Eating Attitudes Test-26 (EAT-26)

Abstract

The Eating Attitudes Test-26 (EAT-26) is one of the most widely utilized, extensively validated screening instruments designed to assess eating disorder risk, abnormal eating attitudes, and preoccupation with food and body weight. Derived from the original 40-item Eating Attitudes Test (EAT-40) developed by David M. Garner and colleagues, the EAT-26 retains robust psychometric qualities while offering enhanced clinical utility, reduced administration burden, and refined factor structure. The measure evaluates three distinct symptom dimensions: Dieting (avoidance of fattening foods and preoccupation with thinness), Bulimia and Food Preoccupation (bulimic behaviors, binge eating, and pervasive intrusive thoughts about food), and Oral Control (self-regulation, environmental pressures to eat, and perceived external control over food intake). The instrument consists of 26 core self-report items administered via a 6-point forced-choice response format, scored nonlinearly on a 0-to-3 metric to selectively capture symptomatic and pathological extremes, alongside supplementary behavioral inquiries assessing compensatory purging, laxative abuse, excessive exercise, and precipitous weight loss over a six-month window.

Extensive psychometric investigations have affirmed the EAT-26’s structural validity, showing internal consistency coefficients (Cronbach’s alpha) ranging from .80 to .90 across diverse clinical cohorts with anorexia nervosa, bulimia nervosa, and other specified feeding or eating disorders, as well as general community populations, high school students, university undergraduates, and elite athletic groups. Using the established clinical cutoff threshold of 20 or higher, the scale demonstrates exceptional criterion sensitivity and specificity for identifying individuals who warrant immediate, comprehensive clinical psychiatric evaluation. This article presents an exhaustive, academically rigorous review of the EAT-26, detailing its historical derivation, psychological constructs, theoretical underpinnings, psychometric validation data, factor-analytic evidence, diagnostic utility, administrative protocols, and complete authentic scale inventory.

Keywords

Eating Attitudes Test-26, EAT-26, eating disorders, anorexia nervosa, bulimia nervosa, psychometrics, screening instrument, body image disturbance, dietary restriction, oral control

Authors

The EAT-26 was developed by a team of prominent psychiatric and psychological researchers led by David M. Garner, Ph.D., Marion P. Olmsted, Ph.D., Yvonne Bohr, Ph.D., and Paul E. Garfinkel, M.D. At the time of the instrument’s design and publication in 1982, the investigative team conducted their foundational psychometric research at the Clarke Institute of Psychiatry (now incorporated into the Centre for Addiction and Mental Health, CAMH) and the Department of Psychiatry at the University of Toronto, Ontario, Canada.

Primary Author Profiles:

• David M. Garner, Ph.D.: Renowned clinical psychologist recognized internationally for pioneering the empirical measurement of eating pathology, including the EAT-40, EAT-26, and the Eating Disorder Inventory (EDI). Dr. Garner subsequently served as the Director of the River Centre Clinic in Sylvania, Ohio, dedicating decades of clinical practice and research to evidence-based assessment and residential treatment of eating disturbances.
• Paul E. Garfinkel, M.D., FRCPC: Eminent academic psychiatrist, author, and global authority on the etiology, neurobiology, and clinical treatment of anorexia nervosa and bulimia nervosa. Dr. Garfinkel served as President and Psychiatrist-in-Chief of CAMH and Chair of the Department of Psychiatry at the University of Toronto.
• Marion P. Olmsted, Ph.D., C.Psych.: Senior clinician and researcher within the Eating Disorder Program at the University Health Network and Associate Professor at the University of Toronto, whose work has advanced longitudinal tracking and cognitive-behavioral paradigms in eating pathology.
• Yvonne Bohr, Ph.D.: Clinical psychologist specializing in developmental psychopathology, infant mental health, and family dynamics, serving on the faculty of the Department of Psychology at York University.

Purpose

The primary purpose of the Eating Attitudes Test-26 is to serve as a standardized, cost-effective, easily interpretable, and psychometrically sound screening tool capable of identifying eating disorder risk in clinical, non-clinical, and epidemiological research contexts. Originally derived from the 40-item scale (Garner & Garfinkel, 1979), the EAT-26 was engineered to isolate the most reliable and clinically informative items, eliminating psychometric redundancies while maintaining near-perfect diagnostic convergence with its parent instrument (r = .98).

