INITIAL DEVELOPMENT AND VALIDATION OF THE PHILLIPS DISSOCIATION SCALE (PDS) OF THE NMI David W. Phillips, Ph.D. David W. Phillips, Ph.D., is Psychological Consultant to the Center for Dissociative Disorders at Ridgeview Institute, Smyrna, Georgia. For reprints write David W. Phillips, Ph.D., 4015 South Cobb Drive, Suite 160, Atlanta, GA 30080-6315. ABSTRACT This paper examined whether certain previously-identified MMPI critical items and whether MMPI post-traumatic stress scales could significantly differentiate dissociative patients from non-dissocia- tive patients. Defying the common belief that the MMPI has little to offer by way of diagnosing dissociative patients, the results of this study suggest that certain critical items and the PTSD scales could indeed detect differences.However, the primary focusof this inves- tigation was to report on the derivation, reliability and validity an MMPI scale of dissociation (Phillips Dissociation Scale - PDS). The 20 item scale, which was constructed for use in all the MMPIs (MMPI, MMPI-2, and MMPI-A) was tested with a dissociative group and a general psychiatry group. Results show the PDS to be inter- nally reliable. The PDS was also found to differentially diagnose dis- sociative disorders at a statistically and clinically significant level. INTRODUCTION TheMinnesota Multiphasic Personality Inventory (MMPI) is one of the most widely used objective psychological tests largely because it efficiently provides the psychologist with an assessment of a variety of clinical symptoms. With the recent increase in suspected cases of dissociative disorders, researchers have explored the utility of this mainstay of psy- chological evaluation in the assessment of dissociative phe- nomenology, the screening for dissociative disorders, and the overall evaluation of dissociative patients. With respect to the general evaluation of dissociative patients, some psychologists have used the not solely for the detection of dissociation, but for the elucidation of various aspects of dissociative patientsnon-dissociative func- tioning (Phillips, 1992) . These clinicians have found the useful in describing both axis 1 and axis II disorders as well as dissociative patientsdegree of depression, anxiety, anger, defense mechanisms, and interpersonal styles and issues. Also it has helped to clarify additional aspects of the polysymp- tomatic pictures, including eating disorders, drug and alco- hol abuse, antisocial behaviors and sexual disorders, prog- nostic indications and ego strength, mania, hallucinations and delusions, and psychosomatic issues. But some have found the MMPI difficult to administer to dissociative patients. Loewenstein (1991) reported many of the true-or-false questions may induce traumatic respons- es in Multiple Personality Disorder (MPD) patients. He dropped theMMPI bitterly about taking it. Other clinicians have found little dif- ficulty administering the MMPI if patientsconcerns are addressed, if they are encouraged to answer the question- naire based on their feelings on the average and if they are assured that other dissociative patients are able to satisfac- torily complete the questionnaire (Phillips, 1992). When dissociative patients were thought to be rare, case studies ofMPDpatientsMMPIs were about the only method available to MMPI investigators. Brandsma and colleagues (Brandsma Ludwig, 1974; Ludwig, Brandsma, Wilbur, Bendfeldt Jameson, 1972) administered MMPIs to a male patient with three alternate personalities. All four profiles were similar on five scales (K, Hs, Pt, D and Ma). The dif- ferences between the remaining scaled scores were gener- ally consistent with interview and observational data. The most notable differences were between the primary personality and the three alters. The altersprofiles appeared more alike than unlike. The discharge profile of the integrated person suggested that the new identity was psychiatrically "sicker" than any of the others. Larmore, Ludwig and Cain (1977) found that shared scale elevations for the primary person- ality and three alters occurred only on the F, Hy, and Sc scales. These scales are entirely different than those found to be abnormal by Brandsma and his colleagues. Wagner and Heise (1974) and Danesino and colleagues (Danesino, Daniels McLaughlin, 1979) found that scale differences between the primary and alternate personalities occurred on scales sensitive to emotional states of depression (D), anxiety (Pt) and mania (Ma) and