The aim of this study is the development and analysis of the psychometric properties of a new instrument to assess the possible psychological abuse experienced in a group setting. The Psychological Abuse Experienced in Groups Scale was administered to 138 people who self-identified as former members of abusive groups. An exploratory factor analysis revealed a common factor, which showed appropriate reliability. The scale scores were correlated with a prior measure aimed to assess group abusiveness, providing evidence of external validity. Participants reported a higher degree of psychological distress than normative samples, and those who requested psychological counseling after leaving the group had suffered group psychological abuse to a greater extent. The scale covers a wide range of subtle and explicit abusive behaviors and overcomes the limitations of previous instruments, being useful in both research and applied settings.
El objetivo de este estudio es el desarrollo y análisis de las propiedades psicométricas de un nuevo instrumento para evaluar el posible abuso psicológico experimentado por una persona en el seno de un grupo. La Escala de Abuso Psicológico Experimentado en Grupos se administró a 138 personas auto-identificadas como exmiembros de grupos abusivos. Mediante un análisis factorial exploratorio se extrajo un factor común que mostró una adecuada fiabilidad. Las puntuaciones de la escala correlacionaron con una medida previa que evalúa el grado de abuso que caracteriza a un grupo, aportando evidencias de validez externa. Los participantes reportaron un mayor grado de malestar psicológico que en muestras normativas y aquellos que buscaron atención psicológica tras abandonar el grupo reportaron haber sufrido abuso psicológico en mayor grado. La escala cubre un amplio rango de conductas abusivas tanto explícitas como sutiles y permite superar las limitaciones de instrumentos previos, resultando útil tanto en el campo académico como en el aplicado.
This study describes a scale designed to assess psychological abuse behaviors that can be experienced in groups. Psychological abuse in group settings has mainly been studied in the so-called cult or abusive groups, where abusive strategies are applied by a group leader or some of the members to recruit and retain other members (Langone & Chambers, 1991). Some sources suggest that there are over 5,000 abusive groups operating within the United States and Canada (Singer, 2003), with a combined membership of over 2,500,000 people (McCabe, Goldberg, Langone, & DeVoe, 2007). Former members of these types of groups report having suffered persistent forms of psychological abuse (e.g., Chambers, Langone, Dole, & Grice, 1994; Matthews & Salazar, 2014), and several studies report negative post-involvement consequences (for a review, see Aronoff, Lynn, & Malinoski, 2000).
The concept group psychological abuse has been proposed to refer to the abusive practices of these types of groups (Langone & Chambers, 1991); it has been defined as a process of systematic and continuous application of pressure, control, manipulation, and coercion strategies for the purpose of dominating other people in order to achieve their submission to the group (Rodríguez-Carballeira et al., 2015). Over the years, researchers have extensively documented psychologically abusive behaviors that may take place in group settings (e.g., Baron, 2000; Coates, 2012; Hassan, 2013; Langone, 1992; Singer, 2003). Examples of these behaviors include isolation from the family, control of sexual life, humiliation, denigration of critical thinking, and imposition of an absolute authority.
Prior studies have classified these behaviors according to the possible effects they can have on group members. In this regard, Langone (1992) classified psychologically abusive behaviors that can occur within groups according to whether they mainly disrespect individuals’ minds, autonomy, identity, or dignity. Hassan (2013) classified them according to whether they are directed toward behaviors, information, thoughts, or emotions. These classifications have limitations because an abusive behavior may influence more than one psychological process. Moreover, some authors have stressed that the definition and assessment of group psychological abuse should focus on the abusive actions rather than on their possible psychological and social effects, which can have a different nature or seriousness and should be evaluated separately (Rodríguez-Carballeira et al., 2015).
Recently, a comprehensive taxonomy of group psychological abuse that focused on the delimitation of psychologically abusive behaviors was developed and validated by a panel of experts through a Delphi study (Rodríguez-Carballeira et al., 2015). It is composed of 26 strategies classified into 6 components: (a) isolation, (b) control and manipulation of information, (c) control over personal life, (d) emotional abuse, (e) indoctrination within an absolute and Manichean belief system, and (f) imposition of a single and extraordinary authority. An operational definition for each component of the taxonomy is also provided.
Evaluation of Group Psychological AbuseEfforts to evaluate group psychological abuse strategies from a rigorous, quantitative and empirical perspective are still recent. The Group Psychological Abuse scale (GPA; Chambers et al., 1994) was the first instrument designed to assess the degree and varieties of psychological abuse practices perceived within a group, and it is the only instrument with repeated evidence of its reliability, validity and utility (Almendros, Gámez-Guadix, Carrobles, & Rodríguez-Carballeira, 2011). Other relevant scales are the Individual Cult Experience index (ICE; Winocur, Whitney, Sorensen, Vaughn, & Foy, 1997), which is designed to assess the extent of exposure to negative cult experiences through 47 items, and the Across Groups Psychological Abuse and Control scale (AGPAC; Wolfson, 2002), which contains 22 items structured in three dimensions: Emotional Abuse, Isolation-Control of Activity, and Verbal Abuse.
