Perceived acceptability of anorexia and bulimia in women with and without eating disorder symptoms

Abstract

Perceptions of the acceptability of eating‐disordered behaviour were examined in young adult women with (n = 44) and without (n = 268) eating disorder symptoms. All participants viewed vignettes of anorexia nervosa (AN) and bulimia nervosa (BN) and responded to the same series of questions—addressing different possible ways in which the conditions described might be seen to be acceptable—in relation to each vignette. Participants with eating disorder symptoms perceived eating‐disordered behaviour to be more acceptable than asymptomatic participants, and this was the case for both AN and BN vignettes and for a range of different items. Differences on items tapping the perception that it ‘might not be too bad’ to have an eating disorder and that an eating disorder is ‘nothing to be concerned about’ were particularly pronounced. The findings could not be accounted for by between‐group differences in body weight. The findings indicate the ambivalence towards eating‐disordered behaviour that exists among a subgroup of young women in the community and the clear association between such ambivalence and actual eating disorder symptoms. The perceived acceptability of eating‐disordered behaviour may need to be addressed in prevention and early‐intervention programs for eating disorders.

Key words:

• acceptability
• anorexia
• bulimia
• eating disorders
• mental health literacy

Conflict of Interest: None.

Although ego‐syntonic aspects of eating‐disordered behaviour are well known to clinicians (Gale, Holliday, Troop, Serpell, & Treasure, 2006; Vitousek, Watson, & Wilson, 1998), little is known about the extent to which the behaviour characteristics of anorexia nervosa (AN) and bulimia nervosa (BN) are perceived to be acceptable or desirable in the general population. Perceptions of this kind are of interest because they can inform the content of programs designed to reduce stigmatisation and to promote early, appropriate treatment seeking among individuals with symptoms (Crisp, Gelder, Rix, Meltzer, & Rowlands, 2000; Mond, Hay, Paxton et al., 2010; Stewart, Keel, & Schiavo, 2006).

Theoretically, it would not be surprising to find that at least some women consider the symptoms of AN to be acceptable, or even desirable, given the extent to which thinness is idealised in contemporary Western societies and given that the low body weight and steadfast control over dietary intake that characterise AN are often linked to fame and wealth in the popular media (Gustafsson, Edlund, Kjellin, & Norring, 2008; Striegel‐Moore & Franko, 2003). It would be more surprising to find that the symptoms of BN are viewed as acceptable, given that BN is characterised by uncontrolled episodes of binge eating, purging behaviours, such as self‐induced vomiting and laxative misuse, and weight cycling (Hay, 2003; Huon, Brown, & Morris, 1998). Nevertheless, certain aspects of BN, such as the regular use of extreme weight‐control behaviours, might be viewed as acceptable or desirable in some sections of the community, particularly given the current concern surrounding the obesity ‘epidemic’ (Austin, 1999; Burns & Gavey, 2004).

In a 1997 study (Vander Wal & Thelen, 1997), in which male and female college students rated the ‘acceptability’ of each of 14 bulimic behaviours (e.g., ‘binge eating on a regular basis’, using laxatives to prevent weight gain’), the perceived acceptability of these behaviours was generally low. Interestingly, however, there was a positive association between the perceived acceptability of bulimic behaviours and the participants' actual levels of eating disorder symptoms. This latter finding raised the possibility that women who consider bulimic behaviours to be acceptable may be more likely to experiment with those behaviours. Alternatively, women with symptoms may adjust their attitudes, perceiving their eating disorder behaviours to be acceptable, to alleviate the dissonance between attitudes and behaviour (Vander Wal & Thelen, 1997).

Findings from more recent studies point to an association between perceptions of the acceptability of both AN‐ and BN‐type disorders and actual symptom levels (Mond, Hay, Rodgers, Owen, & Beumont, 2004a; Mond, Robertson‐Smith, & Vetere, 2006). Thus, in a study of young adult women presented with a vignette of BN (Mond et al., 2004a), 46.2% of participants with eating disorder symptoms reported that they had occasionally, often, or always thought that it ‘might not be too bad’ to have the problem described, whereas 11.8% of asymptomatic participants gave these responses. In a different sample of women (Mond, Hay, Rodgers, & Owen, 2006), presented with a vignette of AN, 60.8% of participants with symptoms reported that they had occasionally, often, or always thought that it might not be too bad to have the problem described, whereas 21.6% of asymptomatic participants gave these responses. However, the number of participants with symptoms in the BN study was small, and the assessment of perceived acceptability was confined to the single item referred to above in both studies. Other items assessed perceived severity and whether and how perceptions of the severity of mental disorders relate to perceptions of acceptability is unclear.

