How Can Men Convicted of Violence Against Women Feel Moral While Holding Sexist and Violent Attitudes? A Homeostatic Moral Model Based on Self-Deception.

A moral model is proposed to understand how men convicted of violence against the partner can feel moral in spite of their past violent behavior and their current violent and sexist attitudes. Because of its appeal to the role of self-deception and its relationship to psychological well-being, it was hypothesized that a rigid conception about what is right and wrong (moral absolutism) is associated with ambivalent outcomes that keep their psychological system in homeostasis. The relationships were specified a priori and tested using path analysis. Several fit indices supported the adequacy of the model and showed that moral absolutism was indirectly related to both psychological well-being and a good moral self-conceptualization through self-deception. At the same time, moral absolutism was related to sexist and violent attitudes and a poor moral self-conceptualization. Future interventions could include strategies to reduce the resistances to change based on the reduction of moral absolutism and self-deception.

Morality and violence are related topics since morality is about what is right and wrong, and intimate partner violence (IPV) is a kind of violent behavior currently considered wrong in most social systems. That does not mean that violent people assume they are immoral. Research shows that people prefer to not self-identify as violent and they are consequently motivated to invoke a variety of affectively and cognitively based strategies, such as self-deception, to feel well and moral regardless of the situation (Barkan, Ayal, Gino, & Ariely, 2012; Effron, Miller, & Monin, 2012; Monin & Jordan, 2009; Sun, Gu, Wu, Wang, & Jin, 2012; Triandis, 2011; Trivers, 2000). In this study, we explore the kind of psychological balance that allows men convicted of violence against the partner to feel well and moral, in spite of their past violent behavior and their current violent and sexist attitudes, and propose a Homeostatic Moral Model (HOMM) in which their moral absolutism plays an ambivalent role, one directly negative on sexist and violent attitudes, and other indirect and positive ones on well-being and moral self-concept through self-deception.

A Case of Motivated Moral Reasoning in Men Convicted of Violence Against the Partners

Although people may use moral reasoning to make decisions about what they and others should do, they also may use moral reasoning to achieve more instrumental ends, such as exerting social influence, exploiting others, and justifying immoral behavior (Ditto, Pizarro, & Tannenbaum, 2009). Most of these ends are intended to be explicitly pursued or inadvertently used by men who resort to this kind of violence, which could be framed as a case of motivated moral reasoning if, as a result, they maintain an immaculate moral self-conceptualization.

This theoretical framing is possible now that the role of affect in moral reasoning is better understood and a prominent focus of research is on moral psychology (Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Haidt, 2001; Krebs & Denton, 2005; Vecina, Marzana, & Paruzel-Czachura, 2015). Current research, in fact, has moved away from considering moral judgment as an essentially rational process (Kohlberg, 1984) to increasingly emphasizing the role of affect and intuition in arriving at moral judgments (Kunda, 1990; Merritt et al., 2012; Monin & Jordan, 2009; Uhlmann, Pizarro, Tannenbaum, & Ditto, 2009). An example of this is the Social Intuitionist Model (Haidt, 2001; Haidt & Bjorklund, 2008; Haidt & Joseph, 2008), which assumes that moral judgments come to mind automatically—not by applying standards of conduct derived from group norms but as the result of one’s intuition regarding an issue. The arguments that are subsequently offered in support of these judgments are usually only rationalizations or post hoc justifications for feeling states.

More specifically, we can appeal to different theories to frame the kind of incoherence in which we hypothesize that violent men live in. In this respect, moral rationalization is a cognitive process that individuals use to convince themselves that their behavior does not violate their moral standards (Tsang, 2002). Such a process can be necessary, as the theory of Motivated Reasoning suggests (Kunda, 1990), because people attempt to be rational and to construct a justification of their desired conclusion that would persuade a dispassionate observer (p. 482). However, moral rationalization can also be necessary when a person’s intelligence and integrity are called into question. The theory of self-affirmation argues that individuals are motivated to reaffirm themselves as good and predicts strong reactions to anything that threatens the sense of self-integrity (Steele, 1988; Steele & Liu, 1983).

