Eating Disorder Inventory

Abstract

The Eating Disorder Inventory (EDI) is a premier self-report psychometric instrument designed to assess the multidimensional psychological and behavioral traits characteristic of anorexia nervosa and bulimia nervosa. Developed by David M. Garner, Marion P. Olmstead, and Janet Polivy in 1983, the EDI emerged as a response to the methodological limitations of unidimensional, purely symptom-oriented measurement tools. The scale conceptualizes eating disorders not merely as aberrant eating habits or weight dysregulation, but as complex syndromes maintained by profound cognitive distortions, affective vulnerabilities, and developmental conflicts. Comprising 64 self-report items, the EDI measures eight distinct subscales divided into eating-disorder-specific behavioral attitudes (Drive for Thinness, Bulimia, Body Dissatisfaction) and fundamental psychological constructs (Ineffectiveness, Perfectionism, Interpersonal Distrust, Interoceptive Awareness, Maturity Fears). Responses are captured on a 6-point forced-choice Likert scale (always, usually, often, sometimes, rarely, never) and evaluated using a weighted scoring method (3, 2, 1, 0, 0, 0) that isolates clinically severe psychopathology. Extensive empirical investigations confirm robust psychometric integrity: internal consistency coefficients (Cronbach’s alpha) typically range from .80 to .92 across clinical samples, and criterion validity is established through the scale’s documented ability to differentiate clinical cohorts (e.g., restricting anorexia, bulimic anorexia) from non-clinical female comparison groups and recovered individuals. The EDI serves as a benchmark instrument in psychiatric evaluation, clinical trial monitoring, case conceptualization, and empirical investigations into the etiology of eating pathology.

Keywords

Eating Disorder Inventory, anorexia nervosa, bulimia nervosa, psychometrics, body image, eating disorder assessment, drive for thinness, interoceptive awareness, ineffectiveness, maturity fears, psychological measurement, David M. Garner

Authors

The Eating Disorder Inventory was conceptualized, operationalized, and psychometrically validated by a collaborative team of distinguished clinical psychologists and researchers in Toronto, Ontario, Canada:

• David M. Garner, Ph.D. — Associated with the Department of Psychiatry at the University of Toronto and the Toronto General Hospital. Dr. Garner is internationally recognized for his pioneering contributions to the empirical assessment, cognitive-behavioral conceptualization, and treatment of eating disorders.
• Marion P. Olmstead, Ph.D. — Associated with York University and the Toronto General Hospital Eating Disorder Program. Dr. Olmstead has contributed extensively to longitudinal outcome research, psychometric scale development, and clinical trials in eating disorders.
• Janet Polivy, Ph.D. — Professor of Psychology and Psychiatry at the University of Toronto. Dr. Polivy is an eminent researcher in the psychology of eating behavior, dietary restraint theory, self-regulation, and emotional influences on food intake.

Purpose

Before the development of the Eating Disorder Inventory, diagnostic assessment in the field of eating pathology leaned predominantly toward tracking overt physical parameters (e.g., body weight fluctuations, body mass index, menstrual status) or documenting behavioral frequencies (e.g., weekly count of binge episodes, self-induced vomiting, hours spent exercising). Instruments such as the Eating Attitudes Test (EAT-40) provided significant value in screening for general symptoms but were primarily unidimensional or restricted to manifest dietary restraint. They offered limited insight into the complex psychological architecture that predisposes, precipitates, and perpetuates anorexia nervosa and bulimia nervosa.