The goal of the present study was to further examine perceptions of the acceptability of eating‐disordered behaviour among individuals with and without eating disorder symptoms, extending the research outlined above in two ways. First, we included several items to address the construct of perceived acceptability. Second, the same participants were presented with vignettes of both AN and BN. Consistent with previous research, it was hypothesised that individuals with a high level of eating disorder symptoms would be more likely to consider eating‐disordered behaviour to be acceptable than would asymptomatic individuals. It was also hypothesised that perceptions of the acceptability of eating‐disordered behaviour would be more common for AN than for BN, given the more conspicuously ego‐syntonic features of AN (Mond, Hay, Rodgers, Owen, & Beumont, 2005; Vitousek et al., 1998).

A secondary aim of the study was to clarify whether and how perceptions of acceptability might relate to perceptions of severity. Although evidence is lacking, it would be reasonable to hypothesise an inverse association between these constructs; that is, greater perceived severity of AN or BN might be expected to be associated with lower perceived acceptability (Mond et al., 2004a).

METHOD

Study design and participants

Participants were young adult women who were enrolled as internal or external (long‐distance) undergraduate or postgraduate students at a regional university in north‐eastern Australia and who chose to complete an Internet‐based survey. Young adult women were chosen so as to permit identification of a subgroup of participants with a high level of eating disorder symptoms without the need for recruitment of a very large sample (Luce, Crowther, & Pole, 2008; Mond, Hay et al., 2006). The inclusion of external students was intended to provide a geographically broader sampling frame than is typically possible with University‐based samples.

A total of 333 women aged 16–56 completed the survey during the 3‐month recruitment period. To confine the study population to young adult women, defined as women aged 18–40 years, data for three participants aged less than 18 years and for 13 participants aged 41 years or more were excluded. Data for a further five participants were omitted because of missing data. Hence, participants in the present study were 312 women aged 18–40 years (mean = 23.3 years, standard deviation (SD) = 4.9). Most participants were born in Australia (84.6%) and had English as their first language (90.7%).

Where appropriate, data for a general population sample of women (n = 495) and for women with eating disorders receiving specialist treatment (n = 118) are given for descriptive purposes. The latter groups were recruited independently, in research conducted in the Australian Capital Territory (ACT) (Mond, Rodgers, Hay et al., 2004; Mond et al., 2005). The clinical sample (n = 118) comprises consecutive referrals to the ACT Eating Disorders Day Program, with diagnoses of AN (n = 35), BN (n = 43), and eating disorder not otherwise specified (n = 40).

Survey composition

The survey included, in addition to the questions addressing perceived acceptability, measures of eating disorder psychopathology, general psychological distress, impairment in role functioning, self‐reported height and weight, and limited socio‐demographic information. The measure of eating disorder psychopathology was included to be able to identify participants with symptoms. Measures of distress and functional impairment were included to be able to determine the ‘clinical significance’ of those symptoms. The study design was approved by the Central Queensland University Human Research Ethics Committee.

The survey was divided into two sections, one relating to AN, the other to BN. In each section, participants first completed socio‐demographic information and measures of distress and functional impairment. A vignette describing a fictional person with AN (‘Lucy’) or with BN (‘Kelly’) was then presented, followed by a series of questions addressing the perceived acceptability of the problem described. The measure of eating disorder symptoms constituted the final section of the survey. The vignettes, which were the same as those employed in previous research (Mond et al., 2004a; Mond, Robertson‐Smith et al., 2006), are given in the Appendix.

Study measures

Assessment of perceived acceptability

Only one of the items employed in previous research (Mond et al., 2004a; Mond, Robertson‐Smith et al., 2006), namely: ‘Have you ever thought it might not be too bad to be like Lucy (Kelly)?’, was considered to assess perceived acceptability and was, therefore, included in the present study. Response options for this question were ‘never’, ‘rarely’, ‘occasionally’, ‘often’, and ‘always’.

Nine additional items assessing perceived acceptability were tested in a subgroup of early respondents. On the basis of the pilot results, redundant items were eliminated, and item content was refined. Six of the new items were subsequently retained, namely: ‘Do you think that having Lucy's (Kelly's) problem might have beneficial aspects?’; ‘How worried or concerned would you be if you thought you were developing Lucy's (Kelly's) problem?’; ‘How happy would you be if you had Lucy's (Kelly's) problem?’; ‘Do you think that people who have Lucy's (Kelly's) problem would find it easier to “fit in”?’; ‘Do you think that having Lucy's (Kelly's) problem would improve someone's confidence/self‐esteem?’; and ‘Do you think that having Lucy's (Kelly's) problem would make someone more popular?’. Response options for these questions were ‘not at all’, ‘a little’, ‘moderately’, ‘very’, and ‘extremely’ or ‘definitely not’, ‘possibly’, ‘mixed feelings/yes and no’, ‘probably’, and ‘definitely’.

