|
Michigan Alcoholism Screening Test (MAST)
Please note that this section is an archive (last updated in June 2006). [disclaimer]
Sections:
Overview |
Instrument Reviews |
Construct Overviews |
Book Compendium Reviews |
Internet Site Reviews
|
|
|
|
Practical Information
|
Instrument Name:
|
Michigan Alcoholism Screening Test (MAST)
|
Instrument Description:
|
The MAST is a 25-item screening instrument used to assess lifetime alcohol abuse/dependence in adults (does not discriminate between lifetime and current alcoholism). It does not contain subscales. None of the questions relate to the amount of alcohol consumption, because the developer found vague responses to such items. (Ref: 1) Questions relate to the frequency, pattern and consequences of drinking. (Ref: 2)
|
Price:
|
Free (available in the public domain)
|
Administration Time:
|
10-15 minutes
|
Publication Year:
|
1971
|
Item Readability:
|
Flesch-Kincaid reading grade of 6.9; all items are no more than one sentence long and are written in language that is easy to comprehend.
|
Scale Format:
|
Dichotomous, yes/no format
|
Administration Technique:
|
Self-administered or interview; no training required to administer
|
Scoring and Interpretation:
|
Most items are scored one or two points per question, however, some questions may receive a score of five points. [Ref: 1] Some items are marked with an asterisk to indicate that they receive more points based on the number of arrests a person has had, or if a person has had Delirium Tremens. Item scores are summed and the total score range depends on the number of Delirium Tremens or arrests accounted for. A score of 3 or less indicates a person is not alcoholic. Four (4) or more points are suggestive of alcoholism, while 5 or more indicates a person who is alcoholic.
|
Forms:
|
Other forms exist, but are not reviewed here: Brief MAST (BMAST, 10 items), Short MAST (SMAST, 13 items), Geriatric version (MAST-G, 24 items), Short Geriatric version (SMAST-G, 10 items), and Malmo Modification (Mm-MAST, 9 items).
|
|
|
Research Contacts
|
Instrument Developers:
|
Melvin L. Selzer, MD
|
Instrument Development Location:
|
Highway Safety Research Institute
University of Michigan
Ann Arbor, Michigan 48104
|
Instrument Developer Email:
|
No information found.
|
Instrument Developer Website:
|
No information found.
|
|
|
Annotated Bibliography
|
1. Selzer ML. The Michigan Alcoholism Screening Test: The quest for a new diagnostic instrument.
Am J Psychiatry. 1971 Jun;127(12):1653-8.[PMID: 5565851]
|
Purpose:
To assess the validity of the newly-devised MAST.
|
|
|
Sample:
The sample consisted of the following five groups:
|
|
1) *hospitalized alcoholics: 75 white males in a nonprofit treatment center
and 41 white males in the Ypsilante State Hospital. Mean age=44 years.
|
|
|
2) *control group: 67 white male blue-collar employees of the University of
Michigan and 36 white males randomly selected from the school’s allergy clinic.
Mean age=41 years.
|
|
|
3) persons convicted of drunk driving: 99 persons interviewed at the Ann
Arbor Municipal Court following conviction. Mean age=35 years, 8% women, 14% blacks.
|
|
|
4) persons convicted of drunk and disorderly behavior: 110 persons interviewed
at the Ann Arbor Municipal Court following conviction. Mean age=34 years, 5% women,
10% blacks.
|
|
|
5) drivers with licenses under review: 98 persons with licenses under review due
to excessive accidents and moving violations. Mean age=25 years, 3% women, 10% blacks.
|
|
|
* In the first two groups, 70% of subjects were in classes 4 and 5 of the Hollingshead
social class scale (refer to: Hollingshead AB. Four factor index of social status. New
Haven (Connecticut): Department of Sociology, Yale University; 1975). Social class was
reported similar in the remaining groups.
|
|
|
|
Methods:
Medical, social, arrest and traffic records were reviewed and compared to subjects’ MAST scores.
|
|
|
Implications:
The MAST revealed more problem drinkers than a review of records and took substantially less time
to complete.
|
2. Zung BJ, Charalampous KD. Item analysis of the Michigan Alcoholism Screening Test.
J Stud Alcohol. 1975 Jan;36(1):127-32. [PMID: 238069]
|
|
Purpose:
To perform an item analysis of MAST responses in a group of DWI offenders.
|
|
|
Sample:
Two groups of 100 persons each arrested for drunk driving and
referred to Baylor College of Medicine for pre-trial evaluation
of alcohol problems.
|
|
Group 1
|
Group 2
|
| Age
|
41.72 years
|
40.30 years
|
| Education
|
10.53 years
|
10.77 years
|
| Gender
|
94% male
|
89% male
|
| Ethnicity
|
6% black
|
21% black
|
| Marital status
|
57% married
|
63% married
|
|
|
|
Methods:
The MAST was administered in a pre-trial interview.
