Opposing torture: Moral conviction and resistance to majority influence

Abstract

Even though nearly every society and moral system condemns the use of torture, and despite recent outrage about abuses at Abu Ghraib and Guantanamo Bay, over half of Americans support the use of torture when interrogating suspected terrorists. Moreover, public support for the use of torture is increasing (Sidoti, 2009). The present study tested the role of people's moral convictions against the use of torture in resisting conforming to a majority of peers who supported the use torture when interrogating suspected terrorists. Results from an Asch-inspired conformity paradigm indicated that after controlling for other indices of attitude strength, strength of moral conviction uniquely predicted the extent that people expressed opposition to torture both publicly and privately. Implications are discussed.

The first and second author contributed equally to the research and to this paper; order of authorship was randomly determined. We would like to thank Amir Jacob and Jared Majerle for their assistance in conducting this research.

Keywords:

• Morality
• Moral conviction
• Majority influence
• Conformity
• Attitude strength

Since the end of the eighteenth century nearly every society and moral system has in principle—if not always in practice—regarded torture as an unmitigated evil (Ross, 2005). However, despite recent outcries about prisoner abuse at Abu Ghraib and forceful interrogation techniques at Guantanamo, many Americans support the use of torture. In fact a 2009 opinion poll showed that 52% of Americans believed that torture is justified when interrogating suspected terrorists, a 14% increase since 2005 (Sidoti, 2009). In summary, support for torture is now a majority opinion in the U.S., and one that seems to be gaining support.

Given this trend, it may be increasingly likely for people to interact with co-workers, friends, and family members who express support for torture. This possibility provides a backdrop for addressing a question of long-standing interest and importance in social psychology: What factors affect whether people resist or conform to majority influence? Although majority influence may operate in large groups (e.g., attitude change in response to viewing the results of a public opinion poll), majority influence is perhaps most prevalent and powerful within small groups due to people's desires to compare their thoughts and opinions to similar others (Festinger, 1954). With this in mind, the goal of the present study was to test the role that people's moral convictions play in resisting a small group of peers expressing support for torture. We predicted that people with stronger moral convictions would be more resistant to majority influence than people with weaker moral convictions (the morality as motivated resistance hypothesis). We provide our rationale for this prediction below.

Moral conviction and resistance to majority influence

People often conform to the majority when faced with the choice to accept or reject the majority position. This occurs because those who oppose the majority risk ridicule and disenfranchisement, whereas those who conform expect acceptance (Asch, 1956). In addition to acceptance, people may conform when they are unsure about the appropriate way to think or behave; they adopt the majority opinion because they believe that this is the correct thing to do (Chaiken & Stangor, 1987; Deutsch & Gerard, 1955). Therefore people conform both to gain acceptance from others as well as to be “right.”

However, people who have strong moral convictions about the unacceptability of torture should be more resistant to majority influence, because the typical motives for conformity—the need for acceptance and to be correct—should be relatively weak. For example, people tend to reject and socially distance themselves from morally dissimilar others (Bauman, 2006; Skitka, Bauman, & Sargis, 2005), and should therefore have no desire to be accepted by them. In addition, people experience moral convictions much like facts: wrong is wrong, right is right, end of story. There is little room for the group to provide information about the proper way to think because moral convictions have little ambiguity (Skitka et al., 2005). Finally, failing to defend one's moral convictions is likely to produce feelings of shame, guilt, or inauthenticity to those who conform, and people are likely experience a failure to defend their core moral beliefs as akin to embracing evil (Bowers, 1984). For these reasons the influence of the majority should be weaker for those with strong compared to weak moral convictions.1