In spite of the strengths of the existing instruments, there is still room for improvement. These prior instruments were based on a variety of theoretical formulations and clinicians’ experiences (Chambers et al., 1994), a list of group practices hypothesized to be related to post-involvement distress (Winocur et al., 1997), or a model of intimate partner violence (Wolfson, 2002). The lack of a comprehensive semantic definition of group psychological abuse that clearly differentiates its operative components, and the lack of a table of specifications, may have led to the misrepresentation of severe abusive strategies and the over-representation of aspects that are not abusive strategies in themselves (Carretero-Dios & Pérez, 2005). Therefore, a semantic definition of the construct, such as the operational definitions of group psychological abuse components provided by Rodríguez-Carballeira et al. (2015), is a necessary step in the development of any instrument, in order to obtain theoretical content validity and ensure that the design process is not ambiguous or imprecise (Nunnally & Berstein, 1994).
Another limitation detected in prior instruments is that the content and wording of some items may not meet the standards specified by several authors (e.g., Haladyna, 2004; Morales, Urosa, & Blanco, 2003). On the one hand, the content of some items does not seem to be congruent with the construct domains. They do not specifically describe psychological abusive behaviors that could be applied within a group, but rather other aspects related to the phenomenon, such as sexual and physical abuse practices, the group's generic characteristics, its purposes, or its relationship with outsiders (e.g., ICE's item 34 is “The leader or members physically injured you or your loved ones to the extent that you or they needed medical care”; GPA's item 6 is “Gaining political power is a major goal of the group”). On the other hand, some items can have a variety of interpretations because they contain unclear and undefined words such as “mind control” or “coercive persuasion” (e.g., GPA's item 24 is “Mind control is used without conscious consent of members”).
In addition, another practical limitation related particularly to the GPA is that this scale was intended to assess perceived group abusiveness, asking about psychologically abusive practices, whether experienced personally or not (Chambers et al., 1994). Thus, the scores on this scale involve judgements about the group as a whole and, therefore, have limited use in research or applied contexts where the interest lies in the individual experience of psychological abuse.
Consequences of Group Psychological AbuseSome studies have hypothesized that the intensity and frequency of the abusive strategies experienced within a group are related to post-involvement psychological distress (Winocur et al., 1997) or other substantial consequences for the affected people (Baron, 2000). Along these lines, several studies have reported psychopathological symptoms and adjustment problems in samples of former members of abusive groups (e.g., Coates, 2010; Malinoski, Langone, & Lynn, 1999; Matthews & Salazar, 2014). Some of the difficulties involve low self-esteem, depression, anxiety, somatization, dissociation, or self-destructive tendencies (Aronoff et al., 2000). However, to date, very few empirical studies have examined the relationship between psychological distress and the abusive behaviors experienced using standardized measures (Winocur et al., 1997). The aforementioned mental health and adjustment problems explain the fact that some former members of abusive groups seek psychological care after leaving the group. To facilitate their empowerment, some authors have proposed that it is necessary to accurately assess their past experiences of abuse (Matthews & Salazar, 2014).
Relationship between Group Psychological Abuse and Biographical VariablesPrior empirical studies with samples of former members of abusive groups have not found differences in the perception of group abusiveness based on gender, age of participation, or other sociodemographic variables (e.g., Almendros et al., 2012). However, former members’ evaluations of their group experiences could be negatively biased, due to the influence of contact with cult-awareness organizations or the method of departure (Lewis, 1986). Thus, the testimonies of people who contacted these organizations or left the group after being expelled or counseled by professionals were labeled as atrocity tales (Bromley, Shupe, & Ventimiglia, 1983), referring to a negative exaggeration of the events in the group. Nevertheless, a recent study with a Spanish sample of 101 former members of abusive groups found no differences in the perceptions of psychological abuse or reported psychological distress based on the influence of organizations and the method of departure (Almendros, Carrobles, Rodríguez-Carballeira, & Gámez-Guadix, 2009).
Study ObjectivesThe relevance of measuring group psychological abuse by focusing on individuals’ experiences and the possibility of overcoming the limitations of existing instruments have led to the need to develop new tools to comprehensively assess the phenomenon. Moreover, a review of the scientific literature suggests the relevance of studying the relationship between group psychological abuse and certain psychological difficulties, as well as the influence of certain biographical variables on the perception of the abuses experienced.
The goal of the present study is to develop and test the psychometric properties of a new instrument designed to assess group psychological abuse, focusing on the frequency of psychologically abusive behaviors experienced within a group. The instrument is intended to evaluate the experiences of former members of groups where abuses might have occurred, and could be useful for both researchers and mental health professionals during the assessment process in clinical settings. Three specific aims were established to evaluate the psychometric properties of the instrument. Thus, the present study analyzes (1) its internal structure, (2) its internal consistency, and (3) its relationship with perceived group abusiveness, psychological distress, and various sociodemographic variables.
MethodParticipantsConvenience and snowball sampling methods were used to select a sample composed of 138 people self-identified as former members of groups that were overly controlling or abusive. In all, they had been involved in a total of 66 different groups of a religious, pseudo-therapeutic, political, or commercial nature. Of the participants, 71% were female and 29% male, most of them were White (93.4%), mainly from the United States (58.7%), Canada (15.2%), and the United Kingdom (13%), and their native language was English. Regarding participants’ educational level, 4.7% of them had finished high school, 25% had some college education, and 70.3% had university studies.
