Abstract
Abstract
Pro-eating disorder (Pro-ED) Websites are widespread and easily accessible. The risks associated with these Websites (e.g., reinforcing and encouraging eating disordered behaviors) have received interest from researchers, media, and mental health professionals. However, little is known about how these Websites are found, which may provide insight into prevention and intervention initiatives. Using Google AdWords Keywords in February 2011, a series of search terms (based on previous research) were entered to generate search-related data regarding actual pro-ED terms used in Google, including the corresponding search results, which were coded for degree of potential harm. Results indicated that Pro-ED search terms are sought out more than 13 million times annually, with pro ana receiving the most searches monthly. Overall, different terms are associated with varying numbers of monthly searches and regional interest. Search terms with references to thinspiration and thinspo are associated with the most harmful Website content. To provide those who seek pro-ED content with helpful, research-supported resources, it may be important to intervene before the point of access by targeting the search results corresponding with pro-ED search terms. Efforts may need to pay particular attention to terms such as thinspiration and thinspo.
Introduction
The nature of pro-eating disorder Websites represents an important public health issue given the potential risks associated with their content.1,3–7 These Websites typically provide suggestions for engaging in eating disorder behaviours (e.g., tips for dieting/purging), and many contain an interactive component (e.g., message board). Also common on these Websites are references to thinspiration, a blend of thin and inspiration (sometimes referred to as thinspo for short), which involves sharing photos/prose intended to inspire eating disorder behaviors. Some pro-ED Websites also provide recovery-oriented information (e.g., links to treatment programs) in addition to pro-ED content. One risk of pro-ED Websites is that many operate under the pretext of a supportive forum (i.e., peer-support in which recovery is, in theory, encouraged by the community) but largely function as a means for users to support engaging in eating disordered behaviors. Thus, other major risks include encouraging and reinforcing eating disordered behaviors in the form of sharing ‘tips and tricks’ to lose weight and a perceived competition among Website visitors to gain an extreme level of thinness. Finally, pro-ED Websites may dissuade individuals from recovery and professional help-seeking behavior. 7
Among adolescents and young adults, social groups have been found to exhibit similar eating behaviors (including dieting and eating disorders), body mass indices, and body-image. 8 Studies suggest that the content of pro-ED Websites (i.e., the techniques, encouragement of eating disorder behaviors) may have a contagion-like effect on individuals who are experimenting with eating disorder behaviors without fully committing to them.1,9 Furthermore, among adolescents, the use of pro-ED Websites is associated with a longer duration of eating disorder illnesses and greater hospitalization rates, indicating that pro-ED Websites may contribute to the perpetuation of eating disorder behaviors. 6
In 2001, Yahoo! shut down several pro-ED Websites—a movement believed to be propelled by the media, clinicians, and professional organizations offering support and advocacy for eating disorders.10,11 Although access to these Websites was temporarily thwarted, many sites re-emerged and new ones surfaced. Indeed, shutting down pro-ED Websites does not solve the problem. One possible way to address the issue of pro-ED content on the Internet is to provide individuals who search for pro-ED material online with helpful resources at the point of their Internet search. Similar initiatives have already started for web-searches about suicide. For example, in the United States, 12 and more recently in the United Kingdom, 13 individuals who search for terms associated with suicide (e.g., suicide methods) in Google are presented with helpful resources at the top of the search page. Since Google represents not just the most popular search engine on the Internet, but the Website garnering the most traffic worldwide, 14 similar initiatives for pro-ED Websites may be beneficial.
The current study used the Google AdWords Keywords 15 program to examine the nature of online Google searches for pro-eating disorder content. This program (described below) allows users to determine the nature of actual search queries used in Google's search engine. Identifying which terms are used, how often they are used, and the regions in which these terms are most frequently sought may help spur efforts to present individuals with helpful eating disorder resources in Google search results—similar to the efforts undertaken for suicide-related searches that have just been mentioned. No study has examined the nature of searches for pro-ED content online. Google AdWords Keywords 15 represents a useful tool in this regard as it permits users to determine the nature of actual search terms used by Internet users, how often each search term is sought over the course of a year, and the geographic regions associated with each search term. It also provides the corresponding Google search results (i.e., Websites) for each search term produced by the program. The current study used this program and is the first to formally examine the nature and scope of online pro-ED searches on Google.
