Articles‎ > ‎

Deprogramming: A Case Study 2

Cultic Studies Journal, 1990, Volume 7, Number 2, pages 174-216

Deprogramming: A Case Study Part II: Conversation Analysis

Steve K. Dubrow-Eichel, Ph.D. RETIRN Philadelphia


This article continues the examination of a successful deprogramming of an International Society for Krishna Consciousness (ISKCON) devotee (cultist) described in Cultic Studies Journal Special Issue Vol. 6, No. 2. The deprogramming was observed and audiotaped. A sample consisting of 1,938 speech fragments (3.8% of all "thoughts') drawn from three audiotape segments representing the beginning, redecision ('snap'), and ending phases were coded using the Deprogramming Statements Checklist  DSC). The cultists receptivity and integration of new information was assessed using the Experiencing (EXP) Scale. Results suggested that this deprogramming was a persuasive conversation and moral discourse in which the primary activities were asking for and receiving information (education), and self- disclosing (affiliation). The cultist's decreased attentional motility and increased ideational activity suggested improved concentration and implied a change in consciousness. Qualitatively, the deprogramming had distinct 'formal' (cultist-focused) and 'casual" (subgroup-focused) modes. Three change process dimensions were identified (Communicative, Cognitive, and Social-Affiliative).

This article presents a quantitative analysis of the deprogramming of "Ken Butler" (a pseudonym), a devotee of the International Society of Krishna Consciousness (ISKCON). An earlier issue of this journal (Dubrow-Eichel, 1989) presented a detailed qualitative analysis of Ken Butler's deprogramming. This article supplements the qualitative analysis by systematically examining conversation, attention, and insight during the deprogramming. First, the rationale behind choosing a case study approach to an unresearched topic is presented. I define a method of population sampling that eschews random sampling and other traditional methods (e.g., time sampling) in favor of deliberately singling out contrasting significant events and closely examining any changes that took place (a form of event sampling). Then, the development and use of a coding strategy for verbal productions during a deprogramming is discussed. I then introduce the Experiencing Scale (a rating scale for assessing subjective experience), and I present my rationale for using this instrument in my study. Finally, I discuss my design, statistics, and results.


Prior to Ken's deprogramming, my experiences suggested several ways of tackling this topic. My most pressing problem was how to balance breadth with depth. With little or no research data on what happens during a deprogramming, how could I design a meaningful experimental or comparative study? My survey of the literature did not find a set of antecedent data to guide me in choosing a narrow and well-defined set of hypotheses to test.

Before attempting to uncover "universal" or generalizable principles that govern deprogramming, it seemed necessary to be able to specify the forces that determined the process of a single instance of deprogramming. Deprogramming seemed to meet Lewin's (1935) criteria for an unusual and uncommon event. Allport (1942) suggested that "if a hypothesis deals with a rare and unusual occurrence" (p. 147), then intense case study is the most appropriate means of initially studying it. Sefltiz, Wrightsman and Cook (1976) seemed to agree when they counseled that 'where there is little experience to serve as a guide, [we] have found the intensive study of selected examples to be a particularly fruitful method for stimulating insights and suggesting hypotheses for [subsequent) research" (pp. 97-98).

I chose as my conceptual framework for this study the examination of conversation from a cognitive/rational viewpoint I developed a method of encoding verbal behavior in a manner that focused on conversation within a “comprehensive typology of persuasive message strategies" (Burgoon and Bettinghaus, 1980) broad enough to allow the examination of several dimensions of information exchange and their possible fluctuations over the course of the deprogramming. I was also interested in comparing these dimensions of information exchange to hypothesized changes over the course of the deprogramming in two other areas: attentional motility (a term I employed to signify a proposed measure of the degree of attentional rigidity) and experiencing (a research construct related to the quality of

information processing and depth of achieved insight during counseling sessions).

The potential pool of data in this study threatened to become unwieldy unless I could arrive at a method for closely examining segments of the deprogramming. Rice and Greenberg (1984) provided a rationale for exploring hypothesized changes in process as a function of specific events or phases of the deprogramming. In considering research on the process of change in psychotherapy, they concluded that change does not occur at a steady rate throughout all sessions, or even within one session. Rather, successful therapeutic process consisted of 'marker' events (related to client change) that alternated with periods of relatively little or no change. To understand the process of change, researchers can focus on events just prior to, including, and following a targeted change in the client's behavior and/or experience. This paradigm seemed eminently consistent with the recommendations made by Selltiz, Wrightsman, and Cook.

The qualitative description of Ken's deprogramming (see Cultic Studies Journal, Vol. 6, No. 2) helped facilitate my analyses of quantitative data I collected to account for affective, interpersonal, and other influences. My notes, along with the audiotapes, became a "personal document' of the deprogramming (Allport, 1942): a "self-revealing record ... [yielding] information regarding the structure, dynamics, and functioning” (p. xii) of the deconversion process. They permitted a detailed reconstruction of the deprogramming, which developed into the behavioral observations section of my study.

Thus, basic research design questions were resolved when I decided to: (a) conduct a naturalistic rather than a cornparative study, (b) concentrate on the cognitive/rational rather than the affective and interpersonal aspects of deprogramming, (c) analyze data collected prior, during, and following a marker change event, and (d) maintain a journal of my observations and subjective impressions to describe the context in which data were collected.


Final Design

To concentrate on cognitive phenomena, I limited my data collection to verbal output. This focus was consistent with my conceptual framework of deprogramming as a specialized form of conversation. In a single case study, a modified cross-lagged panel design was employed in which statements made by the deprogrammers and the cultist were coded and treated as variables (Campbell and Stanley, 1963). The independent variables were, Time (defined as “stage of deprogramming”) and Speaker (cultist vs. deprogrammers). Predictors were Deprogramming Statements Checklist (DSC) codes obtained during the initial stage of the deprogramming. The predicted variables were obtained during the “snap” and at the end of the deprogramming. They consisted of. Deprogramming Statements Checklist (DSC) codes (including a computed speech/thought productivity index); computed scores on the Experiencing Scale (cultist only); tallied focus of attention codes; and a computed attentional motility index (tallied shifts in attention). Figure I is a graphic depiction of this design (see Fig. 1).

I chose the moment in which Ken renounced his allegiance to ISKCON, dramatically symbolized by his request to have his sika (the ISKCON devotee's "ponytail”) cut off, as the point of most pronounced “snapping.” Choosing this symbolic act as the snapping point followed Rice and Greenberg's (1984) recommendations by providing for maximum contrast between the three phases (beginning, snap, and end) of the deprogramming.


The participants in this study consisted of a member of the International Society for Krishna Consciousness (ISKCON, the “Hare Krishnas”), Ken Butler, and a deprogramming team that consisted of five deprogrammers (in order of deprogramming team "rank'): Curt Miller, Sandy, Greg Stem, Dara and occasionally Brian. The participants' names are pseudonyms.


In this study, the actual “subjects” were not the individual participants involved in the deprogramming, but rather their verbal products (complete thoughts, or statements, expressed during the deprogramming). The rationale behind my method was initially suggested by Allport (1942, p. 36) in his review of a study of one individual's personality as revealed by personal documents. Deprogrammers' statements or thoughts were treated as one unit. Thus, although the transcripts identify individual deprogrammers, the statements analyzed belonged to one of only two groups: C, Cultist or DP, Deprogrammer.

I accumulated 37.5 hours (2,250 minutes) of taped conversation, out of which a total of 85 minutes of conversation were transcribed, which yielded a total of 1,938 coded thoughts (1,298 for the deprogrammers, and 640 for the cultist). Extrapolating from the above figures, I estimated the size of this study's universe of audiotaped thoughts to be 51,300. My sample of 1,938 thoughts therefore represented 3.8% of did estimated universe.

Figure I

Research design

Participant                        Stage of Deprogramming    

                                Beginning                        Main Snap                End

                                (Day 1)                                         (Day 2)                        (Day 5)

Conversational Style*

Agree, Disagree, Consider

Hypo., Disagree, Gives Info.,

Orients, Questions, Self

Discloses, Cites Doc., Productivity.

EXP Scale score*

Focus of Attention*


Conversational Style**

Agree, Disagree, Con-

Sider Hypo., Disagree,

Gives Info., Orients,

Questions, Self-

Discloses, Cites Doc.,


EXP Scale score **+

Focus of Attention**+




Agree, Disagree,

Consider Hypo.,

Disagree, Gives

Info., Orients,

Questions, Self-

Discloses, Cites Doc., Productivity.

