«A Dissertation Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of ...»
Independent Variable. The primary independent variable was the type of political referred to, financial (40.74%) or sexual (59.26%). This variable was labeled financial scandal in subsequent modeling procedures, with financial scandals coded high and sex scandals coded low.
Scandals were considered primarily “financial” in nature if they involved unethical activities surrounding public monies or the use of power to obtain or transfer personal monies, and “sexual”
scandal involved congressmen hiring prostitutes, presumably with sexual intent in mind.
Dependent Variable. A count of scandal news stories was the primary outcome variable.
This variable measured the raw frequency of news stories in the Pew NCI pertaining to financial (M = 189.64, SD = 369.78) and sexual scandals (M = 129.44, SD = 138.69). This variable was heavily skewed (X2 = 32.34, p.001), suggesting the need for non-linear modeling techniques.
Procedure. The first goal of the analysis was to determine whether cases of financial or sexual misconduct receive differential press coverage, controlling for variables that may influence the relationship between scandal type and press attention. Due to the dependent variable being comprised of count data and being positively skewed, the appropriate statistical modeling procedure was a Poisson regression (Long & Freese, 2006). This procedure assumes that the dependent variable cannot take on negative or non-integer values. Data are assumed to be raw counts of a particular phenomenon, in this case, counts of news stories. All negative binomial coefficients are interpreted as the effect of a one-unit change in the predictor variable resulting in a B unit change in the difference in the logs of the expected counts of the dependent variable, holding all other variables in the model constant (Long & Freese, 2006).
Additionally, it should be noted that standard errors were clustered by the year the scandal appeared in the Pew NCI, to eliminate any temporal effects on the dependent story count variable. The primary concern here was to eliminate any changes in the media or political environments that could have occurred between the years 2007 to 2012, including increasing cable TV news and Internet penetration among the American population, the decline of newsroom budgets especially among “hard news” sources such as print newspapers, the rise of tablet computers as a news delivery mechanism, and changes in party leadership among the
Patterson, 2013). Any of these changes may influence relationships analyzed in the model, and thus, clustering standard errors by year the scandal began was a necessary precaution.
A second, exploratory analysis examined how different media sectors and outlets covered scandals at the story level (n = 4,157). This analysis used only Pew NCI data and raw counts of scandals stories as the dependent variable. Media sector – newspapers, radio, network TV, cable TV, and online news outlets – and outlet (e.g., Fox News Channel, CNN) were used as the independent variables. Here the goal was to determine simple differences in scandal coverage by sector and outlet, determining whether media sector and partisan bias drove patterns of overall coverage. Since news tends to follow seasonal patterns, with the news agenda “congested” during some periods and not others, this analysis assessed scandal coverage by year.
Effect of Scandal Type. This chapter hypothesized that qualitative differences among political scandals would affect media coverage. Specifically, it was predicted that sexual scandals would garner more media attention than financial scandals (H1). The Poisson regression model shown in Table 1 provides evidence in support of HI. The results shown in Table 1 suggest that sexual scandals garnered more coverage than did financial scandals (b =-1.06, SE = 0.31, p.001). Indeed, the rate of story counts decreases by 85.81 going from the average sex scandal to the average financial scandal, with all other variables in the model held constant.
A number of other control variables were strong predictors of scandal news coverage.
Official resignations within two weeks of scandal revelations related to fewer stories being published about a scandal (b =-2.05, SE = 0.52, p.001). Scandals reflecting on members of the Judicial Branch also generated less coverage than gubernatorial scandals in the excluded baseline
they were covered (b =-1.86, SE = 0.28, p.001), if they involved a criminal misdemeanor or felony charge (b =0.53, SE = 0.21, p.05) and if the accused official actively sought scandal publicity (b =1.75, SE = 0.44, p.001), as was the case with former Illinois Governor Rod Blagojevich, who conducted an aggressive PR campaign to address claims of bribery.