In support of the morality as motivated resistance hypothesis, research by Hornsey and colleagues suggests that strength of moral conviction leads people to resist majority influence (Hornsey, Majkut, Terry, & McKimmie, 2003; Hornsey, Smith, & Begg, 2007). Specifically, students with strong moral convictions about a social issue expressed stronger intentions to speak out in support of this issue when presented with information that a majority of fellow university students opposed their stance than when they believed that they were in the opinion majority. Importantly, however, this counter-conformity effect emerged only for expressed behavioral intentions and not actual behavior (Hornsey et al., 2007). Moreover, the source of influence in the Hornsey et al. study was distant and abstract (information about the majority opinion on campus). It therefore remains unclear whether strength of moral conviction will be associated with resisting conformity when people express their opinions within a group of interacting peers, a less abstract and immediate—and therefore more powerful—source of influence than statistical information about peer opinion. The goal of this study was therefore to test whether moral conviction facilitates resistance to conformity in stronger social situations than have been studied to date.

Study overview

To provide a strong test of the morality as motivated resistance hypothesis we included three indices of attitude strength as controls: attitude certainty, importance, and religious conviction. Attitude certainty is associated with attitude stability and resistance to persuasion (Bassili, 1996), and people are more likely to resist conforming in a group when they are certain of their attitude (Kelley & Lamb, 1957). Moreover, people are less likely to conform when they place high personal importance on a particular issue (Vaughn & Mangan, 1963). Although moral conviction predicts behavior above and beyond these indices of attitude strength in other contexts (Skitka et al., 2005), differentiating the effects of moral conviction from attitude extremity, importance, and certainty is critical given that these latter attitude components could provide reasonable alternative explanations for any effects of moral conviction.

In addition to controlling for these indices of attitude strength we also measured participants’ religious convictions about the issue of torture. Because moral and religious conviction are often viewed as inextricably connected constructs (e.g., Geertz, 1968, 1973; McCready & Greeley, 1976; Shweder, Much, Mahapatra, & Park, 1997), it was important to establish that moral conviction uniquely predicted resistance to majority influence above and beyond religious conviction. A strong test of the morality as motivated resistance hypothesis should show that strength of moral conviction predicts conformity behavior even when controlling for people's attitude strength and strength of religious conviction.

Additionally, because social support is a key factor in reducing conformity (Allen, 1975), we tested whether the effects of moral conviction depended on the presence of social support within the group. Specifically, we tested whether morally motivated resistance to the majority would depend on the presence of another individual who shared a participant's opinion. In line with many years of conformity research we predicted that the presence of social support would reduce conformity. Of particular interest, however, was whether social support was necessary to facilitate standing up for one's convictions, or whether the effect of moral conviction was sufficiently strong that it would emerge regardless of whether participants had social support.

Method

Participants and procedure

Participants were introductory psychology undergraduates who received credit toward class requirements for their participation (N = 170). Prior to participating in the laboratory sessions, participants completed pretest measures assessing their degree of support or opposition to the use of stress techniques on suspected terrorists (the use of torture), as well as the degree of certainty, importance, moral, and religious conviction they felt about this issue. Participants who expressed no opinion were excluded from the experimental phase of the study because we were interested in whether participants would change pre-existing positions, not adopt a new position when confronted with a majority opinion. Moreover, we restricted analyses to participants who opposed torture because there were too few participants who supported torture in our sample.

Participants attended laboratory sessions in groups that ranged from 4 to 10. At the lab they were greeted by an experimenter and seated in close proximity to one another at one of several divided cubicles with computers. Participants learned from the experimenter that they would be participating in a study that (ostensibly) examined the effect of technology on group decision making. They learned that most of the session would involve a small group discussion with four other study participants about the issue of torture and that they would draft a short position statement summarizing their group's opinion on the issue. Participants learned that they would meet fellow group members for the first time over the computer, where they would share their opinions with one another about the use of torture. Furthermore, this interaction would be very brief, lasting only about 5 minutes. Consistent with the cover story, the experimenter explained that the purpose of the computer interaction was for group members to meet each other and learn each others’ opinions about torture to “get the ball rolling” prior to the upcoming face-to-face discussion. After this initial interaction over the computer, each group would be escorted to a second location where they would hold their discussion. To increase the plausibility that participants engaged in a real communication with four other study participants regardless of how many attended any given session, participants were told that the study was being conducted simultaneously in two rooms and that their group might consist of people from either room. This story was enhanced with a feigned phone call 5 minutes into the instructions that presumably coordinated the start time with the experimenter in another room.