Both objectives (appearing to be rational and preserving self-integrity) also are relevant for men convicted of violence against the partners. Although some of them can assume responsibility, feel remorse, and be willing to change, most feel a strong need to defend themselves, their beliefs, and their moral integrity. In addition, we can appeal to the Integrated Theory of Moral Conviction (Mullen & Skitka, 2006; Skitka, Bauman, & Mullen, 2008; Skitka & Mullen, 2008) to understand how moral convictions can be held with strong certainty or absolutism and inspire defensive actions without reasoning. This constitutes another example of motivated moral reasoning—in this case, the willingness to justify violent means to achieve the preferred ends (Skitka, 2002).

Consistently with these theoretical approaches, it has been found that men who have used violence against their own partners feel themselves highly moral persons and present an extraordinary high regard for not harming people or animals and treating them fairly (Vecina & Chacón, 2016). Taking into account this paradoxical evidence, we ask ourselves what kind of psychological balance allows for this possibility.

The Ambivalent Outcomes of Moral Absolutism and the Role of Self-Deception

Moral absolutism is understood to be the degree to which people believe that their own definition of morality is objectively correct (Peterson, Smith, Tannenbaum, & Shaw, 2009). The higher individuals score in terms of moral absolutism, the more they endorse the idea that morals should not vary by culture, person, or situation because they are “true” and are the only correct set to be adopted. Moral absolutism can stem from epistemic motivations for certainty that offer individuals a pleasant sense of security but can also be associated with violent answers (Triandis, 2009, 2011) because diversity is everywhere, including the moral domain, and diversity creates uncertainty and threats that humans are generally motivated to resolve (Haidt, Rosenberg, & Hom, 2003). In this respect, it has been found that moral absolutism strengthens the relationship between religiosity and support for violent warfare (Shaw, Quezada, & Zárate, 2011).

The confrontation between a narrow and rigid vision of morality (moral absolutism) and a wide variety of moral practices can have a negative impact on well-being. A large discrepancy would imply conflict and threat to the self that may engender a motivation to defend and fight for whatever is believed (Baumeister, Smart, & Boden, 1996; Bushman & Baumeister, 1998), even without reasoning (Haidt, 2001). However, such a narrow vision of morality may have other positive effects on well-being and moral self-concept through self-deception. In this respect, self-deception is an important component of well-being (Baumeister, 1989; Diener, Suh, Lucas, & Smith, 1999; Paulhus, 1991; Robinson, Moeller, & Goetz, 2009; Taylor & Brown, 1994) and it is related to having a positive moral identity (Mazar, Amir, & Ariely, 2008; Monin & Jordan, 2009). That is why it has been considered a defensive strategy that is adopted by individuals who have difficulty coping with a threatening world (von Hippel & Trivers, 2011). Thanks to self-deception, people can perceive their self in an unrealistically favorable light and protect their beliefs and desires from a conflicting reality (Vecina, Chacon, & Pérez-Viejo, 2016; Vecina & Chacón, 2016), which is what we assume men convicted of violence against their partners face when they begin a prescribed psychological treatment.

Objective and Hypotheses

The objective of this study is to explore the kind of psychological balance that allows men convicted of violence against the partner to feel well and moral in spite of their past violent behavior and their current violent and sexist attitudes. In this respect, it is hypothesized that self-deception will help to keep their psychological system in homeostasis without apparent contradictions between the moral individuals they feel they are, and their restricted vision about what is right and wrong in general, and their sexist and violent attitudes against women in particular. The specific relationships that will be tested and its rationales are the following:

Moral absolutism will be positively related to sexist and violent attitudes toward women in men convicted of violence against the partner. High certainty about one’s own and narrow way of perceiving morality (moral absolutism) would be incompatible with an open and broad view of women and perfectly compatible with the justification of violence, especially if we take into account that whatever is believed in the moral domain is experienced as a unique combination of factual belief, compelling motive, and justification for action (Mullen & Skitka, 2006; Shaw et al., 2011; Skitka, 2010; Skitka et al., 2008; Skitka & Mullen, 2002).