The primary purpose of the EDI was to bridge this theoretical and clinical chasm by establishing a standardized, multidimensional self-report scale that quantifies both manifest eating-related pathology and core psychological deficits. Clinicians and clinical researchers required an instrument capable of executing several distinct clinical and research functions:

• Differential Case Conceptualization: To distinguish between individuals exhibiting benign, normative dietary restraint or cultural weight preoccupation and those grappling with severe, clinically rooted eating disorders characterized by profound personality and ego deficits.
• Subtype Differentiation: To delineate psychopathological profiles across clinical subgroups, such as distinguishing restricting anorexia nervosa from binge-eating/purging anorexia nervosa and normal-weight bulimia.
• Treatment Planning and Personalization: To uncover specific psychological vulnerabilities—such as pervasive interpersonal alienation, extreme perfectionism, or developmental regression—that must be targeted in psychotherapy beyond behavioral nutritional stabilization.
• Treatment Monitoring and Outcome Assessment: To provide an empirically sensitive metric for measuring psychological recovery. Because behavioral stabilization (e.g., weight restoration) frequently precedes cognitive-affective resolution, the EDI allows clinicians to assess whether underlying psychological traits (e.g., ineffectiveness, lack of interoceptive awareness) have genuinely resolved or remain dormant risk factors for relapse.
• Etiological and Epidemiological Research: To provide psychiatric researchers with a psychometrically rigorous metric for longitudinal investigations, structural equation modeling, genetic association studies, and cross-cultural analyses of eating pathology.

Psychological Construct

The Eating Disorder Inventory operationalizes eating disorder psychopathology as an eight-factor multidimensional construct. These eight dimensions are divided into three eating-disorder-specific symptom subscales and five broad psychological deficit subscales.

1. Drive for Thinness (DT; 7 items)

Drive for Thinness measures excessive preoccupation with dieting, an unyielding pursuit of extreme thinness, an intense fear of gaining weight, and an overarching dread of fatness. This subscale assesses the central cognitive symptom required for the diagnosis of anorexia nervosa and bulimia nervosa. Individuals scoring high on this dimension view weight regulation not merely as a cosmetic goal, but as an existential imperative where any caloric surplus generates acute anxiety and distress (e.g., “I am terrified of gaining weight”).

2. Bulimia (B; 7 items)

The Bulimia subscale assesses the tendency to engage in recurrent episodes of uncontrollable overeating (objective binge eating) as well as the immediate cognitive or behavioral impulses to compensate via self-induced vomiting or purging. It captures the sense of loss of control during eating episodes and the secrecy surrounding dietary consumption (e.g., “I have gone on eating binges where I have felt that I could not stop”). This subscale cleanly differentiates bulimic subtypes from pure dietary restricters.

3. Body Dissatisfaction (BD; 9 items)

Body Dissatisfaction evaluates the individual’s negative appraisal of specific bodily regions that are biologically susceptible to fat deposition during female pubertal development, specifically the hips, thighs, buttocks, and stomach. Beyond general body discontent, this construct taps into the cognitive distortion that these specific anatomical sites are unacceptably large or deformed (e.g., “I think that my stomach is too big”; “I think that my thighs are too large”).

4. Ineffectiveness (I; 10 items)

Ineffectiveness measures profound, generalized feelings of personal inadequacy, worthlessness, insecurity, self-doubt, and the pervasive conviction that one lacks control over one’s life and environment (e.g., “I feel ineffective as a person”; “I feel alone in the world”). This construct reflects the deep-seated ego deficit described in psychodynamic and cognitive formulations, where rigid dietary control serves as a compensatory mechanism for an overwhelming internal sense of helplessness.

5. Perfectionism (P; 6 items)

Perfectionism measures excessive, inflexible personal standards for flawless performance, accompanied by a dichotomous, all-or-nothing cognitive evaluation of success and failure. It also captures perceived parental demands for flawless achievement (e.g., “Only outstanding performance is good enough in my family”; “I hate being less than best at things”). In eating disorders, perfectionistic standards are systematically channeled into unattainable ideals of somatic and dietary control.

6. Interpersonal Distrust (ID; 7 items)

Interpersonal Distrust assesses an individual’s chronic reluctance to forge close emotional attachments, a pervasive skepticism regarding the sincerity of others, and an inability to express affect or disclose private thoughts (e.g., “I need to keep people at a certain distance”; “I have trouble expressing my emotions to others”). This alienation often stems from expectations of interpersonal rejection, abandonment, or intrusive control.