After receiving the overview of the research session, participants initiated the computer program and met the members of their group. Once initiated, the computer program prompted participants to introduce themselves to fellow group members by entering their initials, which were then displayed to everyone in the group. After doing so, the program presented text that informed participants that they would share their opinion about torture to the group in a randomly determined order. Although it appeared to participants that the computer program randomly assigned this order, participants always shared their opinion last.

Depending on experimental condition, participants shared their attitudes after viewing three (social support) or four (no social support) other group members who indicated that they supported the use of torture. The computer program displayed the opinions of the other members on the computer screen at 10- to 11-second intervals. In the no social support condition group members expressed strong support, moderate support, slight support, and strong support for the use of torture (in that order). The degree of support as well as order of presentation was held constant. The social support condition was identical to the no support condition with one exception: instead of expressing slight support for torture, the third group member expressed slight opposition.

After viewing the other members’ opinions, participants shared their own opinion, using a 7-point radio-button scale with the point labels strongly oppose, moderately oppose, slightly oppose, uncertain, slightly support, moderately support, and strongly support. After sharing their opinion, participants were directed to click a button that they believed would disconnect them from the rest of the group. After disconnecting from the group, participants privately reported their opinion about torture for a second time using the same 7-point response scale. After doing so, they were directed to a text box and were asked to share any thoughts that they had about the study thus far. They were then instructed to wait quietly for the experimenter to take them and their group to their discussion room. In reality, after sharing their thoughts about the study the experiment was over, participants did not participate in face-to-face discussions.

To ensure that participants believed that they really interacted with other students, participants were probed for suspicion at two time points. Participants’ written responses regarding their thoughts about the research study served as one of these probes. Additionally, during the verbal debriefing the experimenter directly probed participants for any suspicions that their group was not real and that they did not interact with other real study participants. Seven participants expressed suspicion that the group was not real during one of these probes and were removed from subsequent data analysis, leaving a total N = 163.

The most common reactions from participants when they learned that their group was not real were looks of surprise and confusion, and many participants spontaneously expressed to the experimenter that they truly believed they were interacting with other study participants. Moreover, the majority of participants’ comments to the written probe reflected how it felt to share one's opinion in the group (e.g., “I felt my face change because I disagreed with them”) or expressed their thoughts about the upcoming group discussion (e.g., “There may be aggression [that] occurs when the group meets and discusses the topic”). Therefore there was strong evidence that participants believed that they interacted with other study participants, and that this interaction was psychologically engaging and meaningful.

Measures

Position on torture

Participants’ positions on the use of torture were assessed at three time-points (privately at pretest, publicly when sharing opinions with the group, and then privately after the group interaction) with the item “To what extent do you support or oppose the use of stress techniques when interrogating suspected terrorists, such as sleep deprivation, ‘water boarding’ (strapping detainees to a board and dunking them underwater), long periods of hanging detainees by ropes in painful positions, etc.)?” This item was measured on a 7-point scale with the point labels strongly support (+3), moderately support, slightly support, neither support or oppose (0), slightly oppose, moderately oppose, and strongly oppose (−3). Change in relative support or opposition to torture over the three time periods was the primary dependent variable.

Attitude measures

All the attitude-related measures used a 5-point scale with the point labels of not at all (1), slightly, moderately, much, and very much (5). Attitude certainty was assessed with the question “How certain are you that of all the possible attitudes one might have toward this topic your attitude reflects the right way to think and feel about the issue?” Attitude importance was assessed with the question “To what extent is your attitude about whether stress interrogation techniques should be allowed personally important to you?” Participants’ degree of religious conviction was measured with the item “To what extent is your attitude about whether stress interrogation techniques should be allowed related to your religious beliefs?” Finally, moral conviction was assessed using two items: “To what extent does your attitude about whether stress interrogation techniques should be allowed reflect your core moral values and convictions?” and “To what extent is your attitude about whether stress interrogation techniques should be allowed deeply connected to beliefs about fundamental questions of 'right' and 'wrong'?” These two items correlated strongly (r = .75, p < .001), and were therefore averaged to create a single measure of moral conviction.2

