The Eating Disorder Inventory-3 (EDI-3)
Order Information for the EDI-3 can be found at the end of this article
Internal Publication: River Centre Clinic (2008)
Eating disorders are common, relatively chronic and potentially life-threatening conditions primarily affecting young women. Anorexia nervosa is estimated to be the third most common chronic medical illness in girls aged 15-19 years (Lucas et al., 1991) and bulimia nervosa is even more common (Hoek & Hoeken, 2003). The mortality rates for anorexia nervosa are twice as high as seen in any other psychiatric disorder and more than 12 fold the number of deaths expected from all causes among women 15-24 years of age (Birmingham et al., 2005); mortality rates for bulimia nervosa are lower but still not insignificant (Nielsen, 2003) Medical complications are typical during the acute phase of an eating disorder (Becker et al. 1999) and persist among those not successfully treated (Keel et al., 2002) leading to a wide range of physical and emotional disorders into early adulthood (Johnson et al., 2002). These observations underscore the need for well-established and standardized measures to assess relevant symptom domains and therapeutic outcomes.
The Eating Disorder Inventory-3 (EDI-3) is currently included in the Patient-Reported Outcome and Quality of Life Instruments Database (PROQOLID). The EDI-3 is a revision of the most widely used self-report measure of psychological traits or constructs shown to be clinically relevant in individuals suffering from eating disorders (Garner, 2004). The EDI-3 is a standardized and easily administered measure yielding objective scores and profiles that are useful in case conceptualization and treatment planning for individuals with a confirmed or suspected eating disorder. It is also a valuable research tool for assessing areas of psychopathology of interest in theory-testing, identifying meaningful patient subgroups and assessing treatment outcome
Since the EDI-3 is a revision of an instrument used for more than 20 years (Garner, Olmsted & Polivy, 1983), the original item-set as well as items from the 1991 revision (Garner, 1991) have been carefully preserved so that clinicians and researchers can continue to compare data collected previously with data from the revised EDI-3. At the same time, the EDI-3 has been significantly enhanced to provide scales for measuring constructs that are more congruent with the psychological domains identified by modern theories as relevant to the etiology, maintenance, or key symptom variations in individuals with eating disorders. The conceptual refinement and improved psychometric properties of the EDI-3 represent a significant advancement in psychological assessment in the field of eating disorders.
The EDI-3 is intended for older adolescents (age 14 years and older) and adults; however, the item set has also been used with younger adolescents (11 to 13 years old). The EDI-3 consists of 91 items from the EDI-2 (Garner, 1991) organized onto 12 primary scales consisting of 3 eating-disorder-specific scales and 9 more general psychological scales that are highly relevant, but not specific to eating disorders. The EDI-3 also yields six (6) composite scales, one (1) that is eating disorder-specific (Eating Disorders Risk Composite) and five (5) that are general integrative psychological constructs (Ineffectiveness, Interpersonal Problems, Affective Problems, Overcontrol, and the overall Psychological Maladjustment composite). Some EDI-3 scales remain unchanged from the original EDI and the EDI-2 while other scales have been refined or created to improve the psychometric qualities and conceptual content of the scales and the instrument as a whole. The EDI-3 revision is based on a large clinical sample of eating disorder patients from a number of treatment facilities in the United States, Canada, Europe, and Australia. New norms were created from a multi-site clinical sample from the U.S.. Non-clinical comparison samples have come from the U.S. and Europe. The EDI-3 includes a number of other enhancements listed below.
Summary of the Main Features of the EDI-3
- The EDI-3 assesses psychological domains that have conceptual relevance in understanding and treating eating disorders.
- Easily administered and scored, the EDI-3 yields 12 non-overlapping scale scores and 6 composite scale scores that allow the creation of clinically meaningful profiles that can be linked to treatment plans, specific interventions, and treatment monitoring.
- The EDI-3 is designed for both research and clinical applications.
- New scoring system that improves the EDI-3 psychometric properties.
- Excellent reliability and validity.
- Large U.S. clinical multi-site standardization sample split into four different diagnostic subgroups used to create updated norms.
- Large International clinical comparison sample.
- U.S. and International multi-site non-clinical comparison sample.
- EDI-3 discriminates between clinical and non-clinical groups
- Critical items are included in the interpretive profile to allow the development of a more meaningful clinical picture.
- The EDI-3-Symptom Checklist (EDI-3-SC) is a checklist that allows systematic gathering of symptom data that can aid in determining if DSM-IV-TR diagnostic criteria are met.
