Doi:10.1016/j.jpsychores.2005.07.017

Journal of Psychosomatic Research 60 (2006) 253 – 256 Pro-anorexics and recovering anorexics differ in their Elizabeth J. Lyonsa,T, Matthias R. Mehlb,T, James W. Pennebakerc aDepartment of Health Behavior and Health Education, University of North Carolina, NC, United States bDepartment of Psychology, University of Arizona, Tucson, AZ, United States cDepartment of Psychology, University of Texas at Austin, Austin, TX, United States Objective: Pro-anorexia has emerged as a new and emotionally positive emotions, less anxiety, a lower degree of cognitive charged eating disorder phenomenon. This study explored the reflection, and lower levels of self-directed attention than did linguistic markers of differences in Internet self-presentation of recovering anorexics. Pro-anorexics were also more focused on the self-identified pro-anorexics who defend anorexia as a lifestyle and present and less on the past. Finally, pro-anorexics were more self-identified anorexics in recovery. Method: One hundred sixty- preoccupied with eating and less with school-related issues and two Internet message board entries and 56 homepages originating death. Conclusion: Linguistically, pro-anorexics and recovering from either pro-anorexics or recovering anorexics were analyzed anorexics engage in distinct psychological self-presentation styles.
for linguistic markers of emotional, cognitive, and social function- More research is needed to understand the clinical implications of ing, temporal focus, and anorexia-related psychological concerns.
Results: Across both text sources, pro-anorexics displayed more D 2006 Elsevier Inc. All rights reserved.
Keywords: Anorexia; Language use; LIWC; Text analysis; Coping the body and a life ever fastingQ; Although the realprevalence of bpro-anaQ sites is impossible to determine, Pro-anorexia has received widespread media attention as their number has been estimated to exceed 500 A a new and emotionally charged eating disorder phenomenon concern about these sites is their potential to trigger those on . Pro-anorexics are individuals who consider anorexia a the brink of an eating disorder into full-fledged anorexia.
legitimate alternative lifestyle that they choose to have, Despite media efforts to combat pro-anorexic attitudes, rather than an illness that they cannot control Their websites contain pictures of emaciated models, tips for The purpose of this study was to explore differences in dieting and how to hide weight loss from parents or doctors, self-presentation styles of individuals who publicly defend and community manifestoes such as the bAna CreedQ, a list anorexia as a lifestyle and individuals who identify of beliefs that describe what it means to be pro-anorexic themselves as recovering from anorexia. Methodologically, (e.g., bI believe in a wholly black and white world, the we attempted to complement traditional questionnaire-based losing of weight, recrimination for sins, the abnegation of assessments by sampling information directly and non-reactively from pro-anorexics’ main communication plat- B The preparation of this article was aided by the National Institutes of forms, personal homepages, and message boards Two text sources, homepages and message boards, were selected T Corresponding authors. Department of Psychology, University of to identify self-presentation styles that generalize across E-mail addresses: elyons@email.unc.edu (E.J. Lyons), different online media. Pro-anorexics were contrasted against recovering anorexics as a comparison group to 0022-3999/06/$ – see front matter D 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jpsychores.2005.07.017 E.J. Lyons et al. / Journal of Psychosomatic Research 60 (2006) 253 – 256 Table 1Selected text samples of homepages and message board entries from pro-anorexic and recovering anorexic participants You can never be too thin. . .at least to me you can’t. People might I was 10 years old when I began my eating disorder. My life think it’s disgusting, they might think I’m disgusting because of was out of control. Weight was the one thing I could control.
Ana. . .but what is it that people are always saying? It doesn’t matter And I did it well. My parents encouraged me, my friends what other people think, and be happy with who you are. Well Ana admired me and my siblings envied me. Losing weight made is a part of me, and I’m fine with that. Ana makes me happy.
I am the thinnest. The thinnest. This is how my day was yesterday.
I still have days that I want to give up and curl up and die.
