Women and Visual Depictions of the U.S.-Iraq War in Print and Online Media

Abstract

Both American and Iraqi women were affected by the U.S.-led invasion of Iraq by coalition forces in March 2003. Yet women were shown in less than one-fifth of the 480 war-related photographs in a sample from 18 U.S. daily newspapers, three U.S. news magazines, and those publications’ websites. In addition, Iraqi women were less likely to appear than U.S. women, partly because of the news media's intense focus on injured American soldier Jessica Lynch during the early weeks of the war. This study suggests that these findings may have been the result of both media routines, as described by Shoemaker and Reese, and the tendency of the U.S. media to engage in Orientalism, as described by Said.

Susan Keith is an assistant professor in the Department of Journalism and Media Studies, School of Communication and Information at Rutgers University, New Brunswick, NJ.

Carol B. Schwalbe is an associate professor in the Walter Cronkite School of Journalism and Mass Communication at Arizona State University and a member of this journal's editorial board.

Notes

∗Because of rounding, total adds to more than 100%.

1. In January 2006, Temple University mathematics professor John Allen Paulos, author of Innumeracy and a Mathematician Reads the Newspaper, pointed out that if the Lancet count rose in proportion to the Iraqbodycount.net estimate, which climbed from 17,000 to 30,000 civilian deaths between 2004 and 2006, the Iraqi toll might have been as high in early 2006 as 175,000 (Paulos, 2006).

2. The authors formed the newspaper sample by first selecting the eight general-interest newspapers with average weekday circulations over 500,000, as reported in the Editor & Publisher International Yearbook (2000). To give the sample greater geographical and corporate diversity, the authors then added the nation's largest nonchain newspaper (the St. Petersburg Times) and one large newspaper from each of several newspaper chains not represented by more than two newspapers in the 500,000-plus circulation group. Those newspapers and the corporations that own owned them in 2003 were The Arizona Republic (Gannett Co. Inc.), The Atlanta Journal-Constitution (Cox Newspapers), The Dallas Morning News (Belo Corp.), The Miami Herald (Knight Ridder), The Oregonian (Newhouse), the Rocky Mountain News (E.W. Scripps Co.), The Sacramento Bee (The McClatchy Co.), The Tampa Tribune (Media General), and The Virginian-Pilot (Landmark Communications Inc.).

3. Newsweek, Time, and U.S. News & World Report.

4. Although “women and girls” or “men and boys” is the preferred terminology, according to contemporary gender theory, we use the plural words “females” and “males” as shorthand to represent both adult and juvenile women and men.

5. One of the authors coded 221 newspaper front-page photos and 17 magazine-cover photos; the other author coded 312 web homepage photos. Of those 550 photos, 87% (480) contained images of people. A graduate student coded 10% of each group of images to test intercoder reliability, which was 1.00 for the print images and .92 for the web images using Scott's pi.

6. If several images were the same size, no image on that page was coded as dominant. This happened only rarely, however, as print page designers, especially, are encouraged for aesthetic reasons to make one image larger than the others (Harrower, 2007).

7. U.S. females, Iraqi females, U.S. females with Iraqi females, other females (neither U.S. nor Iraqi), females whose nationality could not be determined, Iraqi females with females of indeterminate nationality/ethnicity, U.S. females with females of indeterminate nationality/ethnicity, Iraqi females with other females (neither U.S. nor Iraqi), U.S. females with other females (neither U.S. nor Iraqi), or females of indeterminate nationality/ethnicity with other females (neither U.S. nor Iraqi).

8. A graduate student coded 36 of the photographs containing women to test intercoder reliability. Given the small population of only 85 photos in which women appeared, Scott's pi should not be based on a 10% sample of only nine photos. Instead, we calculated the sample size needed to match the standard error of the 10% sample used with all 550 photos. That sample size was 36. Scott's pi from the random sample of 36 out of 85 photos was 1.00 for both print and web images.

9. In both cases, newspaper and magazines images were collapsed into one category to avoid unacceptably low expected cell counts. The content categories of females and individuals of unidentifiable gender and males and individuals of unidentifiable gender were collapsed for the same reason.

10. This group was made up of the images that showed exclusively females, males and females, and females and people whose gender could not be determined.

11. This group was made up of the images that showed exclusively males, males and people whose gender could not be determined, and exclusively people whose gender could not be determined.