A further three items, which were employed in previous research (Mond et al., 2004a; Mond, Robertson‐Smith et al., 2006) and which were considered to assess perceived severity, were included to permit analysis of the association between perceptions of acceptability and perceptions of severity. These questions were: ‘How distressing do you think it would be to have Lucy's (Kelly's) problem’? ‘How difficult do you think it would be to treat Lucy's (Kelly's) problem?’ and ‘How sympathetic would you be towards someone with Lucy's (Kelly's) problem?’. Response options for each were ‘not at all’, ‘a little’, ‘moderately’, ‘very’, and ‘extremely’.

Hence, a total of 10 items—seven assessing perceived acceptability and three assessing perceived severity—were included in the present study. All were scored on a 5‐point Likert‐type scale from ‘1’ to ‘5’, with higher scores indicating greater perceived severity or acceptability/desirability of the problem described. Reverse scoring of the item addressing concern about developing Lucy's (Kelly's) problem was employed in order that higher scores indicated greater perceived acceptability (or severity) on all items.

Eating disorder psychopathology

Eating Disorder Examination Questionnaire (EDE‐Q)

The EDE‐Q (Fairburn & Beglin, 1994) is a widely used, 36‐item self‐report measure of eating disorder features that focuses on the past 28 days. Scores on each of four subscales—relating to dietary restraint, eating concerns, weight concerns, and shape concerns—and a global score are derived from 22 items addressing attitudinal components of eating‐disorder psychopathology (Mond, Hay, Rodgers, Owen, & Beumont, 2004b; Mond, Hay et al., 2006). Scores on these scales range from 0 to 6, with higher scores indicating greater symptom frequency and/or severity.

Remaining items of the EDE‐Q assess the occurrence and frequency of various eating disorder behaviours, namely objective overeating (binge eating), subjective overeating, self‐induced vomiting, misuse of laxatives or diuretics, and excessive exercise. Because the EDE‐Q does not provide for the assessment of extreme dietary restriction (fasting) as a specific behaviour, an item of the dietary restraint subscale, which assesses the frequency of ‘going without food for a period of 8 or more waking hours to influence weight or shape’, was used for this purpose.

Symptomatic participants were identified according to an operational definition used in previous research, namely the ‘undue influence of weight or shape on self‐evaluation’ in conjunction with any of the regular occurrence of any eating disorder behaviour (Mond, Hay, Rodgers, & Owen, 2009). The ‘undue influence of weight or shape on self‐evaluation’ was defined as a score of 5 or 6 on either or both of the EDE‐Q items assessing this construct. For objective and subjective overeating, self‐induced vomiting, and misuse of laxatives or diuretics, ‘regular’ was defined as ‘at least weekly’. Regular ‘extreme dietary restriction’ was defined as ‘going without food for a period of 8 or more waking hours to influence weight or shape, on average, three or more times per week’, whereas excessive exercise was defined as ‘exercising hard to influence weight or shape, on average, five or more times per week’ (Mond, Hay et al., 2006).

Kessler Psychological Distress Scale (K‐10)

The K‐10 (Kessler, Andrews, Colpe et al., 2002) is a 10‐item measure of general psychological distress developed for use in general population studies. In Australia, it is also used as an outcome measure among individuals treated within mental health services and in routine population health surveillance (Andrews & Slade, 2001). The frequency of each of 10 depressive or anxiety symptoms is measured on a scale from 1 to 5. In the present study, coding of response options was such that total scale scores range from 10 to 50, with lower scores indicating higher levels of distress. Cronbach's alpha in the present study was 0.88.

Medical Outcomes Study (12‐item) Short‐form (SF‐12)

The SF‐12 (Ware, Kosinski, & Keller, 1996) is a 12‐item, generic measure of health‐related quality of life. Items are summarised into two weighted scales (Physical Component Summary Scale (PCS) and Mental Component Summary Scale (MCS)) designed to assess perceived impairment in role functioning associated with physical and mental health problems. Each scale is scored to have a mean of 50 and SD of 10 (in the US population), with lower scores indicating higher levels of impairment. The SF‐12 has very good psychometric properties and demonstrated validity in the Australian population (Sanderson & Andrews, 2002). Cronbach's alpha in the present study was 0.81.

