Implicit association Self-concept test in studying of violence-related cognitions

Introduction

The problem of studying the mechanisms of formation and change of social attitudes of different social groups of the society is actual. For example, attitudes towards violence.

Violence is defined as the most severe type of physical or nonphysical aggression that is likely to cause serious physical or psychological harm [1]. Violence is a sub­type of aggression, generally used to denote extreme forms of aggression such as murder, rape and assault. Criminal violence is understood as forms of violence that is against the law [16]. It is important to study associations of violence-related cognitions as they play an important role in a wide variety of violent behaviors.

Implicit cognitions are understood as attitudes and motivations that an individual holds, but is not aware of. A particular cognition may exist as implicit cognition at some times and explicit cognition at others, however, some implicit cognitions may never become explicit cognitions [8; 19].

Studies of cognitions related to violence may help explain violent actions and invest in predicting violent behavior [3]. Polas- chek, Calvert, & Gannon [12] by two-part mixed design study tried to explain violence- supportive cognitions (often viewed as distorted and deviant beliefs and desires) of violent male convicts. These beliefs guide the behavior, are stable and resistant to change.

Up to date, violence-related cognitions are mainly measured by explicit measurement methods — directly asking the individual to reflect and rate one’s attitudes via interview or questionnaire. The big issue is social desirability especially when dealing with the sample of violent offenders, who are reluctant to provide answers on sensitive issues as attitudes towards violence and researchers usually blame social desirability when scores of self-report measures do not provide the desired results [13]. Many cognitive processes are not available to introspection. The activation of the attitude process in memory may require effort and can be either controlled or automatic. The first processes are deliberate, conscious and slow; the second processes are fast, automatic and unconscious. In the first case, we are talking about an explicit attitude (can be assessed by explicit measurement methods), and in the second case, about an implicit attitude (needs to be assessed by implicit measurement methods).

Most of the known studies use standard two-category IAT for assessing attitudes towards aggression and violence, including target categories as “violent” and “non-violent” and attributes as “pleasant” and “unpleasant” (e.g., [18]).

The study by Blumenthal and colleagues [2] argued that valence IAT (violence — good or bad) is not the only dimension violence could be assessed in. They designed three IAT procedures — valence IAT (good vs bad), hedonic-value IAT (enjoy vs dislike) and arousal IAT (exciting vs boring) with the target categories (images) of “Violent” and “Peaceful” researching attitudes towards violence of convicted violent male prisoners, groups: murderers and non-murderers. Proactive and reactive aggression questionnaire (RPQ) was used as a self­report procedure and previous convictions were also taken into account. The results revealed that valence-IAT and hedonic val- ue-IAT showed acceptable levels of reliability, thus proved to be reliable measure for assessing violence-related cognitions. He- donic-IAT correlated with measures of the general criminality level (previous convictions) in non-homicide group and arousal- IAT had a relationship to the self-reported aggression level in the non-homicide group. It was found that non-homicide group had more convictions and higher level of self-reported aggression than the homicide group.

There are several different other IAT designs that could be used in assessment of associations of violence-related cognitions. Polaschek et al. [13] additionally to a classic two-category IAT (Violence-Housework IAT) used Personalized Weapons-Enter- tainment/Like—Dislike IAT for measuring attitudes towards violence of violent male convicts before and after 9 months Cognitive-behavioral therapy. Changes were found consistent with a treatment effect according to the gained results of Weapons-Entertainment IAT, but not on the Violence-Housework IAT. Polaschek and colleagues explained that the two IATs differed in many ways — the Weapons- Entertainment IAT used pictorial stimuli as target categories and it was personalized, while the Violence-Housework used all word stimuli, and used traditional evaluations of ’good’ and ’bad’. Other authors have suggested that pictorial stimuli [9; 20] and personalized IATs [10] could increase external validity.

Self-Concept IAT was used in a study done by Richetin, Richardson and Mason [15] with the aim at examining the validity of implicit measures of aggressiveness 178

and predicting aggressive behavior in a situation with or without provocation. Two Self-Concept IATs were constructed for measuring direct and indirect aggressiveness. The target concepts were verbal stimuli Harmful and Harmless and the attribute categories were Self and Other. The results showed that implicitly measured aggressiveness predicted the aggressive behavior, meaning that participants, who showed pro-aggression associations reacted in a more aggressive manner than participants, who did not show pro-aggression associations. Self-report procedures did not show significant correlation with aggressive behavior, meaning that in this case, only implicit measures predicted aggressive behavior.

