Link to a list of papers on dissociative disorder.

Dissociative Experiences in a College Population

Michael J. Angiulo and John F. Kihlstrom

University of Arizona

Abstract

A total of 1,463 subjects completed the Dissociative Experience Scale (DES) and the Questionnaire of Experiences of Dissociation (QED), two instruments intended to assess dissociative tendencies, and the Tellegen Absorption Scale (TAS). The three scales were highly intercorrelated, and the DES and QED appeared to be almost interchangeable. Separate principal-components analyses showed all three scales to be multidimensional, and revealed a strong component of absorption and imaginative involvement in the DES.

The recent escalation of interest in post-traumatic stress disorder (PTSD), multiple personality disorder (MPD), and other forms of dissociative syndromes has led to the development of a number of instruments intended to screen the population for individuals who already suffer from these illnesses, or who may be predisposed to develop them in the future. Although a number of questionnaires, such as the Tellegen Absorption Scale (TAS: Tellegen & Atkinson, 1974) have items reflecting experiences of dissociation and depersonalization, the first instruments expressly designed to assess dissociative tendencies were the Dissociative Experiences Questionnaire (DES) of Bernstein and Putnam (1986; Carlson & Putnam, 1988) and the Perceptual Alterations Scale (PAS) of Sanders (1986). Somewhat later, Riley (1988) introduced the Questionnaire of Experiences of Dissociation (QED), as an alternative to the DES. These self-administered questionnaires, intended for screening purposes and psychometric studies, have been supplemented by two clinician-administered structured clinical interviews, the Dissociative Disorders Interview Schedule (DDIS; Ross, Heber, Norton, Anderson, Anderson, & Barchet, 1989) and the Structured Clinical Interview for DSM-III-R Dissociative Disorders (SCID-D; Steinberg, Rounsaville, & Cicchetti, 1990). The DDIS and SCID-D both yield diagnoses of the various dissociative disorders, which are not included in the other structured interviews such as the Schedule for Affective Disorders and Schizophrenia (SADS; Endicott & Spitzer, 1978), the NIMH Diagnostic Interview Schedule (DIS; Robins, Helzer, Croughan, et al., 1981), or the standard version of the Structured Clinical Interview for DSM-III-R (SCID; Spitzer, Williams, Gibbon, & First, 1990).

Both the DES and the QED have been shown to discriminate between MPD and other conditions (Bernstein & Putnam, 1986; Frischholz, Braun, Sachs, Hopkins, Schaeffer, Lewis, Leavitt, Pasquotto, & Schwartz, 1990; Riley, 1988; Ross, Norton, & Anderson, 1988; Steinberg et al. 1991), whether these are clinically diagnosed or assessed through structured interviews. However, little is known about the distribution of DES and QED scores in the population at large, the internal structure of the scales, the relations between them, or their comparative validity in identifying individuals with (or prone to) dissociative psychopathology. In the following paper, we report a large-scale analysis of DES and QED, providing norms as well as analyses of both the internal structure of these scales and the relations between them, based on a large sample of the college-student population.

A total of 1,463 college students, approximately 41% male and 59% female, enrolled in an introductory psychology course took part in a survey session at the beginning of the semester. In return for their voluntary participation, the subjects received credit toward the research participation component of their course.

During the survey session, the subjects completed a packet of personality questionnaires. The first of these consisted of the 28-item Dissociative Experiences Questionnaire (DES; Bernstein & Putnam, 1986), the 26-item Questionnaire of Experiences of Dissociation (QED; Riley, 1988), and the 34-item Tellegen Absorption Scale (TAS; Tellegen & Atkinson, 1974). Items from the three scales were interleaved.

In their original versions, the three scales employ different response formats. The DES relies on 10-cm visual analogue scales, scored to the nearest 5 cm, and items worded in the second person (see below); the QED and TAS employ the usual first-person, dichotomous true-false format; finally, all the items of the DES and TAS are worded positively, while the QED contains several reversed items to counteract acquiescence tendency on the part of the subjects. For comparative purposes, we put the three instruments in a common format and collect responses on a common scale.

The use of visual analogue scales was deemed inappropriate for a large-scale survey, on the grounds of the sheer labor involved in scoring 88 items in this manner from more than 1,000 subjects. Moreover, we realized that a 10-cm visual analogue scale, scored to the nearest .5 cm, is essentially a 21-point Likert-type scale. As Nunnally (1978) has noted, the reliability of Likert-type scales levels off between 7 and 11 steps. Accordingly, for each item on the questionnaire we presented subjects with a 0-100 scale, divided into 11 segments (thus forming an 11-point Likert-type scale), and asked subjects to indicate "approximately the percentage of time each experience happens to you". Carlson (personal communcation, November 15, 1991) has recently revised the DES along similar lines. Summing ratings across items, the possible raw scores ranged from 0 to 2800 for DES, 0-2600 for QED, and 0-3400 for TAS.