to social interaction pat- terns of masculine-feminine interests (Mf), suspiciousness of the motives of others (Pa) and social isolation (Si) . Confer and Ables (1983) found that the greatest volatility between personalities pre and post integration occurred with the Mf, scale which reflected strongly different attitudes and inter- ests of the different alters. The reasons for conducting these case studies were to provide some objective measure of the existence of alter per- sonalities as well as to provide some understanding of the relationship of the personalities to each other. For those purposes, these research efforts were hampered by the extent to which role playing could produce similar results and by 92 DISSOCIATION. Col. VII. No. 2. June 1994 the difficulty determining whether different alters did in fact solely contribute to their respective MMPIs. Interest in case studies ofMMPI profiles waned as the number of cases increased. With the growing numbers of identified MPD patients, attention turned to examinations of groups of dissociative patients MMPI profiles. A number of authors examined the ways in which profiles differ between dissociative sam- ples and non-dissociative samples as well as how their pro- files appear on the average. The intent was to provide psy- chologists with objective data for differential diagnoses. Solomon and Solomon (1982; Solomon, 1983) pointed out it is a misconception that multiple personality patients have a particularly elevated hysteria scale relative to the other MMPI scales. Bjornson, Reagor and Gaston (1988) found that MPD patients, as compared to other diagnostic groups, scored higher on scales F, Pa, Sc and Ma. They endorsed more obvious content on scales Pa-0 and Ma-O and more subtle content on Pa-S. Scale Sc2 was also elevated. Bliss (1984; 1986) reported a relatively consistent pro- file for 15 female MPD patients on the MMPI. The average patient had marked elevations on the F and Sc scales. There were also elevations on scales Hs, D, Pd, Pa, and Pt. The mean profile was as follows: L-46, F-85, K-45, Hs-79, D-86, Hy-77, Pd-85, Mf-38, Pa-83, Pt-84, Sc-100, Ma-69, Si-71. These findings are consistent with the clinical observations of some of a multitude of symptoms in patients diagnosed with a dis- sociative disorder. Bliss suggested the reason for the high elevations of the F and Sc scales is understandable when the items that comprise those scales are examined. First, thir- teen items are common to both scales. The F scale contains a predominance of statements related to psychotic symp- toms ( hallucinations, delusions, paranoid ideas) , depression, family discord, sociopathy, dissociations, and other hypnotic phenomenon - all commonly found in these patients. In turn, the Sc scale has many items pertaining to social isola- tion, depression, family discord, bad thoughts and urges as well as a preponderance of items of a dissociative or hyp- notic nature - again typical of these patients. Furthermore, according to Bliss, the elevated scores on many of the other scales are consistent with a multiplicity of symptoms char- acteristic of these patients. Bliss believed that the singular- ly high F and Sc scores coupled with other elevations may be alerting but not definitive. Coons(1984) ten multiple personality patients produced an average MMPI with an 8-4-7 profile (L-47, F-84, K 50, 64, D-76, Hy-68, Pd-85, Mf-47, Pa-74, Pt-79, Sc-87, Ma-62, Si- 69) . He noted that numerous colleagues remarked how sim- ilar this average profile and the individual profiles which comprise it are to borderline personality MMPI profiles. Fink and Golinkoff (1990) found more identifiable dif- ferences between MPD and Schizophrenics MMPI profiles than betweenMPD and Borderline Personality Disorder(BPD) MMPI profiles. The only difference between MPD and BPD profiles was a somewhat more elevated HS Scale in the MPD sample. Fink and Golinkoff found that theMPD patients had greater overall elevations than the schizophrenics on all 13 validity and clinical scales except scale Mf. In an attempt to objectify the identification of dissocia- tive patients via the MMPI, Coons and colleagues (Coons Fine, 1988; 1990; Coons Sterne, 1986) identified signs of dissociative profiles which they suggest distinguishes them from non-dissociative profiles. The most frequent high-point pair was F and Sc for MPD patients. The mean number of scales elevated over 70 was six. Only infrequently was scale Hy elevated over 