I sent a five-page report to Deputy Commissioner Conboy on August 10, 1979. This was in response to his request of July 25, 1979. Briefly, I stated my growing concern regarding the effects of alcoholism on the Department. At this time, I had 19 years experience dealing with alcoholics on the job. I had personally disarmed some 1,500 cops of all ranks, without incident to me or to others. So, I should not have been perceived as an alarmist. I had communicated to the Stress Committee that all the work on stress prevention was not going to stop alcohol-related incidents on the job. I believed that if you didn't screen out the problem drinker prior to coming on the job, then nothing will be accomplished. It is similar to pushing water out of a flooded basement while you let the faucets continue to overflow upon the floor. I had proposed the use of the MacAndrew Scale to weed out the hidden alcoholic at the pre-appointment medical.
During September, 1979, I sent the following communication to Deputy Commissioner Conboy:
"Over the past few years, situations have occurred in which alcoholic or mentally ill New York City police officers injured the person or property of innocent citizens. While these events are tragic for both the police officer and the injured party, they are also embarrassing to the Police Department as a whole. Questions such as "Why wasn't the officer's illness detected before the tragedy occurred?" and "Don't they screen these people for psychological problems before they are hired?" are posed by concerned citizens and exploited by the media.
In actuality, prospective police officers are given psychological tests before being hired. One of the tests which has been utilized is the Minnesota Multiphasic Personality Inventory (MMPI). This test has been called the "dean" of the objective measuring instruments. The standardization process for the inventory, developed by McKinley and Hayhaway in 1942, was conducted with a wide range of persons judged to have specific psychological abnormalities, and as a person answers the 550 statements, his or her response patterns can be compared with those of the large standardization sample. Obviously, such comparisons mean that the inventory is oriented toward detecting abnormality. The scale is organized around the true-false statement; the test-taker reads each statement and marks it either true or false. The subjects' responses to the large number of statements then enable the psychologist, utilizing established procedures (or computer analysis), to develop a response profile.
With regard to alcoholism, research utilizing the MMPI has focused on the traits of behavior, acting out, impulsivity and authority conflict forming the major elements of the pre- alcoholic personality. As the person grows older, distress, depression, anxiety, feelings of inadequacy and guilt reactions increase. It is felt that the MMPI could be of predictive value in detecting alcoholism prior to the onset of overt symptoms.
Evidence from review of the literature demonstrates the relevance and primacy of the MMPI alcoholism scales in discriminating alcoholic subjects from controls. The findings show that some scales, such as the MacAndrew, are consistent in separating the two groups, even when the alcoholics are not subdivided into homogeneous categories. They also suggest that alcoholics have personality characteristics which distinguish them from other groups, and thus provide psychometric justification for alcoholism diagnosis. The MacAndrew Alcoholism Scale has been called one of the best of its kind to be applied where a screening test is necessary. It was developed by Craig MacAndrew, Ph.D., Department of Social Science, University of California at Irvine, in 1965.
The absence of face validity on the MacAndrew scale masks the purpose of the scale and, therefore, unlikely to arouse suspicion, hostility and defensiveness. This suggests that the scale may be particularly useful among populations with the need to conceal or deny alcoholism. Furthermore, research has shown that the scale assesses a general dimension of acting out behavior. This would suggest that the people that this scale would detect would be most likely to translate their deviant personality qualities into behavior. This might manifest itself in ways which would be embarrassing and tragic for a police department.
All of the aforementioned information seemed to suggest that the MacAndrew Alcoholism Scale might make an excellent screening device to be applied to candidates for the position of police officer. In order to test this idea, thirty names were drawn at random from the Counseling Service files of persons who had been interviewed by us for the possible existence of alcoholism problem. All of these persons had been appointed to the job in 1973 or 1974. With the cooperation of P.O. Robert Daley, Ph.D., Director of Psychological Services, we obtained the MMPI responses that were given by these persons at the time of their entrance investigation for appointment to the job. Due to the fact that some MMPI's that were requested were unavailable, and the two females in the random drawing were eliminated, we ended up with an experimental set of twenty-two MMPI's.
These MMPI's were then scored on the 49 items that comprise the MacAndrew scale. Part of this scoring involved checking the "F" scale of the MMPI (F is a general scale of "faking" and scores above 16 are usually indicative of an invalid test). In fact, six was the highest raw F score found on these 22 MMPI's. After determining that these particular tests were valid, interpretation was simply a matter of applying the appropriate cutoff score of 24. Thus, those persons having scores of 24 or more can be said to be alcoholic. Scoring can be viewed primarily as a clerical function, and little special training is necessary.