There will be negative relationships between pro-violent and sexist attitudes and the moral self-concept because a prejudicial conception about women implies confrontation with the diverse and changing reality of women and conflict with laws that condemn violence and sexism. It is worth noting that the measure of ambivalent sexism itself includes an ambivalent mechanism to resolve such confrontation. Its two components, hostile and benevolent sexism, are positively related and serve to justify gender inequality (Glick & Fiske, 1996, 2001; Yamawaki, Ostenson, & Brown, 2009). In other words, just because sexist people “adore” women in their more restricted role of mother (benevolent sexism), they think they are not sexist. But the fact is that their benevolent sexism is high and positively related to their hostile sexism. This possibility would nullify the negative relationship between ambivalent sexism and moral self-concept.

There will be a positive relationship between moral absolutism and self-deception, and between self-deception and psychological well-being. Men convicted of domestic violence will need self-deception to face a prescribed psychological treatment intending to change what they did wrong. Then, self-deception will protect their beliefs and desires from the conflicting reality by increasing their well-being (Diener et al., 1999; Robinson et al., 2009; Taylor & Brown, 1994; von Hippel & Trivers, 2011).

Both, self-deception and psychological well-being, will have positive relationships with moral self-concept because feeling good is part of feeling well and individuals are motivated to reaffirm themselves as good (Aquino & Reed, 2002; Jordan & Monin, 2008; Jordan, Mullen, & Murnighan, 2011; Mazar et al., 2008; Merritt et al., 2012; Monin & Jordan, 2009; Steele, 1988; Steele & Liu, 1983).

Method

Participants

The participants comprised 410 males convicted of domestic violence offenses who had begun court-mandated psychological treatment that lasted for 16 weeks. During the second session, they were asked to voluntarily participate in a research project under anonymous conditions. All accepted, although we eliminated seven subjects who submitted incomplete questionnaires. None had psychiatric disorders diagnosed—otherwise they would not have been admitted into the treatment program—but we eliminated 16 participants who had been diagnosed with mental disorders in the past (mostly anxiety and depression). A post hoc analysis was conducted with all subjects to confirm the results.

The final number of participants was 387. Most were Spanish (55%), followed by Latin American (31%), Eastern European (8%), and North African (3%). The average age was 38 years (SD = 10.93), with ages ranging from 19 to 84 years. Most of the participants had an intermediate level of education (55%); 20% had a university degree, and only 1% had no education.

Instruments

Moral absolutism

The Moral Absolutism Scale was used, which contains six items and measures the degree to which participants believe their own moral values are reflections of an objective moral landscape (Peterson et al., 2009; Vecina et al., 2016). Participants were asked to rate the degree to which they agreed with six statements on a scale from 1 (strongly disagree) to 6 (strongly agree). Some sample items are “Right and wrong are not usually a simple matter of black and white,” “There are many shades of gray” (Reverse), and “There is really only one proper way to think and behave morally.” The observed Cronbach’s α was .718.

Self-deception

The Self-Deceptive Enhancement subscale from the Balanced Inventory of Desirable Responding was used (Paulhus, 1991; Vecina et al., 2016). Twenty items measure sincere beliefs of desirable self-descriptions on a 7-point scale, ranging from 0 (completely false) to 6 (completely true; e.g., “I never regret my decisions,” “I am fully in control of my own fate,” “I am a completely rational person”). The observed Cronbach’s α was .831.

Psychological well-being

The Self-Acceptance subscale from the Psychological Well-Being Scales was used (Ryff & Keyes, 1995). It constitutes a eudemonic measure of psychological well-being. High scores signify that the person possesses a positive attitude toward himself or herself; acknowledges and accepts multiple aspects of himself or herself, including good and bad qualities; and feels positively about life to date. The response scale was a 5-point continuum, ranging from 0 (completely disagree) to 4 (completely agree). The observed Cronbach’s α was .750.

Moral self-conceptualization

The Morality Subscale of the Six-Factor Self-Concept Scale was used (Stake, 1994), which contains six adjectives related to morality (i.e., loyal, truthful, law-abiding, faithful, trustworthy, and honest). Participants indicated on a 5-point scale how accurately each adjective described them from 0 (never true of me) to 4 (always true of me). The Cronbach’s α was .849.

