In this Issue...
Over the years, we have frequently received calls to our clinical support service requesting assistance with SASSI profiles that include significantly high or low DEF scores. In response, we have published articles in past issues of News & Reports discussing the clinical implications of these scores and the suggested techniques for conveying feedback to clients (Aug 1994 Vol. 2, No. 3). It is gratifying to note that much of what we have been sharing with you about high and low DEF scores has been consistently supported and verified by callers.
In 1996, we published a newsletter that focused on domestic violence and included an interpretation of a "composite profile" of 15 perpetrators of domestic violence (Nov 1996, Vol. 4, No. 4). One noteworthy feature of the profile was a high DEF score. Many callers continue to verify that the high DEF feature, as well as other aspects of the "composite," typifies many of the perpetrators they have assessed using the SASSI.
In this issue, we would like to revisit the topic of high DEF scores and domestic violence by presenting an actual case called into our clinical support service. The client is a 38-year-old male named Jim (not his real name), who was referred for a substance abuse evaluation following a second arrest for domestic violence. The practitioner calling in the profile reported having collateral evidence substantiating a significant history of alcohol abuse for this client.
The SASSI results indicate that Jim has a low probability of having a substance dependence disorder. He is not acknowledging any significant problematic use of alcohol (FVA=0) or other drugs (FVOD=4). In fact, he denies having any of the symptoms commonly associated with substance dependence (SYM=1). However, note that Jim's responses are highly defensive (DEF=9) and significantly similar to individuals who are instructed to minimize and conceal problems. Given that his report on the FVA and FVOD is in direct conflict with information from other sources, it is likely that he is minimizing the degree to which he has experienced alcohol and other drug problems or related symptoms. This increases the risk that the SASSI classification of low probability may be in error - in other words, the accuracy of the decision rules may be slightly decreased. As in most assessment situations where the client is relatively defensive, augmenting self-reported alcohol and drug history with data from external sources is advisable before ruling out substance abuse problems.
Experienced SASSI users working in criminal justice, EAP, DOT, child protection, and other similar settings will recognize this profile as relatively common for clients who are mandated for assessment. Indeed, Jim has been charged with assaulting his partner for a second time. One possibility is that he fears a harsh punishment may be coming if he does not present himself in a favorable way. He may also be convinced that he is not to blame for his behavior, explaining that his partner provoked him or that he was acting in self-defense. While the SASSI does not reveal the exact cause or reason, the high DEF score is a strong indicator that Jim approached the assessment in a defensive manner.
Notice also that Jim's OAT score is bordering on significance given that it falls right on the 15th Percentile (OAT=2), meaning that only 15% of the general population would score this low. A score in this range usually indicates a person who does not identify with any of the problematic behaviors typically associated with substance abuse (for example, anger management problems, negativity, self-centeredness, etc.). Jim is not likely to acknowledge having these behaviors and probably wants to be viewed as being completely different from people who do. Individuals with a family history of addictive or violent behavior often cope by distancing themselves from the addict or perpetrator as if to say, "I'm nothing at all like my alcoholic mother or physically abusive father." In fact, the caller reported that Jim's mother is an active alcoholic.
Jim's FAM score of 11 is also significantly elevated (above the T 60 line or the 85th percentile). His responses are similar to family members of substance dependent individuals. It is likely that he shares many of the characteristics and traits commonly associated with individuals living in addictive family systems -- obsession with controlling the thoughts, feelings and/or actions of others, lack of adequate or healthy psychological, emotional and physical boundaries in relationships, and inability to trust others. Certainly, one theme for individuals with high FAM scores involves their sense of happiness and self-worth being dependent on fixing or controlling the behavior of others. Jim may have learned early on the false perception that the only way he can have a sense of well-being is when he is in complete control of his partner. This need often can result in the perpetration of violence in cases where poor interpersonal boundaries and lack of trust exist in a person with serious impulse control problems. Thus, like other perpetrators of domestic violence, Jim may feel enmeshed at every level with his partner, seemingly unable to restrain himself when he feels like he is losing control of his partner's behavior.
To summarize, then, Jim's profile is similar in many ways to that of other known perpetrators of domestic violence who have completed the SASSI. Although he is classified as having a low probability of substance dependence, his responses are characterized by a significant degree of defensiveness. This, along with other assessment evidence, increases the risk that he has minimized his alcohol and other drug problems and that the SASSI results of low probability of substance dependence may be inaccurate. Jim does not recognize or accept responsibility for his own behavioral problems. Like other domestic violence offenders, he tends to focus almost exclusively on controlling his partner's behavior as a way of achieving happiness and contentment in life. Jim's family history of alcoholism is likely a significant contributor to his behavioral problems and also increases the risk that he may have, or may be developing, a substance-related disorder.
