Effects of norms among those with moral conviction: Counter‐conformity emerges on intentions but not behaviors

Abstract

Previous research has shown that people with a strong moral basis for their attitude show stronger intentions to publicly act in line with their attitude when they are led to believe they are in a minority as compared to a majority. The current paper examines whether this evidence for counter‐conformity can be replicated on speaking‐out behaviors as well as intentions. Participants were told that they were either in a minority or in a majority in terms of their attitude toward a government apology to Australian Aborigines (Experiment 1: N = 100) and legalizing voluntary euthanasia (Experiment 2: N = 169). On intentions to speak out, participants with a strong moral basis for their attitude counter‐conformed, whereas those with a weak moral basis for their attitude were not influenced by the group norm. On behaviors, however, evidence for counter‐conformity disappeared. In Experiment 2, the influence of norms on both intentions and behaviors were moderated by perceptions of whether the status quo was likely or unlikely to change in line with participants' attitudes. Results are discussed with reference to theory and research on normative influence.

Keywords:

• Conformity
• Counter‐conformity
• Normative influence
• Moral conviction

It has been well documented that people shape and bend their attitudes to fit in with the attitudes of the majority. There are at least two reasons why this might be the case. First, people are afraid that by holding a minority position they will open themselves up to ridicule and social isolation. To increase their chances of belonging, they might then adopt the majority position in public and keep their personal position to themselves (“normative influence”). Second, people might conform to a majority opinion because they feel genuinely uncertain about what attitudes are true and correct. The attitudes of others then becomes a source of information about what is true, resulting in genuine attitude change (“informational influence”). Consistent with these perspectives, it has been shown that people's attitudes often conform to the expressed attitudes of the majority, a tendency that increases when people are uncertain (promoting informational influence; Asch, 1956; Sherif, 1936; Smith, Hogg, Martin, & Terry, in press) and when responses are public (promoting normative influence; Asch, 1952, 1956; Deutsch & Gerard, 1955).

It should be noted, however, that the strongest evidence that has been uncovered for conformity—and normative influence in particular—has occurred when researchers examine attitudes that have little personal relevance. Indeed, the classic studies on conformity and conversion have tended to use perceptual tasks as a way of indexing attitudes: Sherif (1936) used perceptions of light movement, Asch (1952) used judgments of line lengths, and Moscovici (1976) examined people's ability to discriminate between blue and green. These types of paradigms have set the tone for much subsequent work on normative influence, presumably because they allow for exemplary experimental control. One consequence of examining attitudes about real‐world social issues is that they introduce unwanted variance or “noise” associated with sociohistorical and political circumstance.

It is possible, however, that the reliance on perceptual judgments has resulted in a skewed perspective on people's willingness to inhibit minority attitudes in public (Jahoda, 1959). When people are faced with a decision about whether to express their minority attitude, they are faced with two countervailing considerations. On one hand, they might reflect on the consequences of expressing their minority attitude in terms of their chances for social acceptance. People who think and act in ways that differ from the norm can face rejection and ridicule (Asch, 1956; Festinger, 1950), and this provides a disincentive for expressing the attitude. On the other hand, people also need to reflect on the psychological consequences of inhibiting their attitude. People in individualist cultures like to think of themselves as immune to the influence of others (Hornsey & Jetten, 2004, 2005; Jetten, Hornsey, & Adarves‐Yorno, 2006), and we have an intra‐psychic need for consistency. Furthermore, in individualist cultures at least, we are socialized to be authentic, to prioritize self‐expression, to march to the beat of our own drum, and to tolerate dissent (Baumeister, 1991; Hornsey, Jetten, McAuliffe, & Hogg, 2006b; Kim & Markus, 1999; McAuliffe, Jetten, Hornsey, & Hogg, 2003). This is particularly the case if the cause is deemed to be important or valuable: We are counseled to stand up for what we believe in, and to have the courage of our own convictions. By inhibiting our attitudes in the face of majority opposition, we are violating a script that teachers, parents, and media consistently advocate.

Furthermore, people understand that by inhibiting their attitude on a point of principle they are entrenching the perception that their attitudes are unpopular, which in turn makes speaking out more difficult for others. This tendency has been articulated by Noelle‐Neumann (1974, 1993) who identified a “spiral of silence” phenomenon in the media, such that people are less likely to speak out when they believe they hold a minority opinion, which in turn deepens perceptions of being in a minority, theoretically leading to a self‐reinforcing cycle of marginalization. In these circumstances, it is possible that people will be prepared to take on the risk of social ridicule and isolation in order to shore up a position on which they feel a particularly strong conviction.

Consistent with this notion, it is interesting to note that most of the evidence for normative influence has emerged on studies using perceptual tasks, and most disconfirming evidence has emerged on studies using social issues. For example, when judging line lengths, people conform more in public than in private, suggesting normative influence (Abrams, Wetherell, Cochrane, Hogg, & Turner, 1990; Asch, 1956; Deutsch & Gerard, 1955; Insko, Drenan, Soloman, Smith, & Wade, 1983). However, on the issue of legalization of marijuana (Frideres, Warner, & Albrecht, 1971), people were no more likely to conform to the majority in public than in private. Furthermore, Shamir (1997) found that Israelis were no less willing—and sometimes more willing—to express their attitudes on Palestine when the prevailing particular culture was antagonistic rather than sympathetic to their views. In a meta‐analysis of studies on the spiral of silence phenomenon (Glynn, Hayes, & Shanahan, 1997), it was found that the relationship between perceptions of support for an attitude and willingness to express the attitude was positive but very small (r = .05). In other words, most people most of the time appear willing to speak out on issues even when they find themselves holding a minority opinion. Finally, Crutchfield (1955) found that when testing items ranging from factual to attitudinal, structured to ambiguous, and from impersonal to personal, participants yielded significantly to group pressure for all items except the personal judgments, for which no conformity was observed.