Methods
Google AdWords Keywords
Google AdWords Keywords 15 is a free online program enabling users to determine the types of search terms related to a specific topic. One or more keywords are entered into the program, which generates all search terms corresponding with the originally entered keywords (which includes the keywords themselves). The program also produces data related to each search term, including the average number of global monthly searches for, and geographical regions associated with, each search term. These data reflect trends in the past 12 months (and can include a month-by-month analysis). The data produced by Google AdWords Keywords 15 are more expansive than that just noted (e.g., it also provides the estimated costs for developing ads for a search term) and go beyond the scope of the current study. As such, only data relevant to the current study have just been described.
Procedure
In February 2011, terms derived from previous research1,3–7 were entered into the Google AdWords Keywords 15 program. These included pro-ana, pro-anorexia, proana, promia, pro-mia, proanorexia, probulimia, pro-bulimia, pro-ED, proED, pro eating disorder, pro-eating disorder, thin and support, thin and pro acceptance, thin and pro-reality, thinspiration, pro thinspiration, pro-thinspiration, pro-thinspo, pro thinspo. From here, data were consolidated based on those receiving a minimum of 1,000 monthly hits. Data reported next pertain to the top 26 search terms.
Coding pro-ed search results
A coding rubric was developed to examine the level of harm associated with the Websites in the search results for each of the 26 search terms (Table 1). Focus was placed on the first page of the search results, as research suggests that users focus chiefly on the results from the first search page when using search engines. 16 Thus, 260 Websites were coded (i.e., 10 sites for each of the 26 keywords). Three coders carefully and independently examined 20 Websites excluded from the final dataset (i.e., beyond the first page of search results) to generate a coding scheme, and then met to discuss and finalize the rubric. The final rubric had the following anchors: 1=Completely Helpful (e.g., promotes healthy eating/body image, provides resources for eating disorder treatment); 2=Mostly Helpful (e.g., provides recovery information but has a few triggering images of thin celebrities/models); 3=Ambivalent (some aspects of the site are helpful, such as recovery resources; other aspects are harmful, such as dieting tips); 4=Harmful (e.g., contains some triggering images of thin models and has some tips and tricks about Pro-ED behavior); and 5=Very Harmful (contains large amounts of graphic, triggering images including emaciated models and actively encourages Pro-ED behavior with large numbers of tips and tricks). Thus, higher scores indicate more harm. Next, 20 additional sites were selected and independently coded; the rubric was deemed ready if inter-rater reliability exceeded 0.80. Inter-rater agreement exceeded 0.90 across all Websites. As a result, the rubric was applied to all 260 Websites in the final dataset. Inter-rater agreement here again exceeded 0.90.
Results
The top search terms
The following 26 search terms received a minimum of 1,000 monthly hits: pro ana, thinspiration, thinspo, pro mia, pro anorexia, pro ana mia, pro ana tips, tips on anorexia, ana thinspo, pro thinspo, pro ana Websites, pro ana sites, real thinspo, real thinspiration, ana thinspiration, pro ana thinspo, pro thinspiration, real girl thinspiration, thinspiration tips, pro bulimia, pro ana diets, thinspiration blog, anorexia pro ana, thin and support, pro eating disorders, becoming anorexic. Of the 26 top search terms, 12 contain an explicit reference to thinspiration (i.e., these search terms contain the words thinspiration or thinspo), 11 refer to anorexia (i.e., ana or anorexia), and 3 mention bulimia (i.e., mia or bulimia). Only two of the top 26 search terms (thin and support, and pro eating disorders), do not specifically refer to thinspiration, anorexia, or bulimia.