Preliminary (factor) analysis

Principal components analysis with varimax rotation identified three factors—‘severity’, ‘personal acceptability’, and ‘social acceptability’—according to root‐one criterion, accounting for 66.4% and 60.1% of the variance for the AN and BN vignettes, respectively (Table-1).

Table 1 Results (item loadings) of a principal components analysis of items assessing perceptions of the severity and acceptability of anorexia nervosa (AN) and bulimia nervosa (BN) in young adult women (n = 312)

–	Severity		Acceptability ‘personal’		Acceptability ‘social’
Item	AN	BN	AN	BN	AN	BN
How distressing do you think it would be to have Lucy's (Kelly's) problem?	0.74	0.74	–	–	–	–
How sympathetic would you be towards someone with Lucy's (Kelly's) problem?	0.81	0.78	–	–	–	–
How difficult do you think it would be to treat Kelly's (Lucy's) problem?	0.78	0.70	–	–
Have you ever thought it might not be too bad to be like Lucy (Kelly)?	–	–	0.80	0.73	–	–
Do you think that having Kelly's (Lucy's) problem might have beneficial aspects?	–	–	0.73	0.73	–	–
How worried or concerned would you be if you thought you were developing Kelly's (Lucy's) problem?	–	–	0.76	0.65	–	–
Do you think that having Kelly's (Lucy's) problem would make someone more popular?	–	–	–	–	0.83	0.81
Do you think that people who have Kelly's (Lucy's) problem would find it easier to ‘fit in’?	–	–	–	–	0.87	0.85
Do you think that having Kelly's (Lucy's) problem would improve someone's confidence/self‐esteem?	–	–	–	–	0.62	0.54

A total score on each factor was derived as the simple (unweighted) average of scores on the individual items comprising it for use in subsequent analysis. Reliability coefficients (Cronbach's alpha) for the severity, personal acceptability, and social acceptability domains were 0.69, 0.80, and 0.75, respectively, for the AN vignette and 0.64, 0.69, and 0.67, respectively, for the BN vignette. Correlations (Spearman's rho) between age, body mass index (BMI), domain scores for each vignette, and overall levels of eating disorder symptoms, as measured by the EDE‐Q global score, are given in Table-2.

Table 2 Correlations (Spearman's rho) between study variables in the total sample (n = 312)^a

–	Age	BMI	Eating disorder symptoms	AN severity	AN acceptability personal	AN acceptability social	BN severity	BN acceptability personal
Body mass index (BMI)	0.30	–	–	–	–	–	–	–
Eating disorder symptoms^b	0.14	0.45	–	–	–	–	–	–
AN severity	0.02	0.04	0.06	–	–	–	–	–
AN acceptability personal	−0.07	0.05	0.42	−0.18	–	–	–	–
AN acceptability social	−0.08	−0.00	0.28	−0.17	0.45	–	–	–
BN severity	0.05	0.07	0.11	0.64	−0.06	−0.13	–	–
BN acceptability personal	−0.08	0.11	0.44	−0.13	0.54	0.33	−0.24	–
BN acceptability social	−0.08	0.03	0.25	−0.10	0.21	0.52	−0.21	0.34

^a With n = 312, correlations of 0.12 and 0.15 are significant at the 0.05 and 0.01 levels, respectively.

^b As measured by the Eating Disorder Examination Questionnaire global score.

AN = anorexia nervosa; BN = bulimia nervosa.

Statistical analysis

Data are presented as the percentage of participants choosing particular response options for each item. Independent‐samples t‐tests were used to compare women with eating disorder symptoms with asymptomatic participants with respect to age, BMI, and levels of eating disorder psychopathology, general psychological distress, and functional impairment. Chi‐square tests were used to compare the responses of participants with and without eating disorder symptoms on the individual items comprising the severity and acceptability domains, whereas non‐parametric, continuous variable methods (Mann–Whitney U‐tests for independent samples or Wilcoxon signed‐rank tests for within‐subjects analysis) were used to compare scores on each domain as a function of symptom status (symptoms, no symptoms) and vignette type (AN, BN).

RESULTS

Forty‐four participants were classified as symptomatic according to the operational definition outlined previously, whereas the remaining 268 participants were classified as asymptomatic. Table-3 shows mean age, BMI, and scores on measures of eating disorder psychopathology, general psychological distress, and impairment in role functioning for these groups and for women in the general population and eating disorder patients. As can be seen, symptomatic participants were younger, heavier, and had higher levels of eating disorder psychopathology, general psychological distress, and impairment in role functioning than asymptomatic participants. Symptomatic women were more likely to be obese (BMI ≥ 30) (22.0% vs 14.5%, χ = 1.52, p = .22) and less likely to be underweight (BMI < 18.5) (0.0% vs 6.6%, χ = 2.89, p = .09) than asymptomatic women.

