The Group Psychological Abuse Scale

Cultic Studies Journal, Volume 11, Number 1, 1994, pages 88-117

The Group Psychological Abuse Scale: A Measure of the Varieties of Cultic Abuse

William V. Chambers, Ph.D.

University College, Mercer University

Michael D. Langone, Ph.D.

American Family Foundation

Arthur A. Dole, Ph.D.

University of Pennsylvania

James W. Grice, M.A.

University of New Mexico

Abstract

The Group Psychological Abuse (GPA) scale was developed from a factor analysis of 308 former cult members' characterizations of their groups. Four subscales were derived: Compliance, Exploitation, Mind Control, and Anxious Dependency. Reliability and validity findings suggest the GPA should be useful in characterizing the varieties of abuse and in differentiating cults from innocuous groups.

The major apprehension surrounding cults is not that they represent new religious creeds, dissenting political views, or alternative therapeutic methods. The driving concern is that these groups tend to abuse their members, and sometimes nonmembers, unlike bona fide new religious (and other) movements, which treat members and outsiders with relative respect.

How much do we really know about abuse in cults? From experiential, clinical, and philosophical perspectives, we know quite a lot. From the viewpoint of scientific theory, based on quantitative measurements, we know less. A few steps toward quantitative measurement of abuse have been taken, however. Dole and DubrowEichel (1985) used a Delphi strategy to study experts' perceptions of cults. They collected descriptions of numerous dangerous practices found among cults. Ratings of the practices by experts showed that abuses by cults could be quantitatively differentiated. The Dole and DubrowEichel study thus provided precedence for further quantitative studies of abuse in cults.

Some years after the Dole and DubrowEichel (1985) study, Langone and Chambers (1991) conducted a factor analysis of former cult members' interpretations of terms used in cultic studies. Five dimensions of meaningfulness and terminological acceptability were found: Mind Control, Social Manipulation, Group Intensity, Trauma, and Abuse. The former cult members rated the Abuse and Trauma dimensions as most meaningful and acceptable to people leaving cults. The views of the former cult members thus concurred with those of Dole and DubrowEichel in emphasizing the significance of abuse and trauma in cultic groups.

Taking the lead offered by Dole and DubrowEichel (1985), as well as by the subjects in the factor analysis study, Langone (1992) developed a theoretical model of abuse, with the idea of respect for the person as the central concern and contrast to abuse. The respect-versus-abuse model emphasizes four aspects of personhood: Mind, Autonomy, Identity, and Dignity (MAID). These aspects may be construed as needs in a psychological sense or as rights in a philosophical sense.

Disrespect for Mind concerns the violation of the person's right to pursue truth and goodness through reality testing, logical thinking, and objective inquiry. These violations of respect are often described in the literature as mind control, coercive persuasion, or thought reform (Langone, 1993; Lifton, 1961; Schein, Schneier, & Baker, 1961).

The second aspect of respect for the person/Autonomy summarizes the person's right to make decisions independent of the interests and dictates of others. These issues of conformity and compliance are found throughout the literature and reflect the authoritarian features of most cultic groups.

The Identity aspect of personhood concerns the inner integration of the person and integrity as a member of a family, community, and culture. Violation of the rights of people to define themselves is frequently described in the literature (see especially Yeakley, 1988), with the person's identity generally being preempted to serve the group's goals and daily needs.

Dignity, the final aspect of personhood, concerns the person's right to feel worthwhile in the eyes of others, to feel ?equal? in the sense enunciated in the Declaration of Independence (?all men are created equal??that is, with equal inherent worth). The destruction of the person's sense of inherent worth is common to most forms of psychological abuse.

The four aspects of respect, then, appeal to the legitimacy of the person as an end, rather than as a means to be exploited. Langone's theoretical formulation of psychological abuse derives from his work with former members of cultic groups, which, being exploitatively manipulative, place the abuserespect distinction in bold relief. Other clinical and empirical investigations clearly place abuse at the heart of the controversy surrounding the cult phenomenon.

Given this background and our assumption that abuse is the essence of a cult, it is reasonable to proceed to a scientific study of the nature of the abuse in cultic groups. We need a quantitative measure of abuse that can be applied to any group by anyone with experience of the group. Development of such a scale is the purpose of this study.