Sexist attitudes

The Ambivalent Sexism Inventory was used (Glick & Fiske, 1996). Twenty-two items (scale from 0 (totally agree) to 5 (totally disagree) are divided into two subscales: Hostile Sexism and Benevolent Sexism. The first subscale measures conventional prejudice against women, which characterizes them as inferior and legitimizes male social control. Benevolent sexism is understood as an attitude that idealizes women playing the traditional female role, which reflects subjectively positive feelings toward this particular type of woman. The hostile sexism subscale contains items such as “Women are too easily offended” and the benevolent sexism subscale contains items such as “No matter how accomplished he is, a man is not truly complete as a person unless he has the love of a woman.” The observed Cronbach’s α was .910 for both subscales.

Justification of the use of violence

In order to assess violent attitudes without arousing much social desirability, we asked participants how much they agree with the following sentence: “Sometimes one has to resort to violence if one does not want people to think one is dumb.” They indicated their answer on a 6-point scale, ranging from 0 (fully disagree) to 5 (fully agree).

Data Analysis

To test new ideas with a sample of people who are affected by a real problem with important social repercussions, we conducted a cross-sectional study and used path analysis to graphically describe relationships within a somewhat new set of variables. Causal path modeling enables testing of the significance and parsimony of all of the equations that compose the model simultaneously, instead of simply testing the significance and parsimony of each one separately. It is a very useful method for describing complex relationships among variables with a single model. Although cross-sectional research is useful for our exploratory hypotheses, it does not allow causality assertions regardless of the sophistication of the statistical technique. Decisions about causality can only be made with the utilization of appropriate research designs and with adequate theory and sound reasoning. In this respect, the proposed model accounts for knowledge from different fields and consolidated theories that explain the importance of feeling moral regardless of the situation and focuses on a paradoxical configuration of variables in a particular and relevant moment, just before starting a prescribed psychological treatment.

Aware that many models may be plausible, especially in samples as diverse in terms of sociocultural characteristics like ours, a multigroup analysis was conducted to test the assumption of the invariance in the structures across two subsamples: Spanish men and men of other nationalities. This approach involves comparing two models at a time: one with the two groups constrained as equivalent on the parameters of interest (weights) and a second one, less restrictive, in which these parameters were free to any value (Byrne, 1995). This approach may also serve to add plausibility if, in spite of the many and different sociocultural characteristics included between the two groups and even between the immigrants group itself, support is found for a common configuration of variables.

Results

A summary of the means, standard deviations, and intercorrelations of all of the variables is presented in Table 1. It can be observed that men convicted of violence against the partner had fairly high perceptions of their own morality and high levels of psychological well-being. In parallel, they presented a very high level of moral absolutism and high levels of self-deception and ambivalent sexism, especially benevolent sexism. The mean score of justifying violence was low in absolute terms, but problematic enough because it means that at least 35% of the sample explicitly said that they believe violence is sometimes necessary. It is assumed nonetheless that this type of direct question about violence is affected by social desirability, so the real percentage could be greater.

Table 1.

Descriptive Statistics and Intercorrelations Between the Variables Assessed in Men Convicted of Violence Against the Partner .

	M	SD	1	2	3	4	5	6	7
1 Moral absolutism (0–4)	2.88	1.09
2 Self-deception (0–6)	3.62	.87	.335**
3 Psychological well-being (0–4)	2.72	.82	.079	.486**
4 Moral self-concept (0–4)	3.45	.58	.124*	.341**	.346**
5 Ambivalent sexism (0–5)	2.41	.51	.237**	.077	–.064	.011
6 Benevolent sexism (0–5)	2.67	.62	.242**	.044	–.067	.047	.893**
7 Hostile sexism (0–5)	2.14	.54	.169*	.095	–.044	–.033	.859**	.537**
8 Violence’s justification (0–5)	.84	1.05	.160*	.080	–.023	–.079	.108	.041	.157*

p < .05

p < .01

Regarding the correlations between the variables, Table 1 shows that moral absolutism was related to both sets of variables, those concerning a high moral self-conceptualization and those concerning a prejudicial point of view against women. The magnitude of these correlations was small except for that between moral absolutism and self-deception, which was moderate. Specifically, the correlations show that those who enjoyed high levels of self-deception felt well and good (moderate effects), and those who had hostile sexism had also benevolent sexism and pro-violent attitudes (moderate effect and small effect, respectively).

