In an article in the Journal of Family Violence 18: 253-255, Dutton dismissed Gondolf's MCMI subscale results from batterer program participants based on his criticisms of the MCMI and how it was used. Gondolf's main finding, published in 1999, was that less than half of the 840 batterers showed evidence of personality disorders according to the MCMI, and approximately 25 percent showed evidence of a severe mental disorder. Gondolf views Dutton's criticisms of the study as focusing on Gondolf's statement that "there is little evidence for a prevailing 'abusive personality' typified by borderline personality tendencies, and little support for a preponderance of posttraumatic stress disorder among our sample" (Gondolf, 1999, p. 13). In his critique of the study, Dutton argued that the "abusive personality" is not a category in the DSM-IV nor is it deducible from the MCMI. Gondolf replies that the "abusive personality" has been associated with borderline personality tendencies, as have attachment styles, so one might expect some evidence of these tendencies in elevated personality scores associated with those characteristics (e.g., MCMI borderline, anxiety, histrionic, compulsive subscales). Dutton also argued that self-report instruments such as the MCMI are an inadequate and antiquated means of assessing personality disorders. Gondolf replies that the MCMI is a well-established and widely used assessment instrument in general and in batterer research; and it is debatable when and if the MCMI will become obsolete, because of the extensive validity studies and administration advantages associated with it. Gondolf also outlines and then responds to Dutton's criticisms of the study methodology. In addition, Gondolf notes that the ultimate fundamental question regarding the roots of batterer behavior may be the validity of attachment theory, given the serious limitations of the knowledge base for this theory. 15 references