70. The following cutoffs were used to iden- tify MPD patients: F> 80; Hs> 70; D>70; Pd > 80; Mf="low;" Sc >80; Pa > 70; Pt > 70; Ego Strength > 45; and Family Discord > 65. At least three clinical scales need to be greater than 70. The presence of at least one critical item pertain- ing to sex needs to be present. Three out of five obvious- subtle scale pairs need to have obvious scales greater than subtle scales. There were 10 to 15 Grayson Critical Items endorsed. One of the most important MMPI criteria was a positive response on either critical item #156 OR #251 (items 168 and 229 on the MMPI-2). Both of these items indicate the presence of amnesia and identity alteration. Further research is needed to determine the utility and predictability of their criteria in distinguishing dissociative patients from a host of other psychiatric populations. While the descriptive identification of MMPI profiles is useful, Coons and colleaguesapproach begins to take this research to the next step. That is: How well can certain hypothesized crite- ria of the MMPI identify whether a patient has a Dissociative Disorder or the degree to which an individual has clinical dissociative symptoms? Separate from the above research efforts are clinical impressions that have developed in the use of the MMPI. One of these is particularly noteworthy. A number of clini- cians report using the Post-Traumatic Stress Disorder (PTSD) scales of the MMPI, the PK (Keane, Malloy, Fairbank, 1984) and PS scales (Schienger Kulka, 1986) , as indicators of dis- sociation. Since dissociative disorders are thought to be severe reactions to trauma it is believed that the related PTSD symp- toms would be detected by these scales. However, research data is needed to support or refute this assumption. Taken together the following conclusions can be made. First, the MMPI has not yet been shown to be useful in dis- tinguishing dissociative patients from borderlines (Fink Golinkoff, 1990). Many of the MMPI indicators of dissocia- tive patients may only be measuring polysymptomatic psy- chopathology instead of dissociation per se. There are two possible exceptions. One is the critical items noted by Coons and colleagues. MMPI items 156 and 251 measure some amnes- tic phenomenon and identity alteration. A second promis- ing index is either of the PTSD scales. However, both the critical items and the PTSD scales need research to test the degree to which they measure dissociation and to determine if they can assist in differential diagnosis. Given some of the initial problems in establishing the efficacy of the MMPI in the detection of dissociation, the question of why pursue further research in attempting to uncover indices of dissociation in the MMPI is raised. The primary advantage of using the MMPI in the measurement of dissociation is similar to the matter of screening tests for medical diseases. The lower the expense and the higher the 93 DISSOCIATION,Vol, VII.`o. 2. June 1994 TABLE 1 Phillips Dissociation Scale (PDS) MMPI-2 Item Numbers MMP1 MMPI-A F = False T = True Items 23 22 21 T At times I have fits of laughing and crying that I can not control. 24 27 22 T Evil spirits possess me at times. 48 40 45 T Most anytime I would rather sit and daydream than (to)a do anything else. 60 48 433 T When I am with people, I am bothered by hearing very strange things. 72 50 250 T My soul sometimes leaves my body. 159 174 152 F I have never had a fainting spell. 165 178 158 F My memory seems to be alright. 168 156 161 T I have had periods in which I carried on activities without knowing later what I had been doing. 182 194 175 T I have had attacks in which I could not control my movements or speech but in which I knew what was going on around me. 198 184 439 T I often (commonly) hear voices without knowing where they come from. 229 251 214 T I have had blank spells in which my activities were interrupted and I did not know what was going on around me. 247 273 231 T I have numbness in one or more places on (regions of) my skin. 295 330 275 F I have never been paralyzed or had any unusual weakness of any of my muscles. 296 332 276 T Sometimes my voice leaves me or changes even though I have no cold. 308 342 288 T I forget right away what people say to me. 311 345 291 T I often feel as if things are not real. 319 350 299 T 1hear strange things when I am alone. 336 275 315 T Someone has control over my mind. 355 291 332 T At one or more times in my life I felt that someone was making me do things by hypnotizing me. 361 293 337 T Someone has been trying to influence my mind. 