Method

Introductory Comments

Development of a measure of psychological abuse could begin in either of two ways. Descriptive items could be selected that discriminate between groups commonly considered abusive (e.g., cults) and groups commonly considered nonabusive. Discriminant function analysis is the statistical method of choice for this kind of item analysis (Ghiselli, Campbell, & Zedeck, 1981). This method is frequently used, for example, in research on psychiatric diagnosis. The approach assumes, however, that accepted validity criteria exist--for example, the diagnostic decisions of expert clinicians. Unfortunately, there are no equivalent authoritative diagnostic criteria for cults. Therefore, a measure of the psychological abuse associated with cults must first focus on describing what is meant by cultic abuse.

Clinical studies, philosophical analyses, and surveys of personal and expert experience can help explicate definitional issues and assumptions, as well as provide potential items for scales. Accumulated observations must be systematically organized, however, before sophisticated scientific theories can be developed or derivative hypotheses tested empirically. Factor analysis is the appropriate method for this approach.

As Kerlinger (1973) states: ?Factor analysis is a method for determining the number and nature of the underlying variables among larger numbers of measures.... Factor analysis serves the cause of scientific parsimony.... It tells us, in effect, what tests or measures belong together--which ones virtually measure the same thing? (p. 659). Gorsuch (1983) adds that the aim of factor analysis is to ?summarize the interrelationships among the variables in a concise but accurate manner as an aid in conceptualization? (p. 2).

Thus, factor analysis can enable us to distill the varieties of abuse by separating numerical themes. Delineation of the combinations and permutations of these varieties may, in turn, help us to empirically elaborate the dimensions, meaning, and nature of cultic abuse.

Regarding the choice of factor analysis over discriminant analysis, we should find that if the concept abuse is meaningful, then scales derived from factor analysis should discriminate between noncultic groups and the varieties of cultic groups. Discriminant function analysis could, indeed, later be applied to the factor analytically derived scales to predict group membership. The result should be both statistically efficient and theoretically meaningful discrimination of groups.

Instrument

A pool of items was created through several strategies. First, the 20 highest-rated items in the Dole and DubrowEichel (1985) study of cult dangerousness were included. Next, Langone's (1992) model of respect/abuse as well as other theoretical formulations (Lifton, 1991; Ofshe & Singer, 1986; Singer, Temerlin, & Langone, 1990) were used to select additional items. Lastly, the authors' clinical experience and conversations with other experienced clinicians yielded other items. A pilot study that analyzed the responses and comments of 12 former cult members resulted in further modification of the item pool. The items that were ultimately chosen fell into three domains of interest: (1) the purpose of the group, (2) the relationships within the group, and (3) the relationships with others outside the group. In all, 112 items were selected and included in a much larger survey sent to former cult members.

Subjects were instructed: --Please rate the degree to which the following statements ... characterize the group ... according to your experience and observations ... of how [it] ... ACTUALLY functioned ... 1=not at all characteristic; 2=not characteristic; 3= can't say/not sure; 4=characteristic; 5=very characteristic.?

Procedures and Subjects

A 20page questionnaire booklet with explanations and consent forms was sent to 375 members of FOCUS (a national network of former cult members) and approximately 200 excult members on the American Family Foundation's mailing list. (The precise number in the group is not known because duplications to the FOCUS list were unfortunately not removed before mailing.)

In addition, varying numbers of questionnaires were sent to 153 cult educators and organizations, who were asked to distribute the questionnaire. Altogether 800 questionnaires were mailed. Completed questionnaires were returned by 308 persons, a response rate of about 35%. (The precise response rate cannot be determined because it is not known exactly how many questionnaires given to experts and organizations were actually passed on to exmembers.) Most of the respondents had contacted the American Family Foundation (AFF) or the Cult Awareness Network (CAN) seeking information concerning cults. Approximately 37% of the subjects, however, had no or little contact with AFF or CAN and were presumably given questionnaires by others.

This sample could be described as a combination of a network and a snowball sample. It is not necessarily representative of the excult population at large, but probably is reasonably representative of that subgroup of exmembers who come into contact with cult education organizations.

As questionnaires came in, they were given an identification number and separated from the consent forms (which were placed in a safe deposit box) in order to ensure anonymity. Although no subjects were formally interviewed, approximately one dozen initiated contact with the investigators either by phone or mail. The most common and striking aspect of these contacts was subjects saying that the questionnaire helped them better understand their experiences. Most asked for additional questionnaires to give to friends who had also left cults.