As the researchers examined different components of violence related cognitions, and used different versions of experimental IAT procedures, it is difficult to compare the results by common criteria. Also the researches differed in designs and samples used in the studies — students, violent male convicts either with or without known psychopathology.

The researchers underline that the correlation between implicit and self-report procedures of psychological construct is not only connected with the problem of the reliability of self-assessment procedures, but also with the need to account for those variables (conditions, sample, context, personality variables etc.) that effect on the consistency of measurements [5].

Mainly the researches are dedicated to the study of compliance of implicit measurements and self-report procedures with antisocial attitudes. However to our knowledge, there are not any researches so far investigating the relationships between implicit measurements of violence and personality traits. Although there are enough studies of relationship between explicitly measured violence and personality traits. It has been found that certain personality traits are connected to antisocial behavior. Eysenck’s on temperament based PEN (Psychoticism, Extraversion, Neuroticism) model reflects the influence of bio-social factors on the development of antisocial behavior, and it states that the criminals should have to provide high scores in all three dimensions of personality [4]. Some studies at least partly support hypothesis of Eysenck, for example, a research measured the influence of different risk factors on violent crime [14].

The aim of research was to investigate violence-related cognitions with Self-Concept IAT and self-report procedures for sentenced and not-sentenced males.

The research questions: (1) Is there correspondence between the results of measuring implicit associations related to criminal violence using Self-Concept IAT and self­reported criminal attitudes towards violence among sentenced and not sentenced males? (2) Is there a relationship between implicit associations of self with criminal violence measured by Self-Concept IAT with personality traits: psychoticism, neuroticism and extraversion? (3) Is there a relationship between self-reported attitudes towards criminal violence with personality traits: psychoticism, neuroticism and extra­version?

Methods

Participants

141 males. Group “Sentenced” — 77 prisoners, 20—62 years (Mdn = 34), who were all sentenced at least once for violent crimes (murder, assault, burglary or sexual assault) and serving their sentence in high-security prisons. In sum they had 31 different criminal convictions in their lifetime.

Group “Not Sentenced” — 64 previously not sentenced, 18—62 years (Mdn = 32), who have never been sentenced for any criminal offenses and their daily job did not involve violence.

Measures

Violence Self-Concept IAT has been designed on the basis of the Self-Concept IAT by Greenwald and Farnham [7]. The procedure was created in Latvian and Russian languages, and it combined verbal target categories (self-others) and pictorial attributes (violence-peace). The attributes (violent-peaceful) were represented by 12 pictures purchased from “Shutter stock” and “Stock Illustrations”. The images were 10x15 cm and were black and white silhouettes of either an action involving physical non-sexual violence (strangle, stab, fight, hit, shoot, cut) or an action involving peaceful activity (to fish, calm, sleep, play, sail and swim). Black and white pictures were used, because after viewing countless detailed, graphic, colorful pictures of violent acts, researchers decided that pictures like that are too disturbing for the participants and it would be unethical to expose them. Peaceful pictures were picked according to actions that are peaceful and do not involve high adrenalin activities (as skiing, parachute jumping), competitive activities (e.g., team sports, racing or boxing) and women (couple activities). All of the pictorial stimuli were specially selected for a male audience.

Apparatus. Certified licensed software E-Prime 2®.

The Self-Concept IAT’s procedure. Performance of IAT took on average from 10
to 20 minutes (tab. 1). Participant’s RT was registered. Before the start of the research, a participant was given on a computer monitor the general instructions and specific instructions before each of the tasks.

Self-report procedures — linguistically adapted in Latvian and Russian:

Criminal Attitudes towards Violence Scale (CAVS) [11] (a = .89).

The Eysenck Personality Questionnaire Revised-Abbreviated (EPQR-A)” [6] was used (a = .70).

Procedure

All the participants took part in the research voluntarily, they were verbally assured that the participation in the research is anonymous. Confidentiality and anonymity was guaranteed. The research was conducted individually. Participants completed the tasks in the same order: Self­Concept IAT and self-reported procedures, which were brief and specially selected for sample with limited attention span. The prisoners were asked to write down how many times they have been sentenced previously and what criminal codes have they been sentenced under. Not sentenced males were approached in their workplace and the procedure was the same.

Results

Variables:

Group — “Sentenced” and “Not sentenced”.

D(IAT) — implicit associations of self with criminal violence. D > 0.15 — implicit associations with pro-violence, D < — 0.15 — implicit associations with non-violence, -0.15 < D < 0.15 — no effect (that means that implicit associations are either weak or conflict).

CAVs — the strength of the self-reported criminal attitude towards violence.

For the group “Sentenced”: Previously Sentenced, Previously Imprisonment.