EXAMPLES OF COUNTER‐CONFORMITY AMONG THOSE WITH A MORAL BASIS FOR THEIR ATTITUDE

One thing that separates social issues from perceptual judgments is that social issues often possess a moral component. With respect to any single social issue, people might differ in the extent to which their attitude has a moral basis to it, and do so in a way that is independent of attitude strength. For example, a person may oppose the death penalty because they believe in the sanctity of life. Another person might have equally strong attitudes against the death penalty, attitudes grounded in the belief that the death penalty is expensive, unwieldy, and ineffective as a deterrent to crime. Although these people hold equally strong attitudes, the former has a strong moral basis for their attitude in the sense that their attitude is grounded in issues of principle and morality, whereas the latter has a weak moral basis for their attitude in the sense that it is grounded in more practical concerns.

Hornsey and colleagues have argued that when people have a moral basis for their attitude, they might be particularly resistant to conformity effects (Hornsey, Majkut, Terry, & McKimmie, 2003). This argument is consistent with recent theories of moral psychology, which argue that moral judgments are experienced as intuitions that are either innate or developed early in life. Although the externalization of these intuitions is shaped by cultural and social forces, the influence of these forces tends to be gradual and to have its greatest traction in childhood (Haidt, 2001). The argument that moral conviction acts as a buffer against conformity—at least in adults—finds empirical support in Sherif and Hovland's (1961) social judgment theory, which proposes that people who are highly involved in a particular issue are positively influenced by fewer people than are those who are uninvolved (see also Sherif & Cantril, 1947). There is also growing evidence that when people feel a moral compulsion to behave in a certain way, there is a stronger link between their attitudes and their behavior (see Manstead, 2000, for a review). Finally, there is recent evidence that stronger moral conviction leads to greater distancing from those who hold opposing views, lower levels of good will and cooperativeness in groups that feature attitudinally dissimilar others, and a reduced ability to generate solutions to resolve disagreements (Skitka, Bauman, & Sargis, 2005). In short, it appears that people stick stubbornly to views on which they feel moral conviction, and they do so over and above indices of attitude strength. On the basis of this, Hornsey et al. (2003) predicted that people with a strong moral basis for their attitude might be less influenced by group norms than would those with a weak moral basis for their attitude.

To examine this, Hornsey et al. (2003) exposed participants to information suggesting they were either in a minority or a majority with regard to their attitudes to gay law reform (Experiment 1) and making an official apology to Australian Aborigines for extreme assimilationist policies (Experiment 2). They found no conformity effects; in other words, people expressed just as much willingness to engage in attitude‐consistent behavior when they were led to believe they were in a minority than when they were led to believe they were in a majority. However, the effects of the normative information were moderated by the extent to which participants had a moral basis to their attitude. On private behavioral intentions (e.g., intentions to vote in favor of gay law reform), Hornsey et al. found conformity among those with a weak moral basis for their attitude and non‐conformity among those with a strong moral basis for their attitude. On public behavioral intentions (e.g., intentions to attend a rally promoting gay law reform), there was even less evidence for conformity. Specifically, those with a weak moral basis for their attitude displayed non‐conformity and those with a strong moral basis for their attitude displayed counter‐conformity. All of these effects emerged after controlling for (i.e., covarying out) the effects of attitude intensity (i.e., how strongly participants supported gay law reform on a 3‐point scale), so the effects of moral basis appear to be independent of attitude intensity or strength.

In many ways, these data run contrary to the accepted wisdom about conformity and public behavior. First, there was less evidence for conformity in public than in private, the opposite of what might be expected from a normative influence perspective. Second, among people with a strong moral basis for their attitude, people expressed stronger intentions to publicly defend their attitude when led to believe they belonged to a minority than when they believed they belonged to the majority. This finding reinforces the importance of examining conformity effects using real‐world social issues to supplement traditional paradigms using judgments about physical reality. It also provides a counterpoint to the notion that people are inherently conformist and nervous about expressing minority attitudes (see also Nail, MacDonald, & Levy, 2000, who highlight the fact that counter‐conformity can emerge as a function of social pressures, a phenomenon they refer to as “anti‐conformity”). Furthermore, it points to the importance of measuring social influence in public and in private response contexts, a point that has been reinforced by many recent commentators (e.g., MacDonald, Nail, & Levy, 2004; Nail et al., 2000).

There are two major limitations of these studies, however. First, there is little evidence about what is driving the effects of moral basis for attitude, and indeed what is motivating participants to counter‐conform. One possibility is that those with a moral basis for their attitude have a heightened need to convert others to their point of view. This need for conversion is most likely to be activated (a) when they feel they are in a minority, and so need to gain support from others, and (b) when the behavior is public, and thus accessible to others. One of the goals of the current studies is to test this possible mechanism for the effects reported by Hornsey et al. (2003).

Second, and perhaps more importantly, the studies by Hornsey et al. (2003) did not test actual behaviors, but rather behavioral intentions. Although intentions are often used as a proxy for behavior, it is widely recognized that the relationship between intentions and behaviors is moderate (see Sheeran, 2002). This might be particularly the case when examining people's willingness to stand their ground in the face of majority opposition (Kennamer, 1990; Scheufele, Shanahan, & Lee, 2001). Because they did not have to act, it could be that participants were responding as they would like themselves to be: independent creatures, immune to the influence of others, and prepared to have the courage of their convictions. In other words, participants might have been living out a fantasy of what they would do, safe in the knowledge that their responses will not open them up to shame, ridicule, or isolation. It is thus premature to argue that, for those with a strong moral basis for their attitude, the need to be authentic overrode the need for belonging or the fear of social rejection. In the current studies we repeated the methods used by Hornsey and colleagues, but included situations in which participants felt their attitudes would actually reach a public audience. By doing this we were able to provide a truer test of the power of norms to influence behavior.

