Sample Design

Seven media outlets—2 newspaper, 1 news magazine and 4 broadcasts—were monitored for 120 days—two sixty day periods. The first extended from January 21st, 1993 to March 21st, 1993, inclusive; the second extended from January 21st, 2001 to March 21st, 2001, inclusive.

Outlets were selected to develop a sample of coverage provided by the national press: two newspapers, the New York Times and the Washington Post; one weekly magazine, Newsweek, and four broadcast television programs: ABC World News Tonight; the CBS Evening News; NBC Nightly News; and the PBS Newshour.

Inclusion and Screening

Our sample included stories published or broadcast on odd days of the month (e.g., January 21st, January 23rd, etc.), and all stories published or broadcast on Sundays. All Newsweek issues published during the period were used in the study, including editions published the week after March 21st but with stories based on events that occurred during the study period.

Newspaper stories were drawn from section front pages (i.e. national news, style, metro, Sunday review), editorials and op-ed pieces. Complete newscasts and complete issues of Newsweek were the basis for the sample of television and magazine stories. It should be noted that the PBS Newshour only airs from Monday through Friday, while the three broadcast networks air newscasts every day of the week. Also, there were three pre-empted NBC newscasts in 1993: January 31st, February 7th, and March 21.

For 1993, both print and broadcast sources were monitored via the use of the Lexis-Nexis advanced search tool. Search criteria were designed to cast the widest possible net. Any appearance of the name Clinton qualified a story for inclusion in the original sample. Since NBC was not available on Lexis-Nexis in 1993, their transcripts were purchased from Burrelle's and screened for inclusion.

For 2001, hard copies of newspapers and Newsweek were examined and any story with the name Bush qualified for inclusion in the original sample. For Newsweek, the edition carried by Lexis-Nexis in 1993 was the U.S. edition. This is the same edition we examined in 2001, though it is now called the national edition.

Television newscasts were monitored via Lexis-Nexis, and any story with the name Bush qualified for inclusion in the original sample.

Next, those stories were screened for agreement with the project's inclusion rules. Stories less than 50% about the then-President or less than 100 words long, or the purest form of straight news such as were excluded.

The resulting sample of stories from both 1993 and 2001 was then screened again, and all stories that were the purest form of straight news, such as one-on-one interviews or speech transcripts with no dominating narrative frame were removed.

The resulting project sample consisted of 899 stories, all of which were fully coded and are included in the final data analysis.

In some limited cases Lexis-Nexis will not deliver stories or columns produced by news services unaffiliated with a particular outlet. This study includes all relevant stories under the editorial control of the seven media outlets that were monitored. All stories written by staff reporters, OpEd Pieces, and "specials to the news publication" are part of this analysis.

Coding Process

Researchers worked with a detailed, standardized codebook. All stories were first coded for basic inventory variables—source, date, and placement (front page, editorial, etc.; this variable only applied to newspaper stories). Next, stories were coded for topic, frame, and arena. For these three variables, the simple plurality rule was used: researchers coded each paragraph individually, variable by variable, and the code that appeared most often was used to classify the story. Where two codes appeared with equal frequency, the code that occurs earliest in the story was used.

  • Topic refers to the general subject matter of the story: for example, crime, the environment, or the nomination process.
  • Frame describes the narrative technique used by the journalist (whether consciously or not) to inform readers or listeners about the subject at hand.
  • Arena describes to whom the story related the president. For example, a story about the president's budget (topic) might be written in a way that relates him to Congress (e.g., describing White House staff meeting with Senators) or to interest groups (e.g., reporting on the efforts of lobbyists who support or oppose the plan) or to his cabinet (e.g. how closely did Bush work with his cabinet on final figures compared to other president's. Another possible relationship is that of the office of the president. These are stories that focus internally and look at Bush or Clinton in their new role.

Finally, stories were coded for Theme and for Tone. Theme refers to what journalists were assessing about the president. Three possible themes of presidential evaluation were developed: leadership, an assessment of the president's political skills or decision-making ability; ideology, an assessment of the president's agenda; and character, an assessment of the president's personal mores and system of values. Researchers coded comments from sources (including the president himself) and journalists' innuendoes to determine which theme a story belonged to.

Next, researchers coded each comment and innuendoes pertaining to that particular theme for it tone: positive, negative or neutral. Extra weight was given to text in the headline or lead paragraph of a story. When the ratio of positive to negative comments, or negative to positive comments, equaled or exceeded 2:1 a story was coded as a positive or negative assessment of the president. All other stories were classified as neutral.

All subjective variables were reviewed and confirmed by a senior manager.