- The EDI-3-Referall Form (EDI-3-RF) is a Referral Form that can be administered in high school, college, sports team, or other non-clinical settings to assist in determining if an individual should be referred for specialized counseling.
- Improved Scoring System: Some have argued that a 1-6 scoring system is superior to the original 0-3 system. While there has been some evidence that there is a small psychometric advantage to the 1-6 system, there have been others who have failed to find this advantage. However, the EDI-3 uses a 0-4 system that maintains the conceptual integrity of the original system; however, it improves the reliabilities of some scales and yields a wider range of scores.
The aim of the EDI-3 is to provide a comprehensive measure for assessing the presence and intensity of psychological traits or symptom clusters that are clinically relevant in a broad assessment of individuals suffering from eating disorders. This is consistent with the view of eating disorders as multi-determined and heterogeneous in nature (Garner, 1993). Accordingly, these disorders represent final common pathways resulting from the interplay of three broad classes of predisposing or risk factors that include cultural (or social), individual (developmental, psychological, and biological), and familial risk factors combine in different ways, leading to the development of eating disorders. Important predisposing factors include being female, living in Western society, adolescence and early adulthood, low self-esteem, perfectionism, depression, and a family history of any type of eating disorder, obesity, depression, or substance abuse. Significant precipitating factors include dieting to lose weight, occupational or recreational pressures to be slim, critical comments about weight and shape, and sexual abuse. A key perpetuating factor is the psychological, emotional, and physical effects of starvation.
Eating disorder diagnostic subgroups have been formed primarily based on body weight and eating symptoms such as binge eating and self-induced vomiting; however, the remarkable heterogeneity in psychological features within each diagnostic subgroup underscores the clinical utility of evaluating patients on a broad spectrum of meaningful psychosocial variables. It is precisely this variation in patient presentation that is the conceptual hub of the EDI-3.
Goals and Rationale for the EDI-3 Revision
The primary aim in creating the EDI-3, like the original EDI and the EDI-2, was to develop a standardized, self-report measure of psychological traits that were clinically relevant to the eating disorder population. It was based on the fundamental assumption that the psychological themes that cause and maintain eating disorders vary in the heterogeneous clinical population; it was assumed that understanding these themes would aid the understanding and the treatment of eating disorders. Like its predecessors, the EDI-3, was formulated using an approach to construct validation that emphasized both rational and empirical methods of scale development, with constructs chosen based on a review of major theories in the clinical literature and items generated by experienced clinicians to operationally define these constructs. Items included in the final version of the EDI were based on the empirical adequacy as well as scale stability.
The rationale for re-standardizing the EDI-3 was based on several interrelated goals briefly summarized below:
- Earlier versions of the EDI did not provide norms for adolescent patients with eating disorders. Moreover, the relatively small adult clinical samples limited the scope of psychometric evaluation of demographic, diagnostic, clinical, and external correlates of EDI scales. EDI-3 was validated on large enough samples of adult and adolescent eating disorder patients to correct these limitations.
- The EDI-3 revision addresses the relative weakness of the three EDI-2 Provisional Scales introduced in 1991 These Provisional Scales were comprised of items that tapped theoretical meaningful constructs; however, sample size limitations did not permit rigorous analysis of item stability and scale structure. The EDI-3 organized these items onto theoretically meaningful scales, using both rational and empirical methods.
- The original EDI was introduced in 1983 based on constructs relevant at that time. Since then, theoretical developments have led to a refinement of the psychological domains considered most relevant to eating disorders. The EDI-3 re-examined the psychological domains measured by the EDI to determine if the items could be organized into more meaningful constructs to reflect conceptual advances in the eating disorder field.
- The EDI-3 re-standardization sample was drawn from several different treatment sites in the U.S. with comparison samples obtained from centers in Canada, Europe and Australia. This contrast with earlier versions of the EDI which were validated on samples from two geographic regions in the Midwest U.S. and Ontario, Canada. Despite this advancement, the EDI-3 re-standardization aims to achieve geographic, racial and socio-demographic representativeness of the U.S. population.
- There has been some controversy regarding the original 0-3 point scoring system of the EDI that presents items on a 6-point Likert scale but truncates the range (e.g. 000123) by not tallying item scores in the extreme “non-symptomatic direction. The EDI-3 tested several scoring models providing a compromise between systems described by various research groups.
- The EDI-3 incorporates response style indicator scales designed to alert clinicians of response patterns that suggest possible test-taking bias.