Their average age was 48.41 years (SD=11.99), having joined the group with a mean age of 24.18 (SD=11.68), and remaining in the group an average of 12.21 years (SD=9.1). The time elapsed from leaving the group to participating in the present study was an average of 12.15 years (SD=10.26). Concerning the departure method, 52.9% of the participants left the group by walking away after a personal reflection, 17.4% left the group after being counseled by family, friends or professionals, 3.6% after a deprogramming intervention, 7.2% after being expelled by the group, and 18.8% because of the dissolution of the group. Regarding therapy, 30.7% reported having received psychotherapeutic support before their involvement with the group, 35.9% received support while in the group, 78.7% received support after leaving the group, and 42.3% were receiving support when the questionnaire was administered, although not necessarily related to their group experiences. Among the participants, 76.5% reported having sought psychological help in relation to their past group experiences, and 63.1% had contacted cult-awareness organizations.
InstrumentsThe Psychological Abuse Experienced in Groups Scale (PAEGS) is a paper-pencil test developed following the guidelines revised by Carretero-Dios and Pérez (2005). The operational definitions of the group psychological abuse components proposed by Rodríguez-Carballeira et al. (2015) were used as the semantic definition of the construct.
Four researchers with previous experience in the study of the different forms of psychological abuse constructed 92 items using the semantic definition of group psychological abuse and following the guidelines proposed by Haladyna (2004) and Morales et al. (2003). Of these items, 14 were written to represent the component of isolation, 9 for control and manipulation of information, 27 for control over personal life, 21 for emotional abuse, 17 for indoctrination within an absolute and Manichean belief system, and 4 for imposition of a single and extraordinary authority. After a week, the researchers reviewed the written items, as Navas (2001) suggested as an alternative a review by an external expert panel. The items measure components of group psychological abuse in a positive way, and they: (a) represent behaviors, actions, or practices carried out by the group, (b) correspond to the definition of one psychological abuse strategy, (c) refer to the respondent's experience, and (d) do not allude to possible consequences of such practices. The response format for the items is a 5-point Likert-type scale (0=not at all, 1=slightly, 2=quite a lot, 3=a lot, 4=continually). The response labels were selected to assess the frequency of the application of the abusive behaviors experienced.
The responses of a pilot sample composed of 17 participants in a workshop for former cult members were analyzed to select the items with better properties from both a qualitative and quantitative perspective. These subjects were not included for subsequent analyses in the final sample of the study. The item selection process was carried out independently for each group of items related to each group psychological abuse component. First, twenty-seven items were eliminated due to comprehension problems or because their item-dimension correlation score was below the .30 cut-off point established by Nunnally and Bernstein (1994). Second, those items with simpler language and closer to the core meaning of the specific abusive component were selected. In addition, an attempt was made to avoid conceptual overlap and maximize the representation of the different group psychological abuse strategies. The final version of the PAEGS (see Appendix) includes 31 items. This number of items was considered adequate to represent the 6 main components of group psychological abuse proposed by Rodríguez-Carballeira et al. (2015).
The Group Psychological Abuse scale (GPA; Chambers et al., 1994) was used to assess perceived group abusiveness in order to provide evidence of concurrent validity. A modified version of the scale was administered (GPA-M; Almendros, 2006), omitting the middle response labels and excluding the negative wording in item 1. The GPA-M is composed of 28 items rated on a 5-point Likert-type scale ranging from 1 (not at all characteristic of the group) to 5 (very characteristic of the group). In the present study, the overall alpha was .85, and the alphas for each of the four dimensions found in the original study were as follows: Compliance (.83), Anxious-Dependency (.70), Exploitation (.77), and Mind Control (.62). A cut-off point of 74 or above for the total score on the GPA-M has been proposed as an empirical criterion value to distinguish between former members of abusive groups and a community sample.
The Symptom Checklist-90-R (SCL-90-R; Derogatis, 1983) was administered to examine current psychopathological symptoms. It includes 90 symptoms whose pain intensity is rated on a scale from 0 (total lack of discomfort related to the symptom) to 4 (maximum discomfort), and provides nine symptomatic dimensions of psychopathology and three global indices of discomfort. In the present study, the nine symptomatic dimensions and the Global Severity Index (GSI) were taken into account, and the alpha values for the dimensions ranged from .94 to .84.
The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) was administered to examine current personal self-esteem, understood as feelings of self-respect and self-acceptance. It contains 5 positively-worded items and 5 negatively-worded items, all rated on a 4-point Likert-type scale ranging from 1 (totally disagree) to 4 (totally agree). Its one-dimensional factor structure is largely invariant across different nations (Schmitt & Allik, 2005). In the present study, the overall alpha was .91.
The Marlowe-Crowne Social Desirability Scale (SDS; Crowne & Marlowe, 1960) was also administered to examine possible social desirability bias. A short version of the scale, composed of 10 items with a true-false response scale, was used, as recommended, due its acceptable psychometric properties in terms of reliability and dimensionality (Loo & Loewen, 2004). In keeping with traditional practice, the scale was scored so that higher values represented larger numbers of socially desirable responses endorsed.
A biographical questionnaire was also administered to collect information about socio-demographic data and other aspects related to participants’ experiences within the group or after leaving the group. Questions were included about the length of involvement with the group, method of departure, contact with cult-awareness organizations, or psychological counseling received.