The subjects had involvements with a total of 101 groups, including Eastern, Biblebased, political, and other types. The average length of membership was 6.79 years. Fortythree percent reported obtaining leadership status in their group. Sixty percent left the group on their own--that is, without formal outside assistance from exit counselors or mental health professionals and not as a result of legal conservatorship. Ratings on five-point scales, with 2 being important and 1 being very important, revealed the following reasons for leaving their group: time spent away from group, m=1.98; disillusioning experience with leader, m=1.65; becoming aware of being manipulated, m=1.59; and feeling abused and/or exploited, m=1.73.

Fortyseven percent of the subjects were raised Protestant, 24% Catholic, 8% Jewish, and 8% other, with 13% reporting no childhood religious affiliations. Postcult religious affiliations included 30% Protestant, 8% Catholic, 5% Jewish, 46% none, and 12% other.

Sixtyfour percent of the subjects were female, 36% male. Subjects reported an average of five persons in their family of origin. Sixtyseven percent reported that their parents were living together at the time that the subjects joined the group. When asked how well their families got along before the subject's cult involvement, 66% reported ?average? or ?above average.? Sixteen percent reported that their family of origin's income, before taxes, was less than $20,000; 72% reported incomes between $20,000 and $100,000; 12% indicated incomes greater than $100,000.

Eighteen percent were married when they joined their group, with an average of two children per couple. Fortythree percent of those who had been married reported that they did not get along well in their marriage, with 40% reporting that they were separated or divorced at least in part because of their group involvement.

Subjects reported that they had completed an average of 14.84 years of school at the time of completing the survey. The average at the time of joining their group was 13.64 years. Occupations at the time of group membership included 7% clerical, 4% craft, 1% labor, 1% machine or transport operator, 3% managerial or administrative, 16% professional or technical, 4% sales, 11% service, 6% none, 30% student, 16% other. Postcult occupations were similarly distributed, except that fewer were now students (12%), and more were now professionals (34%).

In summary, the sample consisted of individuals who became disenchanted after having made substantial commitments to a wide array of cultic groups. In general, the subjects were raised in the middle class; they grew up in intact, supportive families, with moderate incomes, and with childhood involvement in various mainstream religious denominations.

The subjects represent those people who are most likely to appreciate the nature and extent of abuse in cults--that is, the victims. By factor analyzing their ratings we stand to develop scales that differentiate various kinds of cultic abuse as well as an index of the extent of cultic abuse. We assume this strategy will produce a much narrower range of variance than would be the case if we had included noncult members and current members. But this does not mean that we should not later include such groups.

Having built greater sensitivity into the scales by focusing on variances within former cult members, subsequent applications to samples including other groups should produce even stronger results. Reliability and validity statistics are probably attenuated by the constricted sample and should increase in discriminative power with a broader sample. The choice of selfdescribed victims as the base sample is, therefore, not only theoretically appropriate but also statistically conservative.

Results

Principal components analysis with varimax rotation revealed four interpretable factors, each with eigenvalues greater than 3. The analysis was restricted to components with eigenvalues greater than 3 (as opposed to the Kaiser criterion of 1) because the smaller components contained few items loading substantially (.30 or larger) and because the items failed to be conceptually interpretable. Eigenvalues for the first four components were 20.21, 5.33, 3.72, and 3.64 respectively, accounting for 29.4% of the total variance. Four is probably an underestimate of the number of dimensions of abuse. Our conservative selection, however, provides more confidence in the validity of the dimensions that we do interpret.

The strategy for scale development was to choose items that loaded substantially on the factors, assuming that the scales made up of such items would reflect the factors from which they were chosen. The scale scores were the sum of the ratings originally given on the item by the subject.

Seventyfive of the 112 items loaded substantially on one or more of the four retained factors. The large number of substantial loadings facilitated identification of the factors. The fact that most items loaded on more than one factor, however, considerably narrowed the pool of items for inclusion in the scales. We sought to use items that loaded substantially on only one of the four factors in order to increase the unidimensionality, and thus interpretability, of the scales. An equal number of items per factor/subscale were chosen in order to facilitate scale comparisons.