There are 12 primary scales on the EDI-3 plus five composite scales derived by adding the standardized score of two or more scales together (below). Three of the primary scales are “eating disorder risk” scales since research has shown that high scores on these scales place the individual at an increased risk for having or developing an eating disorder. There are nine primary psychological scales that assess constructs having conceptual relevance to the development and maintenance of eating disorders. The EDI-3 also introduces “item clusters” that are important in the interpretation of the EDI-3 profile. EFAs were first performed on the EDI-2 Eating Disorder Risk Scales and the Psychological Scales to scales used for evaluating the underlying relationship among items. Once preliminary scales were formed, another set of EFA’s were performed on each of the scales to identify conceptually or clinically relevant “item clusters”. These were evaluated by using item analyses and reliability (Alpha) analyses.
Three validity or response style scales (i.e., Inconsistency, Infrequency, and Negative Impression scales) have been developed for the EDI-3. These scales were developed in order to determine the likelihood that the EDI-3 was responded to in a valid manner. In order to develop qualitative classification ranges for the Validity scale, frequency distributions and cumulative percentages for the two U.S. standardization sample were generated. Based on the distribution of scores in both samples, ranges were established for the qualitative classifications. Responses that occurred with a relative frequency that is greater than approximately 95% are classified as Typical, those that are between 1% and 5% are classified as Atypical, and those that are among the most extreme 1% are classified as Very Atypical.
Eating Disorder Inventory-3 Scales
Eating Disorder Specific Scales
Drive for Thinness
Psychological Trait Scales
Eating Concerns Composite
Interpersonal Problems Composite
Emotional Dyscontrol Composite
Global Psychological Maladjustment
Response Style Indicators
Negative Impression Scale
The EDI-3 was developed to aid in the clinical assessment of individuals with eating disorders. The EDI-3 provides normative information for females ages 13 to 53. The normative samples include individuals from a wide range of racial and ethnic backgrounds. All normative protocols were collected in various outpatient and inpatient settings. Normative information is provided for the following DSM-IV diagnostic groups: 1) Anorexia Nervosa Restricting (AN-R); 2) Anorexia Nervosa Bulimic/Purging (AN-B/P); 3) Bulimia Nervosa (BN); and 4) Eating Disorders Not Otherwise Specified (NOS).
It is incumbent upon the examiner to proceed with caution when administering the EDI-3 to individuals whose first or native language is not English. The responses obtained from such individuals may be of questionable validity due to cultural or linguistic differences, and should be reviewed to ensure that the respondent understood each EDI-3 item before scores are generated and interpreted. The EDI-3 Item Booklet, the instrument is written at an overall eighth-grade reading level.
Applications for Clinical Assessment of Eating Disorders
The EDI-3 was developed and standardized for use in the clinical assessment of individuals with eating disorders. As a clinical assessment instrument, the EDI-3 is designed to provide information useful in understanding the patient, planning treatment, and assessing progress. The EDI-3 provides individual patient profiles that may be compared against norms for eating disorder patients and non-patient comparison samples. This type of information is particularly relevant in individual cases because it is recognized that patients vary remarkably on the psychological dimensions assessed by the EDI-3. Administered at several points in time, the EDI-3 can provide valuable information about clinical status and response to treatment. While the EDI-3 does provide information relevant to these purposes, it is important to remember that it is not designed to provide a comprehensive assessment of psychopathology relevant to patients with eating disorder.
The EDI-3 was originally developed and standardized for clinical use on patients 18 years old through adulthood and there is limited information on its psychometric properties with young eating disorder patients. The EDI-3 is the first version of the instrument to include an adolescent eating disorders standardization sample; however, only a small number of younger adolescents were part of this sample. Thus, there are not adequate data to support inferences drawn from the test profiles for patients 12 and younger.
Applications to the Treatment Process
The EDI-3 can be used to provide valuable clinical information to aid in case conceptualization and treatment planning. For example, a patient who has highly elevated scores on the Interpersonal Alienation and Emotional Dysregulation scales at the beginning of therapy may require the therapist to be particularly attuned to potential difficulties in forming a strong and trusting therapeutic relationship. A patient with this profile may be at greater risk to engage in impulsive or uncooperative behaviors and may be more likely to prematurely terminate treatment. Early in treatment, it can be useful to share the EDI-3 profile with a patient to aid in collaborative treatment planning. Some eating disorder patients begin therapy with poor motivation to change. Asking patients to review and comment on the EDI-3 profile form can be a non-threatening way of engage them in the process of identifying problem areas.
Applications in the Non-Clinical Settings
In non-clinical settings, the EDI-3 RF provides an economical means of identifying individuals who may be at risk for developing eating disorders. As described later, it is usually preferable to administer the EDI-3-RF (referral form) as part of a two-stage screening process in schools, athletic programs, and other institutional settings to identify those with eating symptomatology. Other scales can be used to identify other areas of distress.