Results

A one-way repeated-measures ANOVA revealed that the degree of support/opposition to torture that participants expressed differed at pretest, when reporting to the group, and in private, F(2, 320) = 257.45, p < .001. Follow-up analyses revealed that participants reported less opposition to torture to their groups (M = −0.20, SD = 1.69) and in private (M = −0.23, SD = 1.67) than they originally had at pretest (M = −2.28, SD = 0.76), F(1, 320) = 389.52, p < .001 and F(1, 320) = 378.24, p < .001 respectively. Participants reported an equivalent amount of opposition to torture to their groups and in private, F(1, 320) < 1. (See Table 1 for means, standard deviations, and correlations of all study variables.)

Table 1. Means, standard deviations, and correlations among study variables

	1	2	3	4	5	6	7	8
M	−2.28	3.14	2.91	2.58	3.71	0.50	−0.20	−0.23
SD	0.76	1.16	1.19	1.41	1.02	0.50	1.69	1.67
1. Degree of support for torture (pretest)	–
2. Attitude certainty	−.21**	–
3. Attitude importance	−.23**	.48**	–
4. Religious conviction	−.07	.23**	.24**	–
5. Moral conviction	−.29**	.47**	.53**	.42**	–
6. Social support	.08	−.16*	−.12	−.14	−.12	–
7. Degree of support for torture (group)	.36**	−.16*	−.14	.14	−.21**	−.22**	–
8. Degree of support for torture (private)	.35**	−.20**	−.15	.10	−.23**	−.21**	.91**	–
N = 163 for all study variables expect for private degree of opposition where N = 161. Because only participants who opposed torture were included in laboratory sessions, degree of support for torture ranged from −3 to −1 at pretest, but from −3 to +3 when reporting to the group and in private, with positive numbers indicating greater support.
*p < .05, **p < .01.

As these mean levels suggest, conformity to the majority opinion of the group was common. Specifically, when reporting to their groups, 80% of participants reported less opposition to torture than they had reported at pretest, but only 17% reported an equivalent degree of opposition (non-conformity); and 2% reported more opposition (counter-conformity). Additionally, these results suggest that participants were influenced by the majority opinion even after the group interaction. If participants only adjusted their opinions to avoid conflict with their groups in the upcoming discussion, but did not change their real attitudes, the degree of opposition that they reported privately should have been closer to what it was at pretest. Therefore, overall, participants opposed torture less after exposure to a group of peers who unanimously supported torture, and this degree of conformity emerged even when they were given the opportunity to report their attitude privately. We turned next to the question of whether evidence of conformity to the group was equally strong among those whose opposition to torture varied in strength of moral conviction.

Moral conviction and resistance to majority influence

To test the morality as motivated resistance hypothesis we tested whether participants’ strength of moral conviction predicted changes in the degree of support/opposition that participants expressed from (a) pretest to the group, (b) pretest to in private following group interaction, and (c) the group to in private. To do so we created three hierarchical multiple regression models following procedures recommended by Cohen, Cohen, West, and Aiken (2002) for testing change over time in multiple regression. In the first step of all three regression models we entered participants’ prior attitude about torture (e.g., pretest or group-reported), along with attitude certainty, attitude importance, religious conviction (all mean-centered), and the presence of social support (0 = no support, 1 = support). In the second step we entered moral conviction, and in the third step we entered the interaction of moral conviction and social support. Because participants’ prior attitudes about torture were entered into the first step of each regression model, the remaining variance in participants’ attitudes (at the second time point) represented the difference, or change, from what was predicted based on the prior time point. Specifically, positive regression residuals would indicate that participants expressed greater support for torture than would have been predicted by the previous time point (conformity) and negative regression residuals would indicate lower support for torture than would have been predicted by the previous time point (non-conformity). Therefore, if strength of moral conviction was related to resisting majority influence and conformity, we expected that moral conviction would be negatively associated with attitude change at each time point.