THE CURRENT STUDIES

Participants were led to believe that their attitudes on apologizing to Aborigines (Experiment 1) and on legalizing voluntary euthanasia (Experiment 2) were shared by either the majority or a minority of group members. After being exposed to the normative information, we measured whether participants thought they would be comfortable speaking out publicly in line with their attitude (an intention) and the extent to which they did speak out in line with their attitude (behavior). Consistent with Hornsey et al. (2003), we expected that on the measure of intentions, those with a weak moral basis for their attitude would show either conformity or non‐conformity in the face of the normative information. In contrast, those with a strong moral basis for their attitude would show counter‐conformity on intentions; that is, they would express more of an intention to speak out publicly in line with their attitude when in a minority than when in a majority.

On our measure of speaking‐out behavior, at least two outcomes could be expected. One possibility is that the predicted effects on intentions genuinely reflect people's willingness to counter‐conform. If this is the case, the effects on speaking‐out behavior will mirror the effects on intentions. Another possibility is that the intentions to speak out represent an inflated or wishful estimate of people's willingness to counter‐conform. If this is the case, we might expect either conformity or non‐conformity on speaking‐out behaviors, regardless of whether or not participants have a strong moral basis for their attitude.

EXPERIMENT 1

Method

Participants and design

A total of 147 undergraduate students enrolled at the University of Queensland (UQ) participated in the study in return for course credit. Participants were initially required to rate their attitudes on three issues: the legalization of marijuana, capital punishment, and whether or not the federal government should apologize to Australian Aborigines for the policy of assimilation that led to children being removed from their mothers and forced into orphanages (1 = strongly support, 8 = strongly oppose). They also rated the extent to which they had a moral basis for these attitudes. The issues of legalization of marijuana and capital punishment were distracter issues designed to make the study less transparent, and so questions relating to these issues were not analyzed. Consistent with Hornsey et al. (2003), the current paper focused only on participants who held a position contrary to the status quo; that is, participants who identified themselves as supportive of an apology to Aborigines. This decision was taken in both Experiment 1 and Experiment 2 for several reasons: (1) to simplify the design of the experiments; (2) because there were insufficient numbers of status quo supporters to allow for reliable analysis (47 participants in Experiment 1; 66 participants in Experiment 2); and (3) given that the aim of the research was to extend and elaborate on the work of Hornsey et al. (2003), this strategy allows for comparisons with the earlier research to be drawn. Of the total sample, 100 expressed support for a government apology to Aborigines (rating 4 or below) and only this subset of data was analyzed. The subset consisted of 60 females and 40 males (M = 20.35 years old). After recording their attitude, participants were led to believe they had either majority or minority support for their opinion (participants were assigned randomly to normative support conditions).

Materials and procedure

To test whether moral basis for attitude was having an effect over and above attitude strength, it was important to have an index of the intensity with which participants hold their attitude. To obtain such an index, we used participants' responses on the initial measure of the extent to which they supported an apology to Aborigines (1 = strongly support, 8 = strongly oppose). Because we had only pro‐apology participants, this translated to a 4‐item scale of attitude intensity. Consistent with the strategy used by Hornsey et al. (2003), the initial attitude score was recoded such that a score of 4 represented the most intense attitude and 1 represented the least intense attitude.

After indicating their attitude, participants completed three items designed to assess the extent to which they have a moral basis for their attitude (based on Hornsey et al., 2003). On a 9‐point scale (1 = not at all, 9 = very much), participants expressed their agreement with three questions: “To what extent do you feel your opinion is morally correct?”, “To what extent do you think your position is based on strong personal principles?”, and “To what extent do you feel that your position on a government apology to Aborigines is a moral stance?” (α = .76).

Normative support was manipulated using the same technique used by Hornsey et al. (2003). Participants were given graphs that ostensibly represented past and present UQ student opinion on a government apology to Aborigines, the legalization of marijuana, and capital punishment. In all conditions, UQ students were shown to equally support and oppose the legalization of marijuana and capital punishment. However, on the topic of a government apology to Aborigines, UQ student opinion differed depending on condition. In the majority support condition, an average of 85% of UQ student's demonstrated support for an apology, 9% were opposed, and 6% were undecided. The minority support condition showed on average 85% of UQ students opposed an apology, 9% supported an apology, and 6% were undecided. A manipulation check was included which required participants to rate whether UQ students mainly supported, opposed, or equally supported and opposed each issue. Two participants incorrectly completed the manipulation check and so their data were discarded.

Next, motivation to convert was evaluated using two items: “To what extent do you feel a need to convince other UQ students that your opinion on a government apology to Aborigines is correct?” and “To what extent do you feel a need to convert other UQ students to your view on a government apology to Aborigines?” (1 = not at all, 9 = very much; r = .92).

Perceived comfort with speaking out was measured using two scenarios. The first read: “Assume that you're traveling on a bus to university when the student sitting next to you asks about your opinion on the issue of a government apology to Aborigines.” The second scenario read: “Assume that you're at a tutorial in which the issue of a government apology to Aborigines has been raised.” In each case, participants were asked to rate how comfortable (1 = uncomfortable, 9 = comfortable), relaxed (1 = nervous, 9 = relaxed), and safe (1 = unsafe, 9 = safe) they would feel raising their true opinion. Scores on all six items were strongly intercorrelated and were combined as a single scale of intentions to speak out (α = .88).