Crucially, the EAT-26 is intended as an initial screening instrument rather than a definitive standalone diagnostic interview. Its theoretical rationale rests upon the premise that aberrant attitudes toward eating, distorted appraisals of body shape, intense fears of adiposity, and rigid behavioral regimes surrounding food intake represent core cognitive-behavioral vulnerabilities that precede, sustain, and exacerbate clinical eating disorders such as anorexia nervosa, bulimia nervosa, and binge eating disorder. By identifying individuals exhibiting these cognitive and behavioral markers, clinicians and researchers can implement early secondary prevention strategies, initiate targeted interventions, or direct high-risk respondents to structured diagnostic assessments, such as the Eating Disorder Examination (EDE) or the Structured Clinical Interview for DSM Disorders (SCID).

In epidemiological and primary care settings, the instrument provides an efficient method for screening large student populations, adolescent cohorts, military recruits, and artistic or sporting subcultures—such as dancers, gymnasts, wrestling competitors, and long-distance runners—who face heightened socio-environmental pressures regarding lean physical appearance. In clinical treatment settings, the EAT-26 serves as an objective outcome metric administered pre- and post-intervention to quantify therapeutic attenuation in dietary obsession, bulimic symptom severity, and compulsive oral control throughout cognitive-behavioral therapy (CBT), family-based therapy (FBT), or specialized pharmacotherapy.

Psychological Construct

The psychological construct evaluated by the EAT-26 encompasses multidimensional, pathological eating-related cognitions, emotional reactions, and behavioral patterns. These manifestations stem from an overvaluation of shape and weight and an intense fear of weight gain. Through systematic factor analysis of clinical and non-clinical cohorts, Garner et al. (1982) demonstrated that this overarching construct cleanly decomposes into three distinct yet interrelated psychological subscales:

1. Dieting (Factor I, 13 Items)

The Dieting dimension constitutes the largest and most robust subscale of the EAT-26, containing 13 items: 1, 6, 7, 10, 11, 12, 14, 16, 17, 22, 23, 24, and 25. This construct captures a pervasive, rigid preoccupation with thinness, profound anxiety regarding the prospect of becoming overweight, and conscious cognitive avoidance of energy-dense foods, particularly carbohydrates and refined sugars. Individuals scoring high on this dimension display hypervigilance regarding caloric intake (e.g., Item 6: “Aware of the calorie content of foods that I eat”), engage in severe dietary restriction, and consume specialized diet products (Item 17: “Eat diet foods”). Furthermore, this subscale taps into cognitive manifestations of somatosensory distortion and body disparagement (Item 14: “Am preoccupied with the thought of having fat on my body”), alongside profound negative affect, shame, and remorse following food ingestion (Item 10: “Feel extremely guilty after eating”). It also encompasses behavioral manifestations where physical emptiness is pathologically endorsed as comfort or achievement (Item 24: “Like my stomach to be empty”).

2. Bulimia and Food Preoccupation (Factor II, 6 Items)

The Bulimia and Food Preoccupation subscale comprises 6 items: 3, 4, 9, 18, 21, and 26. This construct reflects intrusive cognitive preoccupations with food that disrupt daily functioning, alongside episodic loss-of-control eating and extreme compensatory behaviors. High scores indicate that the individual feels dominated by thoughts of nourishment (Item 18: “Feel that food controls my life”; Item 21: “Give too much time and thought to food”). At the behavioral level, this dimension captures the subjective experience of uncontrollable binge episodes (Item 4: “Have gone on eating binges where I feel that I may not be able to stop”) and active purging via self-induced emesis (Item 9: “Vomit after I have eaten”). Item 26 (“Enjoy trying new rich foods”) serves as a reverse-scored marker of nutritional neophobia, cognitive rigidity, and the inability to experience pleasure in culinary variety, reflecting the anhedonic restriction typical of severe eating pathology.

3. Oral Control (Factor III, 7 Items)

The Oral Control dimension encompasses 7 items: 2, 5, 8, 13, 15, 19, and 20. This construct measures the respondent’s conscious efforts to exercise strict self-control over food intake, often in direct opposition to physiological hunger signals and social expectations. It encapsulates deliberate avoidance behaviors (Item 2: “Avoid eating when I am hungry”) and ritualized eating habits, such as prolonged meal times and cutting sustenance into diminutive portions (Item 5: “Cut my food into small pieces”; Item 15: “Take longer than others to eat my meals”). Furthermore, Oral Control uniquely assesses interpersonal dynamics surrounding eating, capturing an individual’s acute awareness of external concern or pressure from family members, peers, or clinicians regarding their low weight or minimal food consumption (Item 8: “Feel that others would prefer if I ate more”; Item 13: “Other people think that I am too thin”; Item 20: “Feel that others pressure me to eat”). High scores on this subscale reflect both active behavioral self-denial and the perception of an adversarial social environment challenging the respondent’s restrictive habits.