In support of the moral conviction as motivated resistance hypothesis, strength of moral conviction about torture was negatively associated with the degree that participants adjusted their attitudes from pretest toward support for torture both in their groups, b = −0.32, t(156) = 2.13, p = .03, as well as in private following the group interaction, b = −0.32, t(154) = 2.13, p = .04 (see Table 2, Step 2 of Columns 1 and 2). Importantly, these effects were found even after controlling for the effects of attitude certainty, importance, religious conviction, and social support in a prior step. Therefore moral conviction predicted resistance to majority influence and conformity, both in public and in private.3 Moreover, the effect of moral conviction on both public and private resistance to majority influence did not depend on the presence of social support (see Table 2, Step 3 of Columns 1 and 2). Finally, strength of moral conviction was unrelated to the degree that participants adjusted their attitudes about torture from when they expressed their opinions to their groups to when reporting in private (see Table 2, Step 2, Column 3). In other words, participants high in moral conviction persisted in resisting majority influence even when no longer confronted with the group, whereas participants low in moral conviction privately expressed the same degree of conformity as they had to the group. Indeed, 84% of the variance in participants’ private attitudes toward torture following the group interaction was accounted for by the attitude that they expressed to the group.

Table 2. Hierarchical multiple regression analyses predicting degree of support for torture participants expressed to groups and privately following group interaction

	Support reported to group controlling for pretest attitude		Support reported in private controlling for pretest attitude		Support reported in private controlling for group attitude
Variable	ΔR ^2	b	ΔR ^2	B	ΔR ^2	b
Step 1	.24**		.23**		.84**
Previous position on torture		0.78**		0.73**		0.89**
Attitude certainty		−0.19		−0.26*		−0.08
Attitude importance		−0.10		−0.08		0.01
Religious conviction		0.20*		0.17		−0.01
Social support		−0.86**		−0.81**		−0.07
Step 2	.02*		.02*		.00
Moral conviction		−0.32*		−0.32*		−0.03
Step 3	.00		.00		.00
Social support × Moral conviction		0.14		0.16		0.06
*p < .05. **p < .01.

Discussion

The results of the present study provided strong support for the morality as motivated resistance hypothesis. Results indicated that when controlling for other indicators of attitude strength, strength of moral conviction uniquely predicted resistance to majority influence, both publicly and privately. Thus, above and beyond holding otherwise strong beliefs about the acceptability of torture, the strength of people's morally convictions in opposition to torture made them more likely to resist publicly and privately conforming to a group of peers expressing support for this practice.

These results contribute to a growing body of research indicating that people's moral convictions are resistant to social influence. For example, previous research has demonstrated the effect of moral conviction in resisting sources of influence such as authorities or rule of law with adults (Skitka, Bauman, & Lytle, 2009; Skitka & Mullen, 2002; Wisneski, Lytle, & Skitka, 2009), and with children (e.g., Nucci & Turiel, 1978; Smetana, 1981, 1985). Other research has documented moral resistance to peer influences among children and adolescents (e.g., Perkins & Turiel, 2007), and to normative information among college students (Hornsey et al., 2003, 2007). Importantly, however, even though this prior research demonstrated that people often resist those who oppose their moral convictions, resistance is often expressed as a judgment about the source of influence or as an expressed behavioral intention, as opposed to a behavior per se.

The present research contributes to this literature by demonstrating that those with strong moral convictions are willing to resist the influence of interacting peers who do not share their moral beliefs. Considering that the correspondence between people's stated attitudes and their subsequent behaviors tends to be highly variable and often weak (Ajzen, 2001; Glasman & Albarracín, 2006; Wilson & Dunn, 2004), finding that those with stronger moral convictions were less likely than those with weaker moral convictions to capitulate their attitudes despite strong pressures to do otherwise is noteworthy. Future research should examine whether these results would also emerge in non-computer mediated communication, given face-to-face and computer-mediated communication differ in a number of known ways (e.g., Walther, 1996). However, given how much contact people currently have with others in on-line and virtual environments (e.g., Facebook and other social media), studying social influence in computer-mediated communication is both theoretically and practically important in its own right.