EXP Scale score *

Focus of Attention*


Conversational Style*

Agree, Disagree, Consider

Hypo., Disagree, Gives Info.,

Orients, Questions, Self-

Discloses, Cites Doc.,


Focus of Attention*


Conversational Style**+

Agree, Disagree,

Consider Hypo., Dis-

Agree, Gives Info.,

Orients, Questions,

Self Discloses, Cites

Doc., Productivity

Focus of Attention**+


Conversational Style+

Agree, Disagree,

Consider Hypo.,

Disagree, Gives Info,

Orients, Questions,

Self Discloses, Cites

Doc., Productivity

Focus of Attention+






The data for this study were collected between April 17 and April 21, 1984. My two principal contacts in this deprogramming were Curt Miller, a full-time deprogrammer, and Risa Butler (pseudonym), a middle-aged woman whose son, Ken, would be one of my subjects.

Oral permission from the cultist to audiotape was initially obtained by Curt prior to turning on the tape recorder, the cultist gave subsequent written permission to use the tapes for this study later that day (April 17, 1984). The deprogrammers also signed release forms, giving permission to use their statements in my study. The deprogramming began at about 6:00 PM the first day, and ended shortly after 3:00 PM the fifth day.

Data Sources

Ken's deprogramming took place over a 5-day period that covered approximately 95 total hours. Taking into account approximately 36 hours of sleep and personal care time, and approximately 14 hours of video and “private” time, I conservatively estimated that I was able to capture on tape about 83.3% of the "actual” deprogramming.

Following the deprogramming, I transcribed verbatim 20- to 25-minute segments from 4 tapes. Two segments were from two tapes that included the period of most pronounced “snapping.” The remaining two segments included the very beginning of the deprogramming and the near-end of the deprogramming. Thus, the four segments represented three distinct time periods in the deprogramming, drawn for maximum contrast: beginning, snap, and end. After transcribing the selected deprogramming segments, the Deprogramming Statements Checklist (DSC) and related procedures (e.g., a reasonable method for breaking the transcripts into “units” for coding) were refined. Once acceptable reliability was obtained, coders were trained in the use of the DSC, with separate raters chosen later to be trained in the

use of the Experiencing Scale.


Two instruments were used to assess the deprogramming process. Two of the three primary dependent variables, the cultist's level of experiencing and focus of attention, were measured by the Experiencing Scale (Klein, Mathieu, Gendlin, & Kiesler, 1969) and the FoA (Focus of Attention) portion of the DSC, respectively. The cultists pattern of conversation made up the third set of dependent variables. This set of data was obtained computing the proportion of statements within 5-minute blocks that fell into each of the

eight DSC categories; these percentages then became each DSC category's score for a particular time segment. The same procedure was conducted utilizing deprogrammers' DSC category scores, which became the predictors.

To facilitate the coding process, raw transcribed speech was initially divided into “speaking turns." In the Deprogramming Statements Checklist (DSC) manual (Dubrow-Eichel, 1987), a speaking turn is defined as a completed "interchange between one or more deprogrammers and the deprogrammee (the cultist).” Speaking turns were further subdivided into two units: the “speaker within a speaking turn” (SWST) refers to “all statements made by one individual speaker in any given speaking turn." Following a procedure and rationale recommended by Lennard and Bernstein (1969), the 'statement coding unit' (SCU) consisted of the speaker's statement divided into its most basic unit: complete thoughts within that statement. Two coders who independently divided 100 randomly selected statements into SCUs achieved a 93% agreement rate on the number and division of SCUs within the SWSTS.

Deprogramming Statements Checklist

To categon= verbalizations that occurred in deprogrammings, I developed the Deprogramming Statements Checklist. In addition to incorporating my own theory of how cultists change dining deprogrammings, the DSC is based in part on Bales's (1951, 1970) studies of verbal interaction processes in a variety of settings and situations, including counseling sessions, preschooler play groups, married couples and academic discussion groups. I also attempted to incorporate the criteria for verbal coding systems suggested by Allen and Guy (1974). They advised that coding categories be logically related and drawn from broad aspects of conversation, including information exchange, affect projection, reciprocity and social energy. The DSC categories "Question," "Gives Information,” and "Self-Discloses” appeared to assess information exchange. The “Agree” and "Disagree” categories seemed congruent with affect projection. Social energy and reciprocity were examined in my study by quantifying “speech productivity," the proportion of the total conversation attributed to deprogrammers vs. the cultist.

DSC coding categories. A primary goal of this study was to accurately describe the rate, quantity, quality, and nature of information exchange, and shifts in focus of attention as these influenced the cultist's level of experiencing. I was also interested in focus of attention and attentional motility, and oven attempts to direct (orient) behavior. The DSC is designed to code statements into one of eight categories based on the statement's content and expressed purpose or intent. To increase reliability and coding accuracy, most category definitions incorporated or were based on general rules of grammar. The DSC code categories consist of eight main codes, a secondary code related to citing doctrine, and a focus of attention code. The focus of attention codes were also used to calculate a "motility index."

Primary Codes

Agrees. The speaker expresses agreement with a previous statement. Coded “CA” (Cultist Agrees), "DA” (Deprogrammer Agrees), or “OA” (Observer Agrees).

Consider Hypothetically. The speaker requests that the target (of his or her statement) think into the future, consider a possibility, or entertain a hypothesis. Coded “CCH,” "DCH” or “OCH.”

Disagrees. The speaker expresses disagreement with a previous statement. Coded "CD,” “DA," or “OD."

Gives Information. The speaker expresses an impersonal and verifiable fact, explanation, or interpretation. Coded “CGI," “DGI,” or “OGI."

Orients. The speaker directs, or seeks to directly guide or overtly influence, the target's actions, behaviors, or thoughts. Coded “CO," “DO,” or "OO.”

Questions (Asks for Information). ne speaker requests that the target provide information about him/herself or others, or about the cult, cult doctrine, people, beliefs, news events, the physical world, etc. Coded “CQ,” "DQ,” or "O-Q.”

Self-Discloses. The speaker clearly relates personal experiences and/or family information about self. Relates beliefs or thoughts, or indicates what he/she has done, is doing, or will be doing in the future. Coded "CSD,” “DSD,” or “OSD.”

Uncoded. All statements or remarks not clearly failing into one of the above categories; comments and/or remarks that do not constitute complete or coherent thoughts. Coded “CU," "DU," or “OU."

Secondary Codes

Cites Doctrine. In the speaker's statement (usually while giving information or disagreeing), he or she cites cult doctrine. Coded "CCD,” or "OCD.”

Focus of Attention. The speaker appears focused on cult-related material, on noncult-related material, or seems neutral (partly on-, partly off- cult), or the coder is unsure. Coded “-I" (on-cult), "+I" (off-cult) or “O" (neutral).

Motility Index

Motility. This score is a percentage obtained by counting the number of shifts in focus of attention relative to the total number of statements receiving an FoA code in each 5-minute transcript segment. The proportion of shifts then becomes the computed motility index.

I divided each of the 20- to 25-minute transcripts into 5-minute time periods. The frequency of each code's occurrence within a 5-minute time Period was transformed into a percentage of the total of all codes within that same time period, which yielded scores that could range from 0 to 100. The use of percentages allowed me to compensate for fluctuations in the number of SCUs that occurred within each time period.

DSC validity and congruence. DSC validity was initially addressed by adapting previously validated coding systems with similar properties (Allen and Guy, 1974) and categories (Bales, 1951, 1970) to what seemed salient in two deprogrammings I observed prior to this study. Subsequent major revisions of the DSC incorporated comments and suggestions from two experienced exit counselors, Kevin Garvey and Joseph Flanagan. To establish the DSC as a reliable tool for categorizing deprogramming statements, I initially employed two independent coders to pilot the DSC on tape transcripts from another deprogramming. Agreement rates were then computed. Using different coders, I repeated this process with subsequent versions of the DSC until it seemed that agreement rates were acceptable. Finally, transcripts from Ken's deprogramming were coded by the same pair of coders. Rates of agreement for individual codes ranged from 60% to 98%; 92% was the average rate for all codes. Reliability was then assessed using coefficient kappa (Cohen, 1960) on all three DSC coding arm: the eight independent (main) categories, the Cites Doctrine (CD) category (which was scored separately from the eight independent categories), and the Focus of Attention (FoA) category. The results are displayed in Table 1; all obtained z's were significant beyond the .01 level (see Table 1).

