In the Public Interest?

Sample Design

Fourteen media outlets -- 5 newspapers and 9 broadcasts--were monitored for two separate seven-day periods. The first monitoring period began on Thursday, December 9, 1999 and continued through the inclusion of Wednesday, December 15, 1999. The second monitoring period began on Thursday, January 13, 2000, and continued through the inclusion of Wednesday, January 19, 2000.

Newspapers were selected to develop a sample of coverage provided by both national press (New York Times, USA Today, Washington Post) and regional publications (Cleveland Plain Dealer, Orlando Sentinel) that represented geographic and demographic diversity. Broadcast sources included both the flagship program and the morning news show from the three major over-the-air networks; the Newshour to represent public broadcasting; and CNN's The World Today and The Larry King Show as a sample of content on cable all-news networks.

Inclusion and Screening

Both print and broadcast sources were monitored via the use of the LEXIS-NEXIS advanced search tool. Search criteria was designed to cast the widest net possible. Any combination of the root phrase president! and campaign, candidate, candidacy, or election, or any mention of George W. Bush, John McCain, Bill Bradley, or Al Gore qualified a story for inclusion in the original sample. These criteria insured the inclusion of all pertinent stories; yet also produced an initial sample of more than 1,000 articles or broadcast segments, many unrelated to the president campaign. First, the sample was refined by eliminating duplicate stories, photo captions, and letters to the editor. Next, those stories in which less than one-third of the text was devoted to the 2000 presidential campaign were eliminated. (For example, if a story quoted the president of a company re: the advertising campaign of a competitor, it appeared in the original sample, but was then deleted.) In addition, the Vanderbilt Television News archives were downloaded for the dates within the two monitoring periods. The broadcast stories in the sample were compared to the day-by-day log for each of the network evening news programs to insure that all pertinent stories were included.

The resulting project sample consisted of 430 articles, all of which were fully coded and are included in the final data analysis. In some limited cases, the LEXIS-NEXIS database will not deliver stories written by unaffiliated news services. However, all stories under the editorial control of the newspaper are included. All stories written by staff reporters, OpEd pieces, and "stories written by unaffiliated news services. However, all stories under the editorial specials to the newspaper" are part of this analysis.

Coding Process

Researchers worked with a detailed, standardized coding scheme. All stories were first coded for basic inventory variables -- source, dateline, length, etc. For print stories, length was recorded from word counts as provided in the LEXIS-NEXIS database. For broadcast stories, coders timed themselves as they completed their initial reading of the article. Thus, story length for broadcast stories does not reflect the actual minutes/seconds need to present the story within the news show; the figures given serve only comparative purposes. Items read strictly by an anchor that timed out at less than 20 seconds were not included in the analysis.

Next, stories were coded for content variables -- recurring leads, dominant figure, and general topic. Finally, coding examined the intent variables -- story trigger, frame, frame tone, impact and speculativeness.

In all cases, coders worked with a defined set of rules per variable. Of particular note:

• Frame Tone: The 1.5 to 1 Rule
  When calculating Frame Tone, coders must quantify all the pertinent text that is positive for the Dominant Figure, as well as all pertinent text that is negative for the Dominant Figure. Additional weight is given to text within the headline, lead, or first six sentences of the story. In any case where the ratio between positive:negative equals or exceeds 1.5:1.0, the story is coded as positive tone for the Dominant Figure. Likewise, when the ratio between positive:negative equals or exceeds 1.0:1.5, the story is coded as negative tone for the Dominant Figure. All other stories are coded as neutral.

• Story Impact: The One-Half Rule
  When calculating Story Impact, coders identify all text that implies which individual or group interests are at stake/affected by the events in the story. If 50% or more of the text makes this connection, the story is coded as impacting that individual or group. If multiple groups are impacted, but one did not dominate, stories are coded as "No impact implied at 50% or more."

Intercoder Reliability

Intercoder reliability measures the extent to which individual coders, operating independently of one another, reach the same coding decision. Tests were performed throughout the project: no systematic errors were found. In addition, the coding supervisor reviewed all decisions on the intent variables and where necessary, made changes to bring all coders into agreement.