Table 3 Mean (SD) age, body mass index (BMI), impairment in role functioning (SF‐12 PCS, MCS) and general psychological distress (K‐10) for asymptomatic women (‘non‐ED’, n = 268) and women with eating disorder symptoms (‘ED’, n = 44)^a

–	General population sample (n = 495)	Non‐ED (n = 268)	ED (n = 44)	Eating disorder patients (n = 118)	p ^b
Age (years)	33.46(8.91)	23.33(5.05)	22.80(3.79)	22.26(7.32)	<.01
BMI(kg/m^2)	24.85(5.45)	24.48(5.62)	27.52(7.11)	21.27(4.60)	<.01
EDE‐Q subscales^c
Restraint	1.22(1.42)	1.51(1.33)	3.56(1.54)	3.85(1.64)	<.01
Eating concern	0.64(1.00)	0.98(1.10)	3.42(1.56)	3.66(1.54)	<.01
Weight concern	1.65(1.42)	2.51(1.70)	5.38(0.87)	4.27(1.52)	<.01
Shape concern	2.10(1.58)	2.80(1.64)	5.43(0.85)	4.66(1.36)	<.01
Global score	1.40(1.20)	1.96(1.25)	4.44(0.95)	4.09(1.35)	<.01
SF‐12 PCS^d	50.73(8.76)	51.52(7.06)	49.52(8.96)	47.15(10.32)	<.01
SF‐12 MCS^d	47.40(10.30)	44.32(10.60)	33.73(10.97)	30.77(10.57)	<.01
K‐10^e	43.24(6.45)	39.26(6.06)	33.80(7.43)	30.48(7.73)	<.01

^a Data for a general population sample of women aged 18 to 40 years (n = 495) (Mond et al., 2004) and for a clinical sample of eating disorder patients (n = 118) (Mond et al., 2005) are given for descriptive purposes.

^b Significance of the difference between non‐ED and ED subgroups according to an independent samples t‐test.

^c Eating Disorder Examination Questionnaire (EDE‐Q) subscales. Scores range from 0–6 with higher scores indicating greater symptom frequency or severity.

^d Medical Outcomes Study 12‐item Short‐Form Physical (PCS) and Mental (MCS) Component Summary scales. Lower scores indicate higher levels of perceived impairment.

^e Kessler Psychological Distress scale. Lower scores indicate higher levels of distress.

  ED = eating disorder.

Table-4 shows responses to each item for the AN and BN vignettes for the case and non‐case subgroups. As can be seen, the responses of symptomatic participants differed from those of asymptomatic participants for most items and for both vignettes, although differences were more pronounced for items assessing perceived acceptability than for items assessing perceived severity and more pronounced for items loading on the personal acceptability domain than those loading on the social acceptability domain. For example, whereas 18.6% of asymptomatic women had occasionally, often, or always thought that it ‘might not be too bad’ to have AN, 72.7% of women with symptoms gave these responses. Figures for the BN vignette were 26.9% and 70.4%.

Table 4 Responses to items assessing perceptions of the severity and acceptability of anorexia nervosa (AN) and bulimia nervosa (BN) for asymptomatic women (‘non‐ED’, n = 268) and women with eating disorder symptoms (‘ED’, n = 44)

