased 0.15 (p<0.05). Web coverage and print coverage are probably more closely related than this research indicates. The statistics could also indicate that newspapers that have more of an online presence tend to cover public safety issues (including breaking news such as car wrecks and house fires) more than newspapers with less of an online presence.

Table 17: Correlations of Newspaper and City Variables
on 2005-2010 Public Safety Budget Excluding Subregion 4.9

 

 

Mean

Standard deviation

Content analysis

Local ownership

Market saturation (log)

Web coverage

Staff size

Budget change (DV)

11.55

15.49

0.15*

0.16**

0.06

0.07

-0.11

Content analysis

13.37

9.93

 

0.01

-0.16**

0.15**

0.06

Local ownership

0.33

0.47

   

0.13*

-0.06

0.25***

Market saturation (log)

-0.75

0.70

     

-0.06

0.14*

Web coverage

3.59

0.76

       

-0.04

Staff
size

21.46

12.55

         
 

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

In comparison to the other two models, the third model, including all the independent variables and excluding subregion 4.9, provides the best insight into the relationship between public safety budgets and public safety coverage. As table 18 indicates, the model was statistically significant, but only at the p<0.1 level and then with a small effect size (r2 = 0.08). However, the overall model showed three significant independent variables, including coverage (content analysis) (B=0.25, p<0.1), local ownership (B=6.36, p<0.05) and staff size (B=-0.22, p<0.1) in relation to budget change. Given that it is the most significant, the contribution of local ownership to a positive change in public safety coverage is especially worth noting.

Table 18: Regression of Change in Newspaper Coverage and Change in Public Safety Budget, 2005-2010
(Unstandardized Coefficients)
 
  Model 1 Model 2 Model 3
Content analysis 0.23 (0.13) 0.22* (0.13) 0.25* (0.14)
Local ownership   4.50 (3.23) 6.36** (3.04)
Market saturation (log)   1.18 (2.06) 1.84 (2.04)
Web coverage   2.05 (1.95) 1.13 (1.85)
Staff size   -0.27** (0.13) -0.22* (0.12)
Constant 9.10*** (2.20) 7.07 (7.91) 8.10 (7.38)
       
R2 0.02* 0.07* 0.08*
Adjusted R2 0.01 0.03 0.04
N 139 143 122
Standard error of estimate 17.38 17.10 15.16
       

NOTE: Standard errors are in parenthesis

Model 1: No control variables
Model 2: Control variables with all regions
Model 3: Control variables excluding region 4.9

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

By far, public safety was the largest single portion, nearly one-third (32.49 percent) on average, of most town’s budgets. In Middletown, Ohio, a town of only 16,249 people in 2010, public safety made up nearly three-fourths of the city’s budget. In some towns, public safety included everything from police to fire to EMS to the municipal courts while other towns included only police, apparently falling under other jurisdictions for other services. Despite being such a large portion of the budget — or perhaps because public safety was a large portion of the budget — public safety budgets showed less correlation between change in coverage and change in budget over time. That media had little influence on budget changes could also have been because bureaucrats and politicians see the dangers of cutting public safety in a post-9/11 era when public safety is seen as an essential, governmental function.

Still, public safety operations are not immune from recent budget cuts and certainly still warrant coverage in all types of media. Chris Hoene, director of research for the Washington-based National League of Cities, said, “Typically, the public safety sector is sacrosanct. You tend to only see cuts move to public safety when [an economic] downturn’s pretty deep so the fact that we’re seeing layoffs and cuts in funding for public safety services is indicative of just the times that we’re in” (Brock, 2009). While public safety services are generally the last to suffer budget cuts when cities face hard times (Brock, 2010), police and fire departments were cut back in 63 percent of the cities and 39 percent of counties responding to a survey by the Washington-based National Association of Counties, National League of Cities and U.S. Conference of Mayors. “For some communities, this means fire and police stations that are closed and the potential for reduced capacity to respond to emergencies” (U.S. Department of Justice, 2011). In addition, the study cited specifics regarding cuts in public safety — cuts that included the following:

  • By the end of the year, it is expected that nearly 12,000 police officers and sheriff’s deputies will have been laid off.
  • Approximately 30,000 law enforcement jobs are unfilled.
  • An estimated 28,000 officers and deputies have faced week-long furloughs in 2010.
  • An estimated 53 percent of counties are working with fewer staff today than one 
year ago.
  • 2011 could produce the first national decline in law enforcement officer positions in at least the last 25 years.