Theoretical Framework

The conceptual framework underpinning the EAT-26 is rooted in the cognitive-behavioral formulations of eating disorders, initially synthesized by Christopher G. Fairburn, Garner, and Bemis. According to the cognitive-behavioral model, the core psychopathological mechanism underlying both anorexia nervosa and bulimia nervosa is the overvaluation of eating, shape, and weight, alongside an individual’s perceived need to maintain absolute control over them. Whereas psychologically healthy individuals judge their self-worth across multiple domains—such as interpersonal relationships, academic or career achievements, and personal values—individuals vulnerable to eating disorders evaluate their self-worth almost exclusively in terms of their body weight, physical leanness, and capacity to maintain dietary discipline.

This central cognitive disturbance triggers a network of secondary cognitive-behavioral mechanisms:

• Rigid Dietary Restraint: Driven by the catastrophic fear of weight gain, individuals formulate strict, inflexible dietary rules dictating what, when, and how much they may eat. These rigid behavioral boundaries are reflected in the EAT-26 Dieting and Oral Control subscales.
• Dichotomous Thinking and Cognitive Biases: The cognitive architecture of eating pathology is characterized by polarized, “all-or-nothing” cognitive processing. Food items are categorized strictly as “good/safe” or “bad/forbidden” (e.g., carbohydrates and sugars). Minor deviations from these rules are perceived as total failures of willpower, resulting in profound guilt, distress, and subjective loss of control.
• The Binge-Purge Cycle: Under Fairburn’s transdiagnostic model, chronic physiological starvation and intense psychological deprivation increase the biological drive to eat. When individuals eventually breach a rigid dietary rule, cognitive capitulation occurs, frequently precipitating an objective or subjective binge-eating episode. In response to the terror of weight gain and self-recrimination, individuals initiate compensatory behaviors—such as self-induced vomiting, intense exercise, or laxative use—which temporarily alleviate distress but perpetuate the cycle by reinforcing dietary restraint.
• Interpersonal and Social Reinforcement: The Oral Control subscale reflects the interpersonal friction that inevitably arises when significant others observe severe dietary restriction and emaciation. Clinically, as the individual’s eating patterns become visibly aberrant, the social environment responds with pressure, encouragement to eat, or overt monitoring. The individual perceives these well-intentioned interventions as threats to their autonomy, heightening their resolve to maintain behavioral control.

Validity

The EAT-26 has undergone rigorous psychometric validation across clinical, non-clinical, cross-cultural, and epidemiological cohorts, confirming its validity across multiple empirical domains:

Construct and Criterion Validity

In the seminal validation study by Garner, Olmsted, Bohr, and Garfinkel (1982), the EAT-26 was administered to 160 female patients formally diagnosed with anorexia nervosa and 140 female university student controls. The clinical anorexia nervosa group scored markedly higher than the control cohort across all dimensions. The mean total score for the anorexia nervosa cohort was 36.1 (SD = 17.0), compared to a mean of 9.9 (SD = 9.1) in the normal control group, demonstrating clear discrimination between clinical and non-clinical populations (t = 16.3, p < .001). Using a clinical cutoff score of 20 or higher, the EAT-26 correctly identified approximately 85% of patients with anorexia nervosa while classifying only a small minority of the normal control cohort as symptomatic.

Convergent Validity

The convergent validity of the EAT-26 has been established through strong correlations with complementary psychometric measures of eating pathology and psychological distress. Total EAT-26 scores correlate highly with the full 40-item scale (r = .98, p < .0001). Furthermore, investigations assessing the Eating Disorder Inventory (EDI) demonstrate that the EAT-26 Dieting subscale correlates strongly with the EDI Drive for Thinness subscale (r = .70 to .85) and the EDI Body Dissatisfaction subscale (r = .55 to .72). Similarly, the EAT-26 Bulimia subscale shows moderate-to-high correlations with the EDI Bulimia subscale (r = .65 to .80). Statistically significant positive correlations have also been documented with the Beck Depression Inventory (BDI) and the State-Trait Anxiety Inventory (STAI), consistent with known clinical comorbidities between eating disorders, depressive disorders, and trait anxiety.

Discriminant and Known-Groups Validity

Discriminant validity is supported by the tool’s capacity to differentiate between distinct clinical subgroups and non-clinical populations. While female patients with anorexia nervosa and bulimia nervosa systematically score above the clinical threshold (means typically ranging between 30 and 45), cohorts with general psychiatric diagnoses (e.g., major depressive disorder, generalized anxiety disorder, or schizophrenia without comorbid eating pathology) typically register scores below the cutoff of 20. Studies within athletic populations demonstrate that athletes participating in aesthetic or weight-dependent disciplines (e.g., rhythmic gymnastics, figure skating, lightweight rowing) score significantly higher on the EAT-26 than athletes engaged in team or ball sports (e.g., soccer, basketball), reflecting known epidemiological variations in eating disorder prevalence.