To measure speaking‐out behavior, participants were first asked to describe their opinions regarding a government apology to Aborigines in an open‐ended fashion. After reporting their opinion, participants were informed, on the next page of the response booklet, that:

The UQ school of Journalism and Communications publishes a monthly newspaper titled The Queensland Independent which is distributed throughout Queensland. It can also be accessed online. Student journalists from the newspaper have approached the School of Psychology in the interest of compiling a student feature section which represents UQ student attitudes regarding various social issues. Please note that no further work will be required if you nominate for your opinion piece to be published. We can simply use the argument you wrote down earlier in this experiment. However, we will contact you if your opinion piece is selected and give you the opportunity to edit your comments.

Participants then indicated their willingness for an excerpt of their opinion piece—coupled with their full name and degree—to be published (1 = no desire at all, 6 = very strong desire). This measure of speaking‐out behavior was chosen because it eliminated any practical or reality constraints that typically come into play when measuring behaviors (e.g., lack of time, opportunity, or resources). In this case, because participants had already written the essay, engaging in the behavior takes no more effort, time, or resources than not engaging in the behavior. This measure shares similarities with behavioral measures used in past research in which participants are given a pre‐prepared “form letter” and asked to simply sign the letter to express their opinion (see e.g., Martin, Martin, Smith, & Hewstone, 2007; see also DeFleur & Westie, 1958). The letter is then posted by the experimenters, resulting in no expenditure of effort, time or resources for the participants.

Results

The means, standard deviations, and correlations among the predictor and criterion variables are presented in Table 1. Overall, 65% of participants indicated little desire to have their opinion published (scores 1–3 on the 6‐point scale) whereas 35% participants indicated some desire to have their opinion published (scores 4–6 on the 6‐point scale). The correlation between speaking‐out intentions and speaking‐out behavior was not significant, r = .16, p = .11. Separate intention–behavior correlations were also calculated for each of the four cells of the design (with moral basis for attitude dichotomized using a median split procedure). These correlations ranged from −.19 to .26, but no correlation was significantly different from zero (ps>.16).

Table 1. Means, SD, and correlations among the predictor and criterion variables for Experiment 1

	Mean	SD	1	2	3	4	5	6
1. Attitude intensity	2.34	1.11	–	.64∗∗∗	.02	.27∗∗	.19^+	.53∗∗∗
2. Moral basis for attitude	6.80	1.62		–	.17^+	.33∗∗∗	.30∗∗∗	.56∗∗∗
3. Normative support	–	–			–	.23∗	.01	.04
4. Intentions to speak out	6.26	1.70				–	.16	.08
5. Speaking‐out behavior	2.72	1.66					–	.23∗
6. Need to convert	3.49	2.39						–
^+ p<.10, ∗ p<.05, ∗∗ p<.01, ∗∗∗ p<.001.For normative support: 0 = majority support, 1 = minority support.

Data analysis overview

Moderated regression analyses were used to examine the main and interactive effects of normative support and moral basis of attitude on speaking‐out intentions and speaking‐out behavior. To prevent the stability of the regression analyses from being influenced by multicollinearity between the predictor variables and the interaction terms, all of the interaction terms were based on centered scores (see Aiken & West, 1991). In each regression, attitude intensity (centered), level of normative support (coded such that majority = 0, minority = 1), and moral basis for attitude (centered) were entered at Step 1. At Step 2, the interaction term between norm and moral basis for attitude was included. Where the interaction term was significant, simple slopes analysis was performed. Simple slopes were tested by examining the effects of normative support for respondents with a weak (−1 SD) or strong (+1 SD) moral basis for attitude. In all analyses, the effects of attitude intensity were controlled for.

Intentions to speak out

On the measure of speaking‐out intentions, inclusion of the independent effects for attitude intensity, moral basis for attitude, and normative support produced a significant increment in the variance explained, R²ch = .15, F(3, 96) = 5.69, p<.001. Inspection of the beta weights revealed a significant effect of norm only (β = .19, p = .050), such that those who perceived low levels of normative support for their attitude (i.e., the minority support condition) were more likely to express speaking‐out intentions than those who perceived high levels of normative support for their attitude (i.e., the majority support condition). A marginal effect of moral basis for attitude emerged also (β = .21, p = .089), such that people reported stronger speaking‐out intentions the stronger the moral basis for their attitude. Attitude intensity was not a significant predictor over and above the other variables (β = .13, p = .28).

At Step 2 there was a marginal interaction between normative support and moral basis for attitude, R²ch = .03, F(1, 95) = 3.29, p = .073, β = .24 (see Figure 1). Simple slopes analysis revealed that participants who had a strong moral basis for their attitude expressed stronger intentions to speak out when there was a low level of normative support for their attitude (i.e., when they were in a minority) rather than when there was a high level of normative support for their attitude (i.e., when they were in a majority; β = .36, p = .008). However, for participants with a weak moral basis for their attitude, the normative support manipulation had no effect (β = .02, p = .89). Thus, consistent with predictions, there was some evidence participants with a strong moral basis for their attitude were counter‐conforming on intentions and that those with a weak moral basis for their attitude were non‐conforming. Although the interaction term was only marginally significant, it should be noted that the simple slope was significant for those with a strong moral basis for their attitude, and that the pattern of results was identical to that reported by Hornsey et al. (2003).

Figure 1. Interaction between moral basis and support on intentions to speak out(Experiment 1).

Speaking‐out behavior

On the measure of speaking‐out behavior, inclusion of the independent predictors at Step 1 was associated with a significant increase in the variance explained, R²ch = .09, F(3, 96) = 3.27, p = .024. Inspection of the beta weights revealed a main effect for moral basis for attitude only (β = .32, p = .015), such that participants were more likely to agree to being publicly identified with their attitude when they have a strong moral basis for their attitude. Neither attitude intensity (β = .01, p = .92) nor level of normative support (β = −.04, p = .67) was a significant predictor of speaking‐out behavior. Neither was the interaction term between level of normative support and moral basis for attitude significant when entered at Step 2, R²ch = .00, F(1, 95) = .01, p = .95, β = .01.