'Parenthetical words are used in MMPI items but are not included in MMPI-2 or MMPI-A items. Underlined phrases were added to MMPI-2 and MMPI-A items. Source: Minnesota Multiphasic Personality inventory (MMPI). Copyright ? the University of Minnesota 1942, 1943 (renewed 1970). Reproduced by permission of the publisher. Minnesota Multiphasic Personality Inventory-2 (MMPI-2). Copyright? by the Regents of the University of Minnesota 1942, 1943, {renewed 1970), 1989. Reproduced by permission of the publisher. Minnesota Multiphasic Personality Inventory-Adolescent (MMPI-A). Copyright? the Regents of the University of Minnesota 1942, 1943 (renewed 1970), 1992. Reproduced by permission of the publisher. "MMPI," "MMPI-2", "MMPI-A," "Minnesota Multiphasic Personality Inventory", "Minnesota Multiphasic Personality Inventory-2 " and "Minnesota Multiphasic Personality Inventory-Adolescent " are trademarks owned by the University of Minnesota. 94 DISSOCIATION. convenience, the greater likelihood a particular clinical test will be used to regularly screen a condition. Since the is already used worldwide by psychologists, a measure of dis- sociation constructed from MMPI itemswould be more eco- nomical and convenient to use than a separate, specialized clinical test. There may be no reason to suspect dissociative pathology. Thus, a separate dissociative test may not even be administered and dissociative conditions could be over- looked. Whereas each time the MMPI is administered the psychologist would have a measure of dissociation. As a result, dissociative phenomenon could be systematically assessed in routine psychological evaluations. A second benefit has to do with the issue of factitious patient responses. Currently-used dissociation scales are bla- tant and self-evident in their measurement of dissociative phenomena. Their intent is immediately recognizable by their content. Consequently, responses can be easily slant- ed to convey a desired impression (Gilbertson, Torem, Cohen, Newman, Radojicic, Patel, 1992). The MMPI, on the other hand, measures indicators of many different syndromes and characteristics. Its items reflect this diversity. Thus, it is more difficult for a patient to discover the focus of the assessment. This may reduce the likelihood of manipulative response patterns. For patients who are prone to suggestion, who wish to fake MPD, or who wish to avoid having their dissociation detected, a test which clearly intends to measure only dis- sociation will unnecessarily alert the patient to the subject of scrutiny. This may result in a distorted protocol. Using a scale within the total MMPI may dilute this tendency by not so clearly giving away its aim. TheMMPI also provides valid- itymeasures which can alert the psychologist to respondents with exaggerating or suppressing response patterns. Given these potential advantages of using the MMPI to screen dissociative conditions, it was decided to continue in the effort to establish and validate some criteria of dis- sociation. To that aim, this study examined the utility of the PTSD scales in assessing dissociation and evaluated critical items which seem to measure dissociation. The main pur- pose of this study, however, was the construction and vali- dation of a dissociation scale (Phillips Dissociation Scale - PDS) of items selected from the MMPI. METHOD Scale Derivation, Description and Scoring The PDS items were selected by the author based upon the current clinical and theoretical understanding of disso- ciation. Any of theMMPI items which, based on its face valid- ity, gauged some core aspect of dissociation was chosen. Items which assessed symptoms associated with a dissociative diag- nosis but which did not measure dissociation per se were excluded. Another guiding principle of item selection was choos- ing items which could be used interchangeably with all the MMPIs (MMPI, MMPI-2 andMMPI-A [adolescent] ) . Items were picked which exist in the same or substantially the same word- ing on the MMPI as they do on the MMPI-2 andMMPI-A. Of the final items selected, only three are worded differently on one version versus the others. These differences are minor and insubstantial (See Table 1). The scale has 20 items to which the respondent answers true or false. Three items are keyed false. The total score is a tally of the number of items endorsed in the keyed direc- tion (See Table 