Eating Disorders Inventory-3 by

David M. Garner, PhD

Was originally developed in 1984 to test the “continuum model” of anorexia
nervosa proposed by Nylander in 1971. Later revised in 1991 as the EDI-2 and
again in 2004 as the EDI-3.
 
• Assesses risk factors and predicts responses to treatment
 
• It is a “TOOL” to give us a better understanding of individuals with eating
disorders
 
• The most widely used self-report measure of psychological traits or constructs
shown in individuals with eating disorders.
 
• Used in case conceptualization and treatment planning

• It is also a research tool for assessing areas of psychopathology of interest in

theory-testing, identifying meaningful patient subgroups and assessing
 Purpose
•Aimed at the measurement of psychological traits or
symptom clusters relevant to the development and
maintenance of eating disorders.

• Assesses psychological domains of understanding and

treating eating disorders.

• Yields 12 non-overlapping scale scores and 6 composite

scores used to create clinically meaningful profiles linked
to treatment plans, specific interventions and treatment
monitoring.
 EDI-3
• Age/grade levels: adolescents (13yrs and older) and adult females

• Risk Factors - these disorders represent final common pathways resulting

from the interplay of three broad classes of risk factors (Table 4.1)
1. Culture
2. Social
3. Individual (developmental, psychological, biological)
4. Familial

• Type of items:
1. EDI-3 is made up of 6 composites (fig. 1.1)
a. Eating disorder specific (1)
b. General integrative psychological constructs
i. Ineffectiveness
ii. Interpersonal problems
iii. Affective problems
iv. Over control
 EDI-Symptom Checklist
EDI- 3 Symptom Checklist provides data regarding
frequency of symptoms

3. Binge Eating
4. Self-induced Vomiting
5. Exercise Patterns
6. Use of Laxatives, Diet Pills, and Diuretics

Other items included:

9. Weight
10.Weight History
11.Menstrual History
 EDI-Referral Form
EDI- 3 Referral Form - includes 3 scales that compose the
Eating Disorders Risk Composite (EDRC)

3. Drive for Thinness (DT)

4. Bulimia (B)
5. Body Dissatisfaction (BD)

Referral indexes are based on information from the EDI-3 
questions and the individual’s (BMI) and are used to id 
 Scoring & Item Selection
0-4 pt. scoring system that provides a compromise between increasing scale
variability of the original 0-3 pt. scoring system (see Data Collection Site list)
1. Items had to be able to differentiate between the eating-disorder sample
and the non-clinical sample
2. Items were required to have higher item-total scale correlations with the
target scale than with any other scale
3. Item-total scale correlation coefficients above .40 for the eating-disorder
samples were considered desirable.

“Three item-total correlations below .40 were retained because they provided 
meaningful content and had a minimal impact on the total scale reliability.  The 
average item-total scale correlation was .63 indicating substantial within-scale 
common variance among items.  To be included on the original EDI, scales had 
to meet an internal consistency criterion by having an alpha of above .80 for the 
 Scoring & Item Selection
Depending on whether or not the item is positively or negatively
keyed, a forced – choice format was used requiring respondents to
answer whether each item applies ALWAYS, USUALLY, OFTEN,
SOMETIMES, RARELY OR NEVER. The most extreme responses are
given a score of 4. Scoring as follows:
 Standardization & Norms
Samples were drawn from several different treatment sites in the U.S. with 
comparison samples obtained from centers in Canada, Europe, and Australia
“The recruitment of normative samples from the United States, Canada, Europe, 
and Australia is consistent with prevalence of these eating disorders in 
Westernized industrial societies. Females and young adults, the predominant 
clinical demographic of eating disorders, form the principal study samples.”  
(MMY, 2004)
A. DSM-IV-TR DIAGNOSTIC GROUPS
1. Anorexia Nervosa-Restricting Type (AN-R)
2. Anorexia Nervosa-Binge Eating/Purging Type (AN-B/P)
3. Bulimia Nervosa (BN)
4. Eating Disorders Not Otherwise Specified (EDNOS)

D. Samples (Female – Caucasian)

1. Adults (18yrs. and older)
a. U.S. – 983
b. International – 662
2. Adolescent (11-17 yrs.)
a. U.S. - 335
 Reliability
• Reliability was first estimated from a single administration of the test.