American City and County magazine (August, 2009) also reported on cuts in public safety beyond cuts in police departments in some larger cities, cities generally beyond the scope of this study For example, Los Angeles cancelled its July 2009 recruiting class. In June, after closing several stations and implementing brownouts and hiring freezes during the past year, Atlanta found its ISO rating in danger of falling from a 2 to a 4. In July, Boston eliminated two of its 11 fire districts and implemented rotating brownouts for three out of 34 engine companies and one of 22 ladder companies, based on absences. The same study found that less than 10 percent of the responding cities were making cuts in public safety, Hoene said many of the cuts are happening in larger cities, such as Atlanta and Boston, where they are more noticeable (Brock, 2009).

Finally, the change in coverage of public works, like the coverage of parks, recreation and tourism was highly significant and correlated with a change in the budget for public works during a five-year period, excluding subregion 4.9 (table 19). Budget change was correlated with the content analysis (0.39, p<0.01) and with the staff size of the newspaper (-0.16, p<0.05). In addition, the change in newspaper coverage was also, again, related to local ownership (0.15, p<0.05) and (the log of) market saturation (0.17, p<0.05). As discussed above under economic development, local ownership was also related to (log of) market saturation and staff size, and (the log of) market saturation was related to staff size.

Table 19: Correlations of Newspaper and City Variables
on 2005-2010 Public Works Budget Excluding Subregion 4.9

 

 

Mean

Standard deviation

Content analysis

Local ownership

Market saturation (log)

Web coverage

Staff size

Budget change (DV)

7.81

13.98

0.39***

0.07

0.07

-0.02

-0.16**

Content analysis

3.65

5.73

 

0.15**

0.17**

0.09

-0.08

Local ownership

0.33

0.47

   

0.13*

-0.06

0.25***

Market saturation (log)

-0.75

0.70

     

-0.06

0.14*

Web coverage

3.59

0.76

       

-0.04

Staff
size

21.46

12.55

         
 

*Correlation is significant at the 0.10 level (2-tailed)

**Correlation is significant at the 0.05 level (2-tailed)

***Correlation is significant at the 0.01 level (2-tailed)

As a whole, the model examining the relationship between public works coverage and the change in the public works budget was, like the model for parks, recreation and tourism, highly significant (p<0.01) with an adjusted effect size of 0.14. So, coverage of items related to parks, recreation and tourism (everything from swimming pools to little league), explained between 14 (adjusted) and 18 percent of the variance in the budget over a five-year period excluding the states of California, Oregon and Washington, making this a moderate to large correlation. In addition, as table 20 indicates, the coefficients for coverage (0.93) and staff size (-0.17) were significant, coverage at the p<0.01 level and staff size at the p<0.1 level.

Table 20: Regression of Change in Newspaper Coverage and Change in Public Works Budget, 2005-2010
(Unstandardized Coefficients)
 
  Model 1 Model 2 Model 3
Content analysis 0.51*** (0.17) 0.47*** (0.17) 0.93*** (0.21)
Local ownership   2.17 (2.56) 1.30 (2.83)
Market saturation (log)   1.24 (1.83) 0.30 (1.75)
Web coverage   -0.51 (1.53) -1.09 (1.57)
Staff size   -0.24** (0.10) -0.17* (0.10)
Constant 5.32*** (1.29) 12.87** (6.28) 11.70* (6.31)
       
R2 0.07*** 0.10** 0.18***
Adjusted R2 0.06 0.07 0.14
N 138 143 122
Standard error of estimate 13.52 13.51 12.93
       

NOTE: Standard errors are in parenthesis

Model 1: No control variables
Model 2: Control variables with all regions
Model 3: Control variables excluding region 4.9

* Significant at the 0.10 level (2-tailed)
** Significant at the 0.05 level (2-tailed)
*** Significant at the 0.01 level (2-tailed)