We settled on 7 items per subscale, with a total of 28 (or 4x7) items in the summary Group Psychological Abuse (GPA) scale (see Appendix A). Thus, each subscale could produce scores from 7 to 35, while the GPA index, as the summary index of psychological abuse, could range from 28 to 140.

Analysis of the loadings of the items on the factors suggested the following labels: Compliance, Exploitation, Mind Control, and Anxious Dependency (the rationale for these labels will be provided later). Table 1 includes the abbreviated items for each subscale. Included is the loading of each item on its factor, the correlation of the items with the subscale total scores, the correlation of the items with the summary GPA scale, and the communalities (proportion of variance explained) for items across the four factors. (An identical factor analysis based on these 28 items revealed the same factor structure. The four factors accounted for 42% of the variance.)

Alpha coefficients are included for each subscale. Alpha for the GPA summary scale was .81. Alphas for the subscales were .81 for Compliance; .75 for Exploitation; .70 for Mind Control; and .72 for Anxious Dependency. These reliabilities are sufficient for research purposes.

Table 1. Item Descriptions and Statistics by Scale

GPA = Sc1+Sc2+Sc3+Sc4 GPA alpha = .81

Sc1: Compliance alpha = .81

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

01 .54 .69 .01 .20 .31 .46 .36 sex lives not dictated

04 .60 .69 .25 .28 .24 .56 .42 sacrifice own goals

13 .60 .75 .14 .16 .17 .47 .40 group lives together

14 .67 .73 .06 .31 .22 .49 .48 intimacy dictated

18 .47 .59 .20 .23 .31 .51 .30 serve leaders

21 .59 .65 .05 .25 .21 .43 .40 members make decisions

28 .61 .71 .06 .29 .21 .47 .42 members must consult

Sc2: Exploitation alpha = .75

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

02 .47 .12 .52 .13 .21 .40 .26 women seduce for group

03 .61 .17 .65 .20 .22 .50 .42 breaking law okay

06 .51 .07 .65 .02 .06 .34 .27 politics major goal

12 .68 .12 .77 .13 .20 .51 .50 violence to outsiders

19 .67 .06 .69 .04 .10 .39 .46 money major goal

20 .47 .08 .58 .25 .12 .40 .27 critics threatened

27 .43 .06 .52 .12 .17 .21 .34 recruiting major goal

Table 1. (continued)

Sc3: Mind Control alpha = .70

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

09 .43 .22 .08 .40 .28 .34 .26 members part of elite

11 .55 .18 .27 .66 .19 .44 .38 coercive persuasion

15 .56 .24 .21 .68 .32 .50 .42 stay because deceived

22 .50 .20 .15 .59 .20 .38 .30 can think critically

24 .58 .19 .16 .67 .21 .41 .39 mind control used

25 .43 .30 .06 .61 .07 .33 .30 little psycholog'l pressure

26 .45 .18 .02 .63 .05 .24 .27 criticism rare

Sc4: Anxious Dependency alpha = .72

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

05 .39 .15 .11 .21 .62 .42 .23 medical help okay

07 .69 .32 .09 .17 .69 .50 .55 leaving is damnation

08 .42 .25 .05 .07 .53 .36 .22 no negative emotions

10 .64 .28 .14 .13 .65 .50 .46 criticism is evil

16 .37 .12 .09 .20 .56 .37 .17 exercises remove doubt

17 .50 .15 .09 .13 .65 .40 .30 no medical help

23 .45 .22 .18 .15 .60 .45 .28 leader is divine

Note: with n=300, the p<.01 critical value for r is (+,) .14. Column 1, the item number (#), refers to the scale in the appendix. Column 2 is the loading of each item with its factor (F1, etc.--the mathematical abstraction). Columns 3-6 are the correlations of the item with the four scales (the 7 concrete items constituting the scale). Column 7, h, is the communality, that is, the proportion of variance the four factors account for with respect to an individual item. Column 8 is the abbreviated description of the item.

Discussion

Psychometric Rationale and Implications

Because the Cultic Studies Journal is multidisciplinary rather than a strictly psychological journal, and because this article reports on a study that is technically complex, we have chosen in this section to elaborate on the psychometric rationale for the development of the four subscales and the composite GPA index. We hope that readers unfamiliar with the technical aspects of the study might thereby better understand the methods, results, and implications--including theoretical implications to be discussed further in another paper.

