![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
Investigative journalism’s value to democracy is straightforward: it provides useful information to citizens by exposing wrongdoing and holding powerful institutions accountable. But its financial value is questionable. There are indications that this often-expensive form of reporting can enhance audiences and thereby increase revenue, but very few large-scale projects have examined the connection between investigative content and audiences. Numerous studies have established a link between news quality and newspaper circulation, yet television studies of quality and audiences are less definitive. The present study addresses these research gaps, using Lacy's Model of News Demand to explain the interaction between investigative quality and U.S. local television audiences. Results show that investigative quality is strongly associated with larger television audiences, even when controlling for factors such as market competition and organizational structure. Investigative stories that disclosed concealed information were particularly predictive of audience size. However, stations producing a higher quantity of investigative stories were not associated with greater viewership. This suggests that investigative quality might be a more important audience motivator than quantity. The findings are discussed in light of the possibility that quality investigative journalism could provide economic as well as democratic benefits.
Acknowledgements
The authors would like to thank Professor Stephen Lacy for his instrumental help and valuable advice on this project.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1 Lombard and colleagues (Citation2010) and Wimmer and Dominick (Citation2010, 172) recommend a reliability subsample of at least 50 units comprising no less than 10% of the total sample.
2 The principal investigator served as a “tie-breaker” among coders so that units in the reliability subsample could be properly incorporated into the main coding sample (Lombard, Snyder-Duch, and Bracken Citation2010).
3 The affiliate total was greater than 80 because some stations provided news for multiple affiliates in the same market.
4 The final Market Size (M = 2.81, SD = .42), Competition Intensity (M = -2.74, SD = 1.28), and Audience Size (M = .00, SD = 1.00) variables were transformed for analysis purposes (see data analysis strategy).
5 Significant outliers (p < .001) were cases with values that were more than 3.29 standard deviations from the mean (Tabachnick and Fidell Citation2007, 96).
6 Transformations were performed for variable distributions within the main sample (N = 80) and the quality subsample (N = 38). The most appropriate transformation was chosen for each measure (see Tabachnick and Fidell Citation2007, 86–88). The Competition Intensity variable was negatively skewed, and therefore it was reflected before transformation. Square root transformations were performed for the Audience Size and Competition Intensity variables, while a logarithm transformation was performed for the Market Size variable.
7 All cases in the final data set produced Cook’s Distance scores of less than 1.00, and thus no multivariate outliers were suspected (Tabachnick and Fidell Citation2007, 75).
8 The Quality Index was positively correlated with both Concealed Information (r = .70, p < .001) and Public Interest (r = .77, p < .001). This was expected because the index was composed of both individual quality variables. A test indicated low tolerance and high variance proportion scores for all three quality predictors when used in a single regression model, and therefore multicollinearity was suspected.
9 This analysis approach follows Belt and Just (Citation2008), who separated similarly related quality predictors into different regression models.
10 The condition index for one dimension in the two quality regression models was greater than 30, which can be a sign of variance inflation. However, no dimensions in either regression had multiple variance proportion scores greater than .50. Additionally, the minimum tolerance score for any predictor in the quality analysis was .600, and the maximum variance inflation factor score was 1.667. Thus, no multicollinearity was suspected in the final models (Tabachnick and Fidell Citation2007, 88–91).
11 A Levene’s test indicated unequal variances for the two groups in the analysis. This is reflected in the reported t-value and degrees of freedom.
12 For instance, Stone, Stone, and Trotter (Citation1981) used a similar data analysis and found that news quality accounted for just 3% of the variation in audience size.