|
|
|
Implications:
Most items adequately discriminated between groups of problem and nonproblem drinkers;
however, several items assessing severe alcoholism did not (#14, 19, 22, and 23).
|
3. Skinner HA. A multivariate evaluation of the MAST. J Stud Alcohol. 1979 Sep;40(9):831-44. [PMID:513775
]
|
|
Purpose:
To evaluate the MAST using the original weighted scoring method and a simpler unit scoring
method (0 or 1), examine factor structure, assess the degree to which MAST is influenced by
response styles, and evaluate correlations with other measures.
|
|
|
Sample:
418 alcoholics and drug addicts treated at the Clinical Institute of the Addiction Research
Foundation. Sample A (N=208) consisted of 164 men and 44 women, mean age=33.9 years, 43% single,
64% had some high school education, and 66% unemployed. Sample B (N=210) consisted of 160 men
and 50 women, mean age=34.3 years, 37% single, 69% had some high school education, 66% unemployed,
and 40% were laborers.
|
|
|
Methods:
The MAST, Personality Research Form, Basic Personality Inventory, WAIS Vocabulary subscale,
and Raven’s Progressive Matrices were administered.
|
|
|
Implications:
5 factors were found that reflect specific problem areas. MAST scores should be interpreted as
a continuum, with subjects ordered according to the severity of their alcohol related problems.
|
4. Skinner HA, Sheu WJ. Reliability of alcohol use indices. The Lifetime Drinking History and the MAST.
J Stud Alcohol. 1982 Nov;43(11):1157-70. [PMID:7182675
]
|
|
Purpose:
To examine test-retest reliability in the MAST and the Lifetime Drinking History (LDH), determine whether certain aspects of drinking history and problems could be assessed more reliably than others, and to evaluate whether personality characteristics can predict who will give less consistent drinking histories.
|
|
|
Sample:
83 randomly selected persons seeking help at the Clinical Institute of the Addiction Research
Foundation. 78% males, mean age=33 years, mean years of school=11.7, 45.8% single, 45.7%
employed as unskilled or skilled laborers.
|
|
|
Methods:
The first assessment was performed before patients entered treatment programs at the Clinical
Institute. The follow-up was given in patients’ homes. Both times, subjects completed the MAST
and the LDH. At the initial assessment subjects also completed the WAIS Vocabulary and Digit
Symbol subscales, Visual Retention Test, and the Denial scale of the Basic Personality Inventory.
|
|
|
Implications:
Alcohol use patterns can be assessed with high reliability and patients undergoing alcohol
treatment provide consistent responses.
|
5. Magruder-Habib K, Fraker GG, Peterson CL. Correspondence of clinicians' judgments with the Michigan Alcoholism Screening Test in determining alcoholism in Veterans Administration outpatients.
J Stud Alcohol. 1983 Sep;44(5):872-84 [PMID:6645547
]
|
|
Purpose:
To determine the extent to which practitioners’ judgments of VA outpatients’ alcoholism
correspond with an objective measure and to examine factors related to the differences
between the two.
|
|
|
Sample:
Statistics are based on 369 patients at 5 VA medical centers (363 men and 6 women) who
had complete MASTS and clinicians’ judgments.
|
|
|
Methods:
Outpatients who were reporting for treatment were asked to participate in the study.
Those with social security numbers ending in 1 were assigned to Group 1 while those
ending in 5 were assigned to Group 2. Group 1 patients were given the MAST before
seeing the physician and Group 2 patients were administered the MAST afterward. Both
groups of patients were also asked questions about their demographics and alcohol
treatment in the last 30 days. This data and MAST scores were matched with clinicians’
judgments.
|
|
|
Implications:
The MAST was in agreement with clinicians’ judgments 77% of the time. The prevalence of
alcoholism in the VA samples was higher than that found in non-VA samples.
|
6. Mischke HD, Venneri RL. Reliability and validity of the MAST, Mortimer-Filkins Questionnaire
and CAGE in DWI assessment. J Stud Alcohol. 1987 Sep;48(5):492-501. [PMID:3669678
]
|
|
Purpose:
To examine the reliability and validity of the MAST, Mortimer-Filkins Questionnaire
(MFQ) and CAGE.
|
|
|
Sample:
718 persons convicted of driving while intoxicated (DWI). MAST group N=90 (85.6% men,
mean age=34.3 years), MFQ group N=253 (81.4% men, mean age=36.1 years), CAGE group N=375
(79.7% men, mean age=34.7 years).
|
|
|
Methods:
Subjects’ drinking problems were assessed at 34 DWI assessing agencies that were mailed
an instruction packet including the instrument to be given prior to the interview portion.