Table 1

DSC rater congruence

                                                                Possibility of Agreement


DSC Category                Time Period                Obtained                Expected                kappa

All                        Overall                        .94                        .37                        .91

                        Beginning                .95                        .38                        .92

                        Main Snap                .92                        .40                        .87

                        End                        .92                        .33                        .88

Focus of Attention        Overall                        .85                        .38                        .76

                        Beginning                .85                        .38                        .77

                        Main Snap                .88                        .48                        .77

                        End                        .80                        .62                        .47

Cites Doctrine                Overall                        .99                        .87                        .91

                        Beginning                .99                        .92                        .88

                        Main Snap                .99                        .83                        .94

                        End                        .97                        .69                        .90

Agrees                        Overall                        .99                        .90                        .94

Disagrees                Overall                        .99                        .98                        .93

Cons. Hypoth.                Overall                        .98                        .94                        .74

Gives Info                Overall                        .95                        .51                        .90

Orients                        Overall                        .99                        .96                        .80

Questions                Overall                        .99                        .83                        .97

Self-Discloses                Overall                        .97                        .75                        .88

Uncoded                Overall                        .99                        .87                        .96

 Note:  all kappas were significant at p.<.01 (using z-transformation).

The Experiencing (EXP) Scale

The Experiencing (EXP) Scale (Klein, Mathieu, Gendlin, and Kiesler, 1969) was developed as a tool to assess focused insight. According to Klein and his colleagues, experiencing "can be interpreted as tapping the degree of depth of the patient's involvement in the therapeutic task, including his [self- awareness and] openness to the therapist's experiential approach" (P. 26). Theoretically, the EXP Scale seemed based on an information/attentional model of healthy psychological functioning. As a skin, the ability to focus on current events in one's experiential field is the central component of experiencing, as it provides “the means by which painful feelings can be directly faced and changed” (Klein, et. al., 1969, p. 9). The EXP Scale is constructed to include seven “stages” of experiencing, with a range of scores that reflect these stages (i.e., I through 7); Stage 7 denotes the qualitatively highest level of experiencing.

EXP Scale validity. Construct and predictive validation of the EXP Scale has been undertaken within the context of psychotherapy process and outcome studies. FEgh EXP scores have been related to (1) tension relief when the client feels personally involved in therapy, (2) psychotherapy progress stability, and (3) “psychological mindedness,” or insight (Gendlin, E., Beebe, J., Cassens, J., Klein, M., & Oberfander, M., 1968). In a factor analytic study of the EXP Scale, Mintz (1969) and Mintz, Luborsky, and

Auerbach (1969) found experiencing to be related to client receptivity; EXP Scale scores loaded on a factor that was labeled "Interpretive Therapy with a Receptive Patient." In addition, experiencing level has been demonstrated to be related to “the capacity ... to let go of existing constructs” (Klein, et. al., 1969, p. 11). These findings suggest that experiencing level may be related to my hypothesis that deprogramming involves increased receptivity to information input.

EXP Scale reliability. Klein, et. al. (1969, p. 45), summarize seven studies reporting EXP Scale reliability coefficients (using four judges) that ranged from .44 to .91 (average of .66) for EXP Mode and from .43 to 92 (average of .71) for EXP Peaks. Kiesler (1969) reported reliability coefficients of .71, .79, .91, and .94 (Mode), and .73, .79, .91, and .92 (peak) for all judges. The judges in my study achieved excellent EXP Scale agreement rates, which resulted in interrater reliability coefficients of .77 and .95 for Mode and Peak respectively, and a reliability coefficient of .83 across all ratings. A t-test of the difference between the two raters' mean EXP Scale scores was insignificant.

 Use of the EXP Scale in my study. In assessing therapy sessions, the EXP Scale's authors recommend that, for optimal analysis, tapes and transcripts be divided into segments of 5- to 8-minutes duration. Dividing my transcripts into 5 minute segments facilitated analysis using the EXP Scale as well as my own DSC. Scoring statements within each segment yielded two indices of experiencing, both with a range of I (low) to 7 (high) on a Likert-like scale. The "Mode” score is that EXP Scale rating (level) within any segment that is most often scored; the “Peak” score is the highest EXP Scale rating (level) that occurred within any segment. In proposing these two indices, the authors suggested both ratings were necessary, the former to gain a sense of the general experiential “atmosphere” of the segment, the latter as an indicator of the segment's "best moment(s).”

Data Analysis

For the first step in my data analysis, I divided the three time samples into five-minute "Blocks.” The first coded transcript lasted 20 minutes; the second, 45 minutes; and the third, 20 minutes. This yielded time Blocks 1 through 4 (4 blocks total), 5 through 13 (9 blocks total), and 14 through 17 (4 blocks total), for the Beginning, Snap, and End of the deprogramming, respectively (17 blocks total overall). I then calculated frequencies of codes for the deprogrammers and the cultist on each five-minute time block within each of the three time segments. By utilizing proportions rather than raw scores, I was able to account for varying rates of thought and speech productivity during the various time periods. The deprogrammers Contingency tables used proportions rather than raw scores. The Yates correction for continuity (Guilford &- Fruchter, 1973, pp. 204-205) was used whenever the expected frequency of a particular code within a cell (time period) was below 10.

Shifts over time in DSC code occurrences were then investigated by analyzing the level of change in the chi-square distribution of each code as a proportion of each subject's overall thoughts (as expressed in speech) during one time period. Separate analyses were carried out for both the cultist and the deprogrammers. Contingency tables used proportions rather than raw scores. The Yates correction for continuity (Guilford & Fruchter, 1973, pp. 204-205) was used whenever the expected frequency of a particular code within a cell (time period) was below 10.

In my study, shifts in Experiencing Scale scores were also investigated using the chi-square statistic. For the purpose of analysis, Experiencing (EXP) Scale scores were grouped into two categories, reflecting “Low” experiencing levels (scores of I or 2) and “High” levels (scores of 3 and above). The chi-square distributions of EXP Scale scores across time were analyzed to determine whether or not shifts in experiencing level had in fact occurred.

DSC codes were then correlated with Experiencing Scale peak and mode scores; an additional correlation on the straight mew of each time segments EXP Scale scores was also performed. Two additional series of correlations were computed, one utilizing Focus of Attention codes and the other using the computed "attentional motility” index.

Statistical analyses were performed using “StatView 512K Plus" (Abacus Concepts, 1986), a statistical analysis package for the Apple Macintosh series of personal computers with at least 512 kilobytes of random access memory.

Productivity and Level of Participation

During the earliest stage of the deprogramming, the conversation was dominated by the deprogrammers, as indicated by the proportion of all coded thoughts (complete thoughts, with at least implied subject, predicate, and verb) attributed to the deprogrammers. The conversation became more equal during the later time periods. The number of codes obtained for each time period is presented in Table 2, along with the proportion of the total conversation attributed to deprogrammers and the cultist (see Table 2).

Ken's level of participation in the deprogramming increased significantly after the first day. Changes in these proportions over time were evaluated using chi square, with significant changes indicated in Table 2. As expected, he became more actively involved in the deprogramming as time went on, with significant increases in thought productivity (ideational activity) prior to and during the snapping period as well as for the deprogramming as a whole.

The proportion of conversation can be interpreted as an index of each speaker's intellectual involvement in the deprogramming. Figure 2 graphically illustrates one aspect of rapport: the "flow” or “rhythm” of conversation. During the middle stage (the Snap), the conversation became more equalized (see Fig. 2). At the snapping point (Blocks 7 and 8) Ken's productivity became greater than that of the deprogrammers. The last phase of the deprogramming was marked by conversation that tended to “flow" at equivalent rates; in a sense, the speakers had begun to mirror each other's verbal/ideational output.

Table 2

Participation in deprogramming

                                        Number of Coded Thoughts                                        

Participant                Beginning: Day 1        Main Snap: Day 2                End: Day 5        Total

Cultist                                75                        395                        70                640

          Proportion:                18%                        39%++                        34%+*                30%a

Deprogrammers                        338                        630                        330                1298

          Proportion:                82%                        61%                        66%                70%

Note: Each “thought” (i.e., “Statement Coding Unit”) consists of a subject, predicate and verb (these parts of speech may be implied) that expresses one complete thought.

a Average across all time periods.                  *p<.05 overall.