Beyond the examination of the statistics, the coverage of issues surrounding public works got at the heart of what local governments provide to the community. Chief Billy Goldfeder, chairman of the Fairfax, Va.-based International Association of Fire Chiefs’ Safety, Health and Survival Section, said, “Local government was created in its most basic form to provide services to those paying the taxes” (Brock, 2009). Fundamental to those services in most towns includes everything from road maintenance to construction of public buildings to sewage treatment and water purification. Like other areas studied, various topics within the realm of public works warranted coverage. For example, in Oklahoma City, the second largest city in this study, Public Works Director Dennis Clowers told a local newspaper that major personnel cuts and service reductions, likely graffiti removal, have become more possible. “At some point, your budget can get to the point where you simply don’t have the dollars to employ the same number of people,” said City Councilmember Gary Marrs in the article. “It’s going to be a tough year. There are going to be some extremely tough decisions to make, and unfortunately [they are] going to include personnel” (Barkin, 2010).

Coverage of other topics did not necessarily include coverage of budget activities but included everything from the construction of a sewage treatment plant to informing the citizens of Roanoke, Va. about environmental issues in their community related to emissions from a sewage treatment plant. Without newspaper coverage of the topic, it is unlikely the citizens would have ever have known there was a potential problem. Roanoke Times reporter Laurence Hammack (2010) wrote, “Water and sewer treatment plants in Bedford and Franklin counties have been cited by the Virginia Department of Environmental Quality for excessive releases of copper, zinc and nickel. The Western Virginia Water Authority, which operates a water treatment plant in Bedford County, was fined $3,500 for discharging too much copper into Falling Creek. In Franklin County, the Ferrum Water and Sewage Authority was fined $2,200 for too much zinc and nickel in the treated wastewater that it released into Storey Creek.” Maybe Hammack’s work is not the coverage that is going to win a reporter a Pulitzer Prize, but it is exactly the kind of civic journalism the local citizens depend on so they can know what is happening in their community.

Coverage of public works in local government presented two challenges: the long-term nature of the projects and the mandatory nature of the work. Many cities included large public works projects in budgets outside the general fund or departmental budgets, including them instead in large, capital projects that spanned over years or decades and that were beyond the scope of this study. Often a governing board obligated the community to years and years of taxes or bond payments that the current administration was obligated to follow. Hence, there was often little discussion of the items and little media coverage of them unless something went wrong. Public works projects also presented a challenge for local governments because they are necessitated by external conditions and are mandatory. Snow removal, for example, was included in the budgets of towns in northern states. While the council may want to remove this from the budget, they might also find themselves with impassable streets and sidewalks and an irate, paralyzed community. Just as the federal government is obligated to pay, for example, social security payments, and just as cutting back on those payments would not be popular, local governments find many public works projects obligatory.

The Independent Variables

Policy change does not occur in a vacuum. As Rogers and Dearing (1988) noted in their model interpersonal communication, real-word indicators can have an influence upon the agenda and the policy outcomes. With a focus on the media effects of policy outcomes, this research used four independent variables, summarized in table 21, to help account for changes in the policy: local ownership of the newspaper, market saturation (circulation / population), addition of Web coverage and size of the staff.

Table 21: Regression of Change in Newspaper Coverage and Change in Budget, 2005-2010,
Excluding Subregion 4.9, (Unstandardized Coefficients)
 
  Budget Total Economic Development Parks and Recreation Public Safety Public Works
Content analysis -0.00 (0.00) 0.06 (0.07) 0.34*** (0.05) 0.25* (0.14) 0.93*** (0.21)
Local ownership 1.72 (2.86) 0.42 (0.57) 0.46 (0.64) 6.36** (3.04) 1.30 (2.63)
Market saturation (log) -3.88** (1.90) -0.29 (0.37) 0.08 (0.42) 1.84 (2.04) 0.30 (1.75)
Web coverage -3.83** (1.71) -0.16 (0.34) -0.57 (0.38) 1.13 (1.85) -1.09 (1.57)
Staff size -0.11 (0.11) -0.01 (0.02) -0.06** (0.02) -0.22* (0.12) -0.17* (0.10)
Constant 134.74*** (7.02) 0.82 (1.34) 3.57** (1.52) 8.10 (7.38) 11.70* (6.31)
           
R 0.29* 0.13 0.56*** 0.29* 0.42***
R2 0.08 0.02 0.32 0.08 0.18
Adjusted R2 0.04 -0.03 0.29 0.04 0.14
N 121 122 122 122 122
Standard error of estimate 14.21 2.76 3.13 15.16 12.93
           

NOTE: Standard errors are in parenthesis

*Correlation is significant at the 0.10 level (2-tailed)
**Correlation is significant at the 0.05 level (2-tailed)
***Correlation is significant at the 0.01 level (2-tailed)

Size of staff

H2: The size of a newspaper’s staff will correlate positively with policy change. The larger the size of the staff, the larger the positive correlation.