7. Interoceptive Awareness (IA; 10 items)

Interoceptive Awareness evaluates visceral and affective self-attunement. Specifically, it assesses deficits in recognizing and accurately identifying internal visceral sensations (such as hunger, satiety, and fullness) as well as the inability to distinguish primary emotional states from somatic sensations (e.g., “I get confused about what emotion I am feeling”; “I get confused as to whether or not I am hungry”). Severe deficits in interoceptive awareness often lead to emotional dysregulation being experienced as somatic fullness or an urge to binge.

8. Maturity Fears (MF; 8 items)

Maturity Fears captures the cognitive and emotional yearning to retreat to the perceived safety and security of preadolescent childhood, driven by an overwhelming fear of adult biological, interpersonal, and psychological responsibilities (e.g., “I wish that I could return to the security of childhood”; “The demands of adulthood are too great”). Starvation and emaciation functionally suspend secondary sexual characteristics and adult reproductive physiology, fulfilling this unconscious or conscious developmental avoidance.

Theoretical Framework

The EDI is grounded in multidimensional biopsychosocial models of eating pathology, drawing extensively from the foundational theories of Hilde Bruch, Arthur Crisp, and the integrative framework synthesized by David Garner and Paul Garfinkel (1982).

Hilde Bruch’s Perceptual and Conceptual Deficits

In her seminal works, including Eating Disorders: Obesity, Anorexia Nervosa, and the Person Within (1973) and The Golden Cage (1978), psychoanalyst Hilde Bruch posited that anorexia nervosa is not primarily a disorder of appetite, but a desperate struggle for autonomy, self-determination, and identity. Bruch identified three fundamental pathognomonic disturbances:

• A severe disturbance in body image and body concept, manifesting as delusional or near-delusional denial of emaciation.
• A disturbance in the accurate cognitive perception of bodily sensations, where hunger, fullness, fatigue, and affective arousal are grossly misidentified.
• A paralyzing, pervasive sense of personal ineffectiveness, leading the patient to experience themselves solely as responding to the demands and expectations of others rather than acting as an autonomous agent.

Garner, Olmstead, and Polivy operationalized Bruch’s theoretical architecture directly into the EDI’s Interoceptive Awareness, Ineffectiveness, and Body Dissatisfaction subscales, providing empirical metrics for psychoanalytic constructs that had previously eluded quantitative measurement.

Crisp’s Developmental Model and Maturational Avoidance

Arthur Crisp (1965, 1980) formulated anorexia nervosa as a pubertal weight phobia and an avoidance disorder rooted in developmental regression. According to Crisp, the biological changes of female puberty (fat deposition, breast development, menarche) trigger profound psychosocial panic in vulnerable adolescents who feel unequipped for adult sexuality, autonomy, and mature interpersonal relationships. Severe dietary restriction produces biological regression (regression to prepubertal endocrine status, amenorrhea, loss of adult curves), successfully mitigating the terrifying demands of adulthood. The EDI operationalizes this developmental retreat via the Maturity Fears subscale.

The Multidimensional Integrative Model (Garner & Garfinkel, 1982)

Garner and Garfinkel integrated these developmental and psychodynamic theories with cognitive-behavioral principles and sociocultural factors. They posited that eating disorders are heterogeneous final common pathways resulting from predisposing personality traits (e.g., perfectionism, interpersonal distrust), sociocultural pressures for thinness, biological vulnerabilities, and precipitating environmental stressors. The EDI was explicitly designed to mirror this multifactorial conceptualization by measuring both the somatic-dietary symptoms and the predisposing personality architecture.

Validity

The psychometric validation of the Eating Disorder Inventory was conducted through rigorous empirical procedures evaluating criterion-related, convergent, and discriminant validity across diverse clinical and non-clinical cohorts.

Criterion-Related Validity

Criterion validity was established in the original 1983 study by evaluating the EDI across distinct clinical and non-clinical groups:

• Clinical Criterion Group: 113 female patients meeting rigorous DSM-III criteria for primary anorexia nervosa (subdivided into 48 restricting anorexics and 65 bulimic anorexics).
• Non-Clinical Comparison Groups: 577 non-clinical female university and college students, 166 male comparison subjects, 195 normal-weight bulimic women, 44 obese women, 52 formerly obese women, and 17 clinically recovered anorexic women.