ProcedureData were collected through an online questionnaire, given the difficulty of accessing the sample, from June 2008 to March 2009. Contact with participants was made mainly through the mediation of organizations providing information and education about abusive groups (65.9%) and professionals (10.9%). Participants who had already collaborated in the study also invited other former members they knew (23.2%). The questionnaire was divided into two sections. The first included the biographical questionnaire and the psychological abuse measures, while the second section included the SCL-90-R, the RSES, and the SDS. In all, 70.8% of the 195 people who were invited to participate in the study fully completed the first section of the questionnaire, while 65.1% completed the second section. All the participants collaborated in the study voluntarily and without receiving economic compensation, and gave their informed consent. They were also able to choose between an anonymous or confidential collaboration, by providing or not the researchers with an email address.
Data AnalysisAfter checking that there were no missing data, the factor structure of the scale was analyzed by means of an exploratory factor analysis using FACTOR 9.2 (Lorenzo-Seva & Ferrando, 2006), the program recommended by Baglin (2014) to improve EFA practices. Before the analysis, the psychometric adequacy of the items and the correlation matrix were explored. The Unweighted Least Squares (ULS) method was used with the polychoric correlation matrix because it is preferable to other estimation methods due to its robustness with small samples and non-normal data (Ferrando & Anguiano-Carrasco, 2010; Izquierdo, Olea, & Abad, 2014). In selecting the number of factors, the results of the Hull method, the goodness-of-fit index, and the root mean square of residuals were taken into account.
Analyses of descriptive statistics, internal consistency, and convergent validity were carried out using SPSS 20.0 (IBM Corp. Released, 2011). Although the exploratory factor analysis revealed a one-factor structure, psychometric properties and Cronbach's alpha coefficients were computed, both for the total score and for the six psychological abuse components represented in the scale, due to their theoretical and applied interest. Convergent validity was tested by correlating the scores on the PAEGS with those of the GPA-M. Bivariate correlation analyses were conducted between the PAEGS and the measures provided by the SCL-90-R, RSES, and SDS.
To report on the degree of psychological distress, Cohen's ds were computed, taking into account the scores of the present sample and the scores of normative samples for the SCL-90-R (Derogatis, 1983) and the RSES (Schmitt & Allik, 2005). The differences in reported abuse, depending on sex, ethnic background, level of education, received counseling, method of departure, and contact with cult-awareness organizations, were analyzed using the Mann-Whitney U or the Kruskal-Wallis H nonparametric tests, which do not require large normally distributed samples.
ResultsFactor Structure AnalysisResults showed that the data have inherent sufficient correlations to justify an exploratory factor analysis, as the Kaiser-Meyer-Olkin index was .88, and the Bartlett's sphericity test yielded an χ2 value of 2509.5 (p<.001). One major factor was detected by the Hull method, a procedure that outperforms other methods frequently used in applied research (Lorenzo-Seva, Timmerman, & Kiers, 2011). In addition, the GFI (.97) and the RMSR (.09), which was not much larger than the criterion value of .085 proposed by Kelly (1935), also supported the one-factor structure. The retained factor accounted for 45.7% of the total variance. As Table 1 shows, factor loadings ranged from .46 to .80, exceeding in all cases the .4 criterion that leads to including an item in the interpretation of a factor (Izquierdo et al., 2014). The extracted factor was labeled Group Psychological Abuse Experienced because it comprises items that comprehensively represent the different abusive strategies proposed by Rodríguez-Carballeira et al. (2015).
Descriptive Statistics and Factor Loadings for PAEGS Items.
| Item | M [95% CI] | SD | Skew | Kurtosis | rcix | λi1 |
|---|---|---|---|---|---|---|
| 1 | 2.94 [2.68, 3.21] | 1.21 | -0.93 | -0.20 | .64 | .70 |
| 2 | 3.10 [2.82, 3.38] | 1.27 | -1.20 | 0.09 | .50 | .62 |
| 3 | 2.49 [2.15, 2.83] | 1.57 | -0.49 | -1.34 | .61 | .65 |
| 4 | 2.75 [2.44, 3.06] | 1.42 | -0.72 | -0.92 | .64 | .71 |
| 5 | 2.88 [2.60, 3.17] | 1.32 | -0.85 | 0.57 | .63 | .69 |