Need to convert

In summary, we have established that a moral basis for attitude significantly predicts speaking‐out intentions and behaviors, over and above an index of attitude intensity. Furthermore, moral basis for attitude interacted with normative support to predict speaking‐out intentions. To check whether the need to convert was driving these effects, a series of regressions was conducted. The first regression checked whether attitude intensity, level of normative support, and moral basis for attitude independently and interactively predicted need to convert. Two main effects emerged at the first step, R²ch = .36, F(3, 96) = 18.29, p<.001, such that participants had a stronger need to convert the stronger the intensity of their attitude (β = .30, p = .006) and the stronger the moral basis for their attitude (β = .37, p = .001). Normative support had no effect either on its own (β = −.03, p = .72) or as an interaction with moral basis, R²ch = .00, F(1, 95) = .19, p = .66, β = .05. Because there was no interaction on need to convert, we could rule out this variable as a potential mediator of the interactive effect of normative support and moral basis on speaking‐out intentions.

To check whether need to convert was mediating the observed main effects of moral basis for attitude, two regressions were conducted with attitude intensity, normative support, moral basis for attitude, and need to convert entered as predictors simultaneously. Need to convert did not uniquely predict either speaking‐out intentions (β = −.19, p = .12) or actual speaking‐out behavior (β = .09, p = .45). Thus, there was no evidence that need to convert was mediating any of the effects of normative support or moral basis for attitude.

Discussion

For those with a weak moral basis for their attitude, level of normative support appeared to have no effect on their speaking‐out intentions, representing non‐conformity. Those with a strong moral basis for their attitude, however, reported stronger intentions to speak out when they were led to believe they were in a minority than when they were in a majority, representing counter‐conformity. Far from caving in to the pressure associated with holding a minority attitude, these participants appeared to be emboldened by their minority status. On the surface, then, the results appear to suggest that normative influence does not play a strong role in influencing speaking out for those who are morally invested in their attitude. This pattern of effects is identical to that reported by Hornsey et al. (2003), who also used an intention measure.

Obviously, however, an intention measure is not an ideal way of capturing the pressures associated with normative influence, because all participants need do is circle numbers on a questionnaire. The anonymous nature of their response means that participants do not need to fear social ridicule or isolation as a result of holding a minority attitude. Indeed, when participants were put in a position where they could be publicly identified with their attitude, the story was somewhat different. No counter‐conformity was in evidence; instead, participants showed non‐conformity, and the (non)effect of the norm was similar regardless of whether or not participants had a moral basis for their attitude. In other words, evidence for counter‐conformity among those with a strong moral basis for their attitude disappeared when participants were called upon to act.

Across both intentions and behaviors, it should be noted that people were more inclined to speak out the stronger the moral basis for their attitude. This effect emerged over and above the intensity with which participants held their attitude. This raises the question: What is it about having moral conviction that leads to stronger speaking‐out behaviors? One possibility that we examined here is that those with stronger moral conviction feel more motivated to recruit or convert others to their cause. Indeed, if this were the case, it might help explain why those with a strong moral basis for their attitude might counter‐conform. Specifically, learning that one's attitude is shared by only a minority might make more salient the need to speak out and recruit others to the minority cause. However, there was no evidence that the need to convert was producing counter‐conformity, or that the need to convert was driving the main effect of moral conviction.

Before dismissing the role of the need to convert, however, it is worth questioning whether the issue used in this study was suited to capturing this construct. At the time of conducting this study, the federal government had ruled out apologizing to Aborigines and there was no reason to expect that this policy was going to change. As a result, the issue had largely dropped off the political agenda and was no longer being discussed widely in the media. Arguably, then, the need to convert was not playing a central role because the stability of the political status quo meant that there was little need for or use in mobilizing like‐minded others. This was the case also when Hornsey et al. (2003) used this issue, and could apply also to the topic of gay law reform (Experiment 1 of the Hornsey et al. study), for which there was high‐profile political opposition and little optimism for change. Not only does this limit the generalizability of the findings, but there are also theoretical reasons to believe that the results might differ if participants think that the tide of public opinion is turning in their favor. Research on the spiral of silence, for example, shows that when people perceive that public opinion is shifting in their favor, even those who hold a minority opinion become more motivated to speak out, presumably because they sense that people would be candidates for recruitment (Salmon & Neuwirth, 1990; see also Noelle‐Neumann, 1993). One wonders, then, whether the need to convert would play a more central role if we were to use a topic for which there was genuine hope for change. We examined this possibility in Experiment 2.

EXPERIMENT 2

In Experiment 2, participants were led to believe that their attitudes on the legalization of euthanasia were shared either by a majority or a minority of group members. The issue of euthanasia was chosen because it was an issue on which the status quo appeared to be in flux and the issue was still on the political agenda. Before exposing participants to the norm information, we measured the moral basis for participants' attitudes and the extent to which they believed the status quo was likely to change in the future. Predictions for intentions were the same as for Experiment 1. To recap, we predicted that those with a strong moral basis for their attitude would counter‐conform on the intention measure and that those with a weak moral basis for their attitude would be unaffected by the level of normative support for their attitude. On our measure of behavior, we based our predictions on the results of Experiment 1: Participants will be more likely to speak out the stronger the moral basis for their attitude, and the level of normative support will have no effect overall.

We were open, however, to the possibility that these effects would be moderated by perceptions of change. It was predicted that intentions and behavior would be stronger for those who perceive the government policy on euthanasia was liable to change. Furthermore, if we are to assume that need to convert is a driving factor behind people's willingness to act in line with their attitudes, we might expect that evidence for counter‐conformity would be more likely to emerge when there were high rather than low perceptions of change.