Treatment Recommendations:
Ongoing assessment will be necessary to completely rule out the possibility of substance abuse or dependence. Because of the impact that most psychoactive substances tend to have on reducing impulse control, Jim's risk for reoffending is greatly increased if he has a substance-related disorder that is left untreated. Collateral sources of information concerning Jim's alcohol and drug history seem to indicate that his problems with alcohol and other drugs may be more serious than he is reporting on the SASSI. If further assessment results confirm a diagnosis of substance abuse or dependence, his treatment plan would need to include some form of addictions therapy. In addition, a no-use contract and regular toxicological screens could be useful ways to lower his risk of using and support a period of abstinence.
Jim's defensiveness could be a serious barrier to engaging him in a therapeutic relationship, let alone making any significant progress in helping him to change any of his problematic behaviors. Establishing rapport and gaining Jim's trust and confidence would be important steps in creating and maintaining a therapeutic alliance with him. Didactic, cognitively based educational approaches are often viewed by defensive clients as less intrusive and non-threatening. Initially, he may respond more favorably to presentations, films, books, etc., emphasizing the impact of addictions on the individual and their families. This may help to increase Jim's awareness of his own misuse of substances and provide him with some insight into the dynamics of his own family's behavior, including his alcoholic mother. Family involvement in his treatment may also be beneficial.
Referral to a practitioner or program that specializes in treating perpetrators of domestic violence should be strongly considered. Remember that Jim may have little or no awareness that he is responsible for his own violent behavior. His perceptions may be completely dominated by the belief that he has a right to behave in this manner with his partner. Such deeply ingrained patterns of thought and associated impulse control problems are often difficult for clients to begin to recognize, much less change. Support and process groups facilitated by behavioral health professionals trained in the treatment of domestic violence offenders are often an effective approach in helping perpetrators begin to acknowledge their behavioral problems and to effect some healthy changes.
The following report was developed to provide information on the accuracy of the SASSI-3 to people who may not be familiar with the terms that are commonly used to describe the reliability and validity of assessment instruments.
In addition to supporting its use for clinical assessment, the results in the report mentioned below indicate that the SASSI-3 gives highly consistent results over time and is an accurate measurement tool. Most importantly, it was found that the outcome of the SASSI-3 decision rules agree with independent clinical diagnoses 94 percent of the time. Further, classifications on the SASSI-3 proved to be highly accurate in five different treatment settings; and the influence of racial and ethnic group membership, age, and educational level on the accuracy of the SASSI-3 was found to be negligible. Finally, the level of general adjustment or maladjustment of the person taking the questionnaire was found not to have a significant effect on the accuracy of SASSI-3 results.
The SASSI-3 is a psychological questionnaire developed to screen individuals for substance dependence and substance abuse. It currently is used widely by organizations including addictions treatment centers, criminal justice programs, hospitals, other health care organizations, and employee assistance programs. The research conducted to develop and evaluate the effectiveness of the SASSI-3 is reported in an article published in the Journal of Personality Assessment (Lazowski, Miller, Boye, and Miller, 1998).
The effectiveness of a questionnaire may be a result of many things such as the time to complete, the reading level and ease of scoring. The most important aspects are summarized by measures of accuracy (or "validity") and consistency (typically called "reliability"). Although accuracy is much more important, consistency is traditionally mentioned first.
Consistency: The most relevant aspect of consistency of the SASSI is the degree to which an individual gets similar scores when it is administered at different times. This consistency across time is important for a clinician using the SASSI. Forty people were asked to take the SASSI-3 twice, with the two administrations being two weeks apart. The stability coefficient, a measure of how similar the two results were, ranged from .92 to 1.00 (a perfect match) for the various subscales. One measure of consistency is the degree to which the items measure the same thing. "Coefficient alpha," the statistic used to measure this "internal consistency," can range from 0 to 1.00. Using questionnaires from 1,821 individuals, the coefficient alpha for the entire SASSI was found to be .93. Thus, the SASSI is shown to have good reliability in terms of internal consistency as well as stability over time.
Accuracy of Classifications on the SASSI-3: The decision rules of the SASSI-3 classify individuals as having either a high probability or a low probability of substance dependence. The goal of the SASSI-3 was to have these classifications agree with the independent judgments of clinicians on the presence or absence of substance use disorders. The appropriate measure of accuracy is then the percent agreement between the outcome of the SASSI decision rules and the clinicians' diagnoses. The types of samples one selects may influence even this simple percent. If one contrasted very healthy, well-adjusted individuals with no incidence of drinking or drug problems with a sample of substance dependent individuals that clinicians might characterize as "late stage" who are quite willing to acknowledge their disorder and its consequences, it is reasonable to expect high agreement between clinicians and tests. Rather than select such extreme groups, clinical settings were selected that would be more representative of the range of clients that clinicians in general could find when screening for substance dependence. This would make it more likely that the percent agreement found with these samples would reflect the SASSI-3's accuracy in typical clinical practice.
Clinicians who work with substance dependent people were asked to rate a set of clients regarding whether or not they have a dependency problem, without using the SASSI-3 as part of their evaluation. The SASSI-3 was administered separately, and the clinicians' judgment was compared with the SASSI-3 results. The level of accuracy in this comparison was 94%.