| 6 | 3.50 [3.30, 3.70] | 0.94 | -2.12 | 3.97 | .66 | .80 |
| 7 | 3.21 [2.99, 3.43] | 1.03 | -1.21 | 0.56 | .61 | .69 |
| 8 | 1.97 [1.60, 2.35] | 1.72 | -0.04 | -1.72 | .43 | .46 |
| 9 | 2.37 [2.07, 2.67] | 1.38 | -0.22 | -1.28 | .62 | .64 |
| 10 | 3.10 [2.86, 3.35] | 1.12 | -1.08 | 0.17 | .63 | .71 |
| 11 | 2.83 [2.55, 3.10] | 1.25 | -0.78 | -0.59 | .66 | .68 |
| 12 | 3.00 [2.72, 3.28] | 1.31 | -1.15 | 0.09 | .38 | .49 |
| 13 | 2.57 [2.26, 2.89] | 1.44 | -0.56 | -1.11 | .53 | .59 |
| 14 | 3.80 [3.67, 3.92] | 0.58 | -3.79 | 16.87 | .37 | .63 |
| 15 | 3.20 [2.95, 3.44] | 1.11 | -1.29 | 0.65 | .69 | .78 |
| 16 | 2.51 [2.22, 2.80] | 1.32 | -0.29 | -1.27 | .45 | .50 |
| 17 | 3.28 [3.07, 3.50] | 0.99 | -1.33 | 1.00 | .62 | .73 |
| 18 | 2.34 [2.02,2.66] | 1.49 | -0.24 | -1.40 | .58 | .61 |
| 19 | 3.74 [3.59, 3.88] | 0.67 | -3.11 | 10.59 | .48 | .70 |
| 20 | 3.46 [3.25, 3.66] | 0.94 | -1.68 | 1.83 | .64 | .77 |
| 21 | 2.83 [2.55, 3.11] | 1.29 | -0.80 | -0.51 | .42 | .48 |
| 22 | 3.80 [3.66, 3.93] | 0.63 | -3.72 | 14.73 | .42 | .67 |
| 23 | 3.12 [2.83, 3.40] | 1.31 | -1.20 | -0.06 | .65 | .73 |
| 24 | 2.68 [2.37, 2.99] | 1.42 | -0.55 | -1.15 | .43 | .49 |
| 25 | 2.88 [2.60, 3.15] | 1.26 | -0.85 | -0.44 | .61 | .69 |
| 26 | 2.27 [1.95, 2.59] | 1.48 | -0.18 | -1.38 | .70 | .70 |
| 27 | 3.31 [3.07, 3.56] | 1.13 | -1.48 | 0.97 | .67 | .78 |
| 28 | 2.98 [2.70, 3.26] | 1.27 | -1.04 | -0.12 | .70 | .76 |
| 29 | 2.07 [1.74, 2.41] | 1.54 | 0.01 | -1.47 | .50 | .53 |
| 30 | 3.21 [2.95, 3.47] | 1.20 | -1.36 | 0.61 | .56 | .64 |
| 31 | 3.29 [3.05, 3.53] | 1.10 | -1.46 | 1.05 | .67 | .76 |
Note. n=138; rcix=corrected item-total correlation score; λi1=item's factor loadings.
Table 1 shows the descriptive properties of the 31 items on the scale, calculated from the responses of the sample in the present study. The table reveals that most of the items presented negative skewness, and that the corrected item-total correlation was higher than the .30 criterion in all cases. Table 2 shows the number of items, theoretical range, mean, standard deviation, and alpha coefficient for the total score and for the theoretical components of psychological abuse represented by the PAEGS.
Descriptive Statistics and Alpha Coefficient for the PAEGS.
| Measure | Items | Range | M [95% CI] | SD | α | rcdx |
| PAEGS | 31 | 0 - 124 | 91.48 [87.61, 95.35] | 22.99 | .94 | - |
| Emotional abuse | 6 | 0 – 24 | 18.24 [17.31, 19.17] | 5.53 | .83 | .80 |
| Isolation | 6 | 0 – 24 | 15.09 [14.03, 16.14] | 6.25 | .83 | .76 |
| Control of information | 4 | 0 – 16 | 11.02 [10.39, 11.66] | 3.78 | .67 | .56 |
| Control over personal life | 6 | 0 – 24 | 16.79 [15.81, 17.73] | 5.72 | .81 | .80 |
| Indoctrination | 6 | 0 - 24 | 19.03 [18.20, 19.86] | 4.94 | .83 | .81 |
| Imposition of authority | 3 | 0 – 12 | 11.33 [11.07, 11.60] | 1.57 | .78 | .50 |
Note. n=138; range=theoretical range; rcdx=corrected dimension-total correlation score.
Comparing the average of the means of the items included in each component, which have a possible range from 0 to 4, the component with the highest frequency was imposition of a single and extraordinary authority (3.78), followed by indoctrination within an absolute and Manichean belief system (3.17), and emotional abuse (3.04). The components directed toward the immediate context of the affected people, i.e., control over personal life (2.8), control and manipulation of information (2.76), and isolation (2.51) were experienced with less frequency in this sample.
The Cronbach's alpha coefficient obtained by the PAEGS as a whole was .94. Regarding the group psychological abuse components represented in the scale, all the coefficient values were higher than the common criterion value of .70 (Nunnally & Bernstein, 1994), except control and manipulation of information, which was composed of 4 items and had an alpha of .67.
Relationship with Group AbusivenessIn order to provide evidence of convergent validity, the relationship between the experienced psychological abuse measured by the PAEGS and the extent of group abusiveness measured by the GPA-M was examined. Bivariate correlation analyses were conducted, and, as expected, the correlation between the scores on the two measures was positive and significant (rs=.80). The dimensions of the GPA-M with the highest correlations with the PAEGS were Compliance (rs=.75) and Anxious-Dependency (rs=.56), followed by Exploitation (rs=.41) and Mind Control (rs=.40). Regarding the degree of group abusiveness, the mean score on the GPA-M was 104.87, exceeding the cut-off value of 74 proposed by Almendros (2006).
