Embedded Reporters

The study examined morning and evening newscasts on March 21, 22, and 24, 2003.

For the morning newscasts, it examined the period of 7:00 A.M. to 9:00 A.M. on each day. The specific programs studied were ABC's Good Morning America, The CBS Early Show, NBC's Today Show, CNN's American Morning, and Fox News' Fox & Friends. To remain consistent, it did not include a third hour of the Today Show or American Morning.

For the evening newscasts, the study examined the first 30 minutes of ABC World News Tonight, the CBS Evening News and the NBC Nightly News on each night. For CNN and Fox News, on Friday (March 21) and Monday (March 24), it looked at the signature pre-prime time news shows of Wolf Blitzer Reports (5:00 P.M. to 6:00 P.M) and the Fox Special Report With Brit Hume (6:00 P.M. to 7:00 P.M.). On Saturday, March 22, it examined CNN and Fox Coverage from 6:00 P.M. to 7:00 P.M.

The morning of Saturday, March 22nd, the CBS affiliate in Washington D.C. aired local news from 7 a.m. to 8 a.m. and CBS network news from 8 a.m. to 9 a.m. The local news was not included in the study. Also on Saturday, due to the NCAA Basketball Tournament, which aired on CBS, the CBS Evening News was abbreviated to an update totaling 2 minutes and 30 seconds. The study included that update, which contained one embedded report, in our study.

All broadcasts were monitored from videotaped segments of the programs as they aired in the Washington, D.C. market.

Story Selection

Within the hours selected, the study examined all stories from embedded reporters that were 20 seconds or longer. Embedded reports were identified through the station's own on-air identification and secondarily through a list published by the Washington Post, "The Network's Heavy Artillery: Who'll Win the Battles of Tag Lines & Toys," John Maynard, March 21, 2003.

CBS reporter Lee Cowan was not officially categorized by the military or CBS as "embedded." But in two stories, he temporarily attached himself with a unit and produced pieces so similar in nature to other embedded pieces that these pieces were included.

This resulted in a project universe of 108 stories, all of which were fully coded and are included in the final data analysis.

Coding Process

A team of four researchers worked with a detailed, standardized codebook. Each broadcast was initially screened and pre-coded by a single coder. The pre-coding process confirmed whether a story was embedded or not. For embedded stories, the date, outlet and timeslots were determined. Next, stories were fully coded by a single coder via multiple story viewings. All stories were coded for the following variables:

• Live Versus Taped: Whether the report was completely live, had some editing or was fully taped.
• Place of Origin: The country from which the reporter filed his or her report. Embedded reporters on ships were noted as being at sea.
• Topic: The topic that received the most seconds of coverage in the story. If two topics received equal attention, coders deferred to the lead topic.
• Source of Information: This is a hierarchy variable that captured the highest authority used as an on camera source within a story.
• On Camera Action: This is a hierarchy variable that captured how close the story's visual images came to depicting front line of action.
• Level of Reporting: This indicates whether a story is mainly fact, analysis, opinion, or color commentary by tallying the total number of statements for each category, and seeing which category predominated for that story. If two types of statements received equal number of tallies, coders deferred to the lead statement.
• Length in Seconds: This variable captured the total length of a story, rounded to the nearest five seconds. Lead-ins were not included.

Researchers analyzed each news story in its entirety, working through the sequential variables. Project Rules for Coders were established prior to beginning that process, and were applied during all phases of coding. Intercoder reliability was established by training and testing the group beforehand. A series of test stories were fully coded by each member of the coding team and then results were compared until the team was coding each variable identically. In addition, all subjective variables were reviewed and confirmed by a senior manager. No systematic errors were discovered.