1). Thus, scores can range between 0 and 20. All PDS item numbers differ from one versioncom- pared to the others (See Table 1). Thus, item placement varies depending on the form. The items are scattered through- out theMMPIs. The MMPI/PDS includes items numbered 22 through 293 (The MMPI has 566 items.). The MMPI-2/PDS has items numbered 23 through 361 (TheMMPI-2 includes 567 items.). The MMPI-A/PDS has items from number 21 through 439 (The MMPI-A has 478 items.). Another differ- ence pertains to the non-PDS items which surround the PDS. Since these items vary between the MMPIs, the context of the PDS within versions is dissimilar. Whether this affects scores and whether norms for the MMPI-2/PDS can be used for the MMPI/PDS (and so on) are questions thatwill become answerable only by the aggregate data from future studies. The areas of dissociation assessed by the PDS are noted below. After each identified category are the MMPI-2 item numbers associated with each class of symptoms. Since some items are listed in more than one category the total num- ber of items listed below exceeds 20. Identity alteration is questioned by six items (23, 168, 182, 229, 296, 355). Five items measure conversion symptoms (159, 182, 247, 295, 296). The DSM-IV(American Psychiatric Association, 1994) does not list conversion disorder as a dissociative disorder. However, it is listed in the (World Health Organization, 1992) as several syndromes: dissociative disorders ment and sensation, dissociative motor disorders, dissocia- tive convulsions, and dissociative anaesthesia and sensory loss. There are four items assessing amnesia (165, 168, 229, 308). Passive influence phenomena are addressed by three items (361, 355, 336). Hearing voices is a core symptom of MPD which is evaluated by three items (60, 198, 319). One item assesses absorption in fantasy and trance phenomena (48). One item gauges derealization (311). Another item measures depersonalization (72). The sense that one is sessed is addressed by one item (24). Sample Characteristics There are two samples in this study. Both were selected from private practice patients primarily those of the author. Patients were classified as dissociative or non-dissociative on the basisofDSM-lII--Rcriteria for dissociative disorders. While no specific measure of socioeconomic status was taken, it is estimated that most of the patients in both samples were in the middle to upper-middle social classes. The Dissociative Disordered (DD) sample consisted of 20 patients who were diagnosed as either MPD or Dissociative Disorder Not-Otherwise-Specified (DDNOS) . The average age was 37. One was male. All were Caucasian. At the time the test was administered 60% were outpatients. Thirty-five per- cent were employed. Half were married. The general psychiatry (GP) sample consisted of 20 95 DISSOCLtTION. Vol. VII, No. 2. June 1994 5 a I RAI ed highest on a factor of hearingvoic- es. Several items defined the factor trance/depersonalization. They include 24, 48, 311, 336 and 361. Statistical analyses (See Table 2) revealed the DD sample (avg-1 L 1; t-Score sd=4.6) had a significantly higher mean (t= 7.99, p<0.001, df=38) than the GP sample (avg=2.2; sd=2.0). 7.99 (p<0.001, df = 38) The range of scores for the DD sam- ple was from 4 to 19. On the other hand, the range of scores for the GP sample was from 0 to 6. To determine what scores are pathological, an examination ofvar- ious cutoff scores was undertaken to decide the most effi- cient cutoff. A cutoff score of 4 yielded 0% false negatives but 25% false positives. Thus, none of the DD sample was incorrectly identified but a fourth of the GP sample was iden- tified as dissociative. The total percent of subjects identified into their correct sample was 85%. Using a cutoff score of 5 yielded a 5% false-negative rate and a 15% false-positive rate. The percent correctly identified was 90%. A cutoff of 6 produced 15% false negatives and 10% false positives. The percent correctly identified was 88%. Using a cutoff of seven yielded a false-negative rate of 25% and a false-positive rate of 0%. The percent correctly identified was 88%. To provide norms for PDS scores, standardized T-scores (with a mean of 50 and a standard deviation of 10) were cal- culated from the GP sample. The following list provides scores from 0 to 20 along with the respective standardized T-scores: 0-39, 1-44, 249, 3-54, 4-59, 5-64, 6-69, 7-74, 8-79, 9-84, 10-89, 11-94, 12-99, 13-104, 14-109, 15-114, 16-119, 