Implications

The results of the current study potentially shed some light on the recent increase in support for torture in the U.S. One possibility for this trend is that when people discover through conversations with others that they are in the opinion minority, those who are opposed but who have only weak moral convictions regarding the acceptability of torture may express support for what appears to be the majority opinion. False public endorsement of torture may lead others to overestimate the amount of consensus surrounding this issue. Perceptions of consensus have an important influence on how people think about issues; beliefs become more fact-like when they are socially validated (Festinger, 1954) because people believe the consensus opinion signals the appropriate way to think and feel (Chaiken & Stangor, 1987). Moreover, increased public support for torture may spread simply from talking to one's peers (for research on the spread of attitudes through social networks see Fowler & Christakis, 2008; Nowak, Szamrej, & Latané, 1990). Indeed the results of the present research indicated that those with weaker moral convictions were influenced to endorse the practice of torture, even in private.

Even if public opinion is trending toward greater support for torture, there is not yet consensus on its acceptability. Furthermore, this research suggests that as long as there are those with strong moral convictions against the use of torture, complete consensus is unlikely to occur. By speaking up in support of their positions, the morally convicted ensure that their viewpoint stays in the debate, and others may benefit by gaining a more accurate sense of the extent to which there truly is consensus surrounding this issue, and others.

Notes

The first and second author contributed equally to the research and to this paper; order of authorship was randomly determined. We would like to thank Amir Jacob and Jared Majerle for their assistance in conducting this research.

¹The theoretical properties of moral conviction that lead to the prediction that moral convictions are likely to be associated with resistance to majority influence are also features that theoretically distinguish moral convictions from variables such as attitude extremity and importance. For example, one might have an extremely negative attitude about eating insects, and see it as personally important never to eat insects, but these high levels of extremity and importance do not by definition mean that one's attitude about eating insects will be high in moral conviction. Moreover, even though attitudes high in moral conviction are also likely to be high on other dimensions of attitude strength (e.g., extremity, importance), extreme and important attitudes will not necessarily be high in moral conviction (see Skitka et al., 2005, for a more detailed discussion of these issues).

²Because moral conviction was measured using two items, measurement error may have been lower for this variable compared to the other measures of attitude strength, which used single items. Because reduced measurement error increases the likelihood of obtaining statistically significant results, our measurement of moral conviction may have advantaged this variable over the others. To address this concern we ran all reported analyses with a single measure of moral conviction (“reflect your core moral values and convictions”), and found equivalent results. Additionally, one might think that moral conviction should be represented as a dichotomous variable—an all or nothing phenomena. We treat moral conviction as a continuous variable for a number of reasons: (a) classification of an attitude as a moral one, and strength of moral conviction both contribute to explained variance in relevant phenomena (Wright, Cullum, & Schwab, 2008), therefore dichotomizing the variable would lose valuable information, (b) other indices of attitude strength are conceptualized as continuous rather than dichotomous constructs (e.g., attitude importance), (c) dichotomizing continuously measured variables is “seldom defensible and often will lead to misleading results” (MacCallum, Zhang, Preacher, & Rucker, 2002, p. 19).

³We also assessed whether strength of moral conviction was associated with participants maintaining some degree of opposition to torture versus switching their stance to reporting any degree of support or uncertainty about the issue, both to their groups, and in private following the group interaction (0 = switched sides, 1 = maintained opposition). Two binary logistic regressions replicated the pattern of results observed in Table 2, Columns 1 and 2. Specifically, controlling for participants’ prior attitudes about torture at pretest, the other indicators of attitude strength, and the presence of social support, strength of moral conviction was negatively associated with the likelihood of changing one's stance on torture, both in public and in private.