A Closer Look at a Famous Study Linking TV Exposure to Eating Disorders

January 26, 2011

I have researched the links between body image, mass media, dieting, eating disorders, and body dissatisfaction for many years. I first became interested in the topic after reading Naomi Wolf’s best-seller The Beauty Myth , in which Wolf claimed that images of thin women in the mass media caused, or at least contributed to, eating disorders in women. Over the past 20 years this claim has been widely repeated and accepted as fact, though both my background in psychology and my research into the mass media (the subject of my 2003 book Media Mythmakers ) told me that something seemed fishy about this claim.


When you research a subject like this in depth, you find that certain studies keep popping up as offering especially strong evidence. These studies are widely cited and referenced in both popular and scholarly discussions of the subject.


One of the most famous studies of the effects of mass media on girls involved teens on the island of Fiji, and what happened to them after television was introduced to the island in 1995. It is considered especially important and insightful because it allowed a very rare glimpse into a culture that had not been previously exposed to television. The study is often described as concluding that television had created anorexia-like behavior or body image disturbance. Here’s how one writer, Sylvia Lerigo, cast the story:


Compared to the past, when simply “being” was the priority, Fijian teens had become as obsessive about body image and weight as teenage girls in the West who were suffering with conditions like anorexia. Prior to this invasion of television, the Fijian female body was considered more attractive when fuller and naturally feminine, which Fijians considered ideal for childbearing and child rearing. Becker noticed that almost 74% of young women in Fiji considered themselves “too fat” and thinness was what most Fijian teens craved.


What about this hard-hitting study offering solid proof of a connection between media images and eating disorders-written by a Harvard doctor, no less?


Let’s take a closer look at this widely-cited study to examine what, exactly, it found.


In Dr. Anne Becker’s famous 2002 study (“Eating behaviours and attitudes following prolonged exposure to television among ethnic Fijian girls,” British Journal of Psychiatry , 180: 509-514) Becker and colleagues visited Fiji in1995, a few weeks after television was introduced to the island, and returned in 1998, three years later.


Each time they asked school-age Fijian girls (average age 17) about how much TV they watched (and their attitudes about the programming), and questions about their bingeing and purging behaviors. High scores on an eating disorder scale [EAT-26] increased 12.7%, and induced vomiting to control weight increased 11.3% between 1995 and 1998. Becker concluded that “key indicators of disordered eating were significantly more prevalent following exposure to television.”


The Fiji studies are often cited as strong evidence that exposure to thin images in the mass media are linked to eating disorders, and have implications for American teen girls as well. There are several important caveats that should be kept in mind when considering the study’s validity that are rarely mentioned, and only become apparent upon a close reading of the research. For example:


1) Becker’s study involving Fijian girls had a total of only 63 participants in 1995, and 65 in the 1998 follow-up. This is a very small sample size, and makes the study’s conclusions difficult to generalize. A sample size of at least a few hundred subjects would provide much better evidence that the findings were valid. Fiji had a population of around 85,000 people in 1995, and it’s not at all clear that a sample of 64 teen girls represents the population as a whole.


It should be noted that only one-third (29.2%) of the 1998 sample, or 19 girls, scored high on the eating disorder behavior test. So Lerigo’s claim that ” Fijian teens had become as obsessive about body image and weight as teenage girls in the West who were suffering with conditions like anorexia” seems misleading at best.


There’s also the puzzling fact that while Becker takes pains to state that eating disorders were essentially unheard of in Fiji prior to the introduction of television, she found 8 girls (out of 63) who scored high on the 1995 test, a mere two or three weeks after television was introduced to the island. This group is considered the “before TV exposure” group, yet 12.7% of the girls apparently scored high on the test with little or no television exposure. Yet after three years of exposure to the mass media’s constant thin images and unhealthy messages broadcast to tens of thousands of people, only 12 additional girls scored high on the test. While an increase from 12.7% to 29.2% is statistically significant, is seems like a curiously small increase if the mass media is as powerful as often claimed.