Agencies were not sent scoring guidelines so that counselors would not use test results in
making a clinical decision. Alcoholism counselors’ decisions served as the criterion.
Agencies mailed back all of the data to the authors.
|
|
|
Implications:
Counselors found lower rates of problem drinking than are usually found in this population.
The MAST was more reliable and valid than the other two instruments.
|
7. Teitelbaum L, Mullen B. The validity of the MAST in psychiatric settings: a meta-analytic
integration. Michigan Alcoholism Screening Test. J Stud Alcohol. 2000 Mar;61(2):254-61. [PMID:10757136
]
|
|
Purpose:
To integrate available information concerning the validity of the MAST in psychiatric
settings and to investigate factors that may affect performance on the MAST.
|
|
|
Sample:
This meta-analysis involved the review of nine studies with sample sizes ranging from
40 to 400 individuals. Males constituted from 38% to 100% in the various samples.
|
|
|
Methods:
Clinically determined alcohol-related diagnosis was used as the criterion on which to
draw integration between 9 published studies investigating the criterion validity of
the MAST.
|
|
|
Implications:
In psychiatric settings, the MAST is highly valid. Sensitivity was found to be higher
than specificity estimates. Validity of the MAST was greater in psychiatric samples
with a higher number of women and in samples with higher base rates of alcoholism.
|
8. Teitelbaum LM, Carey KB. Temporal stability of alcohol screening measures in a psychiatric setting.
Psychol Addict Behav. 2000 Dec;14(4):401-4. [PMID:11130159
]
|
|
Purpose:
To evaluate the test-retest reliability of the MAST and CAGE in a psychiatric setting.
|
|
|
Sample:
71 people receiving outpatient psychiatric services in New York and 64 people with no
psychiatric history. In the clinical sample (34 male and 37 female), the majority was
European-American (62%), mean age=34 years, with Schizophrenia-spectrum disorder was the
most common diagnosis (63%). The nonclinical sample (35 male and 29 female) consisted
of persons in the community who denied a history of psychiatric services and who
acknowledged drinking alcohol in their lifetime.
|
|
|
Methods:
The MAST and the CAGE were administered twice, with a test-retest interval of one week.
Subjects were screened for sobriety prior to testing. The Structured Clinical Interview
for DSM-IV (SCID), Social Desirability Scale (SDS), Logical Memory I subscale of the
Wechsler Memory Scale Revised, and the Brief Symptom Inventory (BSI) were also administered
initially.
|
|
|
Implications:
Persons with severe mental disorders can give reliable information concerning their alcohol
related problems. The MAST and the CAGE showed acceptable reliability. Younger men who met
criteria for a lifetime alcohol disorder varied most in the MAST scores over time.
|
9. Conley TB. Construct validity of the MAST and AUDIT with multiple offender drunk drivers.
J Subst Abuse Treat. 2001 Jun;20(4):287-95. [PMID:11672645
]
|
|
Purpose:
To evaluate the construct validity of the MAST and AUDIT in screening for alcohol
disorders in those convicted of driving under the influence (DUI).
|
|
|
Sample:
126 DUI offenders at a court-ordered treatment agency. 85% male, 84% white, 87% called
themselves European or American, majority (47%) never married, 34% report earning over $
30,000/year, mean age=38 years.
|
|
|
Methods:
DSM-IV diagnosis of chemical dependency was used as criterion of construct validity,
as obtained through the National Longitudinal Alcohol Epidemiologic Survey questionnaire
items. Subjects were given the MAST, the AUDIT, and 1.5 hours of background questioning
in a classroom setting. Two days later, a Masters level clinician made a DSM-IV diagnosis.
|
|
|
Implications:
The MAST and the AUDIT correlate moderately well with each other, while the MAST correlates
more highly with DSM-IV criteria.
|
10. Thurber S, Snow M, Lewis D, Hodgson JM. Item characteristics of the Michigan Alcoholism Screening Test.
J Clin Psychol. 2001 Jan;57(1):139-44. [PMID:11211282
]
|
|
Purpose:
To evaluate MAST items for disproportional responding, using coefficient alpha and
confirmatory factor analysis.
|
|
|
Sample:
207 patients (156 males, 51 females) admitted to an outpatient drug treatment center.