Reliability

The reliability of the EAT-26 has been consistently replicated across four decades of international psychometric literature, establishing both internal consistency and temporal stability:

Internal Consistency

In the original validation investigation by Garner et al. (1982), the instrument demonstrated high internal consistency, yielding a standardized Cronbach’s alpha coefficient of .90 for the combined clinical anorexia nervosa and control samples. When evaluated exclusively within the clinical cohort (n = 160), alpha remained strong at .83, indicating that the scale items maintain strong internal cohesion even in truncated, high-scoring populations. For the normal control cohort, alpha was calculated at .86.

Subscale reliability analyses reported across diverse clinical and general population samples show high consistency:

• Dieting Subscale: Cronbach’s alpha consistently ranges between .83 and .91, reflecting its cohesive item pool and strong item-total correlations.
• Bulimia and Food Preoccupation Subscale: Cronbach’s alpha ranges between .75 and .84.
• Oral Control Subscale: Cronbach’s alpha ranges between .60 and .72, reflecting its fewer items and inclusion of social interaction inquiries alongside personal behavioral restriction.

Test-Retest Reliability

Investigations assessing temporal stability report solid test-retest reliability coefficients across various retest intervals. Over a standard two-to-three-week interval, test-retest correlations (Pearson’s r and intra-class correlation coefficients) range from .84 to .89 in stable undergraduate and community cohorts. Over longitudinal intervals spanning several months in non-clinical adolescent groups, stability coefficients remain moderate-to-high (r = .65 to .75), capturing both structural measurement stability and natural developmental fluctuations in adolescent dieting interest.

Factor Analysis

The structural dimensionality of the EAT-26 was established through exploratory factor analysis (EFA) and subsequently corroborated using confirmatory factor analysis (CFA) across numerous cultural adaptations:

Exploratory Factor Analysis (EFA)

Garner et al. (1982) conducted a principal components factor analysis with varimax rotation on the combined sample of 160 anorexia nervosa patients and 140 non-clinical controls. The factor extraction criteria (eigenvalues > 1.0, scree plot examination, and clinical interpretability) confirmed a clean three-factor structural model that accounted for 40.2% of the total variance across the item pool:

• Factor 1 (Dieting): Accounted for the largest proportion of total variance (24.7%). Items loading strongly onto this factor include Item 1 (“Am terrified about being overweight”; loading = .61), Item 11 (“Am preoccupied with a desire to be thinner”; loading = .74), Item 14 (“Am preoccupied with the thought of having fat on my body”; loading = .73), and Item 23 (“Engage in dieting behavior”; loading = .69).
• Factor 2 (Bulimia and Food Preoccupation): Accounted for 8.6% of the variance. Items loading strongly onto this dimension include Item 3 (“Find myself preoccupied with food”; loading = .67), Item 4 (“Have gone on eating binges where I feel that I may not be able to stop”; loading = .60), Item 9 (“Vomit after I have eaten”; loading = .63), and Item 21 (“Give too much time and thought to food”; loading = .70).
• Factor 3 (Oral Control): Accounted for 6.9% of the variance. Items loading heavily onto this component include Item 2 (“Avoid eating when I am hungry”; loading = .49), Item 5 (“Cut my food into small pieces”; loading = .53), Item 8 (“Feel that others would prefer if I ate more”; loading = .66), Item 13 (“Other people think that I am too thin”; loading = .70), and Item 20 (“Feel that others pressure me to eat”; loading = .62).

Confirmatory Factor Analysis (CFA) and Cross-Cultural Invariance

Confirmatory factor analyses conducted across multiple languages (including Spanish, French, Portuguese, Turkish, and Chinese adaptations) have generally supported the original three-factor configuration, demonstrating acceptable to excellent goodness-of-fit indices (e.g., Comparative Fit Index [CFI] > .90, Tucker-Lewis Index [TLI] > .90, Root Mean Square Error of Approximation [RMSEA] < .06). While some structural modeling studies in purely non-clinical adolescent groups suggest alternate or nested configurations (such as a 4-factor model splitting food preoccupation from bulimia, or a single broad general factor of eating distress), the original three-factor model remains the gold standard in diagnostic screening and psychometric evaluation.

