![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Political communication scholarship has established the standard operating procedures for national media sourcing of government and politicians. The literature shows a strong reliance by the news media on official and national-level news sources that support the status quo. This paper investigates the less known subject of local newspaper sourcing practices on local House members. House members rely on the local media to communication with constituents but we have little insight into who provides the source material for coverage. Results show that local papers often parallel the national political media by depending on official and nonlocal sources for reporting ostensibly local political angles. Further, members’ press releases, papers’ size, and presence of a Washington bureau help explain local papers’ sourcing practices.
Acknowledgements
This research was supported in part by the National Science Foundation (SES-0318055). I thank Diane Brandt and Qiang Yan for research assistance and Jennifer Wolak for valuable comments.
Notes
1 This paper is considering human sources of news and not other sources such as databases.
2 Obtaining a true random sample of House members is difficult due to restrictions on electronic search engines like Lexis–Nexis. For example, Lexis–Nexis does not include every paper and often researchers are limited to searching only certain time spans.
3 Lexis–Nexis has coverage gaps but it is still the most used means of attaining news stories for analysis (CitationArnold, 2004; CitationEshbaugh-Soha, 2010). To assure a clean data file, coders removed duplicate and early and late version stories, where applicable, from the data. Content analysis was performed by five independent coders. Using a random sample of stories, intercoder reliability agreement score on independent information searching stories was ninety-one percent and on the sources used in the stories was ninety-three percent.
4 See CitationFogarty, 2008.
5 There are also static elements of members that may influence source usage such as party and seniority. Preliminary analysis showed that static elements are not statistically significant predictors of source usage and have been excluded in the final analyses.
6 Press releases were attained from archives of members’ congressional websites.
7 Bill sponsorship data were attained from the House of Representatives website.
8 Circulation information comes from CitationSRDS Circulation (2001).
9 Thanks to an anonymous reviewer for pointing this out.
10 Papers without Washington correspondents also use wire services to acquire information and stories about local members. Yet, only five percent of stories in this study used wire services and there were no systematic differences in their usage.
12 However, they may only be used as sources when the issues at hand present some conflict between the actions of House members and state officials. A confounding factor is that most papers have correspondents located in their respective state capitals, which may negate the effect of papers situated in capitals.
13 Chain newspaper data were attained from individual newspaper and major chain paper websites (e.g., McClatchy and the Tribune Company).
14 Publicly traded newspaper data were attained from individual newspaper company, NYSE, and NASDAQ websites.
15 Since the unit of analysis is individual stories, instead of number of paragraphs, sentences, etc., count models are not appropriate for this dataset. The central concern with using multinomial logit is the assumption of independence of irrelevant alternatives (IIA); an assumption not made by multinomial probit (CitationCameron and Trivedi, 2005). Diagnostic testing using both the Hausmann and the Small-Hsiao tests, showed the model did not violate IIA. This is the case for all three multinomial logit models used in the paper. In addition, multinomial probit is most useful with alternative-specific data and as a typically superior model choice over conditional logit (Long and Freese, 2006). Since the data here are case-specific and do not violate IIA, the multinomial logit model is preferable.
16 The baseline category is “multiple sources,” since it was the largest category of the dependent variable and thus should provide the most reliable results (CitationLong and Freese, 2006).
17 The baseline category is “nonlocal sources” since it is the largest category in the dependent variable.
18 The baseline category is “official sources” since it was the largest category of the dependent variable.
19 The “unofficial sources” versus “official sources” results are not reliable due to the small N for stories with only unofficial sources.
20 Huntsville Item is the smallest paper in the data, while the Los Angeles Times is the largest.