Mean age=33 years, 166 Caucasians, majority court-referred, mean years of schooling=11.
|
|
|
Methods:
A counselor administered the MAST after a general interview.
|
|
|
Implications:
Item18 has a low level of endorsement and a low relationship with other MAST items
and the total score. Items concerned with drug-related arrests detracted from otherwise
homogeneous items.
|
|
|
top
|
Factors and Norms
|
Factor Analysis Work:
|
Principal components analysis ssuggestedfive components: 1) Recognition of alcohol problem by self and others, 2) Legal, work and social problems, 3) Help-seeking, 4) Marital-family difficulties, 5) Liver pathology. However, most variance was accounted for by the first (unrotated) component (41.3%, 6.7%, 6.3%, 5.0%, and 4.6% for the first five extracted components, respectively). (Ref: 3) This suggests that the MAST is dominated by one factor, with each of the remaining four components explaining relatively small amounts of variation among individuals.
|
Normative Information Availability:
|
No information found.
|
|
|
Reliability Evidence
|
Test-retest:
|
Test-retest reliability was 0.84 over an average time of 4.8 months (Ref: 4), and 0.95 over a period of one week. (Ref: 8) No significant increase in test-retest scores was found, t(135)= - 0.43, p=0.66. (Ref: 8)
|
Inter-rater:
|
No information found.
|
Internal Consistency:
|
Point-biserial correlation coefficients generally ranged from 0.30 to 0.79 for each item, with more than half greater than 0.50 (Ref: 2), although one study found the range as low as 0.13 to 0.72, with 10 items greater than 0.50. (Ref: 6) Coefficient alpha ranges from 0.79 (Ref: 4,6,10) to 0.90. (Ref: 3) The split-half correlation coefficient is 0.83. (Ref: 10)
|
Alternate Forms:
|
No information found.
|
|
|
Validity Evidence
|
Construct/ Convergent/ Discriminant:
|
The MAST correlates with the 16 primary scales of the Alcohol Use Inventory at r=0.83. It correlates with lifetime daily average alcohol consumption at r=0.58 and with duration of drinking at r=0.33. (Ref: 3) MAST scores and clinicians’ judgments of alcoholism are significantly related (÷2=73.62, p<=0.01). There was agreement between the two 77% of the time, with Kappa of 0.44. (Ref: 5) MAST is correlated with the AUDIT at r=0.61 (p<0.05), and with DSM-IV diagnosis used as a measure of construct validity at r=0.60 (p<0.05). (Ref: 9)
|
Criterion-related/ Concurrent/ Predictive:
|
Correlation between MAST scores and counselor decisions about patients’ alcoholism (used as a measure of criterion validity) was r=0.65. (Ref: 6) In a meta-analytic review of nine studies, combined results show criterion validity that is significant (z=17.88, p<0.01) and moderate (r=0.52, Fisher’s z=0.58) in psychiatric settings. (Ref: 7)
|
Content:
|
No information found.
|
Responsiveness Evidence:
|
There is a significant (z=13.93, p<0.01), strong (r=0.78, Fisher’s z=1.05) degree of sensitivity when the MAST is used in psychiatric settings. (Ref: 7)
|
Scale Application in VA Populations:
|
Yes. (Ref: 5)
|
Scale Application in non-VA Populations:
|
Yes. (Ref: 1-4, 6, 8-10)
|
|
|
Comments
|
The MAST is “devised to provide a consistent, quantifiable, structured instrument to detect alcoholism.”
Overall Usefulness for a Certain Population: The MAST is frequently chosen for use in alcoholism research as well as in treatment practice; the instrument has a longstanding history of successful use with members of its respective population.
Advantages: The MAST is well established in the research literature on alcoholism. Statistics on the instrument’s internal consistency, content and construct validity, sensitivity and specificity are readily available. Most alcoholism studies that have employed the MAST have reported satisfactory measurement results, and studies that have compared the MAST to other alcoholism tests have found the MAST to perform best.
Disadvantages: Because the MAST explicitly avoids quantification of alcohol consumed, its usefulness in tying behaviors directly to physiological outcomes is limited. Items that attempt to target physiological aspects of alcoholism call upon respondents to self-diagnose symptoms that may not be as easy to discern as the items would suggest (for example, one question asks, “have you ever been told you have liver trouble? Cirrhosis?” presupposing that the respondent has knowledge necessary to glean this information from what was said to him/her by the physician). Lastly, several studies have criticized the MAST for excessive length, criticism reinforced by the MAST itself in that it includes one item, #7, that never counts in the overall score. Further, compared to clinician judgment, the MAST is only modestly accurate in capturing alcoholics (k = .44; Ref: 5); with a review of the literature suggesting that researchers have found the rate of false positives to be 3.5 times that of false negatives. (Ref: Gibbs, 1983) Thus, the MAST does well in screening out non-alcoholics, but may significantly over-diagnose alcohol problems.
Recommendation: Several studies (not reviewed here) suggest that the short forms of these instruments, the SMAST and SMAST-G, work as effectively in identifying subjects at the upper range of the scale (i.e., most indicative of alcoholism) as the longer forms do. Therefore, we suggest employing a combination of the appropriate short form of the MAST and the short form of another alcoholism test validated for the population of interest in order to help reduce the rate of false positives.
|
Updates
No information found.
|
|
|
|