Method

Participants and design

A total of 235 undergraduate students participated in the study in return for course credit. Participants were initially required to rate their attitudes on three issues: legalization of voluntary euthanasia, legalization of marijuana, and capital punishment (1 = strongly support, 8 = strongly oppose). They also rated the extent to which they had a moral basis for these attitudes and the extent to which they expected change on these issues in the future. The issues of legalization of marijuana and capital punishment were distracter issues and were not analyzed. As in Experiment 1, only those participants who held an attitude contrary to the status quo were included in the analyses. Of the original sample, 169 participants expressed support for euthanasia (rating 4 or below) and only this subset of data was analyzed. The subset consisted of 129 females and 40 males (M = 20.87 years old). After recording their attitude, participants were led to believe they had either majority or minority support for their opinion.

Procedure and measures

The procedure was identical to that used in Experiment 1, with two minor exceptions. First, the reference group that was used was not UQ students but rather Queenslanders. Queensland is the state of Australia in which the participants reside, and this was chosen as the reference group because it is at the state government level that laws about voluntary euthanasia are created. Another exception relates to the procedure for collecting the measure of speaking‐out behavior. As in Experiment 1 we asked participants to write an essay about their attitude about legalizing voluntary euthanasia, and then asked how willing they would be to have this essay published. Unlike Experiment 1, however, participants wrote the essay at the start of the experimental session (rather than at the end of the experimental session), but indicated their willingness to have their essay published at the end of the study.

Measures of moral basis for attitude (α = .66) and need to convert (r = .88) were the same as those used in Experiment 1, although the latter items were adapted to refer specifically to participants' need to convert Queenslanders rather than UQ students. Perceptions of change were measured using the following three items: “Do you think that it is a real possibility that voluntary euthanasia will be legalized in the future?” (1 = no, not at all, 9 = yes, certainly), “Do you think that, in the future, support from Queenslanders for legalizing voluntary euthanasia will increase, decrease, or remain the same?” (1 = decrease, 9 = increase), and “How likely is it that the current legal policy on voluntary euthanasia will change in the future?” (1 = not at all likely, 9 = extremely likely). These items were combined into a single scale of perceived likelihood of policy changing in line with their attitude (α = .79).

The measure of intentions that was used in Experiment 1 was adapted slightly in Experiment 2. First, because we were no longer using UQ as a reference group, it was unnecessary to include scenarios that related to students (e.g., speaking out in tutorials). Instead, we presented one scenario: “Assume that you're talking to a group of Queenslanders in which the issue of legalizing voluntary euthanasia has been raised. You are asked to speak.” As in Experiment 1, participants were asked the extent to which they would feel comfortable, relaxed, and safe presenting their true opinion on voluntary euthanasia in front of this audience. We also added a fourth item: “In this context, how willing would you be to express your true opinion about legalizing voluntary euthanasia?” (1 = very unwilling, 9 = very willing). These items were combined into a single scale of intentions to speak out (α = .92).

Results

The means, standard deviations, and correlations among the predictor and criterion variables are presented in Table 2. Overall, 68% of participants indicated little desire to have their opinion published (scores 1–3 on the 6‐point scale) whereas 32% participants indicated some desire to have their opinion published (scores 4–6 on the 6‐point scale). The correlation between speaking‐out intentions and speaking‐out behavior was moderate and significant, r = .27, p<.001. Moral basis for attitude was dichotomized using a median split procedure and separate intention–behavior correlations were calculated for the four cells of the design (created by crossing normative support and moral basis for attitude). These correlations ranged from .11 to .37, but only the correlations for the conditions in which participants were exposed to majority support were significantly different from zero (rs = .36 and .37 for low and high moral basis respectively, ps<.02). The correlations in the minority support condition were not significantly different from zero (rs = .23 and .11 for low and high moral basis respectively, ps>.18).

Table 2. Means, SD, and correlations among the predictor and criterion variables for Experiment 2

	Mean	SD	1	2	3	4	5	6	7
1. Attitude intensity	2.60	1.04	–	−.03	.43∗∗∗	.26∗∗	.16∗	.23∗∗	.41∗∗
2. Moral basis for attitude	6.20	1.52		–	.06	.13	.18∗	.31∗∗∗	.29∗∗∗
3. Normative support	–	–			–	−.07	.02	.09	.06
4. Perceptions of change	6.05	1.39				–	.14^+	.09	.24∗∗
5. Intentions to speak out	6.39	1.91					–	.27∗∗∗	.12
6. Speaking‐out behavior	2.69	1.63						–	.34∗∗∗
7. Need to convert	4.00	2.18							–
^+ p<.10, ∗ p<.05, ∗∗ p<.01, ∗∗∗ p<.001.For normative support: 0 = majority support, 1 = minority support.

Data analysis overview

In each regression, normative support for one's attitude (coded such that majority = 0, minority = 1), moral basis for attitude, attitude intensity, and change perceptions were entered at Step 1. All continuous variables were centered. At Step 2, the two‐way interaction terms were entered, and the three‐way interaction term was entered at Step 3. Significant three‐way interactions were explored by examining the effects of norm and moral basis for attitude at different levels of perceptions of change. When analyzing simple interactions and simple slopes, the effects of attitude intensity were controlled for at all times.