• Cronbach’s Alpha Coefficient was used to estimate the internal consistency.

• Composite reliabilities were calculated with the traditional composite

reliability formula.
1. Eating Disorder Risk Composite ( EDRC) Reliability ranged from .90 to 
.97 (Median =  .94) across the four diagnostic groups and the three 
normative groups.    For the 3 EDR scales, all reliabilities were generally 
in the high .80s to low .90s.
2. Reliabilities for the psychological scales and composites:
a. General Psychological Maladjustment Composite (GPMC) ranged from
.93 to .97 across the three normative groups. All other composite
reliabilities were in the .80s to .90s across the three normative samples.

a. The median reliabilities for the Psychological Scales were .84, .74, and .85
 Reliability
1.Test-Retest Reliability

Was conducted over a short period of time. This was done

to minimize both inherent and extraneous factors that
could potentially impact test scores (i.e. Treatment change,
fluctuations in psychological states). The Sample
consisted of 34 female participants who had undergone
past treatment for Eating Disorder. Ages ranged from 15 to
55 yrs. old and the test-retest interval ranged from 1 to 7
days. The EDRC coefficients was .98 and the GPMC
coefficient was .97. The median test-retest coefficients for
 Validity
• Evidence Based on Internal Structure
1. Intercorrelation Studies
2. Factor Analysis of the Earlier versions of the EDI
3. Exploratory Factor Analysis of the EDI-3 Eating Disorder Risk Scale 
Items
4. Exploratory Factor Analysis of the EDI-3 Psychological Scale Items
5. Confirmatory Factor Analysis of the EDI-3 Psychological Scales

•Evidence Based on Relationship to Other External Variables

1. Correlations with the EDI-2
2. Correlations with External Eating Disorder Measures
a. Eating Attitudes Test-26 (EAT-26)
b. Bulimia Test – Revised (BULIT-R)
3. Correlations with External Measures of Personality and 
Psychopathology
a. Rosenberg Self-Esteem Scale
b. Center for Epidemiologic Studies Depression Scale (CESD)
 Interpretation
Qualification Level: B
A degree from an accredited 4-year college or university in Psychology, 
Counseling, or a closely related field PLUS satisfactory completion of 
coursework in Test Interpretation, Psychometrics and Measurement Theory, 
Educational Statistics, or a closely related area; OR license or certification from 
an agency that requires appropriate training and experience in the ethical and 
competent use of psychological tests.

The EDI-3 profile must be interpreted along with other data

    Clinician should be cautious in interpreting the origin and significance of the 
symptoms.  

  Need to have a broad technical competence in the area of psychological 
evaluation, as well as competence with evaluating specific medical, behavioral 
and psychological symptoms that are comm0on among those with ED.
 Counseling-Clinical Use /
Applications in counseling
ØIncorporated into the stages of counseling

ØSupport, Learning, Action; or

ØAssess problem, conceptualizing & defining

problem, treatments, evaluation
 Strengths
• Thorough and detailed in seeking out disorders based on
behavioral and psychological symptoms.

• Can purchase computer version for scoring, which is less

time consuming and lower risk of scoring errors
The EDI-3 SP is the unlimited-use computer-based scoring
program for the new Eating Disorder Inventory™-3 (EDI™-3).
After administration of the EDI-3, demographic information and
item responses are hand-entered into the software program by
the clinician; the software generates a detailed, individualized
Score Report with raw scores for all EDI-3 scales for each
client. Reports can be edited on-screen to incorporate
additional clinical information or to edit descriptive
statements.

•Samples obtained within the U.S., as well as including various

cultures outside U.S. from Clinical and Nonclinical groups

• Claims to only take 20 minutes to administer

 Strengths
• Cost is practical

•  Cited over 100 times in any given year since 1996 and the
highest number of citations since publishing the manual in
2004 has been 130 in 1999.
 Weaknesses
1. Difficult to understand manual….very in depth and lengthy.  There 
were so many tables and graphs, it was easily confusing.

3. Only has an English-Speaking version

5. “Although test–retest coefficients across scales were excellent, 
ranging from .93 to .98, test–retest data on a nonclinical sample and 
larger eating disorder sample would have been useful to assess.” 
Edward J. Cumella, 2006. 