Instrument / Measurement Tool

• Tool Name: Eating Attitudes Test-26 (EAT-26)
• Primary Author(s): David M. Garner, Marion P. Olmsted, Yvonne Bohr, and Paul E. Garfinkel
• Publication Year: 1982
• Administration Format: Self-report questionnaire; available in paper-and-pencil or computerized/digital administration
• Target Population: Adolescents and adults (ages 12 and older; suitable for clinical groups, primary care screening, community surveys, and high-risk athletic/aesthetic cohorts)
• Completion Time: Approximately 5 to 10 minutes
• Total Number of Items: 31 total (26 core attitudinal/cognitive items contributing to the formal score, plus 5 supplementary behavioral screening questions assessing clinical risk)
• Response Scale (Mandatory Formulation):
  • 6-point forced-choice scale: Always (3), Usually (2), Often (1), Sometimes (0), Rarely (0), Never (0) for items 1-25
  • For item 26, scoring is reversed: Always (0), Usually (0), Often (0), Sometimes (1), Rarely (2), Never (3)
• Scoring and Quantification Protocol:
  • Nonlinear Scoring System: Items 1 through 25 are scored 3, 2, 1, 0, 0, 0 for responses “Always”, “Usually”, “Often”, “Sometimes”, “Rarely”, and “Never”, respectively. This conservative scoring model is intentionally designed to weight clinically extreme, symptomatic responses while assigning zero points to normative, low-level eating concerns.
  • Reverse Scored Item: Item 26 (“Enjoy trying new rich foods”) is inversely coded: “Never” = 3, “Rarely” = 2, “Sometimes” = 1, “Often” = 0, “Usually” = 0, “Always” = 0.
  • Total Score Range: 0 to 78 points across the 26 items.
  • Subscale Breakdown:
    • Dieting Subscale: Items 1, 6, 7, 10, 11, 12, 14, 16, 17, 22, 23, 24, 25 (Score range: 0–39).
    • Bulimia and Food Preoccupation Subscale: Items 3, 4, 9, 18, 21, 26 (Score range: 0–18).
    • Oral Control Subscale: Items 2, 5, 8, 13, 15, 19, 20 (Score range: 0–21).
• Clinical Interpretation and Diagnostic Cutoff:
  • Cutoff Score ≥ 20: A total score of 20 or higher indicates significant concern about dieting, body weight, or eating attitudes and warrants immediate referral for structured clinical diagnostic evaluation.
  • Behavioral Questions (Items 27–31): These five questions evaluate objective bingeing, self-induced emesis, laxative/diuretic/diet-pill misuse, excessive exercise (greater than 60 minutes daily for weight control), and rapid weight loss (≥ 20 pounds/9 kg in the past 6 months). Even if the 26-item total score is below 20, an affirmative response to any behavioral item signifies substantial clinical risk requiring professional appraisal.

Permissions & Fee and Test Year

The EAT-26 was formally published in 1982 in the journal Psychological Medicine. Dr. David M. Garner and his co-investigators made the instrument publicly accessible to encourage wide adoption across clinical and academic settings.

The EAT-26 is widely considered in the public domain for academic, clinical, and non-profit educational purposes. Researchers, healthcare practitioners, universities, and health clinics may administer and score the tool without paying licensing fees or royalties. The copyright is retained by David M. Garner, Ph.D. Complete instructions, scoring spreadsheets, and specialized online screening materials are accessible via the official website maintained by the author: www.eat-26.com. Commercial software developers, corporate health systems, or publishers incorporating the scale into proprietary or for-profit platforms are required to request formal authorization directly from Dr. Garner.

References

• Fairburn, C. G., & Beglin, S. J. (1994). Assessment of eating disorders: Interview or self-report questionnaire? International Journal of Eating Disorders, 16(4), 363–370. https://doi.org/10.1017/s0033291700030762
• 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
• 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/10.1207/S15327752JPA7403_11
• Nunes, M. A., Camey, S., Olinto, M. T., & Mari, J. J. (2005). The validity and 4-factor structure of the EAT-26 in a Brazilian population. Cadernos de Saúde Pública, 21(4), 1145–1153. https://doi.org/10.1590/s0102-311×2005000400017
• Ocker, L. B., Lam, E. T., Jensen, B. E., & Zhang, J. J. (2007). Psychometric properties of the Eating Attitudes Test-26 among high school female athletes. Journal of Applied Sport Psychology, 19(2), 209–229. https://doi.org/10.1080/10413200701230635
• Rogoza, R., Brytek-Matera, A., & Garner, D. M. (2016). A modern approach to the assessment of eating disorder symptomatology: Validation of the new Polish version of the EAT-26. Comprehensive Psychiatry, 70, 37–46. https://doi.org/10.1016/j.comppsych.2016.06.012

13. Items of the Scale (Questionnaire)