Intentions to speak out

On intentions to speak out, no significant main effects or two‐way interactions emerged at Step 1, R²ch = .05, F(4, 163) = 2.15, p = .077, or Step 2, R²ch = .03, F(3, 160) = 1.62, p = .19. There was, however, a significant three‐way interaction between level of normative support, moral basis for attitude, and perceptions of change, R²ch = .04, F(1, 159) = 6.76, p = .01, β = −.28 (see Figure 2). Follow‐up analyses revealed a non‐significant two‐way interaction between level of normative support and moral basis for attitude among those with high perceptions of change (β = −.25, p = .099); instead, a main effect of moral basis for attitude emerged such that intentions to speak out were stronger the stronger the moral basis for their attitude (β = .33, p = .016). There was, however, a significant two‐way interaction between level of normative support and moral basis for attitude among those with low perceptions of change (β = .31, p = .015). For these participants, we then examined the effects of normative support separately for those with a strong and weak moral basis for attitude. For those with a strong moral basis for their attitude, intentions to speak out were stronger when they believed that their attitude was not supported by the group (i.e., when they were in a minority), as opposed to when they believed that their attitude was supported by the group (i.e., when they were in a majority; β = .39, p = .012). For those with a weak moral basis for their attitude, intentions to speak out were unaffected by the norm manipulation (β = −.07, p = .63).

Figure 2. Interaction between moral basis and support on intentions to speak out among those with low perceptions of change(top) and high perceptions of change (bottom).

Speaking‐out behavior

When speaking‐out behavior was used as the criterion, inclusion of the main effects at Step 1 produced a significant increment in the variance explained, R²ch = .12, F(4, 163) = 5.34, p<.001. A significant main effect of moral basis for attitude emerged, such that participants were more likely to speak out the stronger the moral basis for their attitude (β = .26, p = .002). None of the two‐way interactions was significant, R²ch = .04, F(3, 160) = 2.43, p = .067; however the three‐way interaction between level of normative support, moral basis for attitude, and perceptions of change was significant, R²ch = .04, F(1, 159) = 6.77, p = .01, β = −.27 (see Figure 3). In this case, when participants had little expectation of change in the future, there was no interaction between normative support and moral basis for attitude (β = −.04, p = .74). Rather, there was a main effect of moral basis for attitude, such that people spoke out more the stronger the moral basis for their attitude (β = .33, p = .017). However, among those who had high expectations of positive change, the simple interaction between normative support and moral basis for attitude was significant (β = −.47, p<.001). Analyses of simple main effects were then conducted just on those who had high perceptions of change. This showed that those with a strong moral basis for their attitude were more likely to speak out when they believed they were in a majority as opposed to a minority (β = −.32, p = .029). In contrast, those with a weak moral basis for their attitude were more likely to speak out when they believed they were in a minority as opposed to a majority (β = .39, p = .017).

Figure 3. Interaction between moral basis and support on speaking‐out behavior among those with low perceptions of change (top) and high perceptions of change (bottom).

Need to convert

A regression on need to convert revealed a significant increment in the variance explained at Step 1, R²ch = .21, F(4, 164) = 10.71, p<.001. Inspection of the beta weights revealed a main effect of attitude intensity, such that people reported a stronger need to convert the more intensely they held their original attitude (β = .32, p<.001). Participants also reported a stronger need to convert the more they perceived that the status quo was changing in their favor (β = .15, p = .044). Normative support and moral conviction had no significant effects on their own, but they did significantly interact, R²ch = .05, F(3, 161) = 3.91, p = .01; β = −.27, p = .004. Analysis of simple main effects showed that, for people with a weak moral basis for their attitude, there was a stronger need to convert when people were led to believe they were in a minority than a majority (β = .26, p = .010). In contrast, for people with a strong moral basis for their attitude, level of normative support had no effect on need to convert (β = −.16, p = .11). The three‐way interaction was not significant, R²ch = .00, F(1, 160) = .16, p = .69, β = .04. Because these effects did not mirror the effects that were found on speaking‐out intentions and behaviors, we could rule out the possibility that need to convert could explain the effects on speaking out.

Discussion

The results of Experiment 2 suggest a similar but more nuanced picture than that described in Experiment 1. Specifically, it suggests that level of normative support for one's opinion does have an effect on speaking out, but that its effects are moderated by both moral basis for attitude and perceptions of change. First we will discuss the results for those people who have low perceptions of change. We discuss this group first because these participants provide the best match for our Experiment 1 participants who supported an apology for Aborigines despite having little hope that the apology would occur. Among this group, people with a strong moral basis for their attitude showed stronger intentions to speak out when they felt they were in a minority than when they felt they were in a majority. Those with a weak moral basis for their attitude, on the other hand, were apparently unaffected by the norm manipulation. Thus, on speaking‐out intentions, we found counter‐conformity among those with moral conviction and non‐conformity among those without. This pattern of findings seems robust, having now been observed across three experiments using issues on which there was little hope for change (both experiments of Hornsey et al., 2003, and Experiment 1 of the current paper) and again here among those participants who subjectively perceive little hope of positive change.

On speaking‐out behaviors, however, evidence for counter‐conformity disappeared. Participants were more likely to agree to being publicly identified with their comments the more they had a moral basis for their attitude, but this tendency was unaffected by level of normative support. In other words, non‐conformity emerged regardless of whether or not people had moral conviction. Again, this is entirely consistent with the findings of Experiment 1 where evidence for counter‐conformity that emerged on speaking‐out intentions failed to materialize on speaking‐out behaviors.

For people who perceive high likelihood of change, the effects of normative support were quite different. On intentions, there was only a main effect such that speaking‐out intentions were stronger the more participants had a moral basis for their attitude. On speaking‐out behaviors, however, the effects of moral basis for attitude were moderated by level of normative support. For these people, it was those with a weak moral basis for their attitude who counter‐conformed, whereas those with a strong moral basis for their attitude conformed to the group norm. Why we should see counter‐conformity in this case for people with a weak moral basis for their attitude is not entirely clear. One possibility is that people with a weak moral basis for their attitude are making cold, economic decisions about when they need to voice their opinion. It could be that, if they already feel the status quo is shifting in their favor and that the majority of the population support their view, then there is so much momentum that they do not need to speak out. In contrast, if they feel that their attitude is shared by only a minority, they might think that more work needs to be done and so are prepared to articulate their position publicly. If this is the case, what looks like counter‐conformity in the minority condition might be better interpreted as complacency in the majority condition.

