Items of the Michigan Alcoholism Screening Test (MAST) were individually ev
aluated regarding disproportional responding. The item with the lowest degr
ee of endorsement (Item 18. related to liver problems) was found to have co
rrespondingly low relationships with other MAST items and with the total sc
ore. Weighted items concerned with drug-related arrests were found to detra
ct from the otherwise relatively homogeneous MAST items. Traditional statis
tical approaches for homogeneity suggest high level internal consistency wi
th the deletion of the aforementioned items. A factor analytic test of homo
geneity was confirmed. Recommendations are presented concerning modificatio
n of the MAST scoring procedures. (C) 2001 John Wiley & Sons, Inc.