Relationship with Psychological DistressIn order to examine the relationship between the experienced psychological abuse and psychopathological symptoms and self-esteem in the study sample, bivariate correlation analyses were conducted between the PAEGS and the dimensions of the SCL-90-R and the RSES. As Table 3 reveals, no significant correlations were found in any case. Regarding the degree of psychological distress, the mean scores on the SCL-90-R were higher than those found in a normative sample (e.g., Derogatis, 1983), and the mean score on the RSES was lower than the one found in another normative sample (Schmitt & Allik, 2005). Thus, following the criterion values provided by Cohen (1988), the scores on the dimensions of Depression, Interpersonal Sensitivity, Psychoticism, Obsessive-Compulsive, Anxiety, and Paranoia Ideation had a difference with a large magnitude between the sample in the present study and the normative sample (d>0.80). The same large magnitude was found with the GSI and the RSES. On the other hand, the scores on the dimensions of Somatization, Hostility, and Phobic-Anxiety show a difference between samples with a medium magnitude (d>0.50).
Descriptive Statistics, α Coefficient, Cohen's d and Correlation with PAEGS.
| Measure | M [95% CI] | SD | α | rs | d |
| Somatization | 0.68 [0.55, 0.81] | 0.75 | .91 | .08 | 0.68 |
| Obsessive-Compulsive | 1.09 [0.93, 1.24] | 0.88 | .92 | -.11 | 1.35 |
| Interpersonal Sensitivity | 0.98 [0.83, 1.13] | 0.84 | .90 | -.06 | 1.49 |
| Depression | 1.16 [1.00, 1.33] | 0.95 | .94 | -.14 | 1.53 |
| Anxiety | 0.80 [0.65, 0.95] | 0.86 | .93 | -.07 | 1.10 |
| Hostility | 0.60 [0.47, 0.73] | 0.73 | .87 | -.12 | 0.67 |
| Phobic Anxiety | 0.36 [0.26, 0.46] | 0.58 | .84 | -.01 | 0.65 |
| Paranoia Ideation | 0.89 [0.74, 1.04] | 0.84 | .84 | -.08 | 1.09 |
| Psychoticism | 0.63 [0.50, 0.75] | 0.70 | .87 | .03 | 1.47 |
| GSI | 0.85 [0.72, 0.97] | 0.71 | .98 | -.06 | 1.43 |
| RSES | 26.65 [25.09, 28.22] | 8.93 | .91 | .08 | -1.06 |
Note. n=127; rs=correlation with the PAEGS total score, d=Cohen's d comparing the study sample with normative samples.
p<.1, * p<.05, ** p<.01, *** p<.001
No significant differences were found in the PAEGS scores based on the main sociodemographic variables. In the same way, no significant correlations were found with social desirability, time spent in the group, time elapsed since the subjects left the group, method of departure, or contact with cult-awareness organizations. The only significant difference was found in relation to the counseling received after leaving the group (Mann Whitney U=1001.5, Z=2.923, p=.003). Results point out that subjects who sought counseling experienced group psychological abuse to a greater extent (M=95.07, SD=20.17) than those who did not seek help (M=78.69, SD=20.51).
DiscussionThe purpose of this study was to examine the psychometric properties of the measurements provided by the PAEGS using a sample of self-identified former members of abusive groups from Anglo-Saxon countries. This new instrument is intended to assess the degree of group psychological abuse experienced by individuals, focusing on the two main elements that characterize the phenomenon, i.e., the abusive nature of the behaviors and their continued duration (Rodríguez-Carballeira et al., 2015). The PAEGS is based on a model of group psychological abuse that covers a wide range of abusive behaviors, from the most obvious to the most subtle strategies (Rodríguez-Carballeira et al., 2015). The use of a comprehensive semantic definition of the phenomenon as a starting point was a key aspect in ensuring the content validity of the instrument (Barbero, 2006; Haynes, Richard, & Kubany, 1995; Nunnally & Bernstein, 1994). Thus, some of the limitations of previous measures related to the specificity and representativeness of the group psychological abuse behaviors domain have been overcome. In this regard, the items on the PAEGS ask for specific psychologically abusive behaviors experienced by the individual, rather than for sexual abuse practices, physical abuse behaviors, or the purposes of the group, topics found on some items of previous instruments (e.g., Chambers et al., 1994; Winocur et al., 1997).
Regarding dimensionality, the PAEGS presents a one-dimensional factor structure with a factor weight above .40 in all items. The extracted factor was labeled Group Psychological Abuse Experienced. This result revealed that the different group psychological abuse strategies can be considered components of a single phenomenon. Thus, people who suffer psychological abuse usually experience abusive strategies of different nature simultaneously. Nevertheless, we have provided descriptive data for the six group psychological abuse components represented in the scale, so that they can be taken into account in both future research and the applied field.
Focusing on internal consistency, we can conclude that the PAEGS is a reliable instrument. The overall Cronbach's α exceeds the recommended criterion value of .85 when working with psychometric instruments in the clinical setting to make decisions about specific people (Pfeiffer, Heslin, & Jones, 1975). Furthermore, the values of the items’ discrimination indexes are generally very high, exceeding in all cases the conventional limit of .30 (Nunnally & Bernstein, 1994).