+ p<.05 from Beginning to End.                          ++ p<.05 from Beginning to Snap.

+++ p<.05 from Snap to End

DSC Coding of Deprogramming Statements

“Gives Information” (GI) was the DSC coding category used most often for the deprogrammers' statements. During the three time periods sampled (Beginning, Snap, and End), close to or over 50% of their statements received this code. However, the frequency of this score dropped from 71% to 49%, while the Self-Discloses code rose proportionally, from 6% to 33%. For the cultist, Questioning was the category scored most during the Beginning, but Gives Information was the code scored most often for the other two time periods. Table 3 presents the number and proportion of each speaker’s thoughts coded in each DSC category for all three time periods (see Table 3).

Figure 2

Speech Productivity:   Deprogrammers vs. cultist

(Graph not available)

Agrees. Ken, the cultist, tended to agree more with the deprogrammers as the deprogramming progressed, although the change in his level of agreement achieved statistical significance only over the deprogramming as a whole. This trend was expected, and seemed valid upon reviewing all the deprogramming tapes. There was no change in the deprogrammers' agreement levels.

Table 3

Changes in DSC codes over time

                                n                                        proportion                        

DSC Code        Beginning        Snap                End                Beginning        Snap                End

(Day 1)                (Day 2)                (Day 5)                (Day 1)                (Day 2)                (Day 5)


  Dpers                11                70                23                7%                11%                7%

  Cultist                  1                20                16                1                  5                9+

Cites Doc.

  Dpers                14                33                16                4%                5%                5%

  Cultist                  4                54                36                5                14                21+*

Cons. Hypo.

  Dpers                  8                12                13                2%                2%                4%

  Cultist                  1                12                  8                1                3                5


  Dpers                  5                  8                  1                1%                1%                 -

  Cultist                  1                  2                  1                1                1                 1

Gives Info

  Dpers                239                362                161                71%                57%                49%

  Cultist                  16                187                  98                21                47++                58*


  Dpers                    4                  37                    8                  1%                 6%                  2%

  Cultist                    0                  15                    4                   -                 4                  2


  Dpers                  23                  48                  15                7%                8%                  5%

  Cultist                  24                  42                  15                7%                8%                  5%


  Dpers                  20                  77                108                6%                12%+++        33%+*

  Cultist                  22                111                  26                29                28                15+


  Dpers                338                630                330                100%                100%                101%a

  Cultist                  75                395                170                  98                101                100

Note: Cites Doctrine was scored in addition to the main 8 codes; therefore, it is not included in lows.

a because of rounding off, totals may not always equal 100%.

*p < .05 overall.                                              +p < .05 from Beginning to End.

++p < .05 from Beginning to Snap.                     +++p < .05 from Snap to End.

Cites Doctrine. I expected the deprogrammers to cite cult doctrine (to debunk it) about as often during the first stages of the deprogramming as they would toward the end of it; this is in fact what happened. Ken, however, significantly increased his doctrine citing over the entire deprogramming, especially during the period between the beginning and the snap. I expected the opposite, that his doctrine citing would decrease. Qualitative analysis of the tapes did indicate increased doctrine citing throughout most of the deprogramming, although not in the context I had anticipated. Ken's increased doctrine citing occurred as he attempted to analyze and question it; at times, he seemed in essence to be "inviting” the deprogrammers to debunk it.

Considers Hypothetically. Based on my theory of deconversion, I expected there would in general be more hypothetical conjecturing than actually occurred, and I predicted that the deprogrammers would “consider hypothetically" at an even rate. In fact, the CH category was rarely scored, and there were no significant changes in its occurrence for any of the speakers.

Disagrees. I expected disagreement rates to decline as the deprogrammers and cultist began to share more and more common ground as far as the latter's thinking on his cult involvement was concerned. Interestingly, although both cultist's and deprogrammers' agreement rates increased, the disagreement

code was rarely scored and did not significantly vary in rate.

Gives Information. I expected the deprogrammers to maintain a fairly constant level of “objective'” information-transmitting, while I predicted that the cultist's GI rate would increase as he became more active in the information-exchanging process. In fact, the amount of "objective” (i.e., impersonal) information-giving by the deprogrammers decreased gradually throughout the deprogramming; qualitatively, it was after the snapping period that the bulk of this decline appeared to occur (although the

measured decline from the “snap” to the “end" was not statistically significant). As the deprogrammers' GI rate declined, their SD (Self-Disclosure) rate tended to increase, which suggests a possible change in the subjective/objective quality of the information being given, rather than a decline in information-giving in general. As expected, Ken tended to increase the amount of impersonal information-giving he engaged in, especially in the period prior to and during the snap.

Self-Discloses. In part, the increase in Ken's GI rate seemed to occur concurrently with a decrease in his self-disclosure rate. I had predicted the overall increase in Ken's GI level, although I expected the bulk of this increase to occur following the snap rather than before it. Ken's declining self-disclosure rate toward the deprogramming's end was unexpected.

Orients. After being relatively nondirective, I expected the deprogrammers to make increasingly more orienting statements just prior to the snap. In fact, this category was rarely coded, and there were no changes in orienting rates either prior or subsequent to the snap, or from beginning to end. Ken's orienting did not change over time, contrary to my prediction that he would be fairly directive at first and would then decrease his level of directiveness prior to or during the snap.

Questions. Questioning, the overt act of requesting information (both impersonal facts and self-disclosure), was expected to increase over time for Ken, with the bulk of the increase expected to occur following the snap. In contrast, I expected the deprogrammers to increase their requests for information from Ken immediately prior to the snap, and for that rate to remain fairly unchanged subsequent to the snap. What I in fact obtained was contrary to these predictions. Ken's overt questioning decreased from

beginning to end, with the snap being the period of least questioning on his part. The deprogrammers' questioning rates remained fairly constant throughout the three time periods. Moreover, one speaker requesting information did not in general appear related to the other speaker giving


Shifts in Focus of Attention and Motility

Throughout the deprogramming, ft degree to which ft deprogrammers were focused on cult-related material hovered within 10 points of the 50% range. The degree of on-cult focus in the cultist's conversation, however, changed from below 45% in the beginning to above 60% in the later two time periods. The frequency with which the deprogrammers' focus of attention changed (motility) remained fairly constant from the beginning to the snapping periods, and then decreased during the third and last time period. The cultists attentional motility index, however, steadily decreased until it was halved by the end of the deprogramming (see Table 4).

Focus of Attention. Ken's attention did become increasingly focused on his cult just prior to and during his snap, but instead of decreasing after this event, it remained relatively focused on cult-related material. Thus, Ken ended the deprogramming more focused on cult-related material than when he began. Closer analysis of the deprogramming tapes suggested that the quality if not the direction of Ken's on-cult focus had changed however, following the snap, Ken did in fact demonstrate increased attention to cult material

and experiences, but his focus had a skeptical and questioning quality. For Ken, the increase in on-cult focus may have occurred at the expense of a “neutral” focus.

Motility. While the motility rates for both deprogrammers and cultist did not decline prior to or during the snap, they did decline significantly for the deprogrammers following the snap, and they declined in general for the cultist. (The decline was not statistically significant for any one time period.) At the end of the deprogramming, motility rates for all speakers were at significantly lower levels than they were at the beginning of the deprogramming.

Table 4

Focus of attention and motility over time

                                                        Proportion (%) of Coded Thoughts

Variable                Participant                        Beginning        Snap                End

                                                        (Day 1)                (Day 2)                (Day 5)

Focus of Attention

  Off-Cult                Cultist                                40%                27%                30%

                        Deprogrammers                        26                38                45+

  Neutral                Cultist                                18%                11%                  5%

                          Deprogrammers                        16                15+++                  3+*

  On-Cult                Cultist                                41%                62%++                65%+

                        Deprogrammers                        58                47                52

  Motility Index                Cultist                                45%                35%                21%+*

                        Deprogrammers                        39                32+++                  8+*

Note: Because of rounding off, totals may not equal 100%.

a) Motility Index is the number (proportion) of shifts in Focus of Attention

(FoA) relative to all FoA codes in a given time period.

*p < .05 overall.                                                    +p < .05 from Beginning to End.

++p < .05 from Beginning to Snap.                      +++p < .05 from Snap to End.

Shifts in Experiencing

Ken's level of subjective involvement in, and integration of, the deprogramming experience, as measured by the EXP Scale indices, generally increased between the beginning and the middle of the deprogramming (see Table 5). Figure 3 illustrates how his highest EXP Scale Peak occurred immediately following the snapping point, when Ken decided to cut his sika (see Figure 3).