The EDI-3 has utility as a research tool with different applications. It provides descriptive information, allowing comparison of samples in one research setting to those from other research sites. It has been used both as an outcome measure and a prognostic indicator in treatment studies. The EDI-3 also has been used to track psychological functioning in prospective studies of individuals at risk for eating disorders. Subscales of the EDI-3 have been used to select or define criterion groups in studies of body dissatisfaction, weight preoccupation, and perfectionism.
Screening for Eating Disorders: The EDI-3 Referral Form (EDI-3-RF)
The EDI-3-RF is a brief self-report measure designed to assess for eating disorder risk. The EDI-3-RF can be used in group or individual settings and is designed to be administered by mental health professionals, school counselors, coaches, camp counselors, and others with interest gathering information to determine if an individual should be referred to a specialist for evaluation for an eating disorder. The EDI-3-RF can be administered to individuals in a non-clinical setting or a clinical setting not specifically focused on eating disorders. It is ideally suited for school settings, infertility clinics, general practice settings and psychiatric outpatient departments. It is an economical first step in a two-stage screening process. According to this methodology, individuals who score above a particular “cutoff” should be referred to a qualified professional and interviewed to determine if they meet diagnostic criteria for an eating disorder.
The EDI-3-RF consists of 25 items from the Drive for Thinness, Bulimia and Body Dissatisfaction Scales of the EDI-3; however, only the DT and B items are used in calculating the cut-off score used in the referral decision process. The EDI-3-RF also consists of questions pertaining to self-reported weight, height, weight history, menstrual history and behavioral symptoms indicative of eating disorders. The EDI-3 RF should be interpreted using the respondent’s answers to several different sections of the instrument as well as from collateral clinical information derived from other sources. Generally, a referral should be made if a respondent scores “positively” or meets the threshold on one or more criteria related to self-reported: 1) excessive concerns related to dieting, body weight, or problematic eating patterns; 2) low body weight compared to age-matched norms; 3) behavioral symptoms reflective of an eating disorder.
EDI-3 is not intended to yield a diagnosis of an eating disorder. Rather it is aimed at the measurement of psychological traits or symptom clusters relevant to the development and maintenance of eating disorders. The psychological profile provided by the EDI-3 is a rich source of information for generating or confirming clinical impressions that go well beyond simple diagnoses. It is also a valuable tool for generating treatment plans and assessing the effect of treatment on key psychological domains.
Becker, A.E., Grinspoon, S.K., Klibanski, A. & Herzog, D.B. (1999). Eating Disorder. New England Journal of Medicine, 340, 1092-1098.
Birmingham, C. L., Su, J., Hlynsky, J. A., Goldner, E. M., & Gao, M. (2005). The mortality rate from anorexia nervosa. International Journal of Eating Disorders, 38(2), 143-146.
Garner, D. M. (1991). Manual for Eating Disorder Inventory. Odessa, FL: Psychological Assessment Resources, Inc.
Garner, D. M. (1993). Pathogenesis of anorexia nervosa. Lancet, 341, 1631-1635.
Garner, D. M. (2004). The Eating Disorder Inventory-3: Professional manual. Odessa, FL: Psychological Assessment Resources, Inc.
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.
Johnson, J. G., Cohen, P., Kotler, L., Kasen, S., & Brook, J. S. (2002). Psychiatric disorders associated with risk for the development of eating disorders during adolescence and early adulthood. Journal of Consulting and Clinical Psychology, 70(5), 1119-1128.
Keel, P. K., Mitchell, J. E., Davis, T. L., & Crow, S. J. (2002). Long-term impact of treatment in women diagnosed with bulimia nervosa. International Journal of Eating Disorders, 31, 151-158.
Millar, H. R., Wardell, F., Vyvyan, J. P., Naji, S. A., Prescott, G. J., & Eagles, J. M. (2005). Anorexia nervosa mortality in Northeast Scotland, 1965-1999. American Journal of Psychiatry, 162, 753-757.
Nielsen, S. (2003). Standardized mortality ratio in bulimia nervosa. Archives of General Psychiatry, 60(8), 851-851.
David M. Garner, Ph.D.
Founder and Administrative Director,
River Centre Clinic, Sylvania, Ohio, USA
Click Here to Order the EDI-3 from Psychological Assessment Resources, Inc.
16204 North Florida Avenue
Lutz, FL 33549
Return to SPECIAL TOPICS page
The Eating Attitudes Test (EAT-26): A Summary
About Eating Disorders