Global hits
More than 13,245,000 searches for pro-ED Websites are conducted on Google annually, based on hits for the top 26 search terms alone. Frequencies for search terms receiving a minimum of 1,000 monthly hits are found in Table 1. At 550,000 global hits, the search term pro ana received considerably more global hits than any other search term; this was followed in frequency by the search terms thinspiration and thinspo, which received 135,000 and 110,000 global hits, respectively.
Geographic location
The most common search terms used across various countries are found in Table 2. Values do not represent absolute values (i.e., a value of 100 does not reflect 100 searches). Rather, values are normalized and scaled (out of 100) such that highly populated regions do not receive higher values compared with less populated regions. Thus, values refer to the number of searches for a given term (e.g., pro ana) relative to the total number of searches conducted on Google over time. For example, a value of 100 for a region (e.g., Country X) indicates that this region records the highest level of interest for that term; a value of 50 in a second region (e.g., Country Y) would indicate this country receives about half as much interest (vs. Country X). If there is no value for a country, this does not mean that the search term is not used in that country; it is just used less than other search terms on Google in that region. Due to a high volume of data collected, only search term and country pairings with regional interest scores over 50 were used in the analyses below.
Values do not represent absolute values and instead represent normalized and scaled (out of 100) values. Specifically, values refer to the number of searches for a given term (e.g., pro ana) relative to the total number of searches conducted on Google over time.
The search term pro ana had high regional interest in the greatest number of countries (10). Since pro ana also received the largest number of searches (vs. other search terms; see Table 1), a high regional interest in many countries for this term is not surprising. References to thinspiration (e.g., thinspiration, thinspo, pro thinspiration) or anorexia (e.g., anorexia, ana, pro anorexia, pro ana Websites, pro ana diets) were present in all of the search terms with high regional interest in Australia, Canada, and the United Kingdom. Search terms with high regional interest in Finland, Germany, and Sweden contained references to thinspiration or thinspo. The United States was unique in that most (but not quite all) of the search terms with high regional interest referred to thinspiration; this was also the only country in which a search term with high regional interest, namely pro eating disorders, did not contain a reference to thinspiration, anorexia, or bulimia. The Netherlands was the only region in which search terms generating high regional interest referred to anorexia but not thinspiration. Finally, Mexico was the only country where search terms with references to bulimia (i.e., pro mia, pro ana mia) had high regional interest.
Harmful content
Overall, findings indicated that terms with references to thinspo and thinspiration tended to yield a pattern of higher harm scores than those terms without these references (Table 3). Specifically, the term thinspiration tips and real thinspo had the highest average scores. Of the 26 search terms identified, 12 (49 percent) had references to thinspo or thinspiration. Interestingly, the mean harm score for terms containing these references had a significantly higher mean (M=4.18) than those without them (M=3.17), t(23)=7.21; p<0.01; Cohen's d=3.00.
Means are presented in descending order.
Discussion
This is the first study that examines how pro-ED Websites may be found by individuals who search for them using Google—the world's most used search engine and Website. 14 Based on the 26 terms yielding at least 1,000 monthly searches, queries for pro-ED terms occur frequently and exceed 13 million searches annually. Although pro-ana was the most common search term identified, search terms including thinspiration and thinspo were also common and occurred more often than terms with an explicit reference to an eating disorder (i.e., terms including anorexia, ana, bulimia, or mia). This is somewhat surprising, as terms with these references are commonly used in research examining pro-ED Websites.1,3–6 Since thinspiration content often includes images of slim or underweight young women,1,5 and this material may promote body comparisons that exacerbate or encourage eating disorder behaviors, 1 search results corresponding with thinspiration and thinspo may be particularly worrisome. Indeed, search terms with these references associate with more harmful content based on Websites in the first page of search results. Thus, efforts to prevent the effects of, and access to, Pro-ED Websites may need to pay particular attention to these terms.