Figure 3

Ken's EXP Scale scores over time

(Chart unavailable)

Table 5

Ken's EXP Scale scores over time

                                           _                        Experiencing Scale Score:                

Time and Block                                Low                Peak                        Mode                Mean

Beginning Block:        1                1                2                        1                1.17

                        2                1                3                        1                1.35

                        3                1                2                        2                1.70

                        4                1                2                        2                1.75

                        Ave                --                2.25                        1.50                1.49

Snap block:                5                1                3                        2                1.62

                        6                1                4                        2                2.05

                        7                1                4                        2                1.65

                        8                1                4                        2                2.25

                        9                1                5                        3                3.00

                        10                1                4                        2                2.25

                        11                1                5                        3                2.22

                        12                1                2                        1                1.10

                        13                1                4                        1                1.50

                        Ave                -                .89                        2.00                1.96

End Block:                14                1                3                        1                1.65

                        15                1                2                        1                1.40

                        16                1                2                        1                1.60

                        17                1                2                        2                1.60

                        Ave                -                2.25                        1.25                1.56

Note:  Ken decided to cut his sika during Block 8.

Analysis of chi-square distributions of EXP Scale scores. To test the hypothesis that as the deprogramming progressed the cultist would achieve higher Experiencing (EXP) Scale scores, I divided the obtained scores into two categories, Low (scores of “1” and "2') and High (scores "3" and higher). I then calculated the percentage of EXP Scale scores that fell into each category for all three time periods; Table 6 summarizes the subsequent analysis of chi-square distributions. The chi-square statistic for all categories across all time periods was significant beyond the .01 level; qualitatively, however, the changes in EXP Scale scores were bidirectional. Further analysis of the chi-square distributions for each category across the three time periods shows that, while the ratio of 'Low" EXP Scale scores did not change over time, the snapping period was marked by a significant increase in the proportion of EXP Scale scores that fell in the “High” category (see Table 6).

Correlations Between DSC Codes, EXP Scale Scores and Attention

Table 7 presents the obtained correlations between specific DSC codes, FoA codes, and Experiencing Scale scores (peak, mode, and mean).

On the deprogrammers' part, questioning, positing hypotheses, giving information, self-disclosing, and shifting focus of attention were not related to the quality of the cultist's experiencing, focus of attention, or

his attentional motility. The deprogrammers' focus of attention was related to the cultist's focus. Ken's motility was inversely related to the deprogrammers' focus on or off cult-related material. Only the deprogrammers' rate of agreeing with and orienting the cultist were positively related to the latter's EXP Scale scores; the former being related to the “atmosphere” or general level of Ken's experiencing, the latter being more related to his experiencing peaks.

As expected, the cultist's rate of self-disclosing was positively related to his general level of experiencing. Ken's rate of hypothesis generation, disagreeing, citing doctrine, and attentional motility were not related to his EXP Scale scores. None of the expected relationships between Ken's verbal behavior or productivity and his attentional motility were supported. Ken's on-cult focus of attention was positively related to his attentional motility.

Table 6

Changes in EXP Scale scores over time

Score Range                Beginning to Snap        Snap to End        Beginning to End                Overall


(Low & High)                Increase                Decrease                                        **

Low                           ---                        ---                        ---                        --        

High                        Increase                Decrease                ---                        **

Note: “Low” EXP scores are below 3; “High” scores are 3 and above.

**p <.01 (shift is bi-directional)

Table 7

Relationships between DSC coces, EXP Scale scores, and attention


                                        EXP Scale                Ken’s Focus of Attention

DSC Code        Peak        Mode        Mean        Off-Cult                On-Cult                Ken’s



  Cultist                                        .12        .34        .37        -.10                .15                -.10

  Deprogram                                .51*        .36        .42        -.28                .26                .12

Cites Doctrine

  Cultist                                        .09        -.32        -.11        -.43                .47                .25

  Deprogram                                .06        .20        .21         .02                .12                    -.19

Considers Hypoth

  Cultist                                        .09        -.36        -.23        -.04                .16                    -.21

  Deprogram                                .01        .02        ,24         ,07                    -.11                .39


  Cultist                                        -.21        -.27        -.37        -.12                .04                .24

  Deprogram                                -.04        -.17        -.17        -.11                .06                .16

Give Information

  Cultist                                        .05        -.07        -.03        -.21                .26                .01

  Deprogram                                -.40        -.21        -.31        .41                -.40                .02


  Cultist                                        .13        .02        .15        -.07                .04                -.14

  Deprogram                                .54*        .49*        .67**        -.03                -.03                .15


  Cultist                                        -.29        -.27        -.38        .41                -.31                -.21

  Deprogram                                .45        .13        .04        -.12                .11                .01


  Cultist                                        .38        .52*        .55*        -.25                .11                .10

  Deprogram                                -.09        -.10        -.03        -.27                .32                -.26


  Cultist                                        .41        .11        .10        -.42                .46                -.09

  Deprogram                                -.41        -.11        -.10        .42                -.46                .09


  Cultist                                        -.01        -.03        .13        .30                .55*                .—

  Deprogram                                .14        .14        -.06        -.13                .01                .43

Focus (Off-Cult)                                

  Cultist                                        -.23        -.14        -.01        .--                -.93                .30

  Deprogram                                .10        -.19        .05        .72*                -.76**                .46

Focus (On-Cult)                

  Cultist                                        .19        .11        -.03        -.93**                .--                .55*

  Deprogram                                .05        .18        -.02        -.64**                .75**                -.61*

*p < .05            ** p <.01


Analysis of Individual Conversational Styles

Table 8 presents data on individual speaker’s conversational styles. Of the 8 main DSC codes and I additional code (CD), 6 main codes and CD (7 total) were found to appear at least 5% of the time for at least one speaker during the deprogramming (see Table 8). Table 9 presents data comparing speaker styles. Using chi-square analysis, the occurrence of the four most frequently scored DSC categories was compared across all speakers. A significant chi statistic indicates that the speakers' DSC-code patterns differed significantly, and hence that their conversational styles were different (see Table 9).

Several speakers clustered together in terms of conversational style. Cluster I consisted of the cultist, Greg, and Brian; Cluster 2 also included Brian and Greg, and Dara (but not the cultist). Cluster 3 consisted of Dara, Greg, and Curt, while Cluster 4 included Greg, Curt, and Sandy. The only speaker whose conversational style was not significantly different from any other speakers was Greg.

Table 8

Conversational style:   Individual differences between speakers

                                                        Proportion of Coded Thoughts                        

DSC Code                Brian                Dara                Sandy                Ken        Curt                Greg

Agrees                        0%                8%                10%                6%        8%                5%

Cites Doc.a                1                3                6                12        6                1

Cons. Hypoth.                0                6                1                3        3                1

Disagrees                0                0                1                1        1                1

Gives Information        88                57                60                43        60                56

Orients                        0                1                4                2        6                4

Questions                0                5                7                17        7                15

Self-Discloses                0                24                8                25        10                16

Uncoded                12                1                2                3        5                2

Note: Because of rounding off, totals may not always equal 100%.

A cites Doctrine was scored in addition to the main 8 codes; therefore, it is not included in totals.

Table 9

Significant differences between speakers on DSC codes

                                         Chi-square on 4 most frequently coded DSC categories                

                                Dara                Sandy                Curt                Greg                Ken

Brian                                .20                8.81*                6.42*                2.28                .71

Dara                                                8.63*                5.65                7.20                8.73*

Sandy                                                                  .49                6.66                16.40*

Curt                                                                                4.73                13.47*

Greg                                                                                                3.98

Note: Codes used in contingency tables were: GI (Gives Info.), SD (Self-Discloses), Q (Questions) and A (Agrees), with df=3; analysis between Brian and other speakers is based on Gl and SD alone (df=l). *p<.05

Limitations of This Study

The nature of this study presented several problems. By studying only one deprogramming in depth, generalizability is of course severely limited. In addition, there were several uncontrolled variables or factors that may have interfered with my analysis of the obtained results. First, the initial "snatch” of Ken, the cultist, may have been such a significant “intervention" as to render the actual deprogramming itself relatively irrelevant; this factor was not directly assessed or measured in this study. The presence of varying numbers of deprogrammers, family members, the researcher and/or others, at varying times, compounded the study further, as these “comings and goings" may have been highly reactive. History and maturation are uncontrolled. For theoretical reasons (as opposed to reasons based on empirical research), I chose to gather data related to my subjects' cognitive and interactional processes, and I furthermore limited my data to verbalizations. I did not systematically attend to nonverbal or affect-related behavior, although certainly there were subtle (e.g., nonverbal) cognitive and social as well as emotional processes that may have affected the cultist's decision to exit ISKCON. Although I attempted to describe individual deprogrammer styles, in collapsing individual deprogrammers' statements into one data set I in effect attempted to create an artificial aggregate ('all' deprogrammers) that may have “smoothed over” meaningful individual variations. Individual deprogrammers and their statements may have had varying impact (both positive and negative) on Ken's decision to exit ISKCON; Ken may have given one deprogrammer's statements greater 'weight" than anothers (there is anecdotal evidence that statements made by deprogrammers who are also former members of the deprogrammee's cult are perceived as having increased valence, for example). I did not systematically account for this possibility.