When examining the budget by department, the size of the staff had a statistically significant impact on the changes in the budget, at least at the p<0.1 level, in three of the departments: parks and recreation, public safety and public works. It had no impact when the budget was considered as an aggregate and it had no impact on economic development (but neither did any of the other variables). With an increase in the newspaper staff size, the town’s budget sizes goes down, on average, by 0.22 people for public safety, 0.17 people for public works and 0.06 people for parks, recreation and tourism. As the newspaper’s staff size increases, the town’s budget decreases. Of course, it could mean that a larger newspaper staff is pushing for budget cuts and limited government rather than tax and fee increases as towns are forced to look for additional revenue sources to prevent budget cuts as other sources of funding dry up.

After examining industry trends, this relationship is most likely a reflection of the downsizing of newspapers during the last decade. That is, newspaper staff sizes are decreasing over time and budgets are still, generally increasing, so the relationship between the two is due more to external factors than any of the factors in this study. Still it is significant and worth noting.

It is also worth noting larger newspapers are cutting more personnel by sheer number and as a percentage of their staff. For example, in a single month in 2008, the Palm Beach Post cut 130 people from its staff of 300, 43 percent (Walton, 2010). The 170 people left, despite the mantra of doing more with less, simply could not do the volume of reporting the staff once did even though the budgets of the towns in this study, towns that are presumably similar to Palm Beach, increased by 21 percent between 2007 and 2008 as the population increased by 1 percent.

The papers in this study cut one-fifth of their staffs in the five-year period between 2005 and 2010 slightly less than the one-fourth of the cuts that the American Society of Newspaper Editors demonstrated since 2001. ASNE said 13,700 jobs had been lost nationwide since 2007 demonstrating the potential economic impact of the downsizing of the newspaper industry. As staff members are being asked to do more with less, inevitably more generic copy from news services instead of local reporters is creeping into what little news space remains. Smaller staffs have less time to investigate the actions of government officials, ceasing to fulfill one of the primary roles of the news media. So, even as the town’s budget increases due to population increases or other factors, newspaper staff sizes decrease.

Market saturation

H3: The higher the quality of the newspaper as measured through circulation, the higher the correlation between local coverage and policy change.

Of the 144 newspapers used in this study, only nine saw an increase in circulation between 2005 and 2010. Only 10 saw an increase in market saturation. With an increase in the (log of) newspaper market saturation, the town’s budget goes down on average by 3.88 when analyzed against the aggregate budget. Market saturation seemed to have no impact on the departmental budgets. As the budgets for the towns increase, the newspaper’s market saturation decreases. Given the time period of the study, crossing the time of the Great Recession, it could be that external factors were so strong that regardless of any media coverage, the budgets were going to decline and that those declines came in areas of the budget not a part of this study. Or it could be that media outlets were helping convey a public opinion against increased taxes and fees — as citizens in some foreign countries have done — arguing instead for the downsizing of government.

Web presence

H4: Local newspapers with a Web presence will show a higher correlation between local coverage and policy change.

Newspapers are making more use of their online presence not simply to republish content in the print edition but to supplement it with social media such as Facebook and Twitter. The majority of adults (66 percent) still say they prefer reading a printed version of the newspaper to an online version (Rasmussen Reports, 2012). The vast majority of newspapers in the study (73 percent) not only have a website that is updated at least daily but also make use of social media such as Facebook and Twitter to distribute their news. Mark Zukerberg started Facebook in 2004; Twitter came online in 2006, both since work on this study began. Both have had a profound impact on news media reporting.