In current practice, scores on the DES are reported in terms of the average percentage rating given to each item. For example, if a subject reported at each experience sampled by the DES occurred 50% of the time, he or she would receive a score of 50. These percentage scores may be calculated by dividing the total DES score by 28, yielding a scale of 0-100. However, the QED and TAS are scored more conventionally, in terms of the number of items endorsed by subjects (or, where Likert-type scales are used instead of dichotomous scale, the total of the ratings given to each item). For example, if a subject endorsed 50% of the items of the DES, he or she would receive an item score of 14 (if the scale were dichotomously scored) or 1400 (using the 0-100 scoring scheme represented by the present 11-point Likert-type scale). The two scores are, of course mathematically equivalent: the percentage score may be obtained by dividing the item score by 28. In order to put the three scales on a common dimension, norms for the DES, QED, and TAS are reported below in terms of percentage scores, varying from 0-100.

In addition, the DES items were reworded in the first person. For example, Item #1 of the original DES reads: "Some people have the experience of driving a car and suddenly realizing that they don't remember what has happened during all or part of the trip. Mark the line to show what percentage of the time this happens to you." This item was reworded as follows: "Sometimes I have been driving a car and suddenly realized that I didn't remember what had happened during all or part of the trip." The negatively worded QED items were reworded in a positive direction. For example, Item #7 of the original QED reads: "I rarely feel confused, like in a daze." This item was reworded as follows" "I sometimes feel confused, like in a daze."

As with any survey of this nature, some subjects failed to complete every item on the questionnaire. Of the 1,463 subjects, 1,367 provided complete scores on DES, 1,335 on QED, and 1,308 on TAS.

The mean percentage score for DES was 22.40 (SD = 12.73), with a median of 20.36. Figure 1A shows the frequency distribution of scores, which is sharply skewed toward the low end of the continuum. Percentage scores equivalent to salient decile cutpoints were: 80th percentile, 32.50; 90th percentile, 40.36; 95th percentile, 45.00; and 99th percentile, 57.32. The reliability of the DES, estimated by Carmine's theta, was .92.

Figure 1.

Frequency distributions of scores on the Dissociative Experiences Scale (DES; Panel A)

and the Questionnaire of Experiences of Dissociation (QED; Panel B).

An exploratory principal-components analysis yielded five factors accounting for 48.85% of the variance (see Table 1). These factors were subjected to orthogonal rotation by the varimax method.

When these factors were subjected to oblique rotation using the oblimin method, the correlations among them were surprisingly low, -.03 < r .47. In particular, Factor 5 was virtually independent of the others, -.03 < r < .11.

Other investigators, usually employing somewhat smaller samples, have obtained a three-factor solution to the structure of the DES (Carlson & Putnam, 1988; Carlson & Putnam, 1990; Carlson, Putnam, Ross, Anderson, Clark, Torem, Coons, Bowman, Chu, Dill, Loewenstein, & Braun, 1991; Ross, Joshi, & Currie, 1991; but see Fischer & Elnitsky, 1990; Fischer, 1990; Schwartz, Frischholz, Braun, & Sachs, 1991). The largest of these, involving a heterogeneous sample 1,574 neurological and psychiatric patients and normals, obtained factors of amnestic dissociation, absorption and imaginative involvement, and depersonalization and derealization (Carlson et al., 1991). Accordingly, a confirmatory principal components analysis (orthogonal varimax rotation) was carried out on the present data set, extracting only three factors: these corresponded closely to those obtained by Carlson et al.; oblique oblimin rotation showed that the three factors were only moderately intercorrelated:

The mean QED percentage score was 23.03 (SD = 13.32), with a median of 20.77. Figure 1B shows the frequency distribution of scores, which is sharply skewed toward the low end of the continuum. Salient cutpoints were: 80th percentile, 33.46; 90th percentile, 41.15; 95th percentile, 47.69; and 99th percentile, 61.92. The reliability of the scale was .91.

An exploratory principal-components analysis yielded four factors accounting for 48.4% of the variance (see Table 2). These factors were subjected to varimax rotation:

When these factors were subjected to oblique rotation using the oblimin method, the correlations among them were positive but modest, .20 < r < .49.

Although not intended as a clinical screening instrument, the TAS is designed to sample absorbing and self-altering experiences, including dissociative experiences, in the normal population. The mean TAS percentage score was 30.60 (SD = 15.77), with a median of 28.53. Figure 1C shows the frequency distribution of scores, which has a somewhat more normal distribution than the DES or QED. Salient cutpoints were: 80th percentile, 43.82; 90th percentile, 52.06; 95th percentile, 59.71; and 99th percentile, 73.53. The reliability of the scale was .94.