7. “Multiple factor analyses reveal some factor inconsistency across the 
U.S. adult, International adult, and U.S. adolescent samples .” 
Edward J. Cumella, 2006. 
Mental Measurements Yearbook
“The instrument appears 'front-loaded' in the sense that its
preliminary screening components may be more efficient and valid
than the full scale. Empirically, the Drive for Thinness and Bulimia
measures, as well as the Body Mass Index and canvassing of
symptoms, converge with other research and clinical knowledge in
presenting reasonable measures for further assessment. As the
extended EDI-3 shows questionable factor structure and
discriminant subscale utility, with the author's own dictate that it
does not substitute for a structured clinical interview, the question
comes up as to whether the added diagnostic contributions of the
EDI-3 warrant the time and expense….. Despite a reasonable
amount of research on this instrument, further studies correlating
the instrument with other measures of similar constructs may
need to be pursued. Of particular interest is the cross-cultural
applicability of the EDI-3. Although studies in countries other than
the United States have been undertaken, it would also be of
benefit to determine the extent to which the measure may be
 Multi-Cultural Issues
Impossible to cover all cultures but I feel it is pretty representative
of most common westernized cultures. However, race is greatly
under representative as samples were predominantly Caucasian.

Not very representative of male gender; but access to large enough

samples is more difficult.

“Despite a reasonable amount of research on this instrument,

further studies correlating the instrument with other measures of
similar constructs may need to be pursued. Of particular interest is
the cross-cultural applicability of the EDI-3. Although studies in
countries other than the United States have been undertaken, it
would also be of benefit to determine the extent to which the
measure may be applicable in developing countries.” (Spillane et.
Al., 2004)
 Additional Comments
Personal critique
Personally the manual gave me a headache; however, I feel if I
were more educated on the assessment and statistical measures
used, it might have been easier to understand.

Is it a practical test? (see Table 4.37)

Table 4.37 of the manual shows a decrease in the means scores
from pretreatment to post treatment in the psychological scales
and composites.

How might it be improved?

Could include:
1. Male population
2. Multi-racial/ethnic groups from developing countries
less westernized than the U.S., Canada, and Australia
and other races such as Hispanic, Asian, and African
American, etc.
 References
Cumella, E. (2006, August). Review of the Eating Disorder Inventory–3. Journal of Personality
Assessment, 87(1), 116-117. Retrieved October 5, 2008, doi:10.1207/s15327752jpa8701_11

Eberenz, K., & Gleaves, D. (1994, December). An Examination of the Internal Consistency and
Factor Structure of the Eating Disorder Inventory-2 in a Clinical Sample. International Journal of
Eating Disorders, 16(4), 371-379. Retrieved October 5, 2008, from Psychology and Behavioral
Sciences Collection database.

Eberly, C., & Eberly, B. (1985, December). A Review of the Eating Disorder Inventory. Journal of
Counseling & Development, 64(4), 285. Retrieved October 5, 2008, from Psychology and
Behavioral Sciences Collection database.

Espelage, D., Mazzeo, S., Aggen, S., Quittner, A., Sherman, R., & Thompson, R. (2003, March).
Examining the construct validity of the Eating Disorder Inventory. Psychological Assessment,
15(1), 71-80. Retrieved October 5, 2008, doi:10.1037/1040-3590.15.1.71

Garner, D. (1984, 1984-2004). Eating Disorder Inventory-3. Retrieved October 5, 2008, from
Mental Measurements Yearbook database.

Klemchuk, H., Hutchinson, C., & Frank, R. (1990, July). Body dissatisfaction and eating-related
problems on the college campus: Usefulness of the Eating Disorder Inventory with a nonclinical
population. Journal of Counseling Psychology, 37(3), 297-305. Retrieved October 5, 2008,
doi:10.1037/0022-0167.37.3.297

Schoemaker, C., van Strien, T., & van der Staak, C. (1994, May). Validation of the Eating
Disorders Inventory in a Nonclinical Population Using Transformed and Untransformed
Responses. International Journal of Eating Disorders, 15(4), 387-393. Retrieved October 5, 2008,
from Psychology and Behavioral Sciences Collection database.

Spillane, Nichea S., Boerner, Laura M., Anderson, Kristen G., Smith, Gregory T.