I am 5’4 and 51 pounds. Clinically dying; and if that’s what gets But I will not let myself. I am going to win this war if it me the thinnest, thinner than every other anorexic, then so be it.
takes my whole 100 years on this damn planet.
I am going to lose 10–15 pounds by 10/11 . . . that’s about a month How have I been so lucky?! How long will my good luck last? of fasting, 1 hour on treadmill a day. NO LESS than 200 sit ups a I’ve never truly feared death through out six hospitalizations day. 100 leg lifts each leg and each low cal food when needing and five years of therapy. . .others have told me I look like energy. Wish me LUCK. 27 days COUNTING DOWN NOW! death, or that I will die if I continue, but it’s like I feel invisible.
Hey everyone! This site is great. . .but anyway–I’m 5V3W 108 lbs–I I don’t know if this will work and I’m not sure if I’ll ever get know, a big fat cow! You guys all have so much will power, how do better and more often than not I’m not sure if there is anything you do it? I want to be 99 lbs and can’t seem to lose the weight.
bto get betterQ from, but I don’t know if I can trust my ownjudgment right now either.
better understand how attitudinal factors can affect psycho- bpro anorexiaQ, bpro anaQ) and by following links from logical processes that, in turn, may create barriers to major pro-anorexia sites (e.g., bAnorexic NationQ, bFat successful anorexia treatment. Recovering anorexics were Like MeQ, bBloody Brick RoadQ). A minimum of 30 words further considered an appropriate comparison group because was required for a homepage to be retained in the sample.
they also extensively use homepages and message boards to The final sample consisted of 28 pro-anorexia homepages.
share their experiences with others.
The estimated mean age of the site authors was 17.3 years This study analyzed self-presentation from a linguistic (note: 10 authors did not list their ages). In a similar way, perspective The psychological study of language use 28 homepages of self-described anorexics in recovery were has recently received increasing scientific attention retrieved (mean age=21.1 years; seven authors did not list Word-count-based text analysis approaches have been their ages). A special effort was made to make the two shown to reliably capture diagnostic information about a groups comparable with regard to the authors’ sex, age, wide range of psychological phenomena, including psychi- atric disorders suicidal ideations coping with Pro-anorexia and recovering anorexia message boards breast and prostate cancer psychological responses (i.e., online bulletin boards) were sampled in a similar way to a national upheaval and even risk for coronary heart via search engines as well as via links from other anorexia sites. The message boards were hosted either by popular Because of their vehement anorexia-as-a-lifestyle ideol- providers such as bAmerica OnlineQ, bMSNQ, or bYahooQ or ogy, we expected pro-anorexics, compared with recovering by private persons. To minimize sampling bias, only the first anorexics, to show a more pronounced hedonic focus five pertinent entries were taken from each message board.
(references to positive emotions and the present), a higher All entries were checked for length (more than 30 words) level of self-absorption (references to self), and a lower level and authorial intent. The final sample consisted of 82 pro- of cognitive reflection (use of cognitive words).
anorexia and 80 recovering anorexia entries. No ageinformation was available on the authors of these entries.
shows selected homepage and message board The study was a 2 (pro-anorexia vs. recovering anorexia) All words contained in the sampled homepages and  2 (homepage vs. message board) factorial design with message board entries were pasted into a text file. These linguistic markers of basic emotional, cognitive, and social 218 text files were cleaned, spell-checked, and submitted to processes, temporal focus, and anorexia-related psycholog- Linguistic Inquiry and Word Count (LIWC; an ical concerns as dependent variables.
extensively validated word-count based text analysis pro-gram LIWC characterizes text samples on Pro-anorexia homepages (i.e., personal websites) were 1 The complete list of homepage and message board URLs, as well as retrieved from popular Internet search engines (keywords: the original text samples, is available from the authors.
