How MRC Conducted Its Labeling Study

June 25, 2002

     To test Bernard Goldberg's observation that network news reporters label conservatives more frequently than liberals, the MRC conducted a study. MRC researchers used Nexis to search three source files: ABC News, CBS News, and NBC News. The goal was to find all instances when a network reporter used "liberal" or "conservative" in a U.S. political context in a story for one of the three evening news broadcasts during a specified time period, January 1, 1997 to December 31, 2001, a time when all of the relevant programs were available on Nexis.

     A total of ten Nexis searches during this time frame were conducted, checking records when the word "liberal" or "conservative" was paired with the name of the appropriate evening news broadcast. Since ABC's weekend newscasts are given a different name than their weekday counterparts, additional searches were required to ensure that all relevant records were examined.

     In about fifteen minutes, these search criteria produced 2,020 records - 1,501 from the term "conservative" and 519 with the term "liberal." By itself, this quick search indicated that conservative labels were used far more than liberal labels, but each of the 2,020 records were printed and MRC analysts began the time-consuming chore of examining the context in which each label was used.

     Once each story was printed, it was obvious that many of these records could be quickly eliminated. Some of the stories were from programs other than the specified evening news broadcast and therefore fell outside the study parameters. One way this happened was when an anchor for a show such as ABC's This Week would inform viewers about a story that would appear later on, say, World News Tonight. If that file also contained the word "liberal" or "conservative," then Nexis returned it as a valid hit. Discarding these shows eliminated 308 records, or about 15 percent of the total.

     Additionally, many newscasts were entered in Nexis's database two or even three times, producing a substantial number of duplicate records. Another 387 records (19%) were discarded for this reason, bringing the number of valid records down to 1,325 - 999 using "conservative" and 326 mentioning "liberal."

     Looking at the actual usage of the words, the MRC analysts removed any records in which the search terms were not used as ideological labels. CBS anchor Dan Rather, for example, used the word "conservative" on November 23, 1999, in a story about the start of Hillary Clinton's New York Senate campaign, but not to describe the then-First Lady or her GOP opponent: "By the way, the most conservative estimates of what the combined cost of this New York Senate race will be are $50 million or more."

     Similarly, in a story about declining interest rates on November 27, 2001, NBC's Anne Thompson referred to "conservative savers, especially retirees feeling the squeeze, [with] one-year CDs falling from 5 1/2 percent to just a little over 2 percent in the past 12 months." The word liberal also popped up in a number of non-political contexts, such as when ABC's David Wright reported on February 17, 2001 that "some 300 small liberal arts colleges nationwide already have made the SAT optional." By going through the stories individually, MRC's researchers were able to exclude another 105 records, or 5 percent of the original sample.

     Since the MRC study was focused only on the use of liberal and conservative labels within the U.S. political context, researchers identified and excluded labels that were outside that context. On November 19, 2001, for example, CBS's Elizabeth Palmer referred to an Afghan warlord, Haji Abdul Qadeer, as "a deeply conservative Islamist" - a label, but not one that belonged in this particular study. A total of 124 records where the search terms were only used in a foreign context were eliminated, reducing the sample by another 6 percent.

     Finally, since the object of the study was to strictly measure only the reporter's independent use of labels, any use of "conservative" or "liberal" in a sound bite by a news source was excluded, as were instances in which the reporter attributed the label to some other source. For instance, on October 22, 1998, ABC's Dean Reynolds reported on TV commercials "by the Republican Party that paint [Wisconsin Senator Russ] Feingold as a tax-happy liberal." In this story, Reynolds was plainly not labeling Feingold himself, so this and similarly attributed labels were not included in the study.

     Similarly, on February 23, 2000, NBC's Fred Francis showed a sound bite of George Bush saying, "We who stand on conservative principles have got compassionate hearts for the people who live in this country." That self-assigned label didn't count, but when NBC anchor Tom Brokaw announced on July 18, 2000, that "George W. Bush [is] continuing his push for programs in the compassionate conservative mode," it did count, since the NBC anchor himself applied the label. This eliminated another 172 records from the study, or about 8 percent.

     That left a total of 924 records in which network reporters applied the conservative or liberal label to a U.S. politician, policy position, or group. Nearly four-fifths of these stories (78%) included only a single label; others, such as a CBS Evening News story by Bob Schieffer on May 24, 2001, included multiple labels.

     Referring to the consequences of Vermont Senator Jim Jeffords' decision to put Democrats in charge of the U.S. Senate, Schieffer - in a rare model of balanced labeling - announced that, "among expected changes: liberal Pat Leahy, in place of conservative Orrin Hatch, as chairman of the Judiciary Committee, which could mean a tougher time for the President's court appointments; liberal Carl Levin takes over Armed Services from conservative John Warner; liberal Joe Biden replaces conservative Jesse Helms at Foreign Relations." Because the search for liberal labels was conducted separately from the search for conservative ones, this story was credited with three labels in each category.

     When multiple uses of labels were taken into account, the final tally was 247 liberal labels and 992 conservative ones. While the original Nexis search also demonstrated a wide disparity in the use of labels, using those raw numbers would have overstated by nearly 40% the amount of labeling by reporters at the three networks. It also would have understated the degree to which conservatives are more often labeled - while the original count produced a roughly three-to-one disparity, the final count shows that reporters are actually four times more likely to label conservatives than liberals. -- Rich Noyes

• Read the Media Reality Check

• See previous MRC labeling studies