–	Lucy's problem (AN)		χ	p	Kelly's problem (BN)		χ	p
–	Non‐ED	ED			Non‐ED	ED
–	%	%			%	%
How distressing
Not at all	0.7	4.5	17.42	<.01	0.4	4.5	21.56	<.01
A little	4.1	15.9			0.7	9.1
Moderately	9.0	11.4			15.7	15.9
Very	33.6	15.9			47.4	36.4
Extremely	52.6	52.3			35.8	34.1
How sympathetic
Not at all	1.5	2.3	5.80	.22	0.7	2.3	11.84	<.05
A little	9.3	18.2			5.6	13.6
Moderately	26.5	13.6			32.5	22.7
Very	37.7	36.4			42.2	27.3
Extremely	25.0	29.5			19.0	34.1
How difficult to treat
Not at all	0.00	2.3	8.79	.07	0.7	2.3	7.28	.12
A little	2.6	6.8			5.2	13.6
Moderately	17.2	13.6			47.0	31.8
Very	41.8	36.4			35.4	38.6
Extremely	38.4	40.9			11.6	13.6
Might not be too bad
Never	58.6	18.2	85.65	<.01	41.8	11.4	58.57	<.01
Rarely	22.8	9.1			31.3	18.2
Occasionally	16.0	34.1			20.9	29.5
Often	1.5	25.0			4.9	25.0
Always	1.1	13.6			1.1	15.9
Beneficial aspects
Definitely not	75.0	47.7	35.92	<.01	65.3	38.6	25.04	<.01
Possibly	14.9	11.4			19.4	22.7
Mixed feelings/yes and no	7.1	20.5			13.8	25.0
Probably	1.5	15.9			0.7	6.8
Definitely	1.5	4.5			0.7	6.8
How concerned?
Not at all	4.5	13.6	39.76	<.01	4.1	13.6	14.79	<.01
A little	7.1	22.7			9.0	15.9
Moderately	9.0	29.5			13.8	22.7
Very	21.3	11.4			34.7	29.5
Extremely	58.2	22.7			38.4	18.2
How happy?
Not at all	86.5	51.2	54.06	<.01	84.2	58.5	33.73	<.01
A little	11.2	19.5			12.3	19.5
Moderately	1.9	19.5			3.1	9.8
Very	0.4	2.4			0.4	7.3
Extremely	0.0	7.3			0.0	4.9
Make more popular
Definitely not	62.7	36.4	23.37	<.01	67.2	56.8	15.56	<.01
Possibly	20.9	22.7			17.9	27.3
Mixed feelings/yes and no	13.8	29.5			12.7	11.4
Probably	2.2	4.5			2.2	0.0
Definitely	0.4	6.8			0.0	4.5
Easier to fit in
Definitely not	53.7	40.9	10.26	<.05	52.2	38.6	3.06	.55
Possibly	25.0	22.7			20.9	25.0
Mixed feelings/yes and no	14.9	18.2			20.5	27.3
Probably	5.2	11.4			5.2	6.8
Definitely	1.1	6.8			1.1	2.3
Improve confidence/self‐esteem
Definitely not	71.9	38.6	23.76	<.01	79.1	50.0	29.58	<.01
Possibly	19.5	34.1			14.6	25.0
Mixed feelings/yes and no	7.1	18.2			5.2	13.6
Probably	1.1	6.8			1.1	6.8
Definitely	0.4	2.3			0.0	4.5

  ED = eating disorder.

Mann–Whitney U‐tests confirmed that participants with symptoms had higher scores than asymptomatic participants on both acceptability domains for both AN (personal acceptability: z = −6.11, p < .01; social acceptability: z = −3.86, p < .01) and BN (personal acceptability: z = −5.76, p < .01; social acceptability: z = −2.43, p < .05) vignettes, whereas scores on the severity domain did not differ between groups for either vignette (AN: z = −0.36, p = .72; BN: z = −0.05, p = .96).

Within‐subjects analysis indicated that AN was perceived to be a more severe disorder than BN for both asymptomatic (z = −7.5, p < .01) and symptomatic (z = −2.1, p < .05) participants. For asymptomatic participants, scores on the personal acceptability domain were higher for the BN vignette (z = −5.0, p < .01), whereas scores on the social acceptability domain did not differ by vignette type (z = −0.60, p = .54). For symptomatic participants, by contrast, scores on the social acceptability domain were higher for the AN vignette (z = −2.1, p < .05), whereas scores on the personal acceptability domain did not differ by vignette type (z = −0.50, p = .61).

Additional analysis was conducted to determine whether differences between groups in responses to at least some items assessing perceived acceptability might have been due, in part, to between‐group differences in body weight. For this analysis, dichotomous variables were created to indicate responses consistent with high perceived acceptability (e.g., ‘often’, ‘always’; ‘probably’, and ‘definitely’) for each item. Multiple logistic regression analysis was then used to examine the association between symptom status and the likelihood of these responses with and without BMI as a covariate. Results of this analysis confirmed that having eating disorder symptoms was associated with increased likelihood of perceived acceptability for all seven items for the AN vignette and for five of the seven items for the BN vignette. In 10 of these 12 cases, higher odds ratios—indicating a stronger association between symptom status and perceived acceptability—were observed after BMI was included in the analysis. In the remaining two cases, the odds ratios were unchanged.

DISCUSSION

We examined perceptions of the acceptability of eating‐disordered behaviour in young adult women with (n = 44) and without (n = 268) eating disorder symptoms. All participants viewed vignettes of AN and BN and responded to the same series of questions—addressing the perceived acceptability of the problems described—in relation to each vignette. It was hypothesised that perceptions of the acceptability of AN and BN would be more common among women with symptoms. The findings provided strong support for this hypothesis in that responses consistent with greater perceived acceptability were more common among women with symptoms for all but one of the seven items assessing perceived acceptability and for both AN and BN vignettes.