Poisson Regression Predicting Frequency of Scandal News Stories
Note: Standard errors are clustered by the year the scandal first made national media attention to mitigate extraneous temporal effects. The model’s proportional reduction in error (PRE) of 75.90% suggests that the model explained a considerable amount of variation in the outcome variable, above and beyond a model without independent variables entered. A two-tailed t-test also showed a significant effect of the scandal type variable and stories with scandal mentions.
***p.001, **p.01, *p.05, †p.10
Note: Frequencies derived from Pew News Coverage Index (NCI), covering 27 cases of political scandal and 4,157 stories from Jan. 1 2007 to May 31 2012. A similar figure using proportions of stories produced a highly similar pattern of results.
sexual and financial scandal coverage across five media sectors. A series of one-tailed t-tests were used to determine differences in the proportion of coverage a medium (e.g., cable TV) gives to a particular type of scandal. A one-tailed t-test revealed that radio news outlets did
Counter to the prediction, there was no difference in how newspapers covered financial and sexual scandals (p = 0.37). H2A is partially supported.
The result also show that online news t(4,155) = 1.66, p.05 and network television outlets t(4,155) = 1.49, p.10 provided more frequent coverage of sex scandals than financial scandals. Counter to the prediction, there was no significant difference in how cable television news covered financial and sexual scandals (p = 0.80). H2B receives partial support.
Coverage by Media Sources. Financial scandals should receive more frequent coverage from sources that specialize in disseminating financial information and substantive public affairs news (H3A). Sources like The New York Times, The Wall Street Journal, the PBS Newshour with Jim Lehrer, and National Public Radio news programs like Morning Edition should pay more attention to official financial misconduct than sex scandals. These outlets tent to cater to older, high-income audiences that seek “hard news” about elections, public policy, and foreign affairs (Hamilton, 2004). Moreover, the ample newsroom budgets of the Times and Journal may provide enhanced opportunity to do original investigative reporting into the complex, bureaucratic world of official financial misconduct. PBS and NPR, being non-profit media organizations, should not have a profit-seeking motive to frequently report about salacious sexual misbehavior. As predicted, hard news sources like NPR’s Morning Edition and Talk of the Nation provide more coverage to financial scandals than sex scandals (see Figure 3).
A series of one-tailed t-tests comparing the proportion of financial and sexual scandal stories each source provided demonstrates a prevailing focus on financial scandals among “hard news” outlets. The PBS Newshour with Jim Lehrer t(4,155) = -3.71, p.001 and NPR’s Morning Edition and Talk of the Nation t(4,155) = -3.46, p.001 both provided more coverage of financial
revealed that The New York Times (p = 0.55) and The Wall Street Journal (p = 0.32) provided roughly equal amounts of coverage to financial and sex scandals. One potential explanation for these divergent findings is that the lack of a profit motive allows NPR and PBS to focus on “boring” and detailed cases of financial misconduct, while The Times and The Journal’s profit
Frequency of Financial and Sexual Scandal Stories by “Hard News” Source Note: Frequencies derived from Pew News Coverage Index (NCI), covering 27 cases of political scandal and 4,157 stories from Jan. 1 2007 to May 31 2012.
and financial misconduct, how do purely “soft news” sources such as NBC’s The Today Show handle scandal coverage? It was predicted that sex scandals should be particularly appealing fodder for “soft news” sources like NBC’s The Today Show, CBS’s The Early Show, ABC’s Good Morning America, and The Huffington Post news website (H3B).
Sources like The Today Show have substantial news holes that they must fill with original and entertaining programming to attract audiences and maintain ratings (Hamilton, 2004). Often this leads to a focus on provocative and sensational stories about crime, celebrities, gossip and scandal (Patterson, 2013). These “soft news” sources have little incentive to cover substantial government corruption and financial misconduct. Instead, they must provide more infotainment, or entertaining news programming, in order to compete for audience attention among other news outlets and pure entertainment sources such as made-for-TV movies, fiction, online porn, video games and so forth (Fallows, 1997; McChesney, 2004; Patterson, 2013). With so many alternative sources of entertainment vying for audience attention, “soft news” sources must constantly focus on sensational stories, particularly sex scandals to bring in viewers.