As noted earlier, factor analysis distills the various numerical patterns characterizing items that are analyzed. Uncorrelated variances between items cancel one another as each factor is derived, leaving principal themes or factors. These factors are mathematical abstractions. Naming each factor requires a parallel cognitive, or psychological, abstraction (to be discussed further in another section). In looking for the common theme among the items that load highest on a factor, the analyst tries to distill the essence of that factor. Since loadings are correlations between the items and the abstract factor, they point to meanings that the items share with the factor. The higher the loading, the more the item contributes to the meaning of the factor.

Typically, factor analysis produces, as with the GPA, several factors composed of a number of items. When each item loads on only one factor, the data is said to manifest simple structure (Thurstone, 1947). This is the ideal of factor analysis. When a particular item loads highly on more than one factor, the item is considered to be complex, rather than simple. Such items are constructions (rather than abstractions) because they unite the orthogonal (independent) features of the factors on which they load. These complex items are rich in meaning and can aid in factor interpretation. Often they reflect syndromatic combinations of factors that characterize the dynamics of nature's concrete and psychological events. The psychometric unreliabilities of individual complex items, however, make them less useful than combinations of orthogonal subscales in illuminating syndromes because subscales, being composed of numerous items, are more reliable than individual items.

Table 2 (below) includes complex items excluded from the scales because of pronounced lack of simple structure. Also included are abbreviations of these items, major factor loadings, scale correlations, and communalities. Table 2 reflects the relevance of the subscales, for as abstract as they may be, they in combination give meaning to the relatively unreliable, complex items that lack simple structure. In other words, they illuminate the construct validity of psychological abuse by revealing permutations of abuse.

For example, item #224, --devotes all free time to group,? loads substantially on factor 1 (Compliance) and on factor 2 (Exploitation). This item suggests the outcome of compliance to an exploitative leader. Items #231, --confess sins and limitations,-- loads on the Compliance and Mind Control factors, suggesting both a demand for confession and abasement (Compliance) and the use of confession and abasement in mind control strategies. Item #231 may also be interpreted to reflect a syndrome, such as the cult of confession, as described by Lifton (1961). These kinds of interpretations are rather speculative, however, because the meaning of the constructional items is obscured by brevity of expression and measurement errors. A more useful approach, which we have followed in this article, is to begin with simple abstractions (i.e., items loading on only one factor) and build up from these to complex constructions (i.e., combinations of factors or items loading on more than one factor).

Table 2: Relevant Items Excluded from Scales

F1: Compliance

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

215 .35 .48 .16 .29 .35 .48 .23 serving leader #1 goal

219 .37 .38 .16 .33 .14 .36 .28 disagreement respected

224 .63 .58 .31 .27 .21 .53 .49 all free time to group

231 .53 .37 .26 .39 .11 .41 .38 confess sins

235 .48 .50 .36 .25 .36 .57 .46 will die for leader

247 .48 .43 .41 .30 .15 .49 .48 work to exhaustion

249 .49 .43 .11 .40 .26 .44 .37 fail to meet standards

262 .52 .41 .08 .28 .29 .39 .40 feel much guilt

263 .42 .32 .01 .33 .26 .33 .33 feel inadequate

268 .45 .42 .12 .29 .37 .45 .36 fear leaving

275 .48 .42 .16 .42 .22 .43 .31 leaders are obeyed

309 .40 .34 .25 .31 .33 .46 .34 dissent not tolerated

310 .45 .42 .31 .22 .44 .54 .50 allegiance to death

Table 2. (continued)