The data in this study indicated that, at least when looking at the total budget, with an increase in the use of social media, the town’s budget goes down. The direction of change on the variable is interesting. The variable for Web coverage was categorical: 0 indicated no Web coverage (no papers fell into this category); 1 indicated a static website (no papers fell into this category); 2 indicated a website that was updated only as frequently as the print edition with no social media component (12 percent of papers in the study); 3 indicated a website that was updated more frequently than the print edition with no social media component (15 percent of papers in the study); and 4 indicated a website that was updated more frequently than the print edition with a social media component (73 of the papers in the study). In short, more online coverage, including a social media component, resulted in a decrease in the town’s budget. The result could be due to other external factors as indicated with market saturation, but, all other things being equal, it is worth noting that both market saturation of the print edition and online coverage were both statistically significant and correlated with a decline in the overall budget of the cities studied even if the relationship with each individual department in the budget is less clear.

Ownership

H5a: Locally owned newspapers will have a stronger correlation between amount of coverage of local issues and policy change than newspapers owned by national chains.

The study also showed that local newspapers are following the trends of their larger counterparts, being purchased by larger and larger chains. Chains, 25 of them in all, such as Cox Media Group, Freedom Communications and Lee Enterprises, McClatchy, owned all of the papers in the study. Still, 44 (31 percent) of the papers remained under local ownership, about 10 percent more than the local ownership of papers demonstrated by Eli Noam in his 2009 research. While some research has shown that locally owned newspaper behave differently than those owned by chains, an avenue of further research would be to systematically determine whether locally owned newspapers cover local governments in a similar fashion to newspapers owned by conglomerates and what impact that might have on the functioning of government officials.

Except with public safety, local ownership showed no relationship with changes in the town’s budget. However, the relationship between local ownership and the dependent variable, budget change, was significant (p<0.05) with a coefficient of 6.36 for public safety. As the categorical variable for ownership (0 = non-locally owned; 1 = locally owned) approaches 1, the town’s budget goes up on average by 6.36. While the budget increase affirms the hypothesis, it also points out that the value of local ownership warrants further examination as to why local ownership has any impact on budget change at all much less only on one area of the policy outcomes.

H5b: Local newspapers owned by the same company as the local television station will show a higher correlation between local coverage and policy change.

As it turned out, data related to this variable was nearly impossible to obtain and it was subsequently deleted from the study. The Telecommunications Act of 1996, the first major overhaul of telecommunications law in almost 62 years, changed the landscape of media ownership by allowing any communication business to compete in any market against one another. Prior to 1996, there were major restrictions that prevented owners of media outlets from owning other media outlets in the same market. The goal was to prevent a monopoly in the marketplace of media opinion. Now, only 15 years later, it is common for newspapers and television stations (and radio stations and websites for that matter) to be owned by the same company. As indicated in previous discussion, it is also common for television stations to get their stories out of local newspapers owned by the same company and vice versa, limiting the diversity of opinion in a single market.

Conclusion

This research set out to examine the impact of newspaper coverage on policy outcomes. Given that prior research validated the correlation between newspaper coverage and public opinion, determining what impact newspaper coverage has on policy will provide bureaucrats, politicians as well as reporters and editors with, at least, guidance as to the outer boundary of media impact at the local level. While media outlets may have a strong impact on public opinion, they have significantly less impact on local government policy change. Certainly the study indicates there is an outer boundary to the impact that local newspapers can have. In addition, the impact that media outlets have varies within the budget itself with media outlets having an impact more on public works and parks and recreation with almost no impact on public safety and no impact on economic development. Further, geography matters at least when it comes to changes in the budget with the Great Recession hitting Western states harder. Local ownership positively relates to changes in the town budget. Finally, (log of) market saturation, Web coverage (including social media) and staff size seem to have a negative correlation with changes in the budget over time. With an increase in these independent variables, the town’s budget goes down on average. These relationships warrant further examination as discussed in the next chapter.


[1] The towns in subregion 3.6 saw a budget increase of 12.6 percent in 2005-2010. However, when one outlier, Chattanooga, Tenn., a town that saw a substantial decline in its public works budget over one year, was removed, that subregion saw an increase of 23.1 percent between 2005 and 2010.

INTRODUCTION | LITERATURE REVIEW | METHODOLOGY | INITIAL FINDINGS
EXPANDING THE MODEL | CONCLUSIONS | REFERENCES | PDF VERSION