My decision to concentrate on my subjects' overt verbal behavior, and then to examine closely only one dimension of that behavior, narrowed my hypotheses to a small subset of all possible hypotheses related to how and why this deprogramming was successful.


In Part I of this study (Dubrow-Eichel, 1989), 1 stated that:

The deprogramming of Ken Butler was primarily a cognitive and social-affiliative intervention within a specific form of extended, intensive conversation. There were some notable changes in content and group process over time. These changes and shifts, some of which were discernible only through my qualitative analysis of the deprogramming as a whole (see Chapter 111), suggested that Ken's deprogramming had different modalities as well as phases. The information conveyed and eventually accepted, along with the strong trust, rapport, and affiliative bonds developed between the deprogrammers and the cultist, combined to produce a dramatic change in Ken (p. 104).

In that article, I discussed my observations of the deprogramming process in terms of cognitive processes, including: information processing, persuasion, shifts in attention and consciousness, utilization of dissonance, and the role of suggestion. I also summarized how the social-affiliative process, group process, and therapeutic relationship factored in Ken's deprogramming.

In this section, I discuss the results of the quantitative analysis of the deprogramming as a conversation.

Summary of Conversation Analysis

For deprogrammers, the beginning stage was marked primarily by (in order of proportional representation): giving information, questioning, agreeing, and self-disclosing. For the cultist, the beginning was marked by self-disclosing, questioning, giving information, and citing doctrine.

For deprogrammers, the middle stage (the snap) was marked primarily by giving information, self-disclosing, agreeing, and questioning. For the cultist, the middle phase was marked by giving information, self-disclosing, citing doctrine, and questioning.

For the deprogrammers, the end stage was marked primarily by giving information, self-disclosing, agreeing, and citing doctrine. For the cultist, it was marked primarily by giving information, citing doctrine,

self-disclosing, and agreeing.

For all speakers during all stages, there was little disagreeing, little emphasis on overt consideration of alternative hypotheses, and little direct orienting. Over the course of the deprogramming, there were some significant changes in degree of participation in the deprogramming, in discourse (conversation) style, in focus of attention, and in motility. These, were the areas of interest addressed by my research questions. Table 10 presents a summary of these changes for the deprogrammers and Ken. Answers to each specific research question are discussed in the following section (see Table 10).

Answers to Research Questions

Question 1: To what extent, if any, were there differences by (a) stage of deprogramming and (b) cultist vs. deprogrammer in degree of participation in the deprogramming, as measured by the number of thoughts expressed (thought productivity)?

The cultist's level of participation increased from the Beginning to the Snap, and over the three time periods as a whole. He both produced more thoughts and accounted for a significantly greater proportion of the overall conversation over these time periods. This shift indicated an increment in Ken's active involvement in the deprogramming; it was also reflective of increased ideational activity (thinking).

The deprogramrners' level of participation remained fairly constant throughout the deprogramming. The deprograrnrners were fairly consistent in their level of involvement and ideational productivity.

Question 2: To what extent, if any, were there differences by (a) stage of deprogramming and (b) cultist vs. deprogrammer on DSC codes?

The cultist increased his level of information-giving and decreased his questioning from the Beginning to the Snap. Over all three time periods, he agreed more, cited doctrine more, and gave more information. Self-disclosing and questioning decreased from Beginning to End. Again, these shifts suggested increased ideational activity.

The deprogrammers increased their self-disclosing from the Snap to the End, and over all three time periods as a whole. Their level of information-giving decreased from Beginning to End.

Question 3: To what extent, if any, were there differences by (a) stage of deprogramming and (b) cultist vs. deprogrammer in focus of attention (on-cult and off-cult) and attentional motility?

The cultists attention to cult-related material increased from Beginning to the Snap, and over the deprogramming as a whole. His degree of off-cult focus did not change over time. His motility decreased overall during these three time periods. The increase in on-cult focus was related to Ken's increased critical thinking about ISKCON doctrine. The decreased motility was related to a shift in consciousness toward enhanced concentration.

The deprogrammers' attention to material unrelated to the cult increased over time. Like the cultist, their attentional motility decreased over the deprogramming as a whole, possibly in response to Ken's increased concentration (decreased motility). For all participants, the proportion of thoughts coded "neutral” decreased over time. Qualitatively, this decrease did not appear to be reflective of a true shift in attention; instead, it may have been due to increased coder accuracy in this DSC category over the course of the three transcripts.

Question 4: To what extent, if any, were there differences by stage of deprogramming in the cultists level of experiencing?

The cultists level of experiencing increased significantly from the Beginning to the Snap. His highest EXP Scale peak score was obtained immediately following the actual snapping event. Following the Snap, the cultists level of experiencing decreased. The 'snap'-Ken's moment of maximum decisiveness-was the point at which Ken was most focused on integrating his personal experiences. There was no significant difference in Ken's experiencing levels between the Beginning and End of the deprogramming.

Question 5: To what extent, if any, were the cultist's level of experience related to DSC codes, attention, and motility?

The deprogrammers' orienting thoughts were positively related to the cultists EXP Scale scores.  Their guiding comments tended to direct Ken's awareness inward, in an integrative manner. The deprogrammers' focus of attention was positively related to the cultist's focus of attention, which suggested that the deprogramming conversants were listening to and following each other's conversational threads (a sign of rapport).

Ken's level of self-disclosing was positively related to his EXP Scale scores. His self-disclosing was therefore an expression of his internal thought processes. (He was sharing his moments of enhanced self-awareness with his deprogrammers.) Ken's attentional motility was positively related to his on-cult focus of attention; i.e., when focused on cult-related material, his attention tended to shift more. Ken's concentration tended to be enhanced when he was not reviewing ISKCON-related material.

Question 6: How, if at all, did individual deprogrammers differ from each other and from the cultist in their conversational or deprogramming 'styles?"

Table 10

Changes over time in conversation, focus of attention and motility


Time Period                        Cultist                                        Deprogrammers

Begin to Snap

Increased: Productivity,

Giving Info., Attention


Decreased:  Questioning

Snap to End

Increased: Self-Disclosing

Decreased: Attention (neutral)

Begin to End


Increased: Agreeing, Citing

Doctrine, Giving Info., Attention

(on cult), Productivity.

Decreased: Self-Disclosing,

Questioning, Attention (neutral),


Increased: Self-disclosing,

Attention (off-cult)

Decreased: Giving info.,

Attention (neutral), Motility

All the participants in this deprogramming were more involved with exchanging information than any other aspect of conversation. Deprogrammers with similar conversational styles were Brian and Dara, Dara and Curt, and Curt and Sandy. Brian and the cultist, Ken, " demonstrated similar conversational styles.

Significantly different conventional styles were demonstrated between Brian and Curt, and Brian and Sandy. Dara also differed from Sandy, and from the cultist, Ken. Sandy's style was significantly different from Ken's, as was Curts. Ken's style differed significantly from Dara's, Sandy's, and Curt’s.

Ken's style was similar to Brian's and Greg's.

Greg's style was not dissimilar to any participant's style. Greg seemed to be a group "bridge" between deprogramming participants with varying conversational styles. Qualitatively, this finding seemed consistent with my observations of Greg's central role in Ken's deprogramming.

Conclusions and Implications

What is Deprogramming?

In Ken's case, deprogramming was characterized by three overlapping change processes: a counseling-like, therapeutic relationship (with social-affiliative, group, and affective components); a cognitive process

(with information- processing and consciousness components); and a communication process (with conversational or discourse factors).