As shown in Figure 4, all four “soft news” sources gave more overall coverage to cases of political sexual misconduct than financial misconduct. The evidence supports H3B. A series of one-tailed t-tests were conducted to determine differences in financial and sexual scandal coverage by source. The t-tests showed significantly more sex scandal coverage by The Today Show t(4,155) = 2.21, p.05, The Early Show t(4,155) = p.10, Good Morning America t(4,155) = 3.58, p.001, and The Huffington Post news website t(4,155) = 2.69, p.01. Across all sources, sex scandals were given more overall attention than financial scandals. An alternative figure
Frequency of Financial and Sexual Scandal Stories by “Soft News” Source Note: Frequencies derived from Pew News Coverage Index (NCI), covering 27 cases of political scandal and 4,157 stories from Jan. 1 2007 to May 31 2012.
there is partisan bias in coverage (Feldman, 2011; Morris, 2005; 2007; Stroud, 2011). Due to the editorial slant of particular cable TV news outlets, coverage of ordinary political events becomes distorted, interpreted through the partisan lens of the outlet’s producers, reporters, and anchors.
Proportion of CNN and Fox News Stories about Scandal Note: The heavy blue line represents the proportion of CNN stories about the Larry Craig sex scandal and the proportion of CNN stories about the Solyndra scandal. The heavy blue line represents the proportion of Fox News stories about the Larry Craig sex scandal and the proportion of Fox News stories about the Solyndra scandal. The X-axis represents time – Aug.
2007 to Oct. 2007 (Craig), Sep. 2011 to Jan. 2012 (Solyndra).
Democratic and Republican scandals. This test assessed mean differences in Republican and Democratic scandal stories broadcast by Fox News Channel and CNN (n = 1,166). Assuming that Fox typically represents a conservative Republican viewpoint and CNN is more likely to provide a liberal Democratic viewpoint (Feldman, 2011; Morris, 2005; Stroud, 2008; 2011), the
(H4A), and CNN will pay more attention to scandals reflecting poorly on Republicans (H4B).
The results show support for H4A and H4B. A one-tailed t-test shows a significant difference in scandal stories t(1,164) = 3.42, p.001, with CNN providing more coverage (M = 0.31, SD = 0.46) to Republican scandals than Fox (M = 0.22, SD = 0.41). This test also suggests the inverse, that CNN paid less attention to Democratic scandals (M = 0.69, SD = 0.46) than did Fox News (M = 0.78, SD = 0.41). The results suggest that there is partisan bias in scandal coverage, with opposition officials receiving more scandal news than likeminded officials.
Two cases that illustrate partisan bias in scandal coverage are represented in Figure 5.
This figure shows the proportion of all CNN and Fox News stories that specifically discussed former Republican Senator Larry Craig’s sex scandal and the Solyndra controversy plaguing the Democratic administration of President Obama. When the scandal hurt Republicans (Larry Craig), the blue trend line representing CNN scandal stories outpaces the red trend line representing Fox stories. The reverse is true of scandals that hurt Democrats (Solyndra).
In the case of Larry Craig, who was arrested for soliciting sex in an Idaho airport men’s room (New York Times, 2007), CNN provided an initial surge of coverage followed by sporadic attention in the succeeding months. Fox downplayed the Craig scandal relative to CNN, providing fewer initial stories and lighter coverage long term. The Solyndra controversy, which involved a possibly illegal government restructuring of a $535 million loan to the Solyndra solar energy company (Stephens & Leonnig, 2011), demonstrated an opposite pattern. CNN paid little attention to Solyndra relative to Fox, who kept the story alive with continuing coverage.
It is also notable that the initial coverage of the Larry Craig sex scandal was more intense than the initial coverage of Solyndra. While many other factors relating to time and “agenda