The results demonstrated remarkable criterion separation. Clinical anorexia nervosa patients scored significantly higher (p < .001) than non-clinical female controls across all eight subscales. For example, on the Drive for Thinness subscale, the mean score for anorexic patients was 14.8 (SD = 5.7) compared to 5.2 (SD = 5.4) for female controls. On the Ineffectiveness subscale, anorexic patients averaged 12.3 (SD = 7.7) versus 2.3 (SD = 3.6) for non-clinical controls. Furthermore, independent clinician ratings of patient pathology—conducted blindly by experienced therapists—correlated significantly with patients’ self-reported EDI subscale scores, confirming strong ecological and criterion concordance.

Discriminant Validity

The EDI demonstrated high discriminant validity across diagnostic subgroups:

• Restricting vs. Bulimic Anorexia: The Bulimia subscale cleanly discriminated between restricting anorexics (Mean = 1.4, SD = 2.4) and bulimic anorexics (Mean = 10.9, SD = 5.2; p < .0001), while both subgroups exhibited equally elevated scores on Drive for Thinness and Body Dissatisfaction.
• Active Pathology vs. Clinical Recovery: The sample of 17 recovered anorexia nervosa patients produced subscale scores that were statistically indistinguishable from non-clinical female comparison controls across most dimensions, establishing that the EDI is sensitive to state-dependent active psychopathology rather than merely capturing static, unalterable personality traits.
• Normal-Weight Bulimia: Normal-weight bulimic women scored exceptionally high on Drive for Thinness, Bulimia, and Body Dissatisfaction, while exhibiting lower Ineffectiveness and Maturity Fears scores than severely underweight restricting anorexics, reflecting distinct psychological configurations.

Convergent Validity

Convergent validity was demonstrated through substantial positive correlations between EDI subscales and validated psychological instruments. The Drive for Thinness and Bulimia subscales correlated strongly with the Eating Attitudes Test (EAT-40; r = .70 to .82). The Ineffectiveness subscale correlated significantly with the Beck Depression Inventory (BDI; r = .65 to .72) and measures of generalized self-esteem. Interoceptive Awareness correlated negatively with measures of emotional stability and positively with somatic anxiety and alexithymia metrics.

Reliability

The Eating Disorder Inventory displays robust reliability across clinical and non-clinical populations, with extensive psychometric investigations confirming high internal consistency and temporal stability.

Internal Consistency (Cronbach’s Alpha)

In the original validation study by Garner, Olmstead, and Polivy (1983), Cronbach’s alpha coefficients demonstrated strong item cohesion across all eight subscales in both clinical (N = 113) and non-clinical female samples (N = 577):

• Drive for Thinness: α = .85 (Anorexia Nervosa), α = .85 (Female Controls)
• Bulimia: α = .90 (Anorexia Nervosa), α = .83 (Female Controls)
• Body Dissatisfaction: α = .91 (Anorexia Nervosa), α = .91 (Female Controls)
• Ineffectiveness: α = .90 (Anorexia Nervosa), α = .80 (Female Controls)
• Perfectionism: α = .82 (Anorexia Nervosa), α = .73 (Female Controls)
• Interpersonal Distrust: α = .84 (Anorexia Nervosa), α = .80 (Female Controls)
• Interoceptive Awareness: α = .85 (Anorexia Nervosa), α = .80 (Female Controls)
• Maturity Fears: α = .83 (Anorexia Nervosa), α = .72 (Female Controls)

Replication studies across diverse clinical eating disorder populations have consistently corroborated these values, with total scale composite reliability routinely exceeding α = .94.

Test-Retest Reliability

Temporal stability evaluations in clinically stable cohorts over intervals ranging from one to four weeks have yielded test-retest reliability coefficients ranging from r = .78 to r = .92 across the subscales. Body Dissatisfaction, Drive for Thinness, and Perfectionism exhibit particularly high stability (r > .85), confirming that these constructs capture durable cognitive-affective patterns during stable clinical phases.