Regardless of what psychological factors are underpinning this unexpected effect, the data send three clear and unambiguous messages. First, intentions to speak out do not translate into speaking‐out behaviors; specifically, people with a strong moral basis for their attitude do not counter‐conform as much as they think they will. Second, perceptions of whether the status quo is likely to shift in the future have a moderating effect on speaking‐out intentions and behaviors. Third, speaking‐out intentions and behaviors appear to be unrelated to the need to convert. People feel a stronger need to convert the more they perceive the status quo as shifting in line with their attitude. However, the need to convert could not explain any of the effects of normative support, and could not explain why having a strong moral basis for one's attitude motivates one to speak out.

GENERAL DISCUSSION

It has long been established that people are prepared to change their views in public so as to fit in to the views of the majority (Asch, 1952; Deutsch & Gerard, 1955). Recently, however, it was shown that this was less likely to be the case for those who had a moral basis for their attitude; in fact, these people showed evidence of reacting against the group norm, or counter‐conforming (Hornsey et al., 2003). On the surface, this research reinforces many of the prevailing norms in individualist societies; that you should stick up for what you believe in, that you should have the courage of your convictions, that you should be authentic in self‐expression, and that you should resist peer influence. In short, it suggests that on moral issues at least, individual will is resilient in the face of group pressure.

The current study, however, reveals two major qualifications to this conclusion. First, it appears that evidence for counter‐conformity among those with moral conviction emerges on intentions to speak out, but when people are in a position to be publicly identified with their attitude, the evidence disappears. In other words, their intention to counter‐conform is not translating into action. This is despite the fact that the public behavior did not require face‐to‐face interaction in our paradigm, the presence of which might be expected to increase normative pressures even further. This is not to say that the evidence for counter‐conformity that emerged earlier is misleading or tainted. On the contrary, intention measures are very revealing in terms of providing an insight into what participants would like to think they would do in a particular situation, which in turn tells us a great deal about people's self‐image and self‐perceptions. However, in this case, intention measures are clearly not adequate as proxy measures of behavior, and should not be interpreted as such. This reinforces the argument that self‐reports might not be appropriate when measuring speaking‐out behavior (Kennamer, 1990; Scheufele et al., 2001).

The second major qualification to the story told by Hornsey et al. (2003) is that the effects of norms and moral basis for attitude are moderated in complex ways by perceptions of whether the status quo is likely or unlikely to change in line with your attitude. The pattern of results that has been found consistently on intentions—counter‐conformity among those with a strong moral basis for attitude and non‐conformity among those with a weak moral basis for attitude—appears to be specific to those who see little chance that the status quo will change in their favor. When people see a realistic chance that the status quo will change in line with their attitude, this pattern disappears, and in some cases even reverses. On a theoretical level, this finding converges with other researchers who have called for a greater recognition of the temporal context when examining speaking‐out behavior, and in particular perceptions of whether the momentum is swinging toward or away from one's position (Salmon & Neuwirth, 1990; see also Noelle‐Neumann, 1993).

This paper contributes to a growing literature suggesting that moral bases for attitudes predict behavior over and above attitude intensity (Manstead, 2000; Skitka et al., 2005). One question that remains unanswered relates to what it is about moral conviction that causes counter‐conformity (at least on intentions) and is positively correlated overall with speaking‐out behaviors. In this paper we were able to rule out one possibility: that those with a strong moral basis for their attitude are more committed to recruiting other people to their cause. This is surprising in some ways, because implicit in much literature on speaking out—including literature on spiral of silence and collective action—is that people's decision to speak out is partly about persuading others to change their attitude to become more in line with the speaker's. However, the current studies suggest that the need to convert was not a useful construct in terms of explaining the effects of moral basis for attitude and norms on speaking out.

A feasible alternative explanation for the results is that moral attitudes serve a value‐expressive function; that is, they help maintain a positive self‐esteem from one's self as audience (see review of attitude function research by Pratkanis & Greenwald, 1989). It is possible that the most value‐expression could be gained from presenting a minority attitude, but that in highly public contexts this function is overwhelmed by the social‐adjustive function of attitudes (i.e., fitting in). Other intrapsychic functions of expressing attitudes that may help link moral conviction with speaking out include the desire for self‐definition (Tice, 1992), the desire to make a symbolic protest (Hornsey et al., 2006a), and the desire for voice (Folger, 1977; Lind, Kanfer, & Earley, 1990). Teasing apart these motivations—and examining their potential roles of driving mechanisms for the current effects—remains a challenge for future research.

Summary and conclusions

Codol (1984, p. 317) once argued that “both conformity and resistance to conformity are fundamentally linked to the image of oneself that one wishes to present to others (and undoubtedly also to oneself).” The point made in parentheses is particularly relevant to this paper; throughout these studies there was very little evidence overall for conformity, and indeed people with a strong moral basis for their attitude claim that they intend to counter‐conform. Perhaps this reflects the changing standards of our time, where people are increasingly trained to value self‐expression and to have the courage of their convictions, even if this means interrupting group consensus (Baumeister, 1991; Hornsey & Jetten, 2004; Kim & Markus, 1999). Having said this, there is little evidence that counter‐conformity is anything more than a promise that people like to make to themselves. In short, participants with a strong moral basis for their attitudes intended to counter‐conform, but when put in a position to act this resolve disappears. The most likely interpretation of this is that these people want to see themselves as independent social beings, but do not have the courage on the whole to act in line with this self‐image. Ironically, then, the current studies serve as both caveats and endorsements of the traditional notion of normative influence.