An assessment of variable newspaper characteristics on the coverage of Move-on anti Iraq war vigils in 2005: The relative impacts of Iraq war issue attention, location type, newshole size, and proximity to a protest

Open Access
Sullivan, Samuel Michael
Graduate Program:
Master of Arts
Document Type:
Master Thesis
Date of Defense:
April 03, 2015
Committee Members:
  • John David Mccarthy, Thesis Advisor
  • Lee Ann Banaszak, Thesis Advisor
  • Roger Kent Finke, Thesis Advisor
  • John Iceland, Thesis Advisor
  • sociology
  • media
  • social movement
  • issue attention
  • location
  • distance
  • newshole
  • Iraq
  • Move On
  • bias
  • selection bias
With the rise of protest event analysis (PEA) newspaper data has become a widely used source of event data. Scholars rely on newspapers in order to create datasets of where, when, and what type of protest events take place. Alongside this expansion of the usage of newspaper data, scholars have identified selection bias as a serious concern, specifically citing that the print media cover only a small percentage of the total number of protests that actually take place. Issue attention (the proportion of a newspaper’s total output that relates to a single issue) is an overlooked aspect of the newspaper landscape that we demonstrate creates biases in PEA data. Additionally, using a public forums classification, we show that the type of space within which a protest event takes place has implications for its likelihood of gaining coverage. Events that took place in public and semi-public forums had significantly higher odds of being reported than events that took place in non-public forums. Validating previous results, we also report that the distance between a newspaper and an event, and newspaper circulation size are both positively associated with reporting likelihood. Uniquely, the current research project relies upon a very large sample of US daily newspapers (N=424) and a sample of protest events that shows very little variation (N=1498); all events were supported by the same organisation (, took the same form (anti-Iraq War Vigils), and took place at the same time (Summer 2005). Given the dyadic structure of the data, we had a total N of 635,043. We thus utilise a uniquely stable, and uniquely large, platform from which to analyse the dimensions of newspaper selection bias. The dataset was constructed from 1) geocoded reports of where MoveOn supported events were held; 2) reports of these events in US daily newspapers, found through NewsBank archives; and 3) a content analysis of US daily newspapers to determine how much attention they dedicated to the Iraq War.