Deprogramming as a therapeutic relationship.  Ken's deprogramming satisfied most of the criteria used to define “counseling.” In its directiveness and use of persuasion and suggestion, however, Ken's deprogramming also resembled clinical psychotherapy (especially as practiced in clinical crisis-intervention) as it has been traditionally defined. As Blackham (1977, p. 11) noted, the aim of clinical psychotherapy has traditionally been personality change, whereas counseling “tries to help clients make

choices.” Clinical psychotherapists generally work with severe behavior problems, whereas counselors work with clients who are .experiencing educational, vocational, situational, and development problems.' In counseling, the duration of treatment tends to be shorter. Ken's deprogramming resembled counseling in its emphasis on resolving a situational problem and making a choice (exiting vs. remaining in ISKCON),

and it was of short duration. Yet clearly Ken's 'problem' could be construed as a “severe behavior problem," although the deprogrammers' stated goal of reawakening Ken's precult personality did not constitute true “personality change."

The inauguration of Ken's deprogramming was contingent upon the deprogrammers’ establishing rapport with him; its ultimate success depended on their ability to maintain their empathic stance. As a distinctive form of counseling, Ken's deprogramming was also marked by cognitive learning, operant conditioning, modeling (identification), suggestion, persuasion, rehearsal, and repetition. Consistent with Zeig's (1987) report of trends in the evolution of counseling and psychotherapy, Ken's deprogramming utilized humor, and it emphasized mobilizing resources (i.e., critical thinking skills) rather than uncovering pathology. It was client-specific (i.e., tailored to meet the needs of the client), results-oriented, specialized, and used informed paraprofessionals.

As of this writing, most deprogrammers call themselves “exit counselors.” The engagement of 'exit counselors' who may have little or no formal graduate- or even college-level training (but who have themselves 're- covered' from destructive cultism, and typically learn their craft by serving as "apprentices” to more experienced exit counselors) seems analogous to the use of former drug and alcohol addicts as "certified alcoholism counselors' (CACs). In the case of CACs, many state certifying agencies have multitiered certification processes that accept supervised experience (i.e., apprenticeship) in lieu of formal training.

Deprogramming as "snapping:" Altering states of conscious- ness. As I pointed out in my earlier CSJ article, deprogrammers believe their clients were unduly influenced when they became involved with cults. They employ theories of "mind control" to justify their actions. With the possible exception of Lifton's (1961) formulations, mind control models are based in varying degrees on the belief that cult indoctrination induces altered states of consciousness analogous to hypnosis. My own experience, together with my conversations with deprogrammers and review of the literature on cults, supports this concept.

There was a significant and quantifiable change in Ken's attentional motility. Several years after proposing that the cultic state of consciousness involves decreased motility, I discovered anecdotal evidence that, at least in the case of the ISKCON devotee, Krishna (cultic) consciousness may actually involve increased attentional motility in the form of ideational disorganization, “frenzy,” and "flightiness.” (Although Hubner & Gruson, 1988, concentrated on ISKCON's illegal activities and tended to portray the Krishna leadership and many devotees as criminals, they did on occasion describe the "frenzied' mentality of those who took their Krishna practices seriously.) In addition, Singer (1979, p. 79), in describing the tendency of her group therapy clients to “float” into and out of a state of altered (cultic) consciousness, noted that .we often see members float off [and] have difficulty concentrating.“ During his deprogramming, Ken made one very telling remark about his state of consciousness: he referred to his Krishna thinking style as 'flitting." This statement was consistent with Singer's (1979) findings.  As the deprogramming proceeded, Ken became more focused, and his attentional motility decreased. The correlation between Ken's decreased motility and the deprogrammers' increased focus on cult-related material indicated that his enhanced concentration was positively associated with critical thinking about this material.

In addition, Ken's increased productivity and involvement in the deprogramming implied increased ideational activity (thinking) in general. On the other hand, Ken's assessed level of “experiencing'” increased during the snapping period, only to then decrease toward the end of the deprogramming.

Thus, my study has provided limited support for the hypothesis that Ken's deconversion involved an alteration of consciousness.

Deprogramming as discourse.  As a group process, Ken's deprogramming was clearly change- and task-oriented (while in the Formal Deprogramming mode). Since it required specialized roles and had a clear agenda involving persuasion and change, it was more than a conversation. Yet my content and process analysis of this deprogramming suggests that it allowed for considerable spontaneity, self-disclosure, and give-and-take on everyone's pan. Hence, it contained many of the elements typically associated with casual conversation.

I believe my study supports a conceptualization of deprogramming, at least in Ken's case, as a specialized form of discourse involving three characteristics:

  1. Deprogramming as persuasive conversation: As a group process, Ken's deprogramming had a clew and overt agenda: persuading Ken to reevaluate and forsake his membership in ISKCON. The deprogrammers promised Ken that, if he gave them enough time (one week) to present their case, he would be able to make an informed choice about his membership in ISKCON. I have known Curt to make and keep this promise in other cases, even when the cultist decided to return to the group. Ken's deprogrammers would have clearly interpreted a return to ISKCON as a “failed" deprogramming, however. While allowing some choice, the deprogrammers' bias was clearly stated to Ken: a rational reappraisal of cult involvement inevitably leads to deconversion.
  1. Deprogramming as teaching: The single most prevalent activity during this deprogramming was the imparting and discussion of three specialized content areas: facts about ISKCON, about destructive cults in general, and about "mind control.' Therefore, Ken's deprogramming was an educational activity in which the deprogrammers served as instructors and Ken as an informed student.
  1. Deprogramming as moral discourse: The nature of the information provided to Ken was heavily weighted, especially in the first two days of the deprogramming, toward: (a) debunking ISKCON philosophy by pointing out its inherent flaws and contradictions (even when compared with traditional Hinduism); (b) demonstrating that ISKCON leaders have intentionally distorted known fact and therefore invalidated Ken's "Search for Truth"; (c) providing clear and overwhelming evidence of wrong-doing and violation of self- proclaimed ethical standards; (d) demonstrating how ISKCON practices are anti-humanistic and therefore spiritually bankrupt; (e) providing a clear rationale for interpreting the behavior of ISKCON leadership as motivated by personal gain rather than altruism. Therefore, Ken's deprogramming was largely concerned with moral discourse.
Three-Dimensional Model of Deprogramming

These three processes were both independent and interdependent and can be viewed as comprising a three-dimensional model of deprogramming. Figure 3 graphically depicts this model.

Communicative dimension.  Ken's deprogramming could be characterized as a specialized form of persuasive conversation, with moral discourse, teaching, and education being the primary conversational activities. The contents of this conversation primarily involved exchanging information and self-disclosing.

Cognitive dimension. Ken's deprogramming involved a change in consciousness, marked by a shift in attentional motility and increased ideational activity, especially at the time of greatest decision-making (the "Snap"). The "Snap” was also characterized by enhanced self-awareness and increased focus on integration of personal experiences (experiencing). Overall, Ken's deprogramming involved considerable intellectual activity (information processing).

Social-affiliative dimension. Ken's deprogramming involved a counseling-like therapeutic relationship with his deprogrammers, based foremost on the establishment of rapport and an interpersonal process based on empathy. As a small-group process, this deprogramming had two distinct modes: a formal, cultist-focused mode and a casual, subgroup-focused mode.

Figure 4

A three-dimensional model of Ken's deprogramming

(Figure unavailable in this medium)

Issues and Implications for Future Research

Deprogramming, coercive conversion and psychological "doubling.” An ancillary goal of this study was to enhance our understanding of how otherwise intelligent, articulate, educated, and idealistic young people can become fanatically devoted to a movement that, according to its critics, is the very antithesis of the idealism and humanism that inspired its members to seek it out in the first place.

Deprogrammers believe their activities are the antithesis of cultic .programming' and hypnotic-like states of consciousness. My investigation of deprogramming has provided some support for the theory that deprogramming involves altering consciousness. There was clearly an increase in Ken's ideational activity, and a shift in his ability to concentrate. There also was a discernible snapping event, which on a qualitative level involved a moment of intense decision-making and heightened realization, and on a quantitative level involved enhanced integration of self-awareness (experiencing). Ken's spontaneous self-report during the deprogramming suggested that the snap was a 'reintegration" event, a dramatic point when the two halves of Ken--his Krishna and his non-Krishna selves--came together and were reunited. Yet Ken's shift in consciousness seemed subtle. My findings seemed in need of a theoretical framework that might explain what occurred during Ken's deconversion in a more satisfying and complete manner. This framework would also need to shed light on the process of cultic conversion.

