Objective Pro-anorexia, or pro-ana websites advocate the development, continuation and normalisation of an eating disordered life style. Typical content includes the Ana Creed and/or the Thin Commandments which are the central beliefs of pro-ana members; Thinspiration, a collection of prose or images to inspire weight loss; tips and tricks, which includes instructions displayed as facts to guide individuals to engage in eating disordered behaviours; and forum or chat room. This research investigated what factors account for internalisation of the different types of content of two pro-ana websites.

Method Two existing scales the Multidimensional Media Influence Scale and the Internalisation of Pro-anorexia website Content scale were adapted into a 30-item questionnaire called the Pro-anorexia Internalisation Scale (PAIS) to measure internalisation of the thinspiration, tips and tricks and chat room content. An online survey was hosted on two pro-ana sites. It contained this measure and a measure of disordered eating behaviours the Eating Disorders Inventory-3 Referral (EDI-3 R) form; the NEO Five Factor Inventory 3, which is a brief measure of the five major domains of personality; a measure of obsessive compulsiveness from the Dimensional Obsessive Compulsive Scale; the Perfectionism Subscale of the Eating Disorders Inventory 3; the Paulhus Deception Scale, a measure of an individual's tendency to give socially desirable responses on self report inventories; and measures of frequency and duration of visits to the different sections of the website, and demographic measures including weight, weight history and perception of current weight measured on a five point Likert-like scale with anchor points Very Underweight and Very Overweight.

Results A complete data set was obtained from 75 participants (all female, mean age 22.68 years, SD=6.15). All respondents met the criteria for eating disorder risk and referral according to the behavioural questions on the EDI-3 R. Forty four (59%) reported that they did not have a clinically diagnosed eating disorder. Only 20 met the criteria for being underweight (BMI< 18.50). The Cronbach's alpha for the PAIS scale was α=0.90, higher than the two existing scales. Data analysis comprised a model reduction strategy via step-wise backward regression between the scores on the personality and demographic items with the PAIS scores for the three component of the web site content. Scores measuring internalisation of thinspiration content showed a statistically significant relationship with only one personality measure the agreeableness scale on the NEO-FFI-3 but explained only a small proportion of the variance (R2=7.34%). Measures of frequency of visits and duration were better predictors, explaining R2=16.30 and R2=11.40% of the variance respectively. Highest weight had a significant negative association (r=−0.22) explaining R2=7.34% of the variance after adjustment for agreeableness, frequency, duration and perception of current weight. For the measures of internalisation of tips and tricks sections, there was a significant relationship with scores on only one personality variable obsessive compulsiveness but again, this accounted for only a small proportion of the variance (R2=7.30). Frequency of visits was a significant and large predictor (R2=45.60). Scores on the deception impression management subscale of the Paulhus Deception Scale had a significant, negative and moderate sized effect on scores of internalisation of chat room content (R2=23.90%). Agreeableness significantly contributed to chat room internalisation, after adjusting for deception impression management but accounted for only a small amount of the variance (R2=8.30%).

Conclusion The study benefitted from recruiting participants from a particularly understudied, private and rapidly growing community. The finding that frequency and duration of visits to pro-ana websites is associated with internalisation of thinspiration and tips and tricks content is consistent with predictions from Cultivation theory (Gerbner & Gross, 1976). Clinicians might benefit from asking clients about internet use. The data provide preliminary evidence for a relationship between the personality trait of agreeableness and internalisation of the content of pro-ana websites.