The nature of Pro-ED related searches is also widespread and varies geographically. This may be due to a number of factors, including differences in the eating disorder culture across countries, and, in particular, the online pro-ED culture; however, future research is needed before conclusions can be made. The term with the highest number of searches, pro-ana, had high regional interest in many countries; however, it did not have high regional interest in all countries represented in the analyses—including the United States. This variability may indicate a need for different countries to develop uniquely tailored prevention and intervention efforts that reflect these regional differences. For example, in the United States, search terms with references to thinspiration (i.e., real thinspo, pro thinspo, pro ana thinspo, thinspiration tips) had higher regional interest compared with terms such as pro-ana, which suggests that the former may merit more attention in this region.
Limitations and Future Directions
Results from this study may not generalize to searches conducted on other search engines (e.g., Yahoo!, Bing). Nevertheless, our results may correspond to the most frequently sought out pro-ED terms as Google generates more traffic than any other Website online 14 and individuals likely use similar terms in various search engines. At the same time, the current findings pertain to searches conducted in English and future work should determine how searches are conducted in other languages. One limitation related to the nature of data generated is that the number of people searching for pro-ED content online is unknown, as frequency counts for each search term represent total searches and not the number of individuals conducting the search; thus, several searches likely reflect the same individuals conducting multiple searches.
Since the goal of this study was to conduct a preliminary analysis of the nature of online pro-ED search patterns, we only examined the first 10 search results (i.e., first search results page). Future work may need to examine the length of time spent conducting these searches and determine how many search results are examined until people's search needs are met. In addition to this, we do not know the motive behind people's online searches using pro-ED terms. It is surmised that the keywords identified represent searches for pro-ED Websites by individuals who have eating disorder difficulties, but this cannot be assumed and the rationale for the search may vary (e.g., some searches may reflect searches conducted by concerned parents, health professionals or researchers). Demographic data about those conducting the searches were also unavailable. Past research suggests that adolescent and young adult women primarily engage in pro-ED online activity1,3,6 but more research will be needed to confirm whether this is the case for pro-ED online searches. Finally, since search term interest and search term results fluctuate temporally, it will be important to reexamine the nature of pro-ED searches via the Internet prospectively.
Implications and Conclusion
Past efforts have been made to take down pro-ED Websites, but this has led to a resurgence of the same Websites and the development of new ones.10,11 This does not solve the problem, and it is important to find alternative methods to address this issue. The current study provides a possible place to intervene and redirect people who search for potentially pernicious pro-ED content before the point-of access—namely, in the search results. Indeed, by virtue of knowing what terms are used to find pro-ED content on the Internet, researchers and health professionals may be better equipped to redirect those who search for this material to helpful, recovery-focused and research-supported resources. Thus, similar to recent efforts to provide resources in the search results for suicide-related searches,12,13 efforts can be made for pro-ED content as well. Efforts to provide suicide resources for potentially harmful suicide-related Google searches present resources (e.g., crisis lines) in the corresponding search results for queries such as suicide methods. Following this logic, efforts can be made to provide sponsored links to supportive and research-based eating disorder resources (e.g., how to seek help, from where to seek help) for individuals suffering from eating disorders when terms such as thinspiration and thinspo are used. Since the nature of online searches may change over time, it will be important to monitor these efforts and account for changes in search terms used and the nature of search results corresponding with these searches. Given the number of pro-ED search terms, the frequency with which they are sought, their widespread global interest, and the harmful material with which many terms associate, this may represent a critical step in preventing the effects that some pro-ED Websites may have on some vulnerable individuals–namely those who search for them online. To this end, it will be important to investigate the effectiveness of these types of initiatives and monitor this temporally, as search terms (and search results) may vary over time.
Footnotes
Acknowledgments
The authors would like to thank Lyndsay Chapman and Ashley Skinner for their assistance with initial data collection and coding.
Disclosure Statement
No competing financial interests exist.