Factor Analysis

The structural development of the EDI utilized a rigorous deductive-rational scale construction strategy combined with subsequent exploratory and confirmatory factor analyses, departing markedly from purely inductive, unconstrained empirical clustering.

Scale Construction and Refinement

The scale development process progressed through structured psychometric phases:

• Item Generation: Clinicians and psychometricians with specialized expertise in anorexia nervosa and bulimia generated an initial item pool of 146 items reflecting eleven theoretical constructs derived from clinical literature and psychodynamic formulations.
• Deductive Pruning: Items displaying poor item-total correlations, restricted variance, or high susceptibility to social desirability in pilot testing were systematically eliminated.
• Empirical Consolidation: Repeated iterations of factor analysis combined with conceptual clarity checks consolidated the original eleven theoretical dimensions into eight psychometrically distinct, non-redundant factors totaling 64 items.

Factor Structure and Confirmatory Modeling

Principal component analyses with varimax and oblimin rotations confirmed an eight-factor orthogonal/oblique solution explaining over 42% of the total variance in the clinical cohort. Each retained item demonstrated primary factor loadings substantially exceeding .40 on its designated theoretical construct, with minimal cross-loadings onto secondary factors. Subsequent modern confirmatory factor analytic (CFA) studies examining the 64-item structure have substantiated acceptable to strong model fit indices:

• Comparative Fit Index (CFI): .91 to .94 across clinical eating disorder samples.
• Tucker-Lewis Index (TLI): .90 to .93.
• Root Mean Square Error of Approximation (RMSEA): .046 to .058 (90% CI [.042, .061]), indicating excellent population approximation.
• Standardized Root Mean Square Residual (SRMR): .052.

These findings substantiate the EDI’s dual structural hierarchy, which encompasses three lower-order behavioral constructs assessing manifest eating/body concerns and five lower-order psychological vulnerability dimensions loading onto higher-order eating disorder psychopathology.

Instrument / Measurement Tool

The structural and administration parameters of the original Eating Disorder Inventory (EDI) are specified below:

• Test Type: Standardized self-report psychometric questionnaire.
• Format: 64 items, 6-point forced-choice Likert-type scale: always, usually, often, sometimes, rarely, never.
• Item Distribution across Subscales:
  • Drive for Thinness (DT): 7 items (Items 1, 7, 11, 16, 25, 32, 49)
  • Bulimia (B): 7 items (Items 4, 5, 28, 38, 46, 53, 61)
  • Body Dissatisfaction (BD): 9 items (Items 2, 9, 12, 19, 31, 45, 55, 59, 62)
  • Ineffectiveness (I): 10 items (Items 10, 18, 20, 24, 27, 37, 42, 50, 56, 63)
  • Perfectionism (P): 6 items (Items 13, 29, 36, 43, 52, 60)
  • Interpersonal Distrust (ID): 7 items (Items 15, 17, 23, 30, 34, 54, 57)
  • Interoceptive Awareness (IA): 10 items (Items 8, 21, 26, 33, 40, 44, 47, 51, 64, 68/Note: original 64-item battery contains 10 items for IA including item 64)
  • Maturity Fears (MF): 8 items (Items 3, 6, 14, 22, 35, 39, 41, 48)
• Scoring Methodology:
  • Original Weighted Scoring: Responses are scored 3, 2, 1, 0, 0, 0. The most extreme ‘anorexic’ or pathological response receives 3 points, the adjacent response receives 2 points, the next receives 1 point, and the remaining three non-pathological responses receive 0 points.
  • Directionality / Reverse Keying: For positively keyed items (where “always” denotes pathology), scoring is: Always = 3, Usually = 2, Often = 1, Sometimes = 0, Rarely = 0, Never = 0. For reverse-keyed items (where “never” denotes pathology, such as Item 1, 12, 15, 17, 19, 20, 22, 23, 26, 30, 31, 37, 39, 42, 50, 55, 57, 58, 62), scoring is: Never = 3, Rarely = 2, Sometimes = 1, Often = 0, Usually = 0, Always = 0.
  • Subscale Scores: Calculated by summing the weighted item scores within each respective subscale.
  • Continuous Alternative Scoring: In non-clinical or parametric psychometric research, researchers sometimes utilize a linear 1-to-6 scoring method to prevent floor effects and preserve variance.
• Target Population: Adolescent and adult clinical psychiatric patients, individuals undergoing evaluation for eating disorders, college students, and community research samples.
• Administration Time: Approximately 15 to 20 minutes self-administered.