The hypothesized model can be seen in Figure 1. Results provided a good fit to the data, χ2 (6) = 7.820, p = .252. Table 2 shows the absolute fit indices—the goodness-of-fit index (GFI), the adjusted goodness-of-fit index (AGFI), and the root mean square error of approximation (RMSEA)—and the comparative fit indices—the normed fit index (NFI), the comparative fit index (CFI), and the Tucker Lewis index (TLI). The absolute fit indexes were very good and allow us to conclude that the specified model reproduces the sample data very well (Hu & Bentler, 1998; Hair et al., 2010). The comparative fit indexes were equally good relative to the more restricted, nested baseline model (Hu & Bentler, 1998). Table 2 also indicates the same good indexes for the whole sample that included men convicted of violence against the partner who had had a mental disorder in the past.

Figure 1.

The Homeostatic Moral Model (HOMM) in men convicted of violence against the partner: standardized coefficients and percentage of explained variance (N = 387).

Table 2.

Goodness-of-fit Indices for the HOMM in Men Convicted of Violence Against Their Partners.

	GFI	AGFI	RMSEA	NFI	CFI	TLI
The HOMMN = 387	.993	.976	.028	.971	.993	.982
The HOMMN = 403(included men with past mental disorder)	.991	.969	.043	.966	.985	.962

Note. AGFI = adjusted goodness-of-fit index; CFI = comparative fit index; GFI = goodness-of-fit index; NFI = normed fit index; RMSEA = root mean square error of approximation; TLI = Tucker Lewis index; HOMM = Homeostatic Moral Model.

All of the hypothesized relationships were statistically significant except the one between ambivalent sexism and moral self-concept. As anticipated, the two related components of ambivalent sexism (benevolent and hostile) may be resolving the conflict with the moral self-concept and make these men think they are not sexist. Supporting this possibility, we can see in Table 1 the high correlation between hostile and benevolent sexism (.537) in this sample.

The standardized coefficients can be observed in Figure 1 and they correspond to the effect-size estimates. All of them were large enough to be theoretically and practically important and can be described saying that (a) more moral absolutism is related to more self-deception and more sexist and pro-violent attitudes; (b) more self-deception is related to more well-being and better moral self-conceptualization; and (c) more pro-violent attitude is related to a poorer moral self-concept. Regarding the standardized indirect effects, self-deception was a significant mediator of well-being. For every 1 SD increase in moral absolutism, well-being increased by .174 SD. Similarly, well-being was a significant mediator of moral self-concept. For every 1 SD increase in self-deception, moral self-concept increased by .120 SD. It seems worth noting that the highest coefficient (.52) links self-deception and psychological well-being, as a fruitful line of research has showed in other many samples, and the second one links moral absolutism and self-deception as it was hypothesized in this study to counteract the negative effects on moral self-concept (.34).

In order to compare the HOMM across the different nationalities included in the study, two groups were formed, one with the Spanish men (N = 212) and another with men of other nationalities (N = 175), and we compared the model with a fully constrained model in which the paths are constrained equally across the subgroups to an unconstrained model, in which the paths are allowed to vary freely. The results of the χ2 difference test showed that the two groups do not vary at the model level (Δχ2(6) = 4.533, p = .604). Assuming that the unconstrained model is correct, the structural weight model was invariant, CMIN (minimum chi-square) = 10,638, df = 7, p = .155. The fit statistics suggest that the structures for Spanish group and for the other nationalities group appear to be sufficiently similar to justify utilizing the same model for a combined sample of both groups in spite of their many sociocultural differences.

All these results globally support the four hypotheses of this study and offer initial evidence for a specific configuration of relationships between new variables, in a sensitive sample, and just before a psychological treatment. This does not mean that this is the only possible configuration, only that some theoretical assumptions to explain why men who use violence against the partner can feel moral seem plausible and need more research.