2) The girls, though claiming to diet and wanting to emulate the bodies of thin TV actresses, did not actually lose weight between the 1995 and 1998 surveys. Becker notes, “there were no significant differences between the samples in mean age or body weight” (p. 510) (in fact, if anything the girls got slightly fatter, as the average BMI went from a 24.5 to a 24.9). Thus the Fijian girls were a hair’s breadth away from being overweight (overweight is a BMI of 25) both before the introduction of TV and three years later, thus there’s no evidence that exposure to the TV images led to any weight loss or anorexia.


This puts Lerigo’s comment that ” Becker noticed that almost 74% of young women in Fiji considered themselves ‘too fat'” in a whole different light. If you hadn’t read the study, it would seem to suggest that these thin or normal-weight girls had been brainwashed into believing they should be thin like Jennifer Aniston. Instead, the reason that 74% of Fijian girls considered themselves “too fat” is because they probably were too fat! This has nothing to do with thin models in mass media and everything to do with the high-fat typical Fijian diets.



3) Fijians were incredibly naive about television. According to Becker, the girls they interviewed believed that TV shows like Seinfeld , E.R ., Home Improvement , Beverly Hills 90210 , and Friends were news stories and reality shows. They did not understand that the people they were seeing were actors. They could not understand the difference between a scripted sitcom and real life, yet they were being asked complex questions about their attitudes regarding the TV characters. On this important measure alone, American teen girls are polar opposites from the Fijian girls. Young women are very media savvy and certainly don’t mistake Grey’s Anatomy for a news report. This seriously undermines attempts to draw parallels between the Fiji study and the media’s influence in modern America (something most writers on the issue, including Lerigo, specifically do). Becker herself points this out: “Generalization about the impact of television upon Fijians to other populations requires caution” (p. 512).


4) Becker did not determine whether or not the 65 girls in the 1998 follow-up study had disordered eating symptoms or attitudes before 1995. Because the same population of girls was not used (which would control for this variable), there is no way to know whether or not television caused the increase or not. Put another way, for all Becker knows, the girls she interviewed in 1998 might have scored high on the disordered eating test before they ever watched television, if Becker had tested them in 1995. Furthermore, even though television was only introduced to Fiji in 1995, mass media images of thin models and actresses were seen in Fiji many years earlier in the form of magazines, newspapers, films, and so on. So it was not, as often claimed, a culture that had little exposure to the “thin ideal.”


5) Becker et al. apparently did not take into account various other possible confounding factors that
co-occurred in Fiji between 1995 and 1998. The introduction of television was surely one important influence on teens, but was it the only one? What other social, cultural, economic, religious, or other factors were changing at the same time that television was introduced? It’s not clear how Becker et al. determined that other factors did not play a role in self-perception and body dissatisfaction. Correlation does not imply causation, and just because television was introduced before the additional 12 girls scored high on an eating disorder test does not mean that television caused the increase.


6) Dr. Becker has been quoted in several places, including a recent Discovery News article, stating that “[Fijian] girls told us they thought these actresses were role models.” This leads many people to assume that most of the Fijian girls were saying they saw the actresses’ thin bodies as role models.


But that’s not what the 2002 study says: Instead, Becker writes (p. 511) “30% of those interviewed indicated that television characters served as role models concerning work or career issues .” These are two very different things. Instead of an implied majority of girls saying they wanted to have a thin body like Kate Moss, fewer than one-third of the girls said that the actresses were role models- and the ones who did explicitly stated that the actresses were not role models because of their body shape but because of their professional careers. Seeing a person on television and saying “Wow, I wish I could be a news anchor, or interview Brad Pitt, or be a movie star” is not the same as saying, “Wow, I wish I had her thin body.”


All this of course does not mean that the famous Fijian study is invalid, but it’s important to understand the limitations of the research, and what the study actually found. This widely-cited, highly-touted research is far less impressive when you take a closer look at the study-something writers and reporters rarely bother to do. The claim that exposure to TV and its mass media images increased eating disorders is one interpretation, but there are good reasons to be cautious about the validity and generalizability of the study.