An exploratory principal-components analysis yielded four factors accounting for 47.6% of the variance (see Table 3). These factors were subjected to varimax rotation: Factor 1 included synesthesia and other items involving absorption in sensory and perceptual experiences (e.g., "Certain pieces of music remind me of pictures or moving patterns of color"); Factor 2 tapped experiences of intuition and insight (e.g., "I often know what someone is going to say before he/she says it"); Factor 3 involved various sorts of imaginative involvements (e.g., "If I wish, I can imagine (or daydream) some things so vividly that they hold my attention as a good movie or story does"); Factor 4 involved various trance states or mystical experiences (e.g., "I sometimes 'step outside' my usual self and experience an entirely different state of being"); and Factor 5 entailed positive emotional involvement in nature and language (e.g., "I can be greatly moved by eloquent or poetic language"). When these factors were subjected to oblique rotation using the oblimin method, the correlations among them were positive but modest, .22 < r < .45. These results are in considerable agreement with the findings of Tellegen (1987; see also Glisky, Tataryn, Tobias, Kihlstrom, & McConkey, 1991; Tellegen, 1981), who obtained six related factors -- responsiveness to engaging stimuli, synesthesia, enhanced cognition, enhanced awareness, vivid reminiscence, and oblivious/dissociative involvement -- in their large sample of college students.

In view of the frequent observation of sex differences in the incidence of MPD, the three scales were analyzed for sex differences (a few subjects had to be omitted from this analysis because of their failure to indicate their gender). Although women scored higher than men on all three scales, the differences were not significant on either DES, t(1363) = 1.44, or QED, t(1331) = 1.25; for TAS, however, the difference was statistically significant, t(1304) = 3.26, p < .005.

The three scales included in this study were highly intercorrelated (a total of 1,245 subjects provided data on all three scales):

Of course, these high correlations could be carried by the high degree of content overlap: although the TAS has few items relating to pathological dissociation, the DES and QED have many items pertaining to absorption and imaginative involvement.

In order to obtain a measure of categorical agreement in the assessment of dissociative tendencies, the DES and QED distributions were divided into deciles. Table 4 shows the resulting contingency table for those 1,308 subjects who had scores on both scales: the relationship between the two classifications is obviously significant, X2(81) = 2312.67, p < .0001. If the top 10% of the distribution (DES cutoff, 11.23; QED cutoff, 10.65) is assumed to contain those who are at risk for dissociative disorder, then there is a fair amount of agreement between the two procedures. A total of 90 subjects placed in the 10th deciles of both the DES and QED distributions, representing 69.23% of those with top scores on DES, and 67.16% those with top scores on the QED. Thus, there was a 93.6% agreement on the classification of subjects. When the cutoff was moved to the 95th percentile, agreement increased to 96.56%; when the criterion was moved to the 99th percentile, agreement increased again, to 99.31%.

In an attempt to further examine the relationships among the scales, an exploratory principal-components analysis of the 88 questionnaire items yielded 16 factors accounting for 55.8% of the variance. Many factors had only a few items with substantial loadings, and many items failed to show substantial loadings on any factors. In order to reduce the number of factors to manageable size, the analysis was repeated on the 14 subscales (five each from DES and TAS, four from QED) derived from the factor analyses described above, using unit weightings to aggregate items into subscales.

The analysis yielded two factors accounting for 64.6% of the variance (Table 5). Factor 1, labelled absorption, consisted of all five TAS subscales, the DES subscales for absorption and trancelike states, and the QED subscales for fantasy and trancelike states. Factor 2, labelled dissociation, consisted of the DES subscales for blackout, memory lapse, and depersonalization/derealization, and the QED subscales for blackout/disorientation and depersonalization/derealization. Oblique (oblimin) rotation showed that these two factors were highly correlated, r = .56.

While the dissociative spectrum includes a wide variety of absorptions and imaginative involvements that occur in the ordinary course of everyday living, blackouts, memory lapses, depersonalization, and derealization seem to come closest to the symptoms of dissociative disorder per se. Accordingly, the items most representative of these factors -- what might be considered "critical items" -- were selected for closer examination.

Table 6 shows the results, in terms of the frequency with which each critical item was endorsed (i.e., incidence: any rating > 0% of the time), and the frequency with which the corresponding experience was reported, in those subjects in whom it occurred at all. Thus, a total of 44.9% of subjects reported at least one episode where they failed to recognize friends (DES Item #08); and for these subjects, the mean frequency with which that experience occurred was 25.14% of the time. Similarly, 28.5% of subjects reported at least one experience where they found new things among their belongings that they did not remember buying; and for these subjects, the mean frequency with which that experience occurred was 19.54% of the time.