Discussion

In this study, we frame IPV in a moral domain. That allows us to study immoral behaviors in a relevant social context with important practical implications, as it is the psychological state of men convicted of violence against the partners at the beginning of their court-mandated psychological treatment. Following ideas from well-established theories about characteristics and consequences of the moral convictions (Mullen & Skitka, 2006; Shaw et al., 2011; Skitka et al., 2008; Skitka & Mullen, 2002), the importance of moral self-concept for everybody (Aquino & Reed, 2002; Jordan et al., 2011; Mazar et al., 2008; Monin & Jordan, 2009) and the role of self-deception to preserve well-being (Barkan et al., 2012; Effron et al., 2012; Shu & Gino, 2012; Taylor & Brown, 1994; Trivers, 2000; von Hippel & Trivers, 2011), we hypothesized a model in which men who have acted morally wrongly—from the current legal point of view at least—manage to feel well and moral due to self-deception while having morally questionable attitudes about women and violence. That paradoxical situation can be possible if we understand it as a case of motivated moral reasoning (Kunda, 1990; Tsang, 2002; Steele, 1988; Steele & Liu, 1983) in a current context where is believed that affect guides reasoning (Haidt, 2001; Haidt & Bjorklund, 2008; Haidt & Joseph, 2008).

The model proposes that moral absolutism plays an ambivalent role for men convicted of violence against the partner. On the one hand, such a narrow, strong, and rigid conception about what is right and wrong would be related to negative outcomes as sexist and violent attitudes against women and poor moral self-conceptualization. On the other hand, it would be positively related to self-deception, and through self-deception to well-being and a high moral self-conceptualization. The results from the path analysis support this possibility. This can be described as a type of motivated moral reasoning because such individuals reach the conclusion of “We are moral persons regardless of our past violent behavior and our current sexist and pro-violent attitudes.” These results remain the same across the whole sample of 403 participants, that included also those with past mental disorders, the subsample of Spanish men (N = 212), and the subsample of men of other nationalities (N = 175). That means that in spite of the many and different sociocultural variables, the theoretical model seems to fit equally well in all of the groups.

We are aware that the model data fitting only indicates that the causal path model under consideration is more parsimonious than the corresponding newly identified model. Also, that parsimony signifies that the model provides a more elegant description of the relationships between the variables without substantial loss of the variance contained in the original set of data. Because the design used was cross-sectional, the validity of the proposed model rests only on the logic of our arguments and on these first empirical results about the adequacy of model fit to the data.

Assuming that the rationale that links the variables is adequate, it can be concluded that the HOMM could be reflecting a type of mental makeup based on self-deception that men convicted of violence against their partners require to feel well and morally upright while having attitudes that justify violence and restrict the roles of women. This model can be read in terms of their need to appear rational, their need for self-affirmation, or as the distinctive characteristics of the moral convictions that lead people to accept violent means to achieve preferred ends (Kunda, 1990; Shaw et al., 2011; Skitka et al., 2008; Skitka & Mullen, 2002; Steele, 1988; Steele & Liu, 1983; Tsang, 2002). In any case, the practical implication is that this paradoxical configuration has to be disrupted during psychological treatment; otherwise, they will be more motivated to defend themselves and their moral beliefs than to change their sexist and violent attitudes.

Study Limitations and Practical Implications

It is obvious that the main limitation of this study is its cross-sectional design, which prevents us from assigning causality beyond the theoretical rationale used to relate the variables. However, some considerations need to be taken into account to extract the potential usefulness of the results: (a) New and creative approaches to study relevant social problems, such as IPV, requires more descriptive and comprehensive methods; (b) path models are sufficient to explore complex ideas with reliable measures in large samples of people; and (c) due to ethical and bureaucratic reasons, real samples of convicted individuals are not always accessible for experimental research, so they are systematically understudied. With these considerations and the previous precautions about causality in mind, this study offers a new descriptive picture of a considerable sample of men convicted of violence against the partner at the beginning of their court-mandated psychological treatment.

This snapshot shows them feeling happy and moral in spite of their rigid and strong perception about what it is right and wrong in general, and about women in particular, thanks to self-deception. This allows us to understand their strong resistances to change during the psychological treatments. It could be said that changing their sexist beliefs and violent behaviors would imply that they did wrong, making them feel bad and immoral. Future interventions could include strategies to reduce the resistances to change based on the reduction of moral absolutism and self-deception. This would allow for experimental manipulations that could reinforce these research data and improve the efficacy of the treatments.