Doubling  Lifton, 1986) may be the process that Ken experienced during and following his “conversion”; his deprogramming seemed to be an “undoing” of that doubling. Doubling helps explain the violent and antisocial behaviors of many ISKCON leaders (Hubner & Gruson, 1988), and it seems to explain the relative ease with which Ken, who had been a law-abiding student prior to becoming a devotee, could commit crimes involving fraud and theft, and condone even worse among his peers and gurus. Ken could commit fraud and "transcendental trickery' and his cult-self did not feel guilty because the nature of “right'”and “wrong" had been redefined. In fact, doubling explains the irony of Ken's guilty feelings when he felt like walking on a beach rather than carrying out his leader's order to engage in fraud during sankirtan.

Lifton's theory of doubling in large part subsumes his earlier theory of thought reform, and it dovetails with dissociation theories (including the neodissociation theory of hypnosis). In Ken's deprogramming, the barrier between the precult self and the cult 'double' seemed to be removed, and the deprogrammee began to feel .normal' guilt (his apology to his mother is an example of the resurgence of normal guilt). His conscience was reclaimed by his 'old' self. In this study, doubling also permits a reinterpretation of Ken's early denial of fraudulent activity (he later admitted to numerous "cons”). Was he in fact 'lying' when he first denied committing fraud? Or was he telling the truth, as the conscience of his double understood the meaning of the "truth"?

ISKCON clearly distinguishes between the world of the non-devotee (maya, the “world of illusion”) and the world of the devotee (Krishna consciousness); it also distinguishes between the spiritually advanced devotee and the spiritually bankrupt karmi. Moreover, ISKCON leaders have often explained to devotees that it is morally and spiritually acceptable to behave in less than .spiritual" ways in their dealings in maya, and with karmi. This 'splitting' of the world and people into two distinct, polar opposites facilitates the doubling process. Ken often referred to the split he felt when he dressed as a karmi (to perform sanhrtan) instead of as a devotee. He clearly viewed non-devotees as inferior beings.

The parallels between the factors that made Ken vulnerable to ISKCON indoctrination and those that made some doctors vulnerable to Nazi indoctrination are also striking. Doubling was the result of a Faustian bargain made by Nazi doctors, typically for reasons that are similar to those I have found in many cultists, including Ken: initial idealism; the desire to eradicate all “demons," or as Lifton (1986) termed it, the “vision of total cure” (p. 471); the 'quest for transcendence' (pp. 473-476) and a 'continuous 'high state" (p. 473); the manipulation of feelings of omnipotence and impotence (p. 447). These have particular relevance to factors that may predispose some individuals toward involvement in ISKCON. Ken made frequent references to all these reasons. When he first encountered ISKCON, he was an idealistic college student in search of “ultimate truth" and a cure for the alienation, spiritual turmoil, and postadolescent psychosexual conflict he was experiencing. Experimentation with hallucinogenic drugs had given him a “taste” of transcendence, which he then began seeking through intense spiritual experience. His goal was nothing less than the achieving of Krishna consciousness, a state most devotees claim is like being "on a perpetual (drug-like) high,” only better, since devotees share in the omnipotence and omnipresence of Lord Krishna (as expressed through ISKCON leadership).

Implications for future research. My study points to the need for additional research in the area of deprogramming and exit counseling. The modified, intensive case study method I employed proved an excellent procedure for a nascent description and quantification of the complex process of deconversion. At this point, I believe additional intensive case studies are warranted. Deprogrammings may vary more among themselves than they differ from other forms of specialized conversation.

I believe the DSC is a valuable instrument in the classification of deprogramming thoughts. I would like to see if it can be applied to other deprogrammings, exit counselings, and other cults.

I wonder if some of the trends I discovered -- the increased participation of the cultist, the shifts in focus of attention, motility and experiencing -- were unique to Ken's case, or to successful deprogrammings. It may be difficult to study 'failed' deprogrammings because of the difficulty in obtaining the cultist’s informed consent. Still, applying the DSC, attentional indices, and the Experiencing Scale to a wide range of deprogrammings ("successful," “partially successful,” and “failed”) may illuminate some of my preliminary findings.    Some of my questions (e.g., attentional motility, and experiencing) may be better answered by employing different sampling strategies, such as random time sampling.


Abacus Concepts. (1986). StatView 512+ [Computer program]. Calabasas, CA: BrainPower.

Allen, D. E., & Guy, R. F. (1974). Conversation analysis: The sociology of talk. The Hague, Netherlands: Mouton.

Allport, G. (1942). The use of personal documents in psychological science. NewYork: SocialSeienceResearchCouncil.

Bales,R. (1951). Interactionprocessanalysis. Cambridge,MA: Addison-Wesley.

Bales, R. (1970). Personality and interpersonal behavior. New York: Hold, Rinehart and Winston.

Blackham, G. (1977). Counseling: Theory, process and practice. Belmont, CA: Wadsworth.

Burgoon, M., & Bettinghaus, E. (1980). Persuasive message strategies. In M. Roloff & G. Miller (eds.), Persuasion: New directions in theory and research (pp. 141-164). Beverly Hills,  CA: Sage

Campbell, D., & Stanley, J. (1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20, 37-46.

Dubrow-Eichel, S. (1987). Deprogramming statements checklist: Manual version 1.30. Unpublished manual.

Dubrow-Eichel, S. (1989). Deprogramming: A case study. Part 1: Personal observations of the group process [Special issue]. Cultic Studies Journal, 6 (2).

Gendlin, E., Beebe, I., Cassens, J., Klein, M., & Oberlander, M. (1968). Focusing ability in psychotherapy, personality, and creativity. In. J. Shlein (Ed.), Research in psychotherapy, Vol. III (pp. 217-238). Washington, DC: American Psychological Association.

Guilford, J., & Fruchter, B. (1973). Fundamental statistics in psychology and education (5th a). New York: McGraw-Ml.

Hubner, J., & Gruson, L. (1988). Monkey on a stick: Murder, madness, and the Hare Krishnas. San Diego, CA: Harcourt Brace lovanovich.

Kiesler, D. (1969). Refinement to the Experiencing Scale as a counseling tool. Report to the US Department of Health, Education and Welfare, Wice of Education, Bureau of Research, Washington, DC.

Klein, M., Mathieu, P., Gendlin, E., & Kiesler, D. (1969). The experiencing scale: Are search and training manual, Vol.1. Madison, WI: Wisconsin Psychiatric Institute.

Lennard,H.,&Bemstein,A. (1969). Patterns in human interaction. San Francisco: Jossey-Bass.

Lewin, K. (1935). A dynamic theory ofpersonality. New York: McGraw- Hill.

Lifton, R. (1961). Thought reform and the psychology of totalism. New York: W. W. Norton & Co.

Lifton, R. (1986). The nazi doctors: Medical killing and the psychology of genocide. New York: Basic Books

Mintz, J. (1969). Dimensions of psychotherapy and their relation to outcome. Unpublished doctoral dissertation, New York University.

Mintz, J., Luborsk-y, L., & Auerbach, A. (1969). In Klein, M., et. al., The experiencing scale: A research and training manual, Vol. I (pp. 25-26). Madison, WI: Wisconsin Psychiatric Institute.

Rice, L., & Greenberg, L. (1984). The new research paradigm. In L. Rice, & L. Greenberg (Eds.), Patterns of change (pp. 7-25). New York: Guilford.

Rice, L., & Greenberg, L. (Eds.). (1984). Patterns of change . New York: Guilford.

Sefltiz, C., Wrightsman, L., & Cook, S. (1976). Research methods in social relations (3rd ed.). New York: Holt, Rinehart and Winston.

Singer, M. (1979, January). Coming out of cults. Psychology Today, 72- 82. Zeig, 1. (1987). ne evolution of psychotherapy - fundamental issues. In J. Zeig (Ed.), The evolution of psychotherapy (Introduction, pp. xv- xxviii).New York: Brunner/Mazel.

Steve K. Dubrow-Eichel received his B.A. from Columbia University in 1976 and his Ph.D. from the University of Pennsylvania in 1989, where he was honored for having defended his doctoral dissertation on deprogramming with distinction. In 1990, AFF awarded him the John Gordon Clark Award for Distinguished Scholarship in Cultic Studies.