Regarding the relationship between the abusive components’ frequency and their relative severity according to experts reported in Rodríguez-Carballeira et al. (2015), it is interesting to note that the least severe component (i.e., imposition of a single and extraordinary authority) had the highest frequency in the present study. Meanwhile, the components directed toward the person's immediate context, especially isolation, were the most severe, after emotional abuse, according to the experts, but the least frequent in this study. This result can be explained by the difficulty of keeping constant control over the person's contextual elements. As the dispersion statistics for the PAEGS components show, the degree to which the control of the context is achieved is quite variable and heterogeneous across different abusive groups. On the other hand, the imposition of a single and extraordinary authority might be the abusive component that best characterizes abusive groups, as it is applied in most groups in a homogenous way.
Focusing on convergent validity, the correlation between the PAEGS and the GPA-M was high and positive. This result suggests that people who experienced a high extent of psychological abuse strategies also describe the group to which they belonged as abusive (Chambers et al., 1994). Although the four dimensions of the GPA-M correlated significantly with the PAEGS, results show that correlations with both Exploitation and Mind Control were lower. The lower correlation with Exploitation may be due to including some items that were not necessarily relevant to the participant's own experience of psychological abuse, such as those evaluating group objectives like raising money, gaining political power, or recruiting new members. In the case of Mind Control, the lower correlation may be due to a limitation of the GPA-M that has already been pointed out, i.e., the use of ambiguous terms in some items, such as “psychological pressure”, “mind control”, or “coercive persuasion”.
Prior studies found that the perception of group abusiveness is not related to the main socio-demographic variables, method of departure, or contact with cult-awareness organizations (e.g., Almendros et al., 2009; Almendros et al., 2012). Our results also indicate that individuals’ experiences of group psychological abuse do not differ based on the aforementioned sociodemographic and biographical variables. In this regard, our findings add evidence that contradicts the assertion of authors who suggest the existence of a negative bias when people who left the group after being counseled or people who have been in contact with cult-awareness organizations assess their past group experiences (Lewis, 1986).
The results indicate that former members who sought psychological counseling after leaving the group, which was not necessarily provided by professionals with expertise in abusive groups, reported experiencing higher levels of group psychological abuse. This finding suggests that having continuously experienced abusive behaviors could lead people to seek counseling for post-involvement problems more often. These difficulties are not necessarily related to psychopathological symptoms (Baron, 2000; Chambers et al., 1994; Winocur et al., 1997), but rather to difficulties in establishing new social relationships, cognitive deficits, or intense feelings of loss, anger, and blame (Aronoff et al., 2000).
Although a high percentage of participants in the present study reported having sought psychological counseling in relation to their past group experiences, their degree of current distress was not significantly related to the extent of the psychological abuse experienced. One possible explanation would involve the time elapsed between leaving the group and participating in the study, and the possible influence of certain moderating variables, such as social support, coping strategies, or stressful life events (Almendros et al., 2011). Another explanation would be related to the high negative asymmetry and lack of dispersion of the distribution of the scores on the PAEGS in the present sample. Thus, it is possible that after reaching a threshold in psychological abuse experienced, the resulting psychological distress stabilizes at high magnitudes and stops increasing linearly. Future studies, using longitudinal designs and comparison samples, should shed light to this possible explanation.
The magnitude of this distress is seen when comparing the responses of former members of abusive groups with those of normative samples. Participants in the current sample reported greater psychological distress symptoms than those reported by a non-patient sample, especially in terms of depression, interpersonal sensitivity, psychoticism, and obsessive-compulsive symptoms. In the same vein, participants reported a lower level of self-esteem than in another community sample. These results are consistent with those reported in previous studies (e.g., Aronoff et al., 2000; Malinoski et al., 1999; Winocur et al., 1997).
LimitationsIn the present study, the representativeness of the sample could not be verified, a common issue in this field of study, due the difficulty of accessing former members of abusive groups, who may be considered a hard-to-reach population (Shaghaghi, Bhopal, & Sheikh, 2011). This makes it almost impossible to use probabilistic sampling methods and usually leads to small sample sizes (e.g., Almendros et al., 2012; Nishida & Kuroda, 2004; Wolfson, 2002). However, the sample in this study overcomes problems identified by some authors (Lewis, 1986; Lewis & Bromley, 1987; Wright, 1984) because it is a mixed sample in terms of the method of departure, it was not collected only through cult-awareness organizations, and it is not a clinical sample composed only of people in treatment with researchers or related professionals.
A second limitation is related to the retrospective evaluation of experienced abuses, which is also common in instruments designed to assess psychological abuse in the different areas where it is applied (e.g., Porrúa-García et al., 2016; Ureña, Romera, Casas, Viejo, & Ortega-Ruiz, 2015). However, the items on the PAEGS describe specific behaviors performed within the group, carefully avoiding subjective interpretations by participants that could lead to biased responses.
Research ImplicationsIn the research setting, the development of the PAEGS makes a relevant contribution to the study of abusive group dynamics, which is characterized by having few valid instruments compared to other fields where psychological abuse is assessed (Almendros et al., 2011). Likewise, the PAEGS’ approach to the measurement of the phenomenon, focusing on individual experiences rather than on the degree of group abusiveness, could be particularly useful for researchers and professionals interested in the degree of group psychological abuse suffered by a particular person. In this regard, future research designed to assess the consequences of belonging to a group considered abusive would be expected to exceed previous studies, where conclusions were drawn without using standardized instruments to assess the abuses suffered by the participants (e.g., Buxant, Saroglou, Casalfiore, & Christians, 2007).