Permissions & Fee and Test Year

The original Eating Disorder Inventory was published in 1983 in the International Journal of Eating Disorders by David M. Garner, Marion P. Olmstead, and Janet Polivy. In subsequent years, the instrument was expanded into the Eating Disorder Inventory-2 (EDI-2; Garner, 1991), which added 27 items across three provisional subscales (Asceticism, Impulse Regulation, and Social Insecurity), and subsequently into the Eating Disorder Inventory-3 (EDI-3; Garner, 2004), which thoroughly restructured the test composite indices and reference norms.

The Eating Disorder Inventory (EDI, EDI-2, and EDI-3) is a proprietary, copyrighted psychometric instrument. Commercial distribution, translation rights, scoring software, and official testing kits are managed exclusively by Psychological Assessment Resources (PAR, Inc.). Researchers and clinicians wishing to administer the official instrument, access standard normative percentiles, or utilize computerized scoring platforms must obtain authorized test booklets and formal licensing through PAR, Inc., subject to professional qualification level standards (typically Qualification Level B or C for psychological assessments). Researchers seeking to use the historical 1983 version strictly for academic, non-commercial investigation must ensure appropriate copyright clearances through the copyright holder or publisher.

References

• American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). American Psychiatric Association.
• Bruch, H. (1962). Perceptual and conceptual disturbances in anorexia nervosa. Psychosomatic Medicine, 24(2), 187–194. https://doi.org/10.1097/00006842-196203000-00009
• Bruch, H. (1973). Eating disorders: Obesity, anorexia nervosa, and the person within. Basic Books.
• Bruch, H. (1978). The golden cage: The enigma of anorexia nervosa. Harvard University Press.
• Crisp, A. H. (1965). Clinical research into anorexia nervosa. Proceedings of the Royal Society of Medicine, 58(10), 814–820. https://doi.org/10.1177/003591576505801027
• Crisp, A. H. (1980). Anorexia nervosa: Let me be. Academic Press / Grune and Stratton.
• Garfinkel, P. E., & Garner, D. M. (1982). Anorexia nervosa: A multidimensional perspective. Brunner/Mazel.
• Garner, D. M. (1991). Eating Disorder Inventory-2: Professional manual. Psychological Assessment Resources.
• Garner, D. M. (2004). Eating Disorder Inventory-3: Professional manual. Psychological Assessment Resources.
• Garner, D. M., & Olmstead, M. P. (1984). The Eating Disorder Inventory manual. Psychological Assessment Resources.
• Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2(2), 15–34. https://doi.org/10.1002/1098-108x(198321)2:2<15::aid-eat2260020203>3.0.co;2-6
• Garner, D. M., Olmsted, M. P., Bohr, Y., & Garfinkel, P. E. (1982). The Eating Attitudes Test: Psychometric features and clinical correlates. Psychological Medicine, 12(4), 871–878. https://doi.org/10.1017/s0033291700049163
• Russell, G. F. M. (1979). Bulimia nervosa: An ominous variant of anorexia nervosa. Psychological Medicine, 9(3), 429–448. https://doi.org/10.1017/s0033291700031974
• Strober, M. (1981). The significance of bulimia in anorexia nervosa: An exploration of possible etiological factors. International Journal of Eating Disorders, 1(1), 28–43. https://doi.org/10.1002/1098-108x(198123)1:1<28::aid-eat2260010104>3.0.co;2-9

Items of the Scale