The SASSI-3 is less accurate if the question is whether someone is an abuser of alcohol or drugs but not substance dependent (according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, DSM-IV). In the clinical samples used to devise the SASSI-3 decision rules, 67 people were diagnosed by clinicians as having a substance abuse disorder but not a substance dependence disorder. In a sample that included these 67 people and 172 people who were diagnosed as not having any substance use disorder, the SASSI-3 incorrectly identified only 10 of the 172 people who did not have a substance use disorder but missed 20 of the 67 individuals clinicians considered to be substance abusers. Thus, as an additional aid to assist clinicians in identifying potential substance abusers for further assessment, The SASSI Institute has developed a set of substance abuse guidelines that are available upon request.
Different settings in which the SASSI-3 is used: To determine if the SASSI-3 is equally useful in a variety of settings, five different types of treatment agencies were used. The percentage of subjects originating from each setting and the accuracy of the SASSI-3 results were as follows:
| Type of agency | Percent of total* | Correct classification |
|---|---|---|
| Addictions treatment centers: | 32% | 96% |
| General psychiatric hospitals: | 31% | 97% |
| Vocational rehabilitation program: | 22% | 94% |
| Sex offender treatment program: | 9% | 93% |
| Dual diagnosis hospital** | 7% | 98% |
*Percent of total cases adds up to 101% due to rounding error.
**Clients have both substance abuse and psychiatric problems.
As the table shows, accuracy of classification was nearly equal in all settings.
Demographic factors and accuracy: There is always a danger in psychological testing that a questionnaire that works well with one type of population does poorly with another. To check on the SASSI-3's ability to discriminate between substance dependent and non-dependent people of various types - both genders, different occupations, etc.- a procedure called "logistic regression analysis" was used. The results showed that differences in gender, racial or ethnic group membership, occupational status, marital status, age or educational level did not significantly affect the accuracy of the SASSI-3. Accuracy rates were as follows: males, 96%; females, 97%; racial or ethnic groups, varied from 92-100%; occupational status, 94-100%; marital status, 92-98%; age group 93-97%, and educational level, 93-100%. These rates are essentially the same as the SASSI-3's overall accuracy rate.
Functioning Level: One concern when using a measure such as the SASSI-3 is the possibility that reported "accuracy" may be higher than it actually is because it is measuring general maladjustment and not just the specific disorder intended. This could be the result of substance dependent clients being more maladjusted than nondependent clients in the study samples. This was tested as follows:
The DSM-IV diagnostic criteria include a clinician's estimate of general life functioning, known as the Global Assessment of Functioning, or GAF. The GAF can be thought of as giving an indication of how well an individual is functioning. SASSI-3 users need to know if a client's level of functioning might have an effect on the SASSI-3 results. Again, the logistic regression analysis procedure was used, and it was found that there was no significant impact of a client's GAF on the accuracy of SASSI-3 decision rules. There are two meanings to this finding: first, that the SASSI-3 works, regardless of whether the client is in crisis or functioning rather well overall; and second, that the SASSI-3 is not measuring maladjustment itself.
The defensiveness (DEF) scale of the SASSI-3 was designed to identify individuals who respond to the questionnaire in a defensive manner. The items on the DEF scale were selected because they were found to discriminate between people who completed the instrument under standard instructions and substance-dependent people who were given instructions to try to hide any sign of substance misuse. The DEF scale is used in conjunction with other scales on the SASSI-3 as part of the decision rules that classify individuals as having a high or low probability of being diagnosed as substance dependent.
Although the SASSI-3 is designed to identify some people who respond to the questionnaire in a less than candid manner, defensive responding would nonetheless be expected to reduce the accuracy of the SASSI-3 (as would be expected with any assessment tool). The table below presents the accuracy rates of SASSI-3 classifications as a function of scores on the DEF scale.
| DEF Score | Accuracy Rate | n* |
|---|---|---|
| 0-4 | 96% | 465 |
| 5 | 97% | 101 |
| 6 | 92% | 74 |
| 7 | 87% | 71 |
| 8 | 91% | 44 |
| 9 | 83% | 30 |
| 10-11 | 71% | 14 |
* n refers to the number of individuals included in the analysis who obtained a DEF score corresponding to the score or scores indicated in the far left column of the table. The demographics of these subjects and the criteria used for their inclusion are described in Lazowski, Miller, Boye and Miller (1998).
The SASSI-3 is reasonably accurate (i.e., from 83% to 91%) even when DEF scores are somewhat elevated (i.e., 8 or 9) and somewhat less accurate only at extremely high levels of DEF (i.e., 10 or 11). Although elevated DEF scores may increase the possibility of the SASSI missing substance dependent individuals, elevated DEF alone should not be taken as evidence of substance-related problems.
These results should be interpreted with caution, due to the small numbers of respondents at the higher DEF levels. Results obtained in different settings may vary substantially.
Reference: Lazowski, L. E., Miller, F. G., Boye, M. W., & Miller, G. A. (1998). Efficacy of the Substance Abuse Subtle Screening Inventory-3 (SASSI-3) in Identifying Substance Dependence Disorders in Clinical Settings. Journal of Personality Assessment, 71(1), 114-128.