F2: Exploitation

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

222 .52 .45 .50 .16 .32 .57 .52 assets given to group

234 .42 .33 .44 .25 .19 .47 .34 physical punishment

237 .40 .28 .36 .22 .18 .40 .23 live as well as leader

279 .61 .15 .53 .22 .20 .43 .46 lying okay for goals

291 .43 .31 .44 .19 .16 .43 .28 fronts not used

292 .39 .03 .31 .06 .12 .11 .30 structured indoctrination

295 .56 .20 .49 .38 .18 .47 .49 lies to look good

305 .39 .02 .34 .00 .16 .21 .20 secret agency supports

313 .44 .18 .39 .42 .29 .47 .41 lack time to sleep

314 .32 .26 .36 .30 .26 .45 .28 no contemplation

F3: Mind Control

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

225 .47 .26 .09 .42 .11 .30 .30 questions encouraged

243 .39 .31 .12 .39 .18 .35 .31 do not disagree

253 .30 .18 .16 .34 .13 .29 .18 dissenters humiliated

254 .38 .33 .34 .40 .20 .47 .37 alters personality

261 .40 .33 .23 .41 .19 .41 .30 time for recreation

270 .43 .27 .20 .50 .31 .45 .32 stay because of abuse

284 .41 .24 .10 .34 .30 .35 .42 enlightenment outside

287 .37 .30 .24 .30 .17 .37 .28 dissenters respected

290 .49 .24 .39 .40 .16 .43 .41 info denied prospects

301 .40 .45 .15 .50 .32 .51 .41 childlike dependency

312 .47 .18 .39 .42 .29 .47 .50 proselytize by deceit

F4: Anxious Dependency

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

232 .41 .41 .09 .21 .37 .41 .40 severe discipline kids

233 .44 .46 .18 .34 .47 .55 .45 terrible if leave

257 .56 .38 .16 .31 .55 .53 .47 punishment for dissent

274 .37 .30 .10 .30 .36 .39 .30 fear self-expression

278 .37 .38 .09 .23 .37 .40 .29 nothing without group

280 .37 .34 .05 .33 .33 .38 .27 friends and family okay

Table 2. (continued)

F4: Anxious Dependency

Correlations

# Load Sc1 Sc2 Sc3 Sc4 GPA h2 Description

297 .32 .36 .11 .21 .35 .39 .22 nongroup frightening

304 .41 .28 .24 .32 .44 .48 .38 loyalty above all else

306 .35 .17 .36 .31 .34 .45 .28 totalitarian view

311 .45 .36 .43 .32 .50 .62 .51 deceit okay for religion

315 .47 .31 .31 .37 .51 .56 .41 only leaders true

317 .46 .44 .21 .19 .46 .51 .44 avoid family and friends

Summary GPA Index. As the sum of orthogonal subscales, the summary GPA index, or scale, is a construction of the overall extent of abuse. The summary scale combines reliable variances, as reflected in the internal consistency (alpha) coefficients of the subscales. The subscales are refined abstractions, approximating simple structure. Conjugating (i.e., elaborating all the combinations of) these subscales will suggest various profiles that reflect the theoretical varieties of psychological abuse. One possibility, for example, is high Compliance, Anxious Dependency, and Mind Control, with low Exploitation. Another is high Compliance with low scores on the other subscales. Still another is high Mind Control with low scores on the other subscales. Although these profiles, or varieties of abuse, are theoretically possible, some may not occur in nature (e.g., the last profile example). If given to enough people from a wide enough spectrum of groups, the GPA Scale can help determine which profiles nature indeed produces.

With regard to statistical efficiency, constructing a GPA summary scale from the refined subscales has the same rationale as the derivation of orthogonal predictor variables in multiple regression, in that the use of orthogonal subscales (or orthogonal predictor variables) helps avoid redundancies and consequent noncollinearity problems. The extent of abuse reflected in the GPA Summary Index will be unchanged by the conjugation of varieties, much as orthogonal variables can be entered into a regression equation in any order without affecting the summary index. As a result, the GPA index should be a statistically optimal measure of the extent of perceived abuse.

In practice, the above statistical optimization will be slightly compromised because the subscales are modestly correlated (see Table 3). This is the price of imperfect simple structure and of weighting each item in the scale by 1, instead of by the item's more refined factor loading, when summing scores of subscales. Nunnally (1978) suggests, however, that such refined weighting makes little difference in practice. This contention is supported by the fact that the correlations between the subscales are much smaller than those between the items within the subscales (summarized by the substantial Cronbach alphas).

These contrasting patterns of correlation generally support the convergent and discriminant validities of the GPA scales, and suggest that efficient statistical discrimination is highly likely. Furthermore, the predominantly orthogonal nature of the scales allows us to conjugate the varieties of abuse, affording us powerful heuristic strategies for theory development. A subsequent paper will elaborate upon the heuristic value of conjugating the GPA scales.

Table 3. Correlations Between Scales