Psychoticism. Neuroticism. Extraversion. Age.

To answer the first research question the correlation coefficients Pearson (r) and Spearman (rS) are shown in the tab. 2.

To answer the second and the third research questions, the correlation coefficients were calculated (tab. 3—4).

Discussion

Using a specially designed Self-Concept IAT, implicit associations of self with criminal non-sexual violence were measured, as well as attitudes of criminal non-sexual violence were also measured with self-reported procedure in the samples of sentenced and not sentenced males. The most part of the measured implicit associations were associations of self with non-violence in the both groups. This corresponds to previous findings of researches investigating anti-social attitudes with the help of IAT [13; 18].

For sentenced males the correspondence of results of measurements were found for the sub-groups with a low level of psychoti- cism, an average level of neuroticism and participants, who had not previously been sentenced or imprisoned. The tendency to discrepancy of the results of measurements was found for the sentenced who showed explicit attitudes towards non-violence, possibly due to social desirability.

For not sentenced males the correspondence of the results of measurements was also found for participants aged 28— 42 and for those who showed pronounced explicit attitudes towards violence. The discrepancy between the results of implicit and explicit measurements was found for participants aged 18—27. 32% of this sub-group showed implicit associations of self with pro-violence. It is known, that young age is one of the main predictors (a risk factor) of convictions for violent crime [17].

The second research question. A tendency has been found for a negative connection between neuroticism and implicit associations of self with criminal violence. For the sentenced, it is observed among participants with a moderate level of neuroticism. For the not sentenced — in the event that implicit associations of participants are either weak or conflict (tab. 3).

For both groups, in certain subgroups, implicit associations of self with violence and a high level of psychoticism correspond to each other. As well as implicit associations of self with non-violence and a low level of psychoticism correspond to each other. These conditions are in the group of sentenced of 28—42 years old and participants with conflicting or with non-ex- pressed implicit attitudes towards criminal violence. In the group of not sentenced — these are participants over 27 years old, participants with a high level of extraversion, participants with explicit attitudes towards criminal violence of either low or high level, and participants with implicit associations of themselves with non-violence.

The relationship between extraversion and implicit associations of self with criminal violence in the group of not sentenced males was not found. For sentenced participants a tendency to a positive relationship was found.

The third research question. Neuroticism is positively associated with attitudes towards criminal violence, measured by self-reporting procedures. It is possible, that high neuroticism corresponds to a pro­violence explicit attitude and vice versa. This is observed for not sentenced participants with high levels of psychoticism, with high levels of neuroticism, with high and medium extraversion, with strong implicit associations. This is also observed in subgroups of sentenced participants above middle age and with a low level of extraver­sion (tab. 4).

Psychoticism is positively associated with attitudes towards criminal violence measured by self-reporting procedures. In the framework of the Eysenck method used, this means that a high non-normative orientation corresponds to a high explicit attitudes towards violence and vice versa. A high focus on observance of norms corresponds to a high explicit attitude towards non-violence and vice versa. This effect is observed in not sentenced participants in subgroups such as: middle and above middle age, with low psychoticism, with high and low levels of neuroticism, with high and low extraversion rates, with pronounced implicit associations of self with non-violence and low levels with explicit attitude to nonviolence. For Sentenced participants, this effect is observed in subgroups of middle­aged participants and with a low level of extraversion.

Extraversion is positively associated with attitudes towards criminal violence measured by self-reported procedures. For the not sentenced, this is observed only in a subgroup of participants with an average level of neuroticism. For Sentenced, there is a positive relationship in the subgroups of participants with explicit pro-violence and a high level of psychoticism.

If we compare the relationships of personality traits with implicit associations of self with criminal violence and the relationships of personal traits with attitudes towards criminal violence measured by self­reporting procedure, then in both cases there are positive relationships with psychoticism and positive relationships with extraversion. This indicates the validity of IAT.

Conclusions

The conducted research has shown that the Self-Concept IAT measures implicit associations of self with criminal violence. There are groups of participants with specific traits for whom the results of measurements of Self-Concept IAT correspond to measurements of attitudes towards criminal violence by self-report procedures. There are also groups of participants with inconsistent measurement results. The relationship of psychoticism, neuroticism and extraversion with attitudes towards criminal violence measured by self-report procedures, is partly the same as with implicit associations of self with violence.

The limitation of this research was non­homogeneity of the sample.

In future research, it is intended to use other implicit methods to study attitudes towards criminal violence and expand the range of personality traits for the sentenced and not sentenced, including research of the possibilities for overcoming (suppressing) of associations with pro-violence.