E.J. Lyons et al. / Journal of Psychosomatic Research 60 (2006) 253 –256 74 standard grammatical (e.g., first person singular pro- nouns, articles, prepositions, present tense), psychological(e.g., emotion words, cognitive words, social words), and summarizes the results from a series of univariate content (e.g., school, sexuality, eating) dimensions.
two-way analyses of variance for the selected LIWC From the available 74 LIWC variables, three language variables. All effects are based on 1 and 214 degrees dimensions were most relevant for understanding differ- ences in self-presentation styles because they have The analyses of the linguistic indicators of emotional previously been linked to different coping strategies: processes indicated that compared with recovering ano- (1) emotional processes, such as the use of words referring rexics, pro-anorexics used more positive emotional words to positive emotions (e.g., bhappyQ, bgoodQ) and negative in both their homepages as well as their message boards.
emotions (bhateQ, bscaredQ), particularly anxiety (e.g., Pro-anorexics also used anxiety words at a markedly lower bafraidQ, bscaredQ; (2) cognitive processes, such rate than recovering anorexics. Furthermore, as predicted, as the use of words that refer to cognitive mechanisms across both text sources, pro-anorexics used significantly (e.g., bcauseQ, bknowQ), particularly insight into causal fewer cognitive mechanism words, and specifically insight processes (i.e., insight words; e.g., bcauseQ, brealizeQ; words, than recovering anorexics did. Contrary to our (3) social processes, particularly words that hint hypotheses, pro-anorexics made not more but fewer first at a general social awareness (i.e., social words; e.g., person singular self-references than did recovering ano- btalkQ, byouQ) versus self-awareness (i.e., first person rexics, indicating lower levels of self-focused attention.
singular pronouns; e.g., bIQ, bmeQ, bmyQ; Consistent with the predicted focus on the here and now, The study also included an analysis of temporal focus both homepages and message boards of pro-anorexics as indicated by references to the past (i.e., use of past contained more present tense verbs and fewer past tense tense verbs) and the present (i.e., use of present tense verbs than did those of recovering anorexics. Finally, with verbs). Finally, three content categories were included to regard to the prevalence of anorexia-related psychological tap into the prevalence of anorexia-related psychological concerns, pro-anorexics, compared with recovering ano- concerns: references to eating (e.g., bmealQ, bdietQ), rexics, showed a higher degree of preoccupation with references to school (bexamQ, bstudyQ), and references to eating and made fewer references to school- and death- Table 2Differences in the language use of pro-anorexics and recovering anorexics on homepages and message boards Means and standard deviations refer to percentage of total words within each text sample. Results are based on univariate two-way ANOVAs (df=1,214).
Rec-Anorexia = recovering g2p = partial eta squared (sums of squares based effect size estimate that indicates what proportion of the effect and error varianceis T P b.10 (two tailed).
TT P b.05 (two tailed).
TTT P b.01 (two tailed).
E.J. Lyons et al. / Journal of Psychosomatic Research 60 (2006) 253 – 256 self-presentation styles for the early detection and treatmentof anorexia.
The purpose of this study was to explore pro-anorexia as a new, highly emotional, and publicly controversialphenomenon. The findings revealed that pro-anorexics andrecovering anorexics differed reliably in the language that they used in their homepages and on message boards.
Specifically, compared with recovering anorexics, the [Electronic Version]. Retrieved from http://www.usatoday.com/ word use of pro-anorexics indicated a more pronounced news/healthscience/health/2001-07-24-anorexia-sites.htm#more, hedonic focus on positive emotions and the here and now, reduced level of cognitive processing, and a lower degree [2] Dias K. The ana sanctuary: women’s pro-anorexia narratives in cyberspace. J Int Womens Stud 2003;4:31 – 45.