When these 22 subjects were referred for interview to the Counseling Service, anywhere from 1 to 5 years after being hired, 16 of them were diagnosed as alcoholic. Of these 16, ten of them were shown to be alcoholic by our application of the MacAndrew scale to the application of the MMPI's that were given them when they were being hired. It appears that, if the MacAndrew scale had been applied to these MMPI's at that time, sixty-three percent (63%) of those police officers later diagnosed as alcoholic could have been screened out.
In the experimental group used in this study, there randomly appeared among those that would have been correctly identified as alcoholic by the MacAndrew scale at the time of their hiring, subjects who not only proved to be alcoholic, but who got involved in serious disciplinary problems. One of these cases proved to be an embarrassment to this Department when his deviant behavior was reported in the press. Had the MacAndrew Scale been applied to his MMPI five years ago, he might never have been hired.
In addition to the 10 subjects that the MacAndrew scale identified in agreement with the Counseling Service's diagnosis, three subjects were identified as alcoholic by the MacAndrew Scale that were not diagnosed as alcoholic by the Counseling Service. A possible explanation for this might be that those three subjects were able to deceive the Counseling Service interviewer, but were unable to deceive the MMPI. Indeed, the "F" scales on their MMPI's indicated that they were not "faking" on the MMPI. Since the MacAndrew items lack face validity, it is unlikely that these subjects would have been aware that those items were causes for alcoholism and, thus, they would have been unaware of a need to fake. Whereas, when someone is summoned to the Counseling Service for an interview, they know their drinking behavior is being investigated, and faking must always be expected. Though the counselors are skilled in dealing with this, it is not unlikely that the three cases out of the 22 might have been misdiagnosed.
There are also valid possible explanations for the six cases that the Counseling Service diagnosed as alcoholic, which were undetected by the MacAndrew Scale. The most obvious explanation is that these subjects might not have developed the disease of alcoholism at the time they took the MMPI, but that it developed in them sometime later. Thus the Counseling Service interview conducted two to five years later would have detected alcoholism because, by then, the disease had developed; whereas, the disease was not present earlier and, thus, there was nothing for the MacAndrew to detect.
There is another explanation which may be valid. The Counseling Service, at times, treats people who are not classically and chronically alcoholic, but who have resorted to alcohol due to some stressful situation in their lives. These people are said to be situational or reactive drinkers. Their drinking is really an isolated behavior; or acute, rather than chronic. Though, for administrative purposes, they would be diagnosed as alcoholic, they would not be alcoholic in the same way as a person suffering from the classical disease. As such, the MacAndrew Scale might not classify these persons as being alcoholic. That would account for the six cases in our experimental group that were identified as alcoholic by the Counseling Service, but not by the MacAndrew Scale.
In order to test how the MacAndrew would do when applied to subjects who had never been interviewed by the Counseling Service, ten names of male police officers appointed in 1973 or 1974 were selected at random from the medical files stored in the Health Services Division. These persons had never been referred for interview by the Counseling Service and had no record of suspicion of alcoholism in their medical records. Their MMPI's were obtained from Psychological Services and scored according to the MacAndrew procedure.
Of the aforementioned 10 control cases, only two wee identified as alcoholic by the MacAndrew scale. While these two might be regarded as false positives, it seems just as possible that they might be two cases of previously unidentified alcoholism, possibly latent alcoholics. In any event, since most studies of the MacAndrew have found it to produce approximately 20% false identifications, its performance of this control group seems to indicate that it is valid for the population about which the Police Department is concerned. The MacAndrew scale correctly identified eight police officers as being non-alcoholic out of 10 presumably non-alcoholic police officers
This report should be viewed as a preliminary study. Certainly more tests should be conducted with a larger and more diversified experimental group and with a similarly large and diversified control group. Also, necessary steps will have to be followed to ascertain a valid cutoff score for females.
If the results of this preliminary study are confirmed by further study, I believe the MacAndrew Scale should seriously be considered for use as a screening device for candidates for police officers. The Police Department has a moral obligation to the public to employ people with the lowest risk factors. Even if the MacAndrew would yield 20% false positive indications, I believe that would be erring in favor of the public, and would be more than compensated for in the 63% or better true positives that would be screened out." In addition, I sent Conboy a 46-page study.