17-124, 18-129, 19-134, 20-139. To determine convergent validity, correlations were cal- culated with various MMPI-2 scales and subscales. Except for the PDS, all analyses of scales were based on their T- scores. The following scales were most highly correlated with the PDS: Sc6, Sc5, BIZ, Sc, F, PS and PK. In terms of diver- gent validity, the scales most unrelated with thePDS were: R, Mal, TRIN, Pa-S, VRIN, L, Mf and MACR. In assessing the discriminate validity of thePDS, corre- lations were determined between the sample the patient was assigned (0 = General Psychiatry Sample, 1 = Dissociative Sample) and various MMPI-2 scales including the PDS. Out of 86 comparisons, the PDS correlated higher with the sam- ple assignment (r = 0.79) than any other scales except scales Sc3, Sc5 and Sc6. While only two of Sc3 ten items (20%) are also PDS items, four of Sc5 11 items (36%) are and ten of Sc6 20 items (50%) are shared with the PDS. All of these scales achieved correlations as high as thePDS. The PTSD scales also had high correlations (PK r = 0.76; PS r = 0.77). Sc had a correlation of 0.74. Twelve of Sc 78 items (15%) are shared with the PDS. Scale F achieved a correlation of 0.70. Items which correlated highest with the sample assign- ment were: 229 (T), 311 (T), 308 (T), and 168 (T). Their respective coefficients were: 0.74, 0.70, 0.67 and 0.64. TABLE 2 Comparisons of PDS Scores Between the DD and GP Samples Sample X sd Range DD Sample 11.1 4.6 4-19 GP Sample 2.2 2.0 0-6 patients who had diagnoses other than dissociation. The aver- age age was 35. Seventy percent were female. All were Caucasian. Forty percent were outpatients. The percent employed was 79%. Sixty-four percent were married. The DSM-III-R axis I diagnoses of these patients are followed by numbers in brackets indicating the frequency of each diag- nosis: Mood Disorders (Major depression [ 4] , Dysthymia [2], Bipolar Disorder [1] and Cyclothymia [1]), Psychoactive Substance Use Disorders (Alcohol Abuse/Dependence [6] , Cannabis Abuse/Dependence [2] and Cocaine Abuse [1] ), Anxiety Disorders (General Anxiety Disorder [2] and Post- Traumatic Stress Disorder [2] ), Sexual Disorders (Pedophilia [2] and Transvestic Fetishism [1] ) , Eating Disorders (Anorexia [1] , Bulimia [3] and Eating Disorder NOS [3] ) , Adjustment Disorders [2], Schizoaffective Disorder [1], Intermittent Explosive Disorder [1] and Somatization Disorder [1]. Procedure The MMPI-2 was administered to patients either sepa- rately or as part of a battery of psychological tests. Most were administered as part of routine clinical assessments. Some were administered during the course of treatment with the instruction they would be informed of the rationale for the administration of the test after they had completed it. RESULTS The average MMPI-2 profile of the DD group was as fol- lows: L-46, -98, K-40, Hs-75, D-78, Hy-74, Pd-79, Mf-51, Pa- 81, Pt-84, Sc-96, Ma-62, Si-70. The average MMPI-2 profile of the GP group was: L-48, F-60, K 50, Hs-57, D-64, Hy-63, Pd- 65,Mf-51, Pa-64, Pt-63, Sc-62, Ma-54, Si-52. Several measures of in ternal consistency of the scale were evaluated. The split-half reliability coefficient was 0.95. The average inter-item correlation was 0.38. The range of cor- relation coefficients between the total PDS score and indi- vidual items was from 0.41 to 0.80. Various factor analytic strategies revealed four factors. They include amnesia/identity alteration, conversion symp- toms, hearing voices and trance/depersonalization. The items which loaded highest on the amnesia/identity alteration fac- tor were 168, 165 and 229. The highest loading items on the conversion factor were 159 and 295. Items 198 and 319load- 96 D1SS0Ct1T!0N, Vol. All. 1994 TABLE 3 Dissociative Disorder Sample MMPI-2 Validity Scales and Profile Characteristics Validity Scales and Profile Indices Average (Raw Scores) SD Range Correlation With PDS L (Lie) scale 29 L9 0-8 -0.16 F (Infrequency) scale 18.0 8.4 3-31 0.67b K (Defensiveness) scale 11.0 3.5 4-20 -0.20 F - K Dissimulation Index 7.0 9.9 -17-24 0.63b ? (Cannot Say Score) 2.7 5.9 0-22 0.03 Percent True 51.1 7.9 35-63 0.62b Percent False 48.5 8.0 37-65 -0.61b Profile Elevation 77.0 a 11.5 53-95 0.72b FB scale 16.5 8.7 2-28 0.56b True Response Inconsistency (TRIN) 9.1 2.0 6-15 0.14 Variable Response Inconsistency (VRIN) 5.9 2.2 2-10 -0.34 a These figures are based on the average of the T scores on eight of the clinical scales (Hs, D, Hy, Pd, Pa, Pt, Sc and Ma). b y