Future studies are needed to continue to examine the psychometric properties of the PAEGS with larger samples. Moreover, proper comparison samples are needed to examine the diagnostic validity of the scale and analyze the relationship between group psychological abuse and the psychological and social problems usually reported by former members of abusive groups (e.g., Matthews & Salazar, 2014). In this direction, it would be desirable to incorporate measures of non-pathological psychological difficulties and moderating variables that may influence this relationship. Finally, it would also be necessary to adapt the instrument to populations from other cultural contexts and to former members of other groups, such as gangs or terrorists, where research has shown similarities in the abuses experienced (e.g., Trujillo, Ramírez, & Alonso, 2009). These studies would increase our knowledge about an area with great social relevance, but incipient scientific development (Almendros et al., 2011).
Clinical and Policy ImplicationsIn the clinical setting, the use of the PAEGS could improve the initial approach to the extent and nature of the abusive practices experienced by affected people. It may lead to more accurate assessment and better planning of the counseling process where experiences of abuse are reexamined (Coates, 2010; Matthews & Salazar, 2014).
In the legal setting, the use of reliable and validated measures provides more accurate expert appraisals. Thus, the PAEGS will make it possible to obtain evidence of abusive behaviors experienced within a group and discriminate them from other minority, but not abusive, practices (Ordeñana, 2001). Experts will have the responsibility of evaluating the possible relationships between the PAEGS scores and the scores on other instruments intended to assess psychological distress, using it as a tool to facilitate more informed judgments about causality that would complement forensic-clinical interviews (Vilariño, Arce, & Fariña, 2013).
It is important to note that results obtained using the PAEGS should not be generalized to all the members of a group and should not lead to labelling a group as abusive or non-abusive. As noted above, individual experiences within a group might be different (Barker, 1989), and the same group may vary in their practices, group dynamics, and the behaviors of the authority figures, both over time and in the different locations where it is present (Almendros et al., 2011). However, a large number of former members of the same group reporting continuous experiences of abuse would merit accurate legal investigation.
Conflict of InterestThe authors of this article declare no conflict of interest.
Financial SupportThis research was supported by grants from the Ministry of Economy and Competitiveness (Spain) (PSI2013-41865-P and PSI2016-75915-P).
Psychological Abuse Experienced in Groups Scale (PAEGS)
| Instructions: Now you’ll find a series of statements about practices you may have experienced in the group. What we want to know is whether you now consider that you went through what is described in each sentence, regardless of whether or not the group performed it intentionally. Please indicate the degree to which you have experienced each practice by selecting an option from the following scale: 0=not at all, 1=slightly, 2=quite a lot, 3=a lot, 4=continually | |
| 1 | They made me reject my life prior to becoming a member of the group. |
| 2 | They made me feel that I would face considerable danger if I left the group. |
| 3 | Before establishing an intimate relationship, I had to have the group's approval. |
| 4 | I was expected to confess any action or feeling that might not have conformed to the philosophy of the group. |
| 5 | They impressed upon me the idea that outside the group everything was reprehensible. |
| 6 | Behaving in accordance with the ideology of the group should be more important to me than my own self. |
| 7 | If I questioned any aspect of the group's philosophy, I was always discredited. |
| 8 | I was compelled to go and live with other members of the group. |
| 9 | They tried to pull me away from my family. |
| 10 | They made me feel guilty about small and unimportant things. |
| 11 | They made me give up my free time activities. |
| 12 | I had to use a special group jargon that altered the normal meaning of words. |
| 13 | I frequently reached the point of exhaustion due to the number of group activities I was involved in. |
| 14 | The authority of the group's leader was presented to me as unquestionable. |
| 15 | The group tried to control the way I spent my time. |
| 16 | They taught me to deceive other people with regard to specific aspects of the group. |
| 17 | They led me to give up aspects of my identity and way of being in order to adapt to those of the group. |
| 18 | They encouraged me to give up my studies or my work. |
| 19 | They made me see the figure of the leader as an indisputable authority who had to be obeyed. |
| 20 | The group wanted me to be ready to make big sacrifices. |
| 21 | They concealed relevant information from me about who they were and what they were really doing. |
| 22 | They made me see the leader of the group as someone with very special and clearly superior characteristics. |
| 23 | The same authority that punished me could withdraw the punishment and forgive me at will. |
| 24 | They lied to me about the real nature of the group. |
| 25 | When they considered that I had disobeyed some of the group's recommendations, they treated me with contempt. |
| 26 | They made me give up the friends I had before joining. |
| 27 | They tried to make me spend as much time as possible with the group. |
| 28 | They kept a constant watch over my behavior. |
| 29 | They tried to make me explain every aspect of my economic situation to them. |
| 30 | They led me to believe that everything inside the group was good and everything outside the group was bad. |
| 31 | They made me fear I would be discredited or humiliated in some way if I didn’t do what was expected of me. |
Relation between items and components of group psychological abuse: emotional abuse (2, 4, 10, 23, 25, 31); isolation (8, 9, 11, 18, 26, 27); control and manipulation of information (12, 16, 21, 24); control over personal life (3, 13, 15, 20, 28, 29); indoctrination in an absolute and Manichean belief system (1, 5, 6, 7, 17, 30); imposition of a single and extraordinary authority (14, 19, 22).