A similar language use pattern was recently found [3] Hardin PK. Shape-shifting discourses of anorexia nervosa: reconsti- tuting psychopathology. Nurs Inq 2003;10:209 – 17.
among users of an online journal website in the aftermath [4] Chesley EB, Alberts JD, Klein JD, Kreipe RE. Pro or con? Anorexia of September 11, 2001 Compared with a pre-9/11 nervosa and the Internet. J Adolesc Health 2003;32:123 – 4.
baseline, participants evidenced slightly elevated levels of [5] Mond JM, Owen C, Hay PJ, Rodgers B, Beumont PJV. Assessing emotional positivity, reduced levels of cognitive process- quality of life in eating disorder patients. Qual Life Res 2005;14:171 – 8.
ing, and reduced levels of self-preoccupation in the months [6] Vitousek KB, Daly J, Heiser C. Reconstructing the internal world of after the attacks. This raises the possibility that pro- the eating-disordered individual: overcoming denial and distortion inself-report. Int J Eat Disord 1991;10:647 – 66.
anorexics’ language use hints at a coping strategy aimed at [7] Goffman E. Forms of talk. Philadelphia7 University of Pennsylvania stabilizing them emotionally. This stabilization, then, may allow pro-anorexics to sustain their immunity to psycho- [8] Pennebaker JW, Mehl MR, Niederhoffer KG. Psychological aspects of natural language use: our words, our selves. Annu Rev Psychol Methodologically, this study used an innovative data collection strategy, the unobtrusive sampling of language [9] Oxman TE, Rosenberg SD, Schnurr PP, Tucker GJ. Diagnostic classification through content analysis of patients’ speech. Am J from the Internet. A limitation of this strategy is that only restricted information on the participants could be gathered [10] Stirman SW, Pennebaker JW. Word use in the poetry of suicidal and Furthermore, although the text samples originated non-suicidal poets. Psychosom Med 2001;63:517 – 22.
from individuals who publicly presented themselves as pro- [11] Owen JE, Klapow JC, Roth DL, Tucker DC. Use of the Internet for anorexic or recovering anorexic, it is not clear to what information and support: disclosure among persons with breast andprostate cancer. J Behav Med 2004;27:491 – 505.
extent the participants actually met clinical DSM criteria.
[12] Alpers GW, Winzelberg AJ, Classen C, Roberts H, Dev P, Koopman Finally, the absence of precise definitions of pro-anorexia C, Taylor CB. Evaluation of computerized text analysis in an Internet and recovering anorexia and the lack of a healthy control breast cancer support group. Comput Hum Behav 2005;21:361 – 76.
group prevent that strong conclusions can be drawn from [13] Cohn MA, Mehl MR, Pennebaker JW. Linguistic indicators of the data. Yet, as one of the first empirical investigations psychological change after September 11, 2001. Psychol Sci into pro-anorexia, this study draws scientific attention to [14] Scherwitz L, Graham LE, Grandits G, Buehler J, Billings J. Self- this new, clinically relevant social phenomenon. With its involvement and coronary heart disease incidence in the multiple risk word-use approach, the study provided a linguistic charac- factor intervention trial. Psychosom Med 1986;48:187 – 99.
terization of pro-anorexics’ Internet self-presentation. This [15] Pennebaker JW, Francis ME, Booth RJ. Linguistic Inquiry and Word characterization suggests that pro-anorexics may use a Count (LIWC 2001): a computerized text analysis program. Mahwah coping strategy that stabilizes them emotionally and allows them to experience a sense of control over their illness [16] Gosling SD, Vazire S, Srivastava S, John OP. Should we trust Web- based studies? A comparative analysis of six preconceptions about Future research needs to investigate the effects that pro- Internet questionnaires. Am Psychol 2004;59:93 – 104.
anorexia homepages and message boards exert on risk- [17] Kraut R, Olson J, Banaji M, Bruckman A, Cohen J, Couper M.
populations, as well as the implications of pro-anorexics’ Psychological research online. Am Psychol 2004;59:105 – 17.

Source: http://dingo.sbs.arizona.edu/~mehl/eReprints/Proanorexics%20JPR%202006.pdf