As has been noted (Mond et al., 2004a; Vander Wal & Thelen, 1997), at least two explanations might be given for the association between greater perceived acceptability of eating‐disordered behaviour and higher levels of actual eating disorder symptoms. First, individuals who have developed eating disorder symptoms may adjust their attitudes, perceiving eating‐disordered behaviour to be more acceptable, to minimise the dissonance between attitudes and behaviours. Second, the perception that eating disorder behaviours are acceptable may increase the likelihood that individuals choose to engage in those behaviours. We are not aware of any research to test this latter hypothesis, although plausible theoretical accounts of the link between perceptions of the acceptability of eating‐disordered behaviour and the subsequent use of specific behaviours have been proposed (Bergstrom & Neighbors, 2006; Crandall, 1988). It also is possible that the risk posed by established risk factors, such as body dissatisfaction and low self‐esteem, is increased when the perception that eating‐disordered behaviour is acceptable, or beneficial, is present.

Although the cross‐sectional design of the present study precludes any conclusions as to the direction of the observed associations, one implication of the findings is that it may be beneficial to target the perceived acceptability of eating‐disordered behaviour in prevention and early‐intervention programs for eating disorders. A focus of this kind could, for example, be accommodated within cognitive dissonance and/or media literacy approaches to eating disorders prevention, both of which have shown promising results (Stice, Shaw, Becker, & Rohde, 2008; Wilksch & Wade, 2008). As we have argued previously (Mond et al., 2010), efforts of this kind need to target not only individuals with early symptoms but also those with whom they interact and the public as a whole.

A secondary aim of the present research was to examine the association between perceptions of the acceptability of eating‐disordered behaviour and perceptions of the severity of eating‐disordered behaviour. It was hypothesised that an inverse association between these constructs would be observed, such that greater perceived severity of AN and BN would be associated with lower perceived acceptability of these conditions. Although this hypothesis was supported, the association was modest for both AN and BN vignettes. In fact, a substantial minority of participants appeared to consider eating disorder symptoms to be both serious and acceptable. Given the way in which eating disorders are portrayed in the popular media (Striegel‐Moore & Franko, 2003) and given the focus, in recent years, on individuals' need to monitor their energy intake and expenditure to maintain a ‘healthy’ weight (Austin, 1999), this ambivalence is not surprising. However, in view of the clear association between the perceived acceptability of eating‐disordered behaviour and actual eating disorder symptoms, it is concerning. Indeed, ambivalence of this kind is a characteristic of eating disorder patients in clinical practice (e.g., Gale et al., 2006).

By presenting vignettes of both AN and BN to all participants, it was possible to consider the comparative acceptability of the different disorders. It was hypothesised that perceptions of acceptability would be more common for AN than for BN, given the more conspicuously ego‐syntonic features of eating disorders characterised by low body weight (Mond et al., 2005; Vitousek et al., 1998). However, there was only partial support for this hypothesis. Whereas symptomatic participants did indeed view AN as being more acceptable than BN, asymptomatic participants viewed BN as being more acceptable than AN.

These findings should be considered tentative until replicated in a different sample. However, given that admiration from peers is a common response to loss of body weight in women (Schmeck & Poustka, 1998), and to the extent that individuals in the symptomatic subgroup have personal experience of weight cycling (e.g., Field, Manson, Taylor, Willett, & Colditz, 2004), the greater perceived acceptability of AN among symptomatic participants is plausible. Symptomatic women may also be more conscious of the fact that BN is unlikely to confer the same benefits, in terms of social standing, as AN. Healthy women, by contrast, may consider BN to be more acceptable than AN because they perceive the health risks of BN to be lower than those of AN (Hunt & Rothman, 2007). Importantly, the greater perceived acceptability of AN observed among women with symptoms could not be accounted for by between‐group differences in body weight. In fact, associations between symptom status and greater perceived acceptability tended to be stronger after the influence of BMI was statistically controlled.

Inclusion of multiple items assessing perceived acceptability in the present study made it possible to consider the ways in which eating disorder features might be seen to be acceptable. Of interest in this regard is that, for both AN and BN vignettes, items assessing perceived acceptability formed two domains, the first is relating to ways in which eating‐disordered behaviour might be considered to be acceptable, or beneficial at an individual or personal level, the second relating to ways in which eating‐disordered behaviour might be of benefit socially. Further, differences between symptomatic and asymptomatic participants were more pronounced for items tapping the personal acceptability domain than for those tapping the social acceptability domain.

These findings should also be considered tentative until replicated in a different sample. It is possible that the perception that eating disorder behaviours confer social benefits is less closely tied to the development of symptoms because benefits of this kind are seen as less salient, or, perhaps, less likely to be achieved, than those relating to personal acceptability. Alternatively, the experience of women with symptoms may be that social benefits are less pronounced than benefits at the personal level, such that there is less dissonance between attitudes and experience relating to social acceptability and, in turn, less need for attitudes to be modulated. In considering these possible interpretations, it needs to be remembered that in community‐based studies of eating‐disordered behaviour, including the present study, participants with symptoms are typically individuals with variants of BN rather than individuals with AN‐type conditions (Mond et al., 2004).

Study limitations and other methodological considerations

Several limitations need to be considered when interpreting the present findings. First, participants were university students who chose to complete an online assessment, as opposed to a general population sample of women. As a consequence, individuals with eating disorder symptoms were over‐represented among participants. While this proved advantageous in identifying a subgroup of symptomatic participants, it necessarily limits the generalisability of the findings. Second, results of factors analysis should always be interpreted in light of the ‘garbage in, garbage out’ problem (Ismail, 2008). Hence, findings relating to the identification of ‘personal’ and ‘social’ acceptability domains in particular should be considered tentative until replicated in a different sample. Third, individuals' reports of their attitudes and beliefs concerning mental health problems may be subject to a social desirability bias. Participants in the present study may, for example, have been inclined to overstate perceived severity and to understate perceived acceptability (Mond et al., 2004a). Finally, the present study was confined to young adult women. Although this was reasonable, given that we were interested in the association between perceptions of the acceptability of eating‐disordered behaviour and actual eating disorder symptoms, replication of the study in men and in younger and older women would be of interest (Mond et al., 2010). Strengths of the present study included sample size sufficient to permit comparison between symptomatic and asymptomatic participants, inclusion of both AN and BN vignettes and inclusion of multiple items assessing perceived acceptability. In addition, the inclusion of measures of distress and functional impairment—and of general population and clinical comparison groups—made it possible to verify that participants identified as ‘symptomatic’ did, in fact, have high symptom levels.

In sum, the findings indicate the ambivalence towards eating‐disordered behaviour that exists among a subgroup of young women in the community and the clear association between such ambivalence and actual eating disorder symptoms. The perceived acceptability of eating‐disordered behaviour may need to be addressed in prevention and early‐intervention programs for eating disorders.

Notes

Conflict of Interest: None.

Appendix APPENDIX: VIGNETTES USED IN THE MENTAL HEALTH LITERACY SURVEY

AN vignette (‘Lucy's problem’)

Lucy is a 19‐year‐old student in her second year of university. Although she has never been overweight, Lucy had been very conscious of the changes in her body shape that occurred during adolescence and has always wanted to be thinner. During her first year of university, Lucy joined a fitness programme at the gym and started running daily. Through this effort, she gradually began to lose weight. At the same time, Lucy started to ‘diet’, avoiding fatty foods, not eating between meals, and trying to eat set portions of ‘healthy foods’. On some days, she does not eat anything at all. Through this combination of dieting and exercise, Lucy has further reduced her weight, to the point that she is well below average for her age and height, and her periods have stopped. Despite her thin and gaunt appearance, Lucy denies that she is underweight. In fact, she is terrified of becoming ‘fat’ and refuses to make any effort to gain weight. As a result, Lucy's relationship with her family has become strained, and her grades have started to slip.

BN vignette (‘Kelly's problem’)

Kelly is a 19‐year‐old university student. Although mildly overweight when she was younger, Kelly's current weight is within the normal range for her age and height. However, she thinks she is overweight. Upon starting her first year of university, Kelly joined a fitness program at the gym and also started running regularly. Through these efforts, she gradually began to lose weight. Kelly then started to ‘diet’, avoiding fatty foods, not eating between meals, and trying to eat set portions of ‘healthy foods’. Kelly also continued with the exercise program, losing several more kilograms. However, she found it difficult to maintain the weight loss and for the past 18 months, her weight has been continually fluctuating, sometimes as much as 5-kg within a few weeks. Kelly has also found it difficult to control her eating. While able to limit her dietary intake during the day, at night she is often unable to stop eating, bingeing on, for example, a loaf of bread, some ice cream, or chocolate, and several pieces of fruit. To counteract the effects of this bingeing, Kelly takes laxatives. On other occasions, she vomits after overeating. Because of